Adding upstream version 5.10.209.upstream/5.10.209

Signed-off-by: Daniel Baumann <daniel.baumann@progress-linux.org>

250 files changed, 27975 insertions, 0 deletions

diff --git a/tools/testing/selftests/powerpc/Makefile b/tools/testing/selftests/powerpc/Makefile
new file mode 100644
index 000000000..0830e6381
--- /dev/null
+++ b/tools/testing/selftests/powerpc/Makefile
@@ -0,0 +1,77 @@
+# SPDX-License-Identifier: GPL-2.0
+# Makefile for powerpc selftests
+
+# ARCH can be overridden by the user for cross compiling
+ARCH ?= $(shell uname -m)
+ARCH := $(shell echo $(ARCH) | sed -e s/ppc.*/powerpc/)
+
+ifeq ($(ARCH),powerpc)
+
+GIT_VERSION = $(shell git describe --always --long --dirty || echo "unknown")
+
+CFLAGS := -std=gnu99 -O2 -Wall -Werror -DGIT_VERSION='"$(GIT_VERSION)"' -I$(CURDIR)/include $(CFLAGS)
+
+export CFLAGS
+
+SUB_DIRS = alignment		\
+	   benchmarks		\
+	   cache_shape		\
+	   copyloops		\
+	   dscr			\
+	   mm			\
+	   nx-gzip		\
+	   pmu			\
+	   signal		\
+	   primitives		\
+	   stringloops		\
+	   switch_endian	\
+	   syscalls		\
+	   tm			\
+	   eeh			\
+	   vphn         \
+	   math		\
+	   ptrace	\
+	   security
+
+endif
+
+all: $(SUB_DIRS)
+
+$(SUB_DIRS):
+	BUILD_TARGET=$(OUTPUT)/$@; mkdir -p $$BUILD_TARGET; $(MAKE) OUTPUT=$$BUILD_TARGET -k -C $@ all
+
+include ../lib.mk
+
+override define RUN_TESTS
+	@for TARGET in $(SUB_DIRS); do \
+		BUILD_TARGET=$(OUTPUT)/$$TARGET;	\
+		$(MAKE) OUTPUT=$$BUILD_TARGET -C $$TARGET run_tests;\
+	done;
+endef
+
+override define INSTALL_RULE
+	@for TARGET in $(SUB_DIRS); do \
+		BUILD_TARGET=$(OUTPUT)/$$TARGET;	\
+		$(MAKE) OUTPUT=$$BUILD_TARGET -C $$TARGET install;\
+	done;
+endef
+
+override define EMIT_TESTS
+	@for TARGET in $(SUB_DIRS); do \
+		BUILD_TARGET=$(OUTPUT)/$$TARGET;	\
+		$(MAKE) OUTPUT=$$BUILD_TARGET -s -C $$TARGET emit_tests;\
+	done;
+endef
+
+override define CLEAN
+	@for TARGET in $(SUB_DIRS); do \
+		BUILD_TARGET=$(OUTPUT)/$$TARGET;	\
+		$(MAKE) OUTPUT=$$BUILD_TARGET -C $$TARGET clean; \
+	done;
+	rm -f tags
+endef
+
+tags:
+	find . -name '*.c' -o -name '*.h' | xargs ctags
+
+.PHONY: tags $(SUB_DIRS)
diff --git a/tools/testing/selftests/powerpc/alignment/.gitignore b/tools/testing/selftests/powerpc/alignment/.gitignore
new file mode 100644
index 000000000..28bc6ca13
--- /dev/null
+++ b/tools/testing/selftests/powerpc/alignment/.gitignore
@@ -0,0 +1,3 @@
+# SPDX-License-Identifier: GPL-2.0-only
+copy_first_unaligned
+alignment_handler
diff --git a/tools/testing/selftests/powerpc/alignment/Makefile b/tools/testing/selftests/powerpc/alignment/Makefile
new file mode 100644
index 000000000..93e9af374
--- /dev/null
+++ b/tools/testing/selftests/powerpc/alignment/Makefile
@@ -0,0 +1,7 @@
+# SPDX-License-Identifier: GPL-2.0-only
+TEST_GEN_PROGS := copy_first_unaligned alignment_handler
+
+top_srcdir = ../../../../..
+include ../../lib.mk
+
+$(TEST_GEN_PROGS): ../harness.c ../utils.c
diff --git a/tools/testing/selftests/powerpc/alignment/alignment_handler.c b/tools/testing/selftests/powerpc/alignment/alignment_handler.c
new file mode 100644
index 000000000..c25cf7cd4
--- /dev/null
+++ b/tools/testing/selftests/powerpc/alignment/alignment_handler.c
@@ -0,0 +1,689 @@
+// SPDX-License-Identifier: GPL-2.0-or-later
+/*
+ * Test the powerpc alignment handler on POWER8/POWER9
+ *
+ * Copyright (C) 2017 IBM Corporation (Michael Neuling, Andrew Donnellan)
+ */
+
+/*
+ * This selftest exercises the powerpc alignment fault handler.
+ *
+ * We create two sets of source and destination buffers, one in regular memory,
+ * the other cache-inhibited (by default we use /dev/fb0 for this, but an
+ * alterative path for cache-inhibited memory may be provided).
+ *
+ * One way to get cache-inhibited memory is to use the "mem" kernel parameter
+ * to limit the kernel to less memory than actually exists.  Addresses above
+ * the limit may still be accessed but will be treated as cache-inhibited. For
+ * example, if there is actually 4GB of memory and the parameter "mem=3GB" is
+ * used, memory from address 0xC0000000 onwards is treated as cache-inhibited.
+ * To access this region /dev/mem is used. The kernel should be configured
+ * without CONFIG_STRICT_DEVMEM. In this case use:
+ *         ./alignment_handler /dev/mem 0xc0000000
+ *
+ * We initialise the source buffers, then use whichever set of load/store
+ * instructions is under test to copy bytes from the source buffers to the
+ * destination buffers. For the regular buffers, these instructions will
+ * execute normally. For the cache-inhibited buffers, these instructions
+ * will trap and cause an alignment fault, and the alignment fault handler
+ * will emulate the particular instruction under test. We then compare the
+ * destination buffers to ensure that the native and emulated cases give the
+ * same result.
+ *
+ * TODO:
+ *   - Any FIXMEs below
+ *   - Test VSX regs < 32 and > 32
+ *   - Test all loads and stores
+ *   - Check update forms do update register
+ *   - Test alignment faults over page boundary
+ *
+ * Some old binutils may not support all the instructions.
+ */
+
+
+#include <sys/mman.h>
+#include <sys/types.h>
+#include <sys/stat.h>
+#include <fcntl.h>
+#include <unistd.h>
+#include <stdbool.h>
+#include <stdio.h>
+#include <stdlib.h>
+#include <string.h>
+#include <assert.h>
+#include <getopt.h>
+#include <setjmp.h>
+#include <signal.h>
+
+#include "utils.h"
+#include "instructions.h"
+
+int bufsize;
+int debug;
+int testing;
+volatile int gotsig;
+bool prefixes_enabled;
+char *cipath = "/dev/fb0";
+long cioffset;
+
+void sighandler(int sig, siginfo_t *info, void *ctx)
+{
+	ucontext_t *ucp = ctx;
+
+	if (!testing) {
+		signal(sig, SIG_DFL);
+		kill(0, sig);
+	}
+	gotsig = sig;
+#ifdef __powerpc64__
+	if (prefixes_enabled) {
+		u32 inst = *(u32 *)ucp->uc_mcontext.gp_regs[PT_NIP];
+		ucp->uc_mcontext.gp_regs[PT_NIP] += ((inst >> 26 == 1) ? 8 : 4);
+	} else {
+		ucp->uc_mcontext.gp_regs[PT_NIP] += 4;
+	}
+#else
+	ucp->uc_mcontext.uc_regs->gregs[PT_NIP] += 4;
+#endif
+}
+
+#define XFORM(reg, n)  " " #reg " ,%"#n",%2 ;"
+#define DFORM(reg, n)  " " #reg " ,0(%"#n") ;"
+
+#define TEST(name, ld_op, st_op, form, ld_reg, st_reg)		\
+	void test_##name(char *s, char *d)			\
+	{							\
+		asm volatile(					\
+			#ld_op form(ld_reg, 0)			\
+			#st_op form(st_reg, 1)			\
+			:: "r"(s), "r"(d), "r"(0)		\
+			: "memory", "vs0", "vs32", "r31");	\
+	}							\
+	rc |= do_test(#name, test_##name)
+
+#define TESTP(name, ld_op, st_op, ld_reg, st_reg)		\
+	void test_##name(char *s, char *d)			\
+	{							\
+		asm volatile(					\
+			ld_op(ld_reg, %0, 0, 0)			\
+			st_op(st_reg, %1, 0, 0)			\
+			:: "r"(s), "r"(d), "r"(0)		\
+			: "memory", "vs0", "vs32", "r31");	\
+	}							\
+	rc |= do_test(#name, test_##name)
+
+#define LOAD_VSX_XFORM_TEST(op) TEST(op, op, stxvd2x, XFORM, 32, 32)
+#define STORE_VSX_XFORM_TEST(op) TEST(op, lxvd2x, op, XFORM, 32, 32)
+#define LOAD_VSX_DFORM_TEST(op) TEST(op, op, stxv, DFORM, 32, 32)
+#define STORE_VSX_DFORM_TEST(op) TEST(op, lxv, op, DFORM, 32, 32)
+#define LOAD_VMX_XFORM_TEST(op) TEST(op, op, stxvd2x, XFORM, 0, 32)
+#define STORE_VMX_XFORM_TEST(op) TEST(op, lxvd2x, op, XFORM, 32, 0)
+#define LOAD_VMX_DFORM_TEST(op) TEST(op, op, stxv, DFORM, 0, 32)
+#define STORE_VMX_DFORM_TEST(op) TEST(op, lxv, op, DFORM, 32, 0)
+
+#define LOAD_XFORM_TEST(op) TEST(op, op, stdx, XFORM, 31, 31)
+#define STORE_XFORM_TEST(op) TEST(op, ldx, op, XFORM, 31, 31)
+#define LOAD_DFORM_TEST(op) TEST(op, op, std, DFORM, 31, 31)
+#define STORE_DFORM_TEST(op) TEST(op, ld, op, DFORM, 31, 31)
+
+#define LOAD_FLOAT_DFORM_TEST(op)  TEST(op, op, stfd, DFORM, 0, 0)
+#define STORE_FLOAT_DFORM_TEST(op) TEST(op, lfd, op, DFORM, 0, 0)
+#define LOAD_FLOAT_XFORM_TEST(op)  TEST(op, op, stfdx, XFORM, 0, 0)
+#define STORE_FLOAT_XFORM_TEST(op) TEST(op, lfdx, op, XFORM, 0, 0)
+
+#define LOAD_MLS_PREFIX_TEST(op) TESTP(op, op, PSTD, 31, 31)
+#define STORE_MLS_PREFIX_TEST(op) TESTP(op, PLD, op, 31, 31)
+
+#define LOAD_8LS_PREFIX_TEST(op) TESTP(op, op, PSTD, 31, 31)
+#define STORE_8LS_PREFIX_TEST(op) TESTP(op, PLD, op, 31, 31)
+
+#define LOAD_FLOAT_MLS_PREFIX_TEST(op) TESTP(op, op, PSTFD, 0, 0)
+#define STORE_FLOAT_MLS_PREFIX_TEST(op) TESTP(op, PLFD, op, 0, 0)
+
+#define LOAD_VSX_8LS_PREFIX_TEST(op, tail) TESTP(op, op, PSTXV ## tail, 0, 32)
+#define STORE_VSX_8LS_PREFIX_TEST(op, tail) TESTP(op, PLXV ## tail, op, 32, 0)
+
+/* FIXME: Unimplemented tests: */
+// STORE_DFORM_TEST(stq)   /* FIXME: need two registers for quad */
+// STORE_DFORM_TEST(stswi) /* FIXME: string instruction */
+
+// STORE_XFORM_TEST(stwat) /* AMO can't emulate or run on CI */
+// STORE_XFORM_TEST(stdat) /* ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ */
+
+
+/* preload byte by byte */
+void preload_data(void *dst, int offset, int width)
+{
+	char *c = dst;
+	int i;
+
+	c += offset;
+
+	for (i = 0 ; i < width ; i++)
+		c[i] = i;
+}
+
+int test_memcpy(void *dst, void *src, int size, int offset,
+		void (*test_func)(char *, char *))
+{
+	char *s, *d;
+
+	s = src;
+	s += offset;
+	d = dst;
+	d += offset;
+
+	assert(size == 16);
+	gotsig = 0;
+	testing = 1;
+
+	test_func(s, d); /* run the actual test */
+
+	testing = 0;
+	if (gotsig) {
+		if (debug)
+			printf("  Got signal %i\n", gotsig);
+		return 1;
+	}
+	return 0;
+}
+
+void dumpdata(char *s1, char *s2, int n, char *test_name)
+{
+	int i;
+
+	printf("  %s: unexpected result:\n", test_name);
+	printf("    mem:");
+	for (i = 0; i < n; i++)
+		printf(" %02x", s1[i]);
+	printf("\n");
+	printf("    ci: ");
+	for (i = 0; i < n; i++)
+		printf(" %02x", s2[i]);
+	printf("\n");
+}
+
+int test_memcmp(void *s1, void *s2, int n, int offset, char *test_name)
+{
+	char *s1c, *s2c;
+
+	s1c = s1;
+	s1c += offset;
+	s2c = s2;
+	s2c += offset;
+
+	if (memcmp(s1c, s2c, n)) {
+		if (debug) {
+			printf("\n  Compare failed. Offset:%i length:%i\n",
+			       offset, n);
+			dumpdata(s1c, s2c, n, test_name);
+		}
+		return 1;
+	}
+	return 0;
+}
+
+/*
+ * Do two memcpy tests using the same instructions. One cachable
+ * memory and the other doesn't.
+ */
+int do_test(char *test_name, void (*test_func)(char *, char *))
+{
+	int offset, width, fd, rc, r;
+	void *mem0, *mem1, *ci0, *ci1;
+
+	printf("\tDoing %s:\t", test_name);
+
+	fd = open(cipath, O_RDWR);
+	if (fd < 0) {
+		printf("\n");
+		perror("Can't open ci file now?");
+		return 1;
+	}
+
+	ci0 = mmap(NULL, bufsize, PROT_WRITE | PROT_READ, MAP_SHARED,
+		   fd, cioffset);
+	ci1 = mmap(NULL, bufsize, PROT_WRITE | PROT_READ, MAP_SHARED,
+		   fd, cioffset + bufsize);
+
+	if ((ci0 == MAP_FAILED) || (ci1 == MAP_FAILED)) {
+		printf("\n");
+		perror("mmap failed");
+		SKIP_IF(1);
+	}
+
+	rc = posix_memalign(&mem0, bufsize, bufsize);
+	if (rc) {
+		printf("\n");
+		return rc;
+	}
+
+	rc = posix_memalign(&mem1, bufsize, bufsize);
+	if (rc) {
+		printf("\n");
+		free(mem0);
+		return rc;
+	}
+
+	rc = 0;
+	/*
+	 * offset = 0 is aligned but tests the workaround for the P9N
+	 * DD2.1 vector CI load issue (see 5080332c2c89 "powerpc/64s:
+	 * Add workaround for P9 vector CI load issue")
+	 */
+	for (offset = 0; offset < 16; offset++) {
+		width = 16; /* vsx == 16 bytes */
+		r = 0;
+
+		/* load pattern into memory byte by byte */
+		preload_data(ci0, offset, width);
+		preload_data(mem0, offset, width); // FIXME: remove??
+		memcpy(ci0, mem0, bufsize);
+		memcpy(ci1, mem1, bufsize); /* initialise output to the same */
+
+		/* sanity check */
+		test_memcmp(mem0, ci0, width, offset, test_name);
+
+		r |= test_memcpy(ci1,  ci0,  width, offset, test_func);
+		r |= test_memcpy(mem1, mem0, width, offset, test_func);
+		if (r && !debug) {
+			printf("FAILED: Got signal");
+			rc = 1;
+			break;
+		}
+
+		r |= test_memcmp(mem1, ci1, width, offset, test_name);
+		if (r && !debug) {
+			printf("FAILED: Wrong Data");
+			rc = 1;
+			break;
+		}
+	}
+
+	if (rc == 0)
+		printf("PASSED");
+
+	printf("\n");
+
+	munmap(ci0, bufsize);
+	munmap(ci1, bufsize);
+	free(mem0);
+	free(mem1);
+	close(fd);
+
+	return rc;
+}
+
+static bool can_open_cifile(void)
+{
+	int fd;
+
+	fd = open(cipath, O_RDWR);
+	if (fd < 0)
+		return false;
+
+	close(fd);
+	return true;
+}
+
+int test_alignment_handler_vsx_206(void)
+{
+	int rc = 0;
+
+	SKIP_IF(!can_open_cifile());
+	SKIP_IF(!have_hwcap(PPC_FEATURE_ARCH_2_06));
+
+	printf("VSX: 2.06B\n");
+	LOAD_VSX_XFORM_TEST(lxvd2x);
+	LOAD_VSX_XFORM_TEST(lxvw4x);
+	LOAD_VSX_XFORM_TEST(lxsdx);
+	LOAD_VSX_XFORM_TEST(lxvdsx);
+	STORE_VSX_XFORM_TEST(stxvd2x);
+	STORE_VSX_XFORM_TEST(stxvw4x);
+	STORE_VSX_XFORM_TEST(stxsdx);
+	return rc;
+}
+
+int test_alignment_handler_vsx_207(void)
+{
+	int rc = 0;
+
+	SKIP_IF(!can_open_cifile());
+	SKIP_IF(!have_hwcap2(PPC_FEATURE2_ARCH_2_07));
+
+	printf("VSX: 2.07B\n");
+	LOAD_VSX_XFORM_TEST(lxsspx);
+	LOAD_VSX_XFORM_TEST(lxsiwax);
+	LOAD_VSX_XFORM_TEST(lxsiwzx);
+	STORE_VSX_XFORM_TEST(stxsspx);
+	STORE_VSX_XFORM_TEST(stxsiwx);
+	return rc;
+}
+
+int test_alignment_handler_vsx_300(void)
+{
+	int rc = 0;
+
+	SKIP_IF(!can_open_cifile());
+
+	SKIP_IF(!have_hwcap2(PPC_FEATURE2_ARCH_3_00));
+	printf("VSX: 3.00B\n");
+	LOAD_VMX_DFORM_TEST(lxsd);
+	LOAD_VSX_XFORM_TEST(lxsibzx);
+	LOAD_VSX_XFORM_TEST(lxsihzx);
+	LOAD_VMX_DFORM_TEST(lxssp);
+	LOAD_VSX_DFORM_TEST(lxv);
+	LOAD_VSX_XFORM_TEST(lxvb16x);
+	LOAD_VSX_XFORM_TEST(lxvh8x);
+	LOAD_VSX_XFORM_TEST(lxvx);
+	LOAD_VSX_XFORM_TEST(lxvwsx);
+	LOAD_VSX_XFORM_TEST(lxvl);
+	LOAD_VSX_XFORM_TEST(lxvll);
+	STORE_VMX_DFORM_TEST(stxsd);
+	STORE_VSX_XFORM_TEST(stxsibx);
+	STORE_VSX_XFORM_TEST(stxsihx);
+	STORE_VMX_DFORM_TEST(stxssp);
+	STORE_VSX_DFORM_TEST(stxv);
+	STORE_VSX_XFORM_TEST(stxvb16x);
+	STORE_VSX_XFORM_TEST(stxvh8x);
+	STORE_VSX_XFORM_TEST(stxvx);
+	STORE_VSX_XFORM_TEST(stxvl);
+	STORE_VSX_XFORM_TEST(stxvll);
+	return rc;
+}
+
+int test_alignment_handler_vsx_prefix(void)
+{
+	int rc = 0;
+
+	SKIP_IF(!can_open_cifile());
+	SKIP_IF(!have_hwcap2(PPC_FEATURE2_ARCH_3_1));
+
+	printf("VSX: PREFIX\n");
+	LOAD_VSX_8LS_PREFIX_TEST(PLXSD, 0);
+	LOAD_VSX_8LS_PREFIX_TEST(PLXSSP, 0);
+	LOAD_VSX_8LS_PREFIX_TEST(PLXV0, 0);
+	LOAD_VSX_8LS_PREFIX_TEST(PLXV1, 1);
+	STORE_VSX_8LS_PREFIX_TEST(PSTXSD, 0);
+	STORE_VSX_8LS_PREFIX_TEST(PSTXSSP, 0);
+	STORE_VSX_8LS_PREFIX_TEST(PSTXV0, 0);
+	STORE_VSX_8LS_PREFIX_TEST(PSTXV1, 1);
+	return rc;
+}
+
+int test_alignment_handler_integer(void)
+{
+	int rc = 0;
+
+	SKIP_IF(!can_open_cifile());
+
+	printf("Integer\n");
+	LOAD_DFORM_TEST(lbz);
+	LOAD_DFORM_TEST(lbzu);
+	LOAD_XFORM_TEST(lbzx);
+	LOAD_XFORM_TEST(lbzux);
+	LOAD_DFORM_TEST(lhz);
+	LOAD_DFORM_TEST(lhzu);
+	LOAD_XFORM_TEST(lhzx);
+	LOAD_XFORM_TEST(lhzux);
+	LOAD_DFORM_TEST(lha);
+	LOAD_DFORM_TEST(lhau);
+	LOAD_XFORM_TEST(lhax);
+	LOAD_XFORM_TEST(lhaux);
+	LOAD_XFORM_TEST(lhbrx);
+	LOAD_DFORM_TEST(lwz);
+	LOAD_DFORM_TEST(lwzu);
+	LOAD_XFORM_TEST(lwzx);
+	LOAD_XFORM_TEST(lwzux);
+	LOAD_DFORM_TEST(lwa);
+	LOAD_XFORM_TEST(lwax);
+	LOAD_XFORM_TEST(lwaux);
+	LOAD_XFORM_TEST(lwbrx);
+	LOAD_DFORM_TEST(ld);
+	LOAD_DFORM_TEST(ldu);
+	LOAD_XFORM_TEST(ldx);
+	LOAD_XFORM_TEST(ldux);
+	STORE_DFORM_TEST(stb);
+	STORE_XFORM_TEST(stbx);
+	STORE_DFORM_TEST(stbu);
+	STORE_XFORM_TEST(stbux);
+	STORE_DFORM_TEST(sth);
+	STORE_XFORM_TEST(sthx);
+	STORE_DFORM_TEST(sthu);
+	STORE_XFORM_TEST(sthux);
+	STORE_XFORM_TEST(sthbrx);
+	STORE_DFORM_TEST(stw);
+	STORE_XFORM_TEST(stwx);
+	STORE_DFORM_TEST(stwu);
+	STORE_XFORM_TEST(stwux);
+	STORE_XFORM_TEST(stwbrx);
+	STORE_DFORM_TEST(std);
+	STORE_XFORM_TEST(stdx);
+	STORE_DFORM_TEST(stdu);
+	STORE_XFORM_TEST(stdux);
+
+#ifdef __BIG_ENDIAN__
+	LOAD_DFORM_TEST(lmw);
+	STORE_DFORM_TEST(stmw);
+#endif
+
+	return rc;
+}
+
+int test_alignment_handler_integer_206(void)
+{
+	int rc = 0;
+
+	SKIP_IF(!can_open_cifile());
+	SKIP_IF(!have_hwcap(PPC_FEATURE_ARCH_2_06));
+
+	printf("Integer: 2.06\n");
+
+	LOAD_XFORM_TEST(ldbrx);
+	STORE_XFORM_TEST(stdbrx);
+
+	return rc;
+}
+
+int test_alignment_handler_integer_prefix(void)
+{
+	int rc = 0;
+
+	SKIP_IF(!can_open_cifile());
+	SKIP_IF(!have_hwcap2(PPC_FEATURE2_ARCH_3_1));
+
+	printf("Integer: PREFIX\n");
+	LOAD_MLS_PREFIX_TEST(PLBZ);
+	LOAD_MLS_PREFIX_TEST(PLHZ);
+	LOAD_MLS_PREFIX_TEST(PLHA);
+	LOAD_MLS_PREFIX_TEST(PLWZ);
+	LOAD_8LS_PREFIX_TEST(PLWA);
+	LOAD_8LS_PREFIX_TEST(PLD);
+	STORE_MLS_PREFIX_TEST(PSTB);
+	STORE_MLS_PREFIX_TEST(PSTH);
+	STORE_MLS_PREFIX_TEST(PSTW);
+	STORE_8LS_PREFIX_TEST(PSTD);
+	return rc;
+}
+
+int test_alignment_handler_vmx(void)
+{
+	int rc = 0;
+
+	SKIP_IF(!can_open_cifile());
+	SKIP_IF(!have_hwcap(PPC_FEATURE_HAS_ALTIVEC));
+
+	printf("VMX\n");
+	LOAD_VMX_XFORM_TEST(lvx);
+
+	/*
+	 * FIXME: These loads only load part of the register, so our
+	 * testing method doesn't work. Also they don't take alignment
+	 * faults, so it's kinda pointless anyway
+	 *
+	 LOAD_VMX_XFORM_TEST(lvebx)
+	 LOAD_VMX_XFORM_TEST(lvehx)
+	 LOAD_VMX_XFORM_TEST(lvewx)
+	 LOAD_VMX_XFORM_TEST(lvxl)
+	*/
+	STORE_VMX_XFORM_TEST(stvx);
+	STORE_VMX_XFORM_TEST(stvebx);
+	STORE_VMX_XFORM_TEST(stvehx);
+	STORE_VMX_XFORM_TEST(stvewx);
+	STORE_VMX_XFORM_TEST(stvxl);
+	return rc;
+}
+
+int test_alignment_handler_fp(void)
+{
+	int rc = 0;
+
+	SKIP_IF(!can_open_cifile());
+
+	printf("Floating point\n");
+	LOAD_FLOAT_DFORM_TEST(lfd);
+	LOAD_FLOAT_XFORM_TEST(lfdx);
+	LOAD_FLOAT_DFORM_TEST(lfdu);
+	LOAD_FLOAT_XFORM_TEST(lfdux);
+	LOAD_FLOAT_DFORM_TEST(lfs);
+	LOAD_FLOAT_XFORM_TEST(lfsx);
+	LOAD_FLOAT_DFORM_TEST(lfsu);
+	LOAD_FLOAT_XFORM_TEST(lfsux);
+	STORE_FLOAT_DFORM_TEST(stfd);
+	STORE_FLOAT_XFORM_TEST(stfdx);
+	STORE_FLOAT_DFORM_TEST(stfdu);
+	STORE_FLOAT_XFORM_TEST(stfdux);
+	STORE_FLOAT_DFORM_TEST(stfs);
+	STORE_FLOAT_XFORM_TEST(stfsx);
+	STORE_FLOAT_DFORM_TEST(stfsu);
+	STORE_FLOAT_XFORM_TEST(stfsux);
+	STORE_FLOAT_XFORM_TEST(stfiwx);
+
+	return rc;
+}
+
+int test_alignment_handler_fp_205(void)
+{
+	int rc = 0;
+
+	SKIP_IF(!can_open_cifile());
+	SKIP_IF(!have_hwcap(PPC_FEATURE_ARCH_2_05));
+
+	printf("Floating point: 2.05\n");
+
+	LOAD_FLOAT_DFORM_TEST(lfdp);
+	LOAD_FLOAT_XFORM_TEST(lfdpx);
+	LOAD_FLOAT_XFORM_TEST(lfiwax);
+	STORE_FLOAT_DFORM_TEST(stfdp);
+	STORE_FLOAT_XFORM_TEST(stfdpx);
+
+	return rc;
+}
+
+int test_alignment_handler_fp_206(void)
+{
+	int rc = 0;
+
+	SKIP_IF(!can_open_cifile());
+	SKIP_IF(!have_hwcap(PPC_FEATURE_ARCH_2_06));
+
+	printf("Floating point: 2.06\n");
+
+	LOAD_FLOAT_XFORM_TEST(lfiwzx);
+
+	return rc;
+}
+
+
+int test_alignment_handler_fp_prefix(void)
+{
+	int rc = 0;
+
+	SKIP_IF(!can_open_cifile());
+	SKIP_IF(!have_hwcap2(PPC_FEATURE2_ARCH_3_1));
+
+	printf("Floating point: PREFIX\n");
+	LOAD_FLOAT_DFORM_TEST(lfs);
+	LOAD_FLOAT_MLS_PREFIX_TEST(PLFS);
+	LOAD_FLOAT_MLS_PREFIX_TEST(PLFD);
+	STORE_FLOAT_MLS_PREFIX_TEST(PSTFS);
+	STORE_FLOAT_MLS_PREFIX_TEST(PSTFD);
+	return rc;
+}
+
+void usage(char *prog)
+{
+	printf("Usage: %s [options] [path [offset]]\n", prog);
+	printf("  -d	Enable debug error output\n");
+	printf("\n");
+	printf("This test requires a POWER8, POWER9 or POWER10 CPU ");
+	printf("and either a usable framebuffer at /dev/fb0 or ");
+	printf("the path to usable cache inhibited memory and optional ");
+	printf("offset to be provided\n");
+}
+
+int main(int argc, char *argv[])
+{
+
+	struct sigaction sa;
+	int rc = 0;
+	int option = 0;
+
+	while ((option = getopt(argc, argv, "d")) != -1) {
+		switch (option) {
+		case 'd':
+			debug++;
+			break;
+		default:
+			usage(argv[0]);
+			exit(1);
+		}
+	}
+	argc -= optind;
+	argv += optind;
+
+	if (argc > 0)
+		cipath = argv[0];
+	if (argc > 1)
+		cioffset = strtol(argv[1], 0, 0x10);
+
+	bufsize = getpagesize();
+
+	sa.sa_sigaction = sighandler;
+	sigemptyset(&sa.sa_mask);
+	sa.sa_flags = SA_SIGINFO;
+	if (sigaction(SIGSEGV, &sa, NULL) == -1
+	    || sigaction(SIGBUS, &sa, NULL) == -1
+	    || sigaction(SIGILL, &sa, NULL) == -1) {
+		perror("sigaction");
+		exit(1);
+	}
+
+	prefixes_enabled = have_hwcap2(PPC_FEATURE2_ARCH_3_1);
+
+	rc |= test_harness(test_alignment_handler_vsx_206,
+			   "test_alignment_handler_vsx_206");
+	rc |= test_harness(test_alignment_handler_vsx_207,
+			   "test_alignment_handler_vsx_207");
+	rc |= test_harness(test_alignment_handler_vsx_300,
+			   "test_alignment_handler_vsx_300");
+	rc |= test_harness(test_alignment_handler_vsx_prefix,
+			   "test_alignment_handler_vsx_prefix");
+	rc |= test_harness(test_alignment_handler_integer,
+			   "test_alignment_handler_integer");
+	rc |= test_harness(test_alignment_handler_integer_206,
+			   "test_alignment_handler_integer_206");
+	rc |= test_harness(test_alignment_handler_integer_prefix,
+			   "test_alignment_handler_integer_prefix");
+	rc |= test_harness(test_alignment_handler_vmx,
+			   "test_alignment_handler_vmx");
+	rc |= test_harness(test_alignment_handler_fp,
+			   "test_alignment_handler_fp");
+	rc |= test_harness(test_alignment_handler_fp_205,
+			   "test_alignment_handler_fp_205");
+	rc |= test_harness(test_alignment_handler_fp_206,
+			   "test_alignment_handler_fp_206");
+	rc |= test_harness(test_alignment_handler_fp_prefix,
+			   "test_alignment_handler_fp_prefix");
+	return rc;
+}
diff --git a/tools/testing/selftests/powerpc/alignment/copy_first_unaligned.c b/tools/testing/selftests/powerpc/alignment/copy_first_unaligned.c
new file mode 100644
index 000000000..db4e8c680
--- /dev/null
+++ b/tools/testing/selftests/powerpc/alignment/copy_first_unaligned.c
@@ -0,0 +1,67 @@
+// SPDX-License-Identifier: GPL-2.0-or-later
+/*
+ * Copyright 2016, Chris Smart, IBM Corporation.
+ *
+ * Calls to copy_first which are not 128-byte aligned should be
+ * caught and sent a SIGBUS.
+ */
+
+#include <signal.h>
+#include <string.h>
+#include <unistd.h>
+#include "utils.h"
+#include "instructions.h"
+
+unsigned int expected_instruction = PPC_INST_COPY_FIRST;
+unsigned int instruction_mask = 0xfc2007fe;
+
+void signal_action_handler(int signal_num, siginfo_t *info, void *ptr)
+{
+	ucontext_t *ctx = ptr;
+#ifdef __powerpc64__
+	unsigned int *pc = (unsigned int *)ctx->uc_mcontext.gp_regs[PT_NIP];
+#else
+	unsigned int *pc = (unsigned int *)ctx->uc_mcontext.uc_regs->gregs[PT_NIP];
+#endif
+
+	/*
+	 * Check that the signal was on the correct instruction, using a
+	 * mask because the compiler assigns the register at RB.
+	 */
+	if ((*pc & instruction_mask) == expected_instruction)
+		_exit(0); /* We hit the right instruction */
+
+	_exit(1);
+}
+
+void setup_signal_handler(void)
+{
+	struct sigaction signal_action;
+
+	memset(&signal_action, 0, sizeof(signal_action));
+	signal_action.sa_sigaction = signal_action_handler;
+	signal_action.sa_flags = SA_SIGINFO;
+	sigaction(SIGBUS, &signal_action, NULL);
+}
+
+char cacheline_buf[128] __cacheline_aligned;
+
+int test_copy_first_unaligned(void)
+{
+	/* Only run this test on a P9 or later */
+	SKIP_IF(!have_hwcap2(PPC_FEATURE2_ARCH_3_00));
+
+	/* Register our signal handler with SIGBUS */
+	setup_signal_handler();
+
+	/* +1 makes buf unaligned */
+	copy_first(cacheline_buf+1);
+
+	/* We should not get here */
+	return 1;
+}
+
+int main(int argc, char *argv[])
+{
+	return test_harness(test_copy_first_unaligned, "test_copy_first_unaligned");
+}
diff --git a/tools/testing/selftests/powerpc/benchmarks/.gitignore b/tools/testing/selftests/powerpc/benchmarks/.gitignore
new file mode 100644
index 000000000..c9ce13983
--- /dev/null
+++ b/tools/testing/selftests/powerpc/benchmarks/.gitignore
@@ -0,0 +1,8 @@
+# SPDX-License-Identifier: GPL-2.0-only
+gettimeofday
+context_switch
+fork
+exec_target
+mmap_bench
+futex_bench
+null_syscall
diff --git a/tools/testing/selftests/powerpc/benchmarks/Makefile b/tools/testing/selftests/powerpc/benchmarks/Makefile
new file mode 100644
index 000000000..a32a6ab89
--- /dev/null
+++ b/tools/testing/selftests/powerpc/benchmarks/Makefile
@@ -0,0 +1,20 @@
+# SPDX-License-Identifier: GPL-2.0
+TEST_GEN_PROGS := gettimeofday context_switch fork mmap_bench futex_bench null_syscall
+TEST_GEN_FILES := exec_target
+
+TEST_FILES := settings
+
+CFLAGS += -O2
+
+top_srcdir = ../../../../..
+include ../../lib.mk
+
+$(TEST_GEN_PROGS): ../harness.c
+
+$(OUTPUT)/context_switch: ../utils.c
+$(OUTPUT)/context_switch: CFLAGS += -maltivec -mvsx -mabi=altivec
+$(OUTPUT)/context_switch: LDLIBS += -lpthread
+
+$(OUTPUT)/fork: LDLIBS += -lpthread
+
+$(OUTPUT)/exec_target: CFLAGS += -static -nostartfiles
diff --git a/tools/testing/selftests/powerpc/benchmarks/context_switch.c b/tools/testing/selftests/powerpc/benchmarks/context_switch.c
new file mode 100644
index 000000000..96554e279
--- /dev/null
+++ b/tools/testing/selftests/powerpc/benchmarks/context_switch.c
@@ -0,0 +1,508 @@
+// SPDX-License-Identifier: GPL-2.0-or-later
+/*
+ * Context switch microbenchmark.
+ *
+ * Copyright (C) 2015 Anton Blanchard <anton@au.ibm.com>, IBM
+ */
+
+#define _GNU_SOURCE
+#include <errno.h>
+#include <sched.h>
+#include <string.h>
+#include <stdio.h>
+#include <unistd.h>
+#include <stdlib.h>
+#include <getopt.h>
+#include <signal.h>
+#include <assert.h>
+#include <pthread.h>
+#include <limits.h>
+#include <sys/time.h>
+#include <sys/syscall.h>
+#include <sys/sysinfo.h>
+#include <sys/types.h>
+#include <sys/shm.h>
+#include <linux/futex.h>
+#ifdef __powerpc__
+#include <altivec.h>
+#endif
+#include "utils.h"
+
+static unsigned int timeout = 30;
+
+static int touch_vdso;
+struct timeval tv;
+
+static int touch_fp = 1;
+double fp;
+
+static int touch_vector = 1;
+vector int a, b, c;
+
+#ifdef __powerpc__
+static int touch_altivec = 1;
+
+/*
+ * Note: LTO (Link Time Optimisation) doesn't play well with this function
+ * attribute. Be very careful enabling LTO for this test.
+ */
+static void __attribute__((__target__("no-vsx"))) altivec_touch_fn(void)
+{
+	c = a + b;
+}
+#endif
+
+static void touch(void)
+{
+	if (touch_vdso)
+		gettimeofday(&tv, NULL);
+
+	if (touch_fp)
+		fp += 0.1;
+
+#ifdef __powerpc__
+	if (touch_altivec)
+		altivec_touch_fn();
+#endif
+
+	if (touch_vector)
+		c = a + b;
+
+	asm volatile("# %0 %1 %2": : "r"(&tv), "r"(&fp), "r"(&c));
+}
+
+static void start_thread_on(void *(*fn)(void *), void *arg, unsigned long cpu)
+{
+	int rc;
+	pthread_t tid;
+	cpu_set_t cpuset;
+	pthread_attr_t attr;
+
+	CPU_ZERO(&cpuset);
+	CPU_SET(cpu, &cpuset);
+
+	rc = pthread_attr_init(&attr);
+	if (rc) {
+		errno = rc;
+		perror("pthread_attr_init");
+		exit(1);
+	}
+
+	rc = pthread_attr_setaffinity_np(&attr, sizeof(cpu_set_t), &cpuset);
+	if (rc)	{
+		errno = rc;
+		perror("pthread_attr_setaffinity_np");
+		exit(1);
+	}
+
+	rc = pthread_create(&tid, &attr, fn, arg);
+	if (rc) {
+		errno = rc;
+		perror("pthread_create");
+		exit(1);
+	}
+}
+
+static void start_process_on(void *(*fn)(void *), void *arg, unsigned long cpu)
+{
+	int pid, ncpus;
+	cpu_set_t *cpuset;
+	size_t size;
+
+	pid = fork();
+	if (pid == -1) {
+		perror("fork");
+		exit(1);
+	}
+
+	if (pid)
+		return;
+
+	ncpus = get_nprocs();
+	size = CPU_ALLOC_SIZE(ncpus);
+	cpuset = CPU_ALLOC(ncpus);
+	if (!cpuset) {
+		perror("malloc");
+		exit(1);
+	}
+	CPU_ZERO_S(size, cpuset);
+	CPU_SET_S(cpu, size, cpuset);
+
+	if (sched_setaffinity(0, size, cpuset)) {
+		perror("sched_setaffinity");
+		CPU_FREE(cpuset);
+		exit(1);
+	}
+
+	CPU_FREE(cpuset);
+	fn(arg);
+
+	exit(0);
+}
+
+static unsigned long iterations;
+static unsigned long iterations_prev;
+
+static void sigalrm_handler(int junk)
+{
+	unsigned long i = iterations;
+
+	printf("%ld\n", i - iterations_prev);
+	iterations_prev = i;
+
+	if (--timeout == 0)
+		kill(0, SIGUSR1);
+
+	alarm(1);
+}
+
+static void sigusr1_handler(int junk)
+{
+	exit(0);
+}
+
+struct actions {
+	void (*setup)(int, int);
+	void *(*thread1)(void *);
+	void *(*thread2)(void *);
+};
+
+#define READ 0
+#define WRITE 1
+
+static int pipe_fd1[2];
+static int pipe_fd2[2];
+
+static void pipe_setup(int cpu1, int cpu2)
+{
+	if (pipe(pipe_fd1) || pipe(pipe_fd2))
+		exit(1);
+}
+
+static void *pipe_thread1(void *arg)
+{
+	signal(SIGALRM, sigalrm_handler);
+	alarm(1);
+
+	while (1) {
+		assert(read(pipe_fd1[READ], &c, 1) == 1);
+		touch();
+
+		assert(write(pipe_fd2[WRITE], &c, 1) == 1);
+		touch();
+
+		iterations += 2;
+	}
+
+	return NULL;
+}
+
+static void *pipe_thread2(void *arg)
+{
+	while (1) {
+		assert(write(pipe_fd1[WRITE], &c, 1) == 1);
+		touch();
+
+		assert(read(pipe_fd2[READ], &c, 1) == 1);
+		touch();
+	}
+
+	return NULL;
+}
+
+static struct actions pipe_actions = {
+	.setup = pipe_setup,
+	.thread1 = pipe_thread1,
+	.thread2 = pipe_thread2,
+};
+
+static void yield_setup(int cpu1, int cpu2)
+{
+	if (cpu1 != cpu2) {
+		fprintf(stderr, "Both threads must be on the same CPU for yield test\n");
+		exit(1);
+	}
+}
+
+static void *yield_thread1(void *arg)
+{
+	signal(SIGALRM, sigalrm_handler);
+	alarm(1);
+
+	while (1) {
+		sched_yield();
+		touch();
+
+		iterations += 2;
+	}
+
+	return NULL;
+}
+
+static void *yield_thread2(void *arg)
+{
+	while (1) {
+		sched_yield();
+		touch();
+	}
+
+	return NULL;
+}
+
+static struct actions yield_actions = {
+	.setup = yield_setup,
+	.thread1 = yield_thread1,
+	.thread2 = yield_thread2,
+};
+
+static long sys_futex(void *addr1, int op, int val1, struct timespec *timeout,
+		      void *addr2, int val3)
+{
+	return syscall(SYS_futex, addr1, op, val1, timeout, addr2, val3);
+}
+
+static unsigned long cmpxchg(unsigned long *p, unsigned long expected,
+			     unsigned long desired)
+{
+	unsigned long exp = expected;
+
+	__atomic_compare_exchange_n(p, &exp, desired, 0,
+				    __ATOMIC_SEQ_CST, __ATOMIC_SEQ_CST);
+	return exp;
+}
+
+static unsigned long xchg(unsigned long *p, unsigned long val)
+{
+	return __atomic_exchange_n(p, val, __ATOMIC_SEQ_CST);
+}
+
+static int processes;
+
+static int mutex_lock(unsigned long *m)
+{
+	int c;
+	int flags = FUTEX_WAIT;
+	if (!processes)
+		flags |= FUTEX_PRIVATE_FLAG;
+
+	c = cmpxchg(m, 0, 1);
+	if (!c)
+		return 0;
+
+	if (c == 1)
+		c = xchg(m, 2);
+
+	while (c) {
+		sys_futex(m, flags, 2, NULL, NULL, 0);
+		c = xchg(m, 2);
+	}
+
+	return 0;
+}
+
+static int mutex_unlock(unsigned long *m)
+{
+	int flags = FUTEX_WAKE;
+	if (!processes)
+		flags |= FUTEX_PRIVATE_FLAG;
+
+	if (*m == 2)
+		*m = 0;
+	else if (xchg(m, 0) == 1)
+		return 0;
+
+	sys_futex(m, flags, 1, NULL, NULL, 0);
+
+	return 0;
+}
+
+static unsigned long *m1, *m2;
+
+static void futex_setup(int cpu1, int cpu2)
+{
+	if (!processes) {
+		static unsigned long _m1, _m2;
+		m1 = &_m1;
+		m2 = &_m2;
+	} else {
+		int shmid;
+		void *shmaddr;
+
+		shmid = shmget(IPC_PRIVATE, getpagesize(), SHM_R | SHM_W);
+		if (shmid < 0) {
+			perror("shmget");
+			exit(1);
+		}
+
+		shmaddr = shmat(shmid, NULL, 0);
+		if (shmaddr == (char *)-1) {
+			perror("shmat");
+			shmctl(shmid, IPC_RMID, NULL);
+			exit(1);
+		}
+
+		shmctl(shmid, IPC_RMID, NULL);
+
+		m1 = shmaddr;
+		m2 = shmaddr + sizeof(*m1);
+	}
+
+	*m1 = 0;
+	*m2 = 0;
+
+	mutex_lock(m1);
+	mutex_lock(m2);
+}
+
+static void *futex_thread1(void *arg)
+{
+	signal(SIGALRM, sigalrm_handler);
+	alarm(1);
+
+	while (1) {
+		mutex_lock(m2);
+		mutex_unlock(m1);
+
+		iterations += 2;
+	}
+
+	return NULL;
+}
+
+static void *futex_thread2(void *arg)
+{
+	while (1) {
+		mutex_unlock(m2);
+		mutex_lock(m1);
+	}
+
+	return NULL;
+}
+
+static struct actions futex_actions = {
+	.setup = futex_setup,
+	.thread1 = futex_thread1,
+	.thread2 = futex_thread2,
+};
+
+static struct option options[] = {
+	{ "test", required_argument, 0, 't' },
+	{ "process", no_argument, &processes, 1 },
+	{ "timeout", required_argument, 0, 's' },
+	{ "vdso", no_argument, &touch_vdso, 1 },
+	{ "no-fp", no_argument, &touch_fp, 0 },
+#ifdef __powerpc__
+	{ "no-altivec", no_argument, &touch_altivec, 0 },
+#endif
+	{ "no-vector", no_argument, &touch_vector, 0 },
+	{ 0, },
+};
+
+static void usage(void)
+{
+	fprintf(stderr, "Usage: context_switch2 <options> CPU1 CPU2\n\n");
+	fprintf(stderr, "\t\t--test=X\tpipe, futex or yield (default)\n");
+	fprintf(stderr, "\t\t--process\tUse processes (default threads)\n");
+	fprintf(stderr, "\t\t--timeout=X\tDuration in seconds to run (default 30)\n");
+	fprintf(stderr, "\t\t--vdso\t\ttouch VDSO\n");
+	fprintf(stderr, "\t\t--no-fp\t\tDon't touch FP\n");
+#ifdef __powerpc__
+	fprintf(stderr, "\t\t--no-altivec\tDon't touch altivec\n");
+#endif
+	fprintf(stderr, "\t\t--no-vector\tDon't touch vector\n");
+}
+
+int main(int argc, char *argv[])
+{
+	signed char c;
+	struct actions *actions = &yield_actions;
+	int cpu1;
+	int cpu2;
+	static void (*start_fn)(void *(*fn)(void *), void *arg, unsigned long cpu);
+
+	while (1) {
+		int option_index = 0;
+
+		c = getopt_long(argc, argv, "", options, &option_index);
+
+		if (c == -1)
+			break;
+
+		switch (c) {
+		case 0:
+			if (options[option_index].flag != 0)
+				break;
+
+			usage();
+			exit(1);
+			break;
+
+		case 't':
+			if (!strcmp(optarg, "pipe")) {
+				actions = &pipe_actions;
+			} else if (!strcmp(optarg, "yield")) {
+				actions = &yield_actions;
+			} else if (!strcmp(optarg, "futex")) {
+				actions = &futex_actions;
+			} else {
+				usage();
+				exit(1);
+			}
+			break;
+
+		case 's':
+			timeout = atoi(optarg);
+			break;
+
+		default:
+			usage();
+			exit(1);
+		}
+	}
+
+	if (processes)
+		start_fn = start_process_on;
+	else
+		start_fn = start_thread_on;
+
+	if (((argc - optind) != 2)) {
+		cpu1 = cpu2 = pick_online_cpu();
+	} else {
+		cpu1 = atoi(argv[optind++]);
+		cpu2 = atoi(argv[optind++]);
+	}
+
+	printf("Using %s with ", processes ? "processes" : "threads");
+
+	if (actions == &pipe_actions)
+		printf("pipe");
+	else if (actions == &yield_actions)
+		printf("yield");
+	else
+		printf("futex");
+
+	if (!have_hwcap(PPC_FEATURE_HAS_ALTIVEC))
+		touch_altivec = 0;
+
+	if (!have_hwcap(PPC_FEATURE_HAS_VSX))
+		touch_vector = 0;
+
+	printf(" on cpus %d/%d touching FP:%s altivec:%s vector:%s vdso:%s\n",
+	       cpu1, cpu2, touch_fp ?  "yes" : "no", touch_altivec ? "yes" : "no",
+	       touch_vector ? "yes" : "no", touch_vdso ? "yes" : "no");
+
+	/* Create a new process group so we can signal everyone for exit */
+	setpgid(getpid(), getpid());
+
+	signal(SIGUSR1, sigusr1_handler);
+
+	actions->setup(cpu1, cpu2);
+
+	start_fn(actions->thread1, NULL, cpu1);
+	start_fn(actions->thread2, NULL, cpu2);
+
+	while (1)
+		sleep(3600);
+
+	return 0;
+}
diff --git a/tools/testing/selftests/powerpc/benchmarks/exec_target.c b/tools/testing/selftests/powerpc/benchmarks/exec_target.c
new file mode 100644
index 000000000..c14b0fc1e
--- /dev/null
+++ b/tools/testing/selftests/powerpc/benchmarks/exec_target.c
@@ -0,0 +1,16 @@
+// SPDX-License-Identifier: GPL-2.0+
+
+/*
+ * Part of fork context switch microbenchmark.
+ *
+ * Copyright 2018, Anton Blanchard, IBM Corp.
+ */
+
+#define _GNU_SOURCE
+#include <unistd.h>
+#include <sys/syscall.h>
+
+void _start(void)
+{
+	syscall(SYS_exit, 0);
+}
diff --git a/tools/testing/selftests/powerpc/benchmarks/fork.c b/tools/testing/selftests/powerpc/benchmarks/fork.c
new file mode 100644
index 000000000..d312e638c
--- /dev/null
+++ b/tools/testing/selftests/powerpc/benchmarks/fork.c
@@ -0,0 +1,325 @@
+// SPDX-License-Identifier: GPL-2.0+
+
+/*
+ * Context switch microbenchmark.
+ *
+ * Copyright 2018, Anton Blanchard, IBM Corp.
+ */
+
+#define _GNU_SOURCE
+#include <assert.h>
+#include <errno.h>
+#include <getopt.h>
+#include <limits.h>
+#include <linux/futex.h>
+#include <pthread.h>
+#include <sched.h>
+#include <signal.h>
+#include <stdio.h>
+#include <stdlib.h>
+#include <string.h>
+#include <sys/shm.h>
+#include <sys/syscall.h>
+#include <sys/time.h>
+#include <sys/types.h>
+#include <sys/wait.h>
+#include <unistd.h>
+
+static unsigned int timeout = 30;
+
+static void set_cpu(int cpu)
+{
+	cpu_set_t cpuset;
+
+	if (cpu == -1)
+		return;
+
+	CPU_ZERO(&cpuset);
+	CPU_SET(cpu, &cpuset);
+
+	if (sched_setaffinity(0, sizeof(cpuset), &cpuset)) {
+		perror("sched_setaffinity");
+		exit(1);
+	}
+}
+
+static void start_process_on(void *(*fn)(void *), void *arg, int cpu)
+{
+	int pid;
+
+	pid = fork();
+	if (pid == -1) {
+		perror("fork");
+		exit(1);
+	}
+
+	if (pid)
+		return;
+
+	set_cpu(cpu);
+
+	fn(arg);
+
+	exit(0);
+}
+
+static int cpu;
+static int do_fork = 0;
+static int do_vfork = 0;
+static int do_exec = 0;
+static char *exec_file;
+static int exec_target = 0;
+static unsigned long iterations;
+static unsigned long iterations_prev;
+
+static void run_exec(void)
+{
+	char *const argv[] = { "./exec_target", NULL };
+
+	if (execve("./exec_target", argv, NULL) == -1) {
+		perror("execve");
+		exit(1);
+	}
+}
+
+static void bench_fork(void)
+{
+	while (1) {
+		pid_t pid = fork();
+		if (pid == -1) {
+			perror("fork");
+			exit(1);
+		}
+		if (pid == 0) {
+			if (do_exec)
+				run_exec();
+			_exit(0);
+		}
+		pid = waitpid(pid, NULL, 0);
+		if (pid == -1) {
+			perror("waitpid");
+			exit(1);
+		}
+		iterations++;
+	}
+}
+
+static void bench_vfork(void)
+{
+	while (1) {
+		pid_t pid = vfork();
+		if (pid == -1) {
+			perror("fork");
+			exit(1);
+		}
+		if (pid == 0) {
+			if (do_exec)
+				run_exec();
+			_exit(0);
+		}
+		pid = waitpid(pid, NULL, 0);
+		if (pid == -1) {
+			perror("waitpid");
+			exit(1);
+		}
+		iterations++;
+	}
+}
+
+static void *null_fn(void *arg)
+{
+	pthread_exit(NULL);
+}
+
+static void bench_thread(void)
+{
+	pthread_t tid;
+	cpu_set_t cpuset;
+	pthread_attr_t attr;
+	int rc;
+
+	rc = pthread_attr_init(&attr);
+	if (rc) {
+		errno = rc;
+		perror("pthread_attr_init");
+		exit(1);
+	}
+
+	if (cpu != -1) {
+		CPU_ZERO(&cpuset);
+		CPU_SET(cpu, &cpuset);
+
+		rc = pthread_attr_setaffinity_np(&attr, sizeof(cpu_set_t), &cpuset);
+		if (rc) {
+			errno = rc;
+			perror("pthread_attr_setaffinity_np");
+			exit(1);
+		}
+	}
+
+	while (1) {
+		rc = pthread_create(&tid, &attr, null_fn, NULL);
+		if (rc) {
+			errno = rc;
+			perror("pthread_create");
+			exit(1);
+		}
+		rc = pthread_join(tid, NULL);
+		if (rc) {
+			errno = rc;
+			perror("pthread_join");
+			exit(1);
+		}
+		iterations++;
+	}
+}
+
+static void sigalrm_handler(int junk)
+{
+	unsigned long i = iterations;
+
+	printf("%ld\n", i - iterations_prev);
+	iterations_prev = i;
+
+	if (--timeout == 0)
+		kill(0, SIGUSR1);
+
+	alarm(1);
+}
+
+static void sigusr1_handler(int junk)
+{
+	exit(0);
+}
+
+static void *bench_proc(void *arg)
+{
+	signal(SIGALRM, sigalrm_handler);
+	alarm(1);
+
+	if (do_fork)
+		bench_fork();
+	else if (do_vfork)
+		bench_vfork();
+	else
+		bench_thread();
+
+	return NULL;
+}
+
+static struct option options[] = {
+	{ "fork", no_argument, &do_fork, 1 },
+	{ "vfork", no_argument, &do_vfork, 1 },
+	{ "exec", no_argument, &do_exec, 1 },
+	{ "timeout", required_argument, 0, 's' },
+	{ "exec-target", no_argument, &exec_target, 1 },
+	{ NULL },
+};
+
+static void usage(void)
+{
+	fprintf(stderr, "Usage: fork <options> CPU\n\n");
+	fprintf(stderr, "\t\t--fork\tUse fork() (default threads)\n");
+	fprintf(stderr, "\t\t--vfork\tUse vfork() (default threads)\n");
+	fprintf(stderr, "\t\t--exec\tAlso exec() (default no exec)\n");
+	fprintf(stderr, "\t\t--timeout=X\tDuration in seconds to run (default 30)\n");
+	fprintf(stderr, "\t\t--exec-target\tInternal option for exec workload\n");
+}
+
+int main(int argc, char *argv[])
+{
+	signed char c;
+
+	while (1) {
+		int option_index = 0;
+
+		c = getopt_long(argc, argv, "", options, &option_index);
+
+		if (c == -1)
+			break;
+
+		switch (c) {
+		case 0:
+			if (options[option_index].flag != 0)
+				break;
+
+			usage();
+			exit(1);
+			break;
+
+		case 's':
+			timeout = atoi(optarg);
+			break;
+
+		default:
+			usage();
+			exit(1);
+		}
+	}
+
+	if (do_fork && do_vfork) {
+		usage();
+		exit(1);
+	}
+	if (do_exec && !do_fork && !do_vfork) {
+		usage();
+		exit(1);
+	}
+
+	if (do_exec) {
+		char *dirname = strdup(argv[0]);
+		int i;
+		i = strlen(dirname) - 1;
+		while (i) {
+			if (dirname[i] == '/') {
+				dirname[i] = '\0';
+				if (chdir(dirname) == -1) {
+					perror("chdir");
+					exit(1);
+				}
+				break;
+			}
+			i--;
+		}
+	}
+
+	if (exec_target) {
+		exit(0);
+	}
+
+	if (((argc - optind) != 1)) {
+		cpu = -1;
+	} else {
+		cpu = atoi(argv[optind++]);
+	}
+
+	if (do_exec)
+		exec_file = argv[0];
+
+	set_cpu(cpu);
+
+	printf("Using ");
+	if (do_fork)
+		printf("fork");
+	else if (do_vfork)
+		printf("vfork");
+	else
+		printf("clone");
+
+	if (do_exec)
+		printf(" + exec");
+
+	printf(" on cpu %d\n", cpu);
+
+	/* Create a new process group so we can signal everyone for exit */
+	setpgid(getpid(), getpid());
+
+	signal(SIGUSR1, sigusr1_handler);
+
+	start_process_on(bench_proc, NULL, cpu);
+
+	while (1)
+		sleep(3600);
+
+	return 0;
+}
diff --git a/tools/testing/selftests/powerpc/benchmarks/futex_bench.c b/tools/testing/selftests/powerpc/benchmarks/futex_bench.c
new file mode 100644
index 000000000..017057090
--- /dev/null
+++ b/tools/testing/selftests/powerpc/benchmarks/futex_bench.c
@@ -0,0 +1,43 @@
+// SPDX-License-Identifier: GPL-2.0-only
+/*
+ * Copyright 2016, Anton Blanchard, Michael Ellerman, IBM Corp.
+ */
+
+#define _GNU_SOURCE
+
+#include <stdio.h>
+#include <sys/syscall.h>
+#include <time.h>
+#include <unistd.h>
+#include <linux/futex.h>
+
+#include "utils.h"
+
+#define ITERATIONS 100000000
+
+#define futex(A, B, C, D, E, F)	 syscall(__NR_futex, A, B, C, D, E, F)
+
+int test_futex(void)
+{
+	struct timespec ts_start, ts_end;
+	unsigned long i = ITERATIONS;
+
+	clock_gettime(CLOCK_MONOTONIC, &ts_start);
+
+	while (i--) {
+		unsigned int addr = 0;
+		futex(&addr, FUTEX_WAKE, 1, NULL, NULL, 0);
+	}
+
+	clock_gettime(CLOCK_MONOTONIC, &ts_end);
+
+	printf("time = %.6f\n", ts_end.tv_sec - ts_start.tv_sec + (ts_end.tv_nsec - ts_start.tv_nsec) / 1e9);
+
+	return 0;
+}
+
+int main(void)
+{
+	test_harness_set_timeout(300);
+	return test_harness(test_futex, "futex_bench");
+}
diff --git a/tools/testing/selftests/powerpc/benchmarks/gettimeofday.c b/tools/testing/selftests/powerpc/benchmarks/gettimeofday.c
new file mode 100644
index 000000000..6b4156833
--- /dev/null
+++ b/tools/testing/selftests/powerpc/benchmarks/gettimeofday.c
@@ -0,0 +1,31 @@
+// SPDX-License-Identifier: GPL-2.0-only
+/*
+ * Copyright 2015, Anton Blanchard, IBM Corp.
+ */
+
+#include <sys/time.h>
+#include <stdio.h>
+
+#include "utils.h"
+
+static int test_gettimeofday(void)
+{
+	int i;
+
+	struct timeval tv_start, tv_end;
+
+	gettimeofday(&tv_start, NULL);
+
+	for(i = 0; i < 100000000; i++) {
+		gettimeofday(&tv_end, NULL);
+	}
+
+	printf("time = %.6f\n", tv_end.tv_sec - tv_start.tv_sec + (tv_end.tv_usec - tv_start.tv_usec) * 1e-6);
+
+	return 0;
+}
+
+int main(void)
+{
+	return test_harness(test_gettimeofday, "gettimeofday");
+}
diff --git a/tools/testing/selftests/powerpc/benchmarks/mmap_bench.c b/tools/testing/selftests/powerpc/benchmarks/mmap_bench.c
new file mode 100644
index 000000000..2525adf64
--- /dev/null
+++ b/tools/testing/selftests/powerpc/benchmarks/mmap_bench.c
@@ -0,0 +1,90 @@
+// SPDX-License-Identifier: GPL-2.0-only
+/*
+ * Copyright 2016, Anton Blanchard, Michael Ellerman, IBM Corp.
+ */
+
+#include <stdio.h>
+#include <stdlib.h>
+#include <sys/mman.h>
+#include <time.h>
+#include <getopt.h>
+
+#include "utils.h"
+
+#define ITERATIONS 5000000
+
+#define MEMSIZE (1UL << 27)
+#define PAGE_SIZE (1UL << 16)
+#define CHUNK_COUNT (MEMSIZE/PAGE_SIZE)
+
+static int pg_fault;
+static int iterations = ITERATIONS;
+
+static struct option options[] = {
+	{ "pgfault", no_argument, &pg_fault, 1 },
+	{ "iterations", required_argument, 0, 'i' },
+	{ 0, },
+};
+
+static void usage(void)
+{
+	printf("mmap_bench <--pgfault> <--iterations count>\n");
+}
+
+int test_mmap(void)
+{
+	struct timespec ts_start, ts_end;
+	unsigned long i = iterations;
+
+	clock_gettime(CLOCK_MONOTONIC, &ts_start);
+
+	while (i--) {
+		char *c = mmap(NULL, MEMSIZE, PROT_READ|PROT_WRITE,
+			       MAP_PRIVATE|MAP_ANONYMOUS, -1, 0);
+		FAIL_IF(c == MAP_FAILED);
+		if (pg_fault) {
+			int count;
+			for (count = 0; count < CHUNK_COUNT; count++)
+				c[count << 16] = 'c';
+		}
+		munmap(c, MEMSIZE);
+	}
+
+	clock_gettime(CLOCK_MONOTONIC, &ts_end);
+
+	printf("time = %.6f\n", ts_end.tv_sec - ts_start.tv_sec + (ts_end.tv_nsec - ts_start.tv_nsec) / 1e9);
+
+	return 0;
+}
+
+int main(int argc, char *argv[])
+{
+	signed char c;
+	while (1) {
+		int option_index = 0;
+
+		c = getopt_long(argc, argv, "", options, &option_index);
+
+		if (c == -1)
+			break;
+
+		switch (c) {
+		case 0:
+			if (options[option_index].flag != 0)
+				break;
+
+			usage();
+			exit(1);
+			break;
+		case 'i':
+			iterations = atoi(optarg);
+			break;
+		default:
+			usage();
+			exit(1);
+		}
+	}
+
+	test_harness_set_timeout(300);
+	return test_harness(test_mmap, "mmap_bench");
+}
diff --git a/tools/testing/selftests/powerpc/benchmarks/null_syscall.c b/tools/testing/selftests/powerpc/benchmarks/null_syscall.c
new file mode 100644
index 000000000..579f0215c
--- /dev/null
+++ b/tools/testing/selftests/powerpc/benchmarks/null_syscall.c
@@ -0,0 +1,153 @@
+// SPDX-License-Identifier: GPL-2.0-or-later
+/*
+ * Test null syscall performance
+ *
+ * Copyright (C) 2009-2015 Anton Blanchard, IBM
+ */
+
+#define NR_LOOPS 10000000
+
+#include <string.h>
+#include <stdio.h>
+#include <stdlib.h>
+#include <unistd.h>
+#include <time.h>
+#include <sys/types.h>
+#include <sys/time.h>
+#include <signal.h>
+
+static volatile int soak_done;
+unsigned long long clock_frequency;
+unsigned long long timebase_frequency;
+double timebase_multiplier;
+
+static inline unsigned long mftb(void)
+{
+	unsigned long low;
+
+	asm volatile("mftb %0" : "=r" (low));
+
+	return low;
+}
+
+static void sigalrm_handler(int unused)
+{
+	soak_done = 1;
+}
+
+/*
+ * Use a timer instead of busy looping on clock_gettime() so we don't
+ * pollute profiles with glibc and VDSO hits.
+ */
+static void cpu_soak_usecs(unsigned long usecs)
+{
+	struct itimerval val;
+
+	memset(&val, 0, sizeof(val));
+	val.it_value.tv_usec = usecs;
+
+	signal(SIGALRM, sigalrm_handler);
+	setitimer(ITIMER_REAL, &val, NULL);
+
+	while (1) {
+		if (soak_done)
+			break;
+	}
+
+	signal(SIGALRM, SIG_DFL);
+}
+
+/*
+ * This only works with recent kernels where cpufreq modifies
+ * /proc/cpuinfo dynamically.
+ */
+static void get_proc_frequency(void)
+{
+	FILE *f;
+	char line[128];
+	char *p, *end;
+	unsigned long v;
+	double d;
+	char *override;
+
+	/* Try to get out of low power/low frequency mode */
+	cpu_soak_usecs(0.25 * 1000000);
+
+	f = fopen("/proc/cpuinfo", "r");
+	if (f == NULL)
+		return;
+
+	timebase_frequency = 0;
+
+	while (fgets(line, sizeof(line), f) != NULL) {
+		if (strncmp(line, "timebase", 8) == 0) {
+			p = strchr(line, ':');
+			if (p != NULL) {
+				v = strtoull(p + 1, &end, 0);
+				if (end != p + 1)
+					timebase_frequency = v;
+			}
+		}
+
+		if (((strncmp(line, "clock", 5) == 0) ||
+		     (strncmp(line, "cpu MHz", 7) == 0))) {
+			p = strchr(line, ':');
+			if (p != NULL) {
+				d = strtod(p + 1, &end);
+				if (end != p + 1) {
+					/* Find fastest clock frequency */
+					if ((d * 1000000ULL) > clock_frequency)
+						clock_frequency = d * 1000000ULL;
+				}
+			}
+		}
+	}
+
+	fclose(f);
+
+	override = getenv("FREQUENCY");
+	if (override)
+		clock_frequency = strtoull(override, NULL, 10);
+
+	if (timebase_frequency)
+		timebase_multiplier = (double)clock_frequency
+					/ timebase_frequency;
+	else
+		timebase_multiplier = 1;
+}
+
+static void do_null_syscall(unsigned long nr)
+{
+	unsigned long i;
+
+	for (i = 0; i < nr; i++)
+		getppid();
+}
+
+#define TIME(A, STR) \
+
+int main(void)
+{
+	unsigned long tb_start, tb_now;
+	struct timespec tv_start, tv_now;
+	unsigned long long elapsed_ns, elapsed_tb;
+
+	get_proc_frequency();
+
+	clock_gettime(CLOCK_MONOTONIC, &tv_start);
+	tb_start = mftb();
+
+	do_null_syscall(NR_LOOPS);
+
+	clock_gettime(CLOCK_MONOTONIC, &tv_now);
+	tb_now = mftb();
+
+	elapsed_ns = (tv_now.tv_sec - tv_start.tv_sec) * 1000000000ULL +
+			(tv_now.tv_nsec - tv_start.tv_nsec);
+	elapsed_tb = tb_now - tb_start;
+
+	printf("%10.2f ns %10.2f cycles\n", (float)elapsed_ns / NR_LOOPS,
+			(float)elapsed_tb * timebase_multiplier / NR_LOOPS);
+
+	return 0;
+}
diff --git a/tools/testing/selftests/powerpc/benchmarks/settings b/tools/testing/selftests/powerpc/benchmarks/settings
new file mode 100644
index 000000000..e7b941753
--- /dev/null
+++ b/tools/testing/selftests/powerpc/benchmarks/settings
@@ -0,0 +1 @@
+timeout=0
diff --git a/tools/testing/selftests/powerpc/cache_shape/.gitignore b/tools/testing/selftests/powerpc/cache_shape/.gitignore
new file mode 100644
index 000000000..b385eee30
--- /dev/null
+++ b/tools/testing/selftests/powerpc/cache_shape/.gitignore
@@ -0,0 +1,2 @@
+# SPDX-License-Identifier: GPL-2.0-only
+cache_shape
diff --git a/tools/testing/selftests/powerpc/cache_shape/Makefile b/tools/testing/selftests/powerpc/cache_shape/Makefile
new file mode 100644
index 000000000..689f6c8eb
--- /dev/null
+++ b/tools/testing/selftests/powerpc/cache_shape/Makefile
@@ -0,0 +1,7 @@
+# SPDX-License-Identifier: GPL-2.0
+TEST_GEN_PROGS := cache_shape
+
+top_srcdir = ../../../../..
+include ../../lib.mk
+
+$(TEST_GEN_PROGS): ../harness.c ../utils.c
diff --git a/tools/testing/selftests/powerpc/cache_shape/cache_shape.c b/tools/testing/selftests/powerpc/cache_shape/cache_shape.c
new file mode 100644
index 000000000..171b6c948
--- /dev/null
+++ b/tools/testing/selftests/powerpc/cache_shape/cache_shape.c
@@ -0,0 +1,121 @@
+// SPDX-License-Identifier: GPL-2.0-or-later
+/*
+ * Copyright 2017, Michael Ellerman, IBM Corp.
+ */
+
+#include <elf.h>
+#include <errno.h>
+#include <fcntl.h>
+#include <link.h>
+#include <stdio.h>
+#include <stdlib.h>
+#include <string.h>
+#include <sys/stat.h>
+#include <sys/types.h>
+#include <sys/wait.h>
+#include <unistd.h>
+
+#include "utils.h"
+
+#ifndef AT_L1I_CACHESIZE
+#define AT_L1I_CACHESIZE	40
+#define AT_L1I_CACHEGEOMETRY	41
+#define AT_L1D_CACHESIZE	42
+#define AT_L1D_CACHEGEOMETRY	43
+#define AT_L2_CACHESIZE		44
+#define AT_L2_CACHEGEOMETRY	45
+#define AT_L3_CACHESIZE		46
+#define AT_L3_CACHEGEOMETRY	47
+#endif
+
+static void print_size(const char *label, uint32_t val)
+{
+	printf("%s cache size: %#10x %10dB %10dK\n", label, val, val, val / 1024);
+}
+
+static void print_geo(const char *label, uint32_t val)
+{
+	uint16_t assoc;
+
+	printf("%s line size:  %#10x       ", label, val & 0xFFFF);
+
+	assoc = val >> 16;
+	if (assoc)
+		printf("%u-way", assoc);
+	else
+		printf("fully");
+
+	printf(" associative\n");
+}
+
+static int test_cache_shape()
+{
+	static char buffer[4096];
+	ElfW(auxv_t) *p;
+	int found;
+
+	FAIL_IF(read_auxv(buffer, sizeof(buffer)));
+
+	found = 0;
+
+	p = find_auxv_entry(AT_L1I_CACHESIZE, buffer);
+	if (p) {
+		found++;
+		print_size("L1I ", (uint32_t)p->a_un.a_val);
+	}
+
+	p = find_auxv_entry(AT_L1I_CACHEGEOMETRY, buffer);
+	if (p) {
+		found++;
+		print_geo("L1I ", (uint32_t)p->a_un.a_val);
+	}
+
+	p = find_auxv_entry(AT_L1D_CACHESIZE, buffer);
+	if (p) {
+		found++;
+		print_size("L1D ", (uint32_t)p->a_un.a_val);
+	}
+
+	p = find_auxv_entry(AT_L1D_CACHEGEOMETRY, buffer);
+	if (p) {
+		found++;
+		print_geo("L1D ", (uint32_t)p->a_un.a_val);
+	}
+
+	p = find_auxv_entry(AT_L2_CACHESIZE, buffer);
+	if (p) {
+		found++;
+		print_size("L2  ", (uint32_t)p->a_un.a_val);
+	}
+
+	p = find_auxv_entry(AT_L2_CACHEGEOMETRY, buffer);
+	if (p) {
+		found++;
+		print_geo("L2  ", (uint32_t)p->a_un.a_val);
+	}
+
+	p = find_auxv_entry(AT_L3_CACHESIZE, buffer);
+	if (p) {
+		found++;
+		print_size("L3  ", (uint32_t)p->a_un.a_val);
+	}
+
+	p = find_auxv_entry(AT_L3_CACHEGEOMETRY, buffer);
+	if (p) {
+		found++;
+		print_geo("L3  ", (uint32_t)p->a_un.a_val);
+	}
+
+	/* If we found none we're probably on a system where they don't exist */
+	SKIP_IF(found == 0);
+
+	/* But if we found any, we expect to find them all */
+	FAIL_IF(found != 8);
+
+	return 0;
+}
+
+int main(void)
+{
+	return test_harness(test_cache_shape, "cache_shape");
+}
diff --git a/tools/testing/selftests/powerpc/copyloops/.gitignore b/tools/testing/selftests/powerpc/copyloops/.gitignore
new file mode 100644
index 000000000..994b11af7
--- /dev/null
+++ b/tools/testing/selftests/powerpc/copyloops/.gitignore
@@ -0,0 +1,15 @@
+# SPDX-License-Identifier: GPL-2.0-only
+copyuser_64_t0
+copyuser_64_t1
+copyuser_64_t2
+copyuser_p7_t0
+copyuser_p7_t1
+memcpy_64_t0
+memcpy_64_t1
+memcpy_64_t2
+memcpy_p7_t0
+memcpy_p7_t1
+copyuser_64_exc_t0
+copyuser_64_exc_t1
+copyuser_64_exc_t2
+copy_mc_64
diff --git a/tools/testing/selftests/powerpc/copyloops/Makefile b/tools/testing/selftests/powerpc/copyloops/Makefile
new file mode 100644
index 000000000..3095b1f1c
--- /dev/null
+++ b/tools/testing/selftests/powerpc/copyloops/Makefile
@@ -0,0 +1,58 @@
+# SPDX-License-Identifier: GPL-2.0
+# The loops are all 64-bit code
+CFLAGS += -m64
+CFLAGS += -I$(CURDIR)
+CFLAGS += -D SELFTEST
+CFLAGS += -maltivec
+CFLAGS += -mcpu=power4
+
+# Use our CFLAGS for the implicit .S rule & set the asm machine type
+ASFLAGS = $(CFLAGS) -Wa,-mpower4
+
+TEST_GEN_PROGS := copyuser_64_t0 copyuser_64_t1 copyuser_64_t2 \
+		copyuser_p7_t0 copyuser_p7_t1 \
+		memcpy_64_t0 memcpy_64_t1 memcpy_64_t2 \
+		memcpy_p7_t0 memcpy_p7_t1 copy_mc_64 \
+		copyuser_64_exc_t0 copyuser_64_exc_t1 copyuser_64_exc_t2
+
+EXTRA_SOURCES := validate.c ../harness.c stubs.S
+
+top_srcdir = ../../../../..
+include ../../lib.mk
+
+$(OUTPUT)/copyuser_64_t%:	copyuser_64.S $(EXTRA_SOURCES)
+	$(CC) $(CPPFLAGS) $(CFLAGS) \
+		-D COPY_LOOP=test___copy_tofrom_user_base \
+		-D SELFTEST_CASE=$(subst copyuser_64_t,,$(notdir $@)) \
+		-o $@ $^
+
+$(OUTPUT)/copyuser_p7_t%:	copyuser_power7.S $(EXTRA_SOURCES)
+	$(CC) $(CPPFLAGS) $(CFLAGS) \
+		-D COPY_LOOP=test___copy_tofrom_user_power7 \
+		-D SELFTEST_CASE=$(subst copyuser_p7_t,,$(notdir $@)) \
+		-o $@ $^
+
+# Strictly speaking, we only need the memcpy_64 test cases for big-endian
+$(OUTPUT)/memcpy_64_t%:	memcpy_64.S $(EXTRA_SOURCES)
+	$(CC) $(CPPFLAGS) $(CFLAGS) \
+		-D COPY_LOOP=test_memcpy \
+		-D SELFTEST_CASE=$(subst memcpy_64_t,,$(notdir $@)) \
+		-o $@ $^
+
+$(OUTPUT)/memcpy_p7_t%:	memcpy_power7.S $(EXTRA_SOURCES)
+	$(CC) $(CPPFLAGS) $(CFLAGS) \
+		-D COPY_LOOP=test_memcpy_power7 \
+		-D SELFTEST_CASE=$(subst memcpy_p7_t,,$(notdir $@)) \
+		-o $@ $^
+
+$(OUTPUT)/copy_mc_64: copy_mc_64.S $(EXTRA_SOURCES)
+	$(CC) $(CPPFLAGS) $(CFLAGS) \
+		-D COPY_LOOP=test_copy_mc_generic \
+		-o $@ $^
+
+$(OUTPUT)/copyuser_64_exc_t%: copyuser_64.S exc_validate.c ../harness.c \
+		copy_tofrom_user_reference.S stubs.S
+	$(CC) $(CPPFLAGS) $(CFLAGS) \
+		-D COPY_LOOP=test___copy_tofrom_user_base \
+		-D SELFTEST_CASE=$(subst copyuser_64_exc_t,,$(notdir $@)) \
+		-o $@ $^
diff --git a/tools/testing/selftests/powerpc/copyloops/asm/asm-compat.h b/tools/testing/selftests/powerpc/copyloops/asm/asm-compat.h
new file mode 100644
index 000000000..e69de29bb
--- /dev/null
+++ b/tools/testing/selftests/powerpc/copyloops/asm/asm-compat.h
diff --git a/tools/testing/selftests/powerpc/copyloops/asm/export.h b/tools/testing/selftests/powerpc/copyloops/asm/export.h
new file mode 100644
index 000000000..e6b80d5fb
--- /dev/null
+++ b/tools/testing/selftests/powerpc/copyloops/asm/export.h
@@ -0,0 +1,4 @@
+/* SPDX-License-Identifier: GPL-2.0 */
+#define EXPORT_SYMBOL(x)
+#define EXPORT_SYMBOL_GPL(x)
+#define EXPORT_SYMBOL_KASAN(x)
diff --git a/tools/testing/selftests/powerpc/copyloops/asm/feature-fixups.h b/tools/testing/selftests/powerpc/copyloops/asm/feature-fixups.h
new file mode 100644
index 000000000..e69de29bb
--- /dev/null
+++ b/tools/testing/selftests/powerpc/copyloops/asm/feature-fixups.h
diff --git a/tools/testing/selftests/powerpc/copyloops/asm/kasan.h b/tools/testing/selftests/powerpc/copyloops/asm/kasan.h
new file mode 100644
index 000000000..e69de29bb
--- /dev/null
+++ b/tools/testing/selftests/powerpc/copyloops/asm/kasan.h
diff --git a/tools/testing/selftests/powerpc/copyloops/asm/ppc_asm.h b/tools/testing/selftests/powerpc/copyloops/asm/ppc_asm.h
new file mode 100644
index 000000000..58c1cef3e
--- /dev/null
+++ b/tools/testing/selftests/powerpc/copyloops/asm/ppc_asm.h
@@ -0,0 +1,48 @@
+/* SPDX-License-Identifier: GPL-2.0 */
+#ifndef __SELFTESTS_POWERPC_PPC_ASM_H
+#define __SELFTESTS_POWERPC_PPC_ASM_H
+#include <ppc-asm.h>
+
+#define CONFIG_ALTIVEC
+
+#define r1	1
+
+#define R14 r14
+#define R15 r15
+#define R16 r16
+#define R17 r17
+#define R18 r18
+#define R19 r19
+#define R20 r20
+#define R21 r21
+#define R22 r22
+#define R29 r29
+#define R30 r30
+#define R31 r31
+
+#define STACKFRAMESIZE	256
+#define STK_REG(i)	(112 + ((i)-14)*8)
+
+#define _GLOBAL(A) FUNC_START(test_ ## A)
+#define _GLOBAL_TOC(A) _GLOBAL(A)
+#define _GLOBAL_TOC_KASAN(A) _GLOBAL(A)
+
+#define PPC_MTOCRF(A, B)	mtocrf A, B
+
+#define EX_TABLE(x, y)			\
+	.section __ex_table,"a";	\
+	.8byte	x, y;			\
+	.previous
+
+#define BEGIN_FTR_SECTION		.if test_feature
+#define FTR_SECTION_ELSE		.else
+#define ALT_FTR_SECTION_END_IFCLR(x)	.endif
+#define ALT_FTR_SECTION_END_IFSET(x)	.endif
+#define ALT_FTR_SECTION_END(x, y)	.endif
+#define END_FTR_SECTION_IFCLR(x)	.endif
+#define END_FTR_SECTION_IFSET(x)	.endif
+
+/* Default to taking the first of any alternative feature sections */
+test_feature = 1
+
+#endif /* __SELFTESTS_POWERPC_PPC_ASM_H */
diff --git a/tools/testing/selftests/powerpc/copyloops/asm/processor.h b/tools/testing/selftests/powerpc/copyloops/asm/processor.h
new file mode 100644
index 000000000..e69de29bb
--- /dev/null
+++ b/tools/testing/selftests/powerpc/copyloops/asm/processor.h
diff --git a/tools/testing/selftests/powerpc/copyloops/copy_mc_64.S b/tools/testing/selftests/powerpc/copyloops/copy_mc_64.S
new file mode 120000
index 000000000..dcbe06d50
--- /dev/null
+++ b/tools/testing/selftests/powerpc/copyloops/copy_mc_64.S
@@ -0,0 +1 @@
+../../../../../arch/powerpc/lib/copy_mc_64.S
\ No newline at end of file
diff --git a/tools/testing/selftests/powerpc/copyloops/copy_tofrom_user_reference.S b/tools/testing/selftests/powerpc/copyloops/copy_tofrom_user_reference.S
new file mode 100644
index 000000000..3363b8640
--- /dev/null
+++ b/tools/testing/selftests/powerpc/copyloops/copy_tofrom_user_reference.S
@@ -0,0 +1,24 @@
+#include <asm/ppc_asm.h>
+
+_GLOBAL(copy_tofrom_user_reference)
+	cmpdi	r5,0
+	beq	4f
+
+	mtctr	r5
+
+1:	lbz	r6,0(r4)
+2:	stb	r6,0(r3)
+	addi	r3,r3,1
+	addi	r4,r4,1
+	bdnz	1b
+
+3:	mfctr	r3
+	blr
+
+4:	mr	r3,r5
+	blr
+
+.section __ex_table,"a"
+	.llong	1b,3b
+	.llong	2b,3b
+.text
diff --git a/tools/testing/selftests/powerpc/copyloops/copyuser_64.S b/tools/testing/selftests/powerpc/copyloops/copyuser_64.S
new file mode 120000
index 000000000..f1c418a25
--- /dev/null
+++ b/tools/testing/selftests/powerpc/copyloops/copyuser_64.S
@@ -0,0 +1 @@
+../../../../../arch/powerpc/lib/copyuser_64.S
\ No newline at end of file
diff --git a/tools/testing/selftests/powerpc/copyloops/copyuser_power7.S b/tools/testing/selftests/powerpc/copyloops/copyuser_power7.S
new file mode 120000
index 000000000..478689598
--- /dev/null
+++ b/tools/testing/selftests/powerpc/copyloops/copyuser_power7.S
@@ -0,0 +1 @@
+../../../../../arch/powerpc/lib/copyuser_power7.S
\ No newline at end of file
diff --git a/tools/testing/selftests/powerpc/copyloops/exc_validate.c b/tools/testing/selftests/powerpc/copyloops/exc_validate.c
new file mode 100644
index 000000000..c896ea9a7
--- /dev/null
+++ b/tools/testing/selftests/powerpc/copyloops/exc_validate.c
@@ -0,0 +1,124 @@
+#include <stdlib.h>
+#include <string.h>
+#include <stdio.h>
+#include <signal.h>
+#include <unistd.h>
+#include <sys/mman.h>
+
+#include "utils.h"
+
+extern char __start___ex_table[];
+extern char __stop___ex_table[];
+
+#if defined(__powerpc64__)
+#define UCONTEXT_NIA(UC)	(UC)->uc_mcontext.gp_regs[PT_NIP]
+#elif defined(__powerpc__)
+#define UCONTEXT_NIA(UC)	(UC)->uc_mcontext.uc_regs->gregs[PT_NIP]
+#else
+#error implement UCONTEXT_NIA
+#endif
+
+static void segv_handler(int signr, siginfo_t *info, void *ptr)
+{
+	ucontext_t *uc = (ucontext_t *)ptr;
+	unsigned long addr = (unsigned long)info->si_addr;
+	unsigned long *ip = &UCONTEXT_NIA(uc);
+	unsigned long *ex_p = (unsigned long *)__start___ex_table;
+
+	while (ex_p < (unsigned long *)__stop___ex_table) {
+		unsigned long insn, fixup;
+
+		insn = *ex_p++;
+		fixup = *ex_p++;
+
+		if (insn == *ip) {
+			*ip = fixup;
+			return;
+		}
+	}
+
+	printf("No exception table match for NIA %lx ADDR %lx\n", *ip, addr);
+	abort();
+}
+
+static void setup_segv_handler(void)
+{
+	struct sigaction action;
+
+	memset(&action, 0, sizeof(action));
+	action.sa_sigaction = segv_handler;
+	action.sa_flags = SA_SIGINFO;
+	sigaction(SIGSEGV, &action, NULL);
+}
+
+unsigned long COPY_LOOP(void *to, const void *from, unsigned long size);
+unsigned long test_copy_tofrom_user_reference(void *to, const void *from, unsigned long size);
+
+static int total_passed;
+static int total_failed;
+
+static void do_one_test(char *dstp, char *srcp, unsigned long len)
+{
+	unsigned long got, expected;
+
+	got = COPY_LOOP(dstp, srcp, len);
+	expected = test_copy_tofrom_user_reference(dstp, srcp, len);
+
+	if (got != expected) {
+		total_failed++;
+		printf("FAIL from=%p to=%p len=%ld returned %ld, expected %ld\n",
+		       srcp, dstp, len, got, expected);
+		//abort();
+	} else
+		total_passed++;
+}
+
+//#define MAX_LEN 512
+#define MAX_LEN 16
+
+int test_copy_exception(void)
+{
+	int page_size;
+	static char *p, *q;
+	unsigned long src, dst, len;
+
+	page_size = getpagesize();
+	p = mmap(NULL, page_size * 2, PROT_READ|PROT_WRITE,
+		MAP_PRIVATE|MAP_ANONYMOUS, -1, 0);
+
+	if (p == MAP_FAILED) {
+		perror("mmap");
+		exit(1);
+	}
+
+	memset(p, 0, page_size);
+
+	setup_segv_handler();
+
+	if (mprotect(p + page_size, page_size, PROT_NONE)) {
+		perror("mprotect");
+		exit(1);
+	}
+
+	q = p + page_size - MAX_LEN;
+
+	for (src = 0; src < MAX_LEN; src++) {
+		for (dst = 0; dst < MAX_LEN; dst++) {
+			for (len = 0; len < MAX_LEN+1; len++) {
+				// printf("from=%p to=%p len=%ld\n", q+dst, q+src, len);
+				do_one_test(q+dst, q+src, len);
+			}
+		}
+	}
+
+	printf("Totals:\n");
+	printf("  Pass: %d\n", total_passed);
+	printf("  Fail: %d\n", total_failed);
+
+	return 0;
+}
+
+int main(void)
+{
+	return test_harness(test_copy_exception, str(COPY_LOOP));
+}
diff --git a/tools/testing/selftests/powerpc/copyloops/memcpy_64.S b/tools/testing/selftests/powerpc/copyloops/memcpy_64.S
new file mode 120000
index 000000000..cce33fb6f
--- /dev/null
+++ b/tools/testing/selftests/powerpc/copyloops/memcpy_64.S
@@ -0,0 +1 @@
+../../../../../arch/powerpc/lib/memcpy_64.S
\ No newline at end of file
diff --git a/tools/testing/selftests/powerpc/copyloops/memcpy_power7.S b/tools/testing/selftests/powerpc/copyloops/memcpy_power7.S
new file mode 120000
index 000000000..0d6fbfaf3
--- /dev/null
+++ b/tools/testing/selftests/powerpc/copyloops/memcpy_power7.S
@@ -0,0 +1 @@
+../../../../../arch/powerpc/lib/memcpy_power7.S
\ No newline at end of file
diff --git a/tools/testing/selftests/powerpc/copyloops/stubs.S b/tools/testing/selftests/powerpc/copyloops/stubs.S
new file mode 100644
index 000000000..ec8bcf2bf
--- /dev/null
+++ b/tools/testing/selftests/powerpc/copyloops/stubs.S
@@ -0,0 +1,19 @@
+#include <asm/ppc_asm.h>
+
+FUNC_START(enter_vmx_usercopy)
+	li	r3,1
+	blr
+
+FUNC_START(exit_vmx_usercopy)
+	li	r3,0
+	blr
+
+FUNC_START(enter_vmx_ops)
+	li	r3,1
+	blr
+
+FUNC_START(exit_vmx_ops)
+	blr
+
+FUNC_START(__copy_tofrom_user_base)
+	blr
diff --git a/tools/testing/selftests/powerpc/copyloops/validate.c b/tools/testing/selftests/powerpc/copyloops/validate.c
new file mode 100644
index 000000000..0f6873618
--- /dev/null
+++ b/tools/testing/selftests/powerpc/copyloops/validate.c
@@ -0,0 +1,100 @@
+// SPDX-License-Identifier: GPL-2.0
+#include <malloc.h>
+#include <string.h>
+#include <stdlib.h>
+#include <stdbool.h>
+
+#include "utils.h"
+
+#define MAX_LEN 8192
+#define MAX_OFFSET 16
+#define MIN_REDZONE 128
+#define BUFLEN (MAX_LEN+MAX_OFFSET+2*MIN_REDZONE)
+#define POISON 0xa5
+
+unsigned long COPY_LOOP(void *to, const void *from, unsigned long size);
+
+static void do_one(char *src, char *dst, unsigned long src_off,
+		   unsigned long dst_off, unsigned long len, void *redzone,
+		   void *fill)
+{
+	char *srcp, *dstp;
+	unsigned long ret;
+	unsigned long i;
+
+	srcp = src + MIN_REDZONE + src_off;
+	dstp = dst + MIN_REDZONE + dst_off;
+
+	memset(src, POISON, BUFLEN);
+	memset(dst, POISON, BUFLEN);
+	memcpy(srcp, fill, len);
+
+	ret = COPY_LOOP(dstp, srcp, len);
+	if (ret && ret != (unsigned long)dstp) {
+		printf("(%p,%p,%ld) returned %ld\n", dstp, srcp, len, ret);
+		abort();
+	}
+
+	if (memcmp(dstp, srcp, len)) {
+		printf("(%p,%p,%ld) miscompare\n", dstp, srcp, len);
+		printf("src: ");
+		for (i = 0; i < len; i++)
+			printf("%02x ", srcp[i]);
+		printf("\ndst: ");
+		for (i = 0; i < len; i++)
+			printf("%02x ", dstp[i]);
+		printf("\n");
+		abort();
+	}
+
+	if (memcmp(dst, redzone, dstp - dst)) {
+		printf("(%p,%p,%ld) redzone before corrupted\n",
+		       dstp, srcp, len);
+		abort();
+	}
+
+	if (memcmp(dstp+len, redzone, dst+BUFLEN-(dstp+len))) {
+		printf("(%p,%p,%ld) redzone after corrupted\n",
+		       dstp, srcp, len);
+		abort();
+	}
+}
+
+int test_copy_loop(void)
+{
+	char *src, *dst, *redzone, *fill;
+	unsigned long len, src_off, dst_off;
+	unsigned long i;
+
+	src = memalign(BUFLEN, BUFLEN);
+	dst = memalign(BUFLEN, BUFLEN);
+	redzone = malloc(BUFLEN);
+	fill = malloc(BUFLEN);
+
+	if (!src || !dst || !redzone || !fill) {
+		fprintf(stderr, "malloc failed\n");
+		exit(1);
+	}
+
+	memset(redzone, POISON, BUFLEN);
+
+	/* Fill with sequential bytes */
+	for (i = 0; i < BUFLEN; i++)
+		fill[i] = i & 0xff;
+
+	for (len = 1; len < MAX_LEN; len++) {
+		for (src_off = 0; src_off < MAX_OFFSET; src_off++) {
+			for (dst_off = 0; dst_off < MAX_OFFSET; dst_off++) {
+				do_one(src, dst, src_off, dst_off, len,
+				       redzone, fill);
+			}
+		}
+	}
+
+	return 0;
+}
+
+int main(void)
+{
+	return test_harness(test_copy_loop, str(COPY_LOOP));
+}
diff --git a/tools/testing/selftests/powerpc/dscr/.gitignore b/tools/testing/selftests/powerpc/dscr/.gitignore
new file mode 100644
index 000000000..1d08b15af
--- /dev/null
+++ b/tools/testing/selftests/powerpc/dscr/.gitignore
@@ -0,0 +1,8 @@
+# SPDX-License-Identifier: GPL-2.0-only
+dscr_default_test
+dscr_explicit_test
+dscr_inherit_exec_test
+dscr_inherit_test
+dscr_sysfs_test
+dscr_sysfs_thread_test
+dscr_user_test
diff --git a/tools/testing/selftests/powerpc/dscr/Makefile b/tools/testing/selftests/powerpc/dscr/Makefile
new file mode 100644
index 000000000..845db6273
--- /dev/null
+++ b/tools/testing/selftests/powerpc/dscr/Makefile
@@ -0,0 +1,13 @@
+# SPDX-License-Identifier: GPL-2.0
+TEST_GEN_PROGS := dscr_default_test dscr_explicit_test dscr_user_test	\
+	      dscr_inherit_test dscr_inherit_exec_test dscr_sysfs_test	\
+	      dscr_sysfs_thread_test
+
+TEST_FILES := settings
+
+top_srcdir = ../../../../..
+include ../../lib.mk
+
+$(OUTPUT)/dscr_default_test: LDLIBS += -lpthread
+
+$(TEST_GEN_PROGS): ../harness.c ../utils.c
diff --git a/tools/testing/selftests/powerpc/dscr/dscr.h b/tools/testing/selftests/powerpc/dscr/dscr.h
new file mode 100644
index 000000000..13e9b9e28
--- /dev/null
+++ b/tools/testing/selftests/powerpc/dscr/dscr.h
@@ -0,0 +1,122 @@
+/* SPDX-License-Identifier: GPL-2.0-only */
+/*
+ * POWER Data Stream Control Register (DSCR)
+ *
+ * This header file contains helper functions and macros
+ * required for all the DSCR related test cases.
+ *
+ * Copyright 2012, Anton Blanchard, IBM Corporation.
+ * Copyright 2015, Anshuman Khandual, IBM Corporation.
+ */
+#ifndef _SELFTESTS_POWERPC_DSCR_DSCR_H
+#define _SELFTESTS_POWERPC_DSCR_DSCR_H
+
+#include <unistd.h>
+#include <stdio.h>
+#include <stdlib.h>
+#include <string.h>
+#include <fcntl.h>
+#include <dirent.h>
+#include <pthread.h>
+#include <sched.h>
+#include <sys/types.h>
+#include <sys/stat.h>
+#include <sys/wait.h>
+
+#include "utils.h"
+
+#define THREADS		100	/* Max threads */
+#define COUNT		100	/* Max iterations */
+#define DSCR_MAX	16	/* Max DSCR value */
+#define LEN_MAX		100	/* Max name length */
+
+#define DSCR_DEFAULT	"/sys/devices/system/cpu/dscr_default"
+#define CPU_PATH	"/sys/devices/system/cpu/"
+
+#define rmb()  asm volatile("lwsync":::"memory")
+#define wmb()  asm volatile("lwsync":::"memory")
+
+#define READ_ONCE(x) (*(volatile typeof(x) *)&(x))
+
+/* Prilvilege state DSCR access */
+inline unsigned long get_dscr(void)
+{
+	unsigned long ret;
+
+	asm volatile("mfspr %0,%1" : "=r" (ret) : "i" (SPRN_DSCR_PRIV));
+
+	return ret;
+}
+
+inline void set_dscr(unsigned long val)
+{
+	asm volatile("mtspr %1,%0" : : "r" (val), "i" (SPRN_DSCR_PRIV));
+}
+
+/* Problem state DSCR access */
+inline unsigned long get_dscr_usr(void)
+{
+	unsigned long ret;
+
+	asm volatile("mfspr %0,%1" : "=r" (ret) : "i" (SPRN_DSCR));
+
+	return ret;
+}
+
+inline void set_dscr_usr(unsigned long val)
+{
+	asm volatile("mtspr %1,%0" : : "r" (val), "i" (SPRN_DSCR));
+}
+
+/* Default DSCR access */
+unsigned long get_default_dscr(void)
+{
+	int fd = -1, ret;
+	char buf[16];
+	unsigned long val;
+
+	if (fd == -1) {
+		fd = open(DSCR_DEFAULT, O_RDONLY);
+		if (fd == -1) {
+			perror("open() failed");
+			exit(1);
+		}
+	}
+	memset(buf, 0, sizeof(buf));
+	lseek(fd, 0, SEEK_SET);
+	ret = read(fd, buf, sizeof(buf));
+	if (ret == -1) {
+		perror("read() failed");
+		exit(1);
+	}
+	sscanf(buf, "%lx", &val);
+	close(fd);
+	return val;
+}
+
+void set_default_dscr(unsigned long val)
+{
+	int fd = -1, ret;
+	char buf[16];
+
+	if (fd == -1) {
+		fd = open(DSCR_DEFAULT, O_RDWR);
+		if (fd == -1) {
+			perror("open() failed");
+			exit(1);
+		}
+	}
+	sprintf(buf, "%lx\n", val);
+	ret = write(fd, buf, strlen(buf));
+	if (ret == -1) {
+		perror("write() failed");
+		exit(1);
+	}
+	close(fd);
+}
+
+double uniform_deviate(int seed)
+{
+	return seed * (1.0 / (RAND_MAX + 1.0));
+}
+#endif	/* _SELFTESTS_POWERPC_DSCR_DSCR_H */
diff --git a/tools/testing/selftests/powerpc/dscr/dscr_default_test.c b/tools/testing/selftests/powerpc/dscr/dscr_default_test.c
new file mode 100644
index 000000000..e76611e60
--- /dev/null
+++ b/tools/testing/selftests/powerpc/dscr/dscr_default_test.c
@@ -0,0 +1,126 @@
+// SPDX-License-Identifier: GPL-2.0-only
+/*
+ * POWER Data Stream Control Register (DSCR) default test
+ *
+ * This test modifies the system wide default DSCR through
+ * it's sysfs interface and then verifies that all threads
+ * see the correct changed DSCR value immediately.
+ *
+ * Copyright 2012, Anton Blanchard, IBM Corporation.
+ * Copyright 2015, Anshuman Khandual, IBM Corporation.
+ */
+#include "dscr.h"
+
+static unsigned long dscr;		/* System DSCR default */
+static unsigned long sequence;
+static unsigned long result[THREADS];
+
+static void *do_test(void *in)
+{
+	unsigned long thread = (unsigned long)in;
+	unsigned long i;
+
+	for (i = 0; i < COUNT; i++) {
+		unsigned long d, cur_dscr, cur_dscr_usr;
+		unsigned long s1, s2;
+
+		s1 = READ_ONCE(sequence);
+		if (s1 & 1)
+			continue;
+		rmb();
+
+		d = dscr;
+		cur_dscr = get_dscr();
+		cur_dscr_usr = get_dscr_usr();
+
+		rmb();
+		s2 = sequence;
+
+		if (s1 != s2)
+			continue;
+
+		if (cur_dscr != d) {
+			fprintf(stderr, "thread %ld kernel DSCR should be %ld "
+				"but is %ld\n", thread, d, cur_dscr);
+			result[thread] = 1;
+			pthread_exit(&result[thread]);
+		}
+
+		if (cur_dscr_usr != d) {
+			fprintf(stderr, "thread %ld user DSCR should be %ld "
+				"but is %ld\n", thread, d, cur_dscr_usr);
+			result[thread] = 1;
+			pthread_exit(&result[thread]);
+		}
+	}
+	result[thread] = 0;
+	pthread_exit(&result[thread]);
+}
+
+int dscr_default(void)
+{
+	pthread_t threads[THREADS];
+	unsigned long i, *status[THREADS];
+	unsigned long orig_dscr_default;
+
+	SKIP_IF(!have_hwcap2(PPC_FEATURE2_DSCR));
+
+	orig_dscr_default = get_default_dscr();
+
+	/* Initial DSCR default */
+	dscr = 1;
+	set_default_dscr(dscr);
+
+	/* Spawn all testing threads */
+	for (i = 0; i < THREADS; i++) {
+		if (pthread_create(&threads[i], NULL, do_test, (void *)i)) {
+			perror("pthread_create() failed");
+			goto fail;
+		}
+	}
+
+	srand(getpid());
+
+	/* Keep changing the DSCR default */
+	for (i = 0; i < COUNT; i++) {
+		double ret = uniform_deviate(rand());
+
+		if (ret < 0.0001) {
+			sequence++;
+			wmb();
+
+			dscr++;
+			if (dscr > DSCR_MAX)
+				dscr = 0;
+
+			set_default_dscr(dscr);
+
+			wmb();
+			sequence++;
+		}
+	}
+
+	/* Individual testing thread exit status */
+	for (i = 0; i < THREADS; i++) {
+		if (pthread_join(threads[i], (void **)&(status[i]))) {
+			perror("pthread_join() failed");
+			goto fail;
+		}
+
+		if (*status[i]) {
+			printf("%ldth thread failed to join with %ld status\n",
+								i, *status[i]);
+			goto fail;
+		}
+	}
+	set_default_dscr(orig_dscr_default);
+	return 0;
+fail:
+	set_default_dscr(orig_dscr_default);
+	return 1;
+}
+
+int main(int argc, char *argv[])
+{
+	return test_harness(dscr_default, "dscr_default_test");
+}
diff --git a/tools/testing/selftests/powerpc/dscr/dscr_explicit_test.c b/tools/testing/selftests/powerpc/dscr/dscr_explicit_test.c
new file mode 100644
index 000000000..32fcf2b32
--- /dev/null
+++ b/tools/testing/selftests/powerpc/dscr/dscr_explicit_test.c
@@ -0,0 +1,70 @@
+// SPDX-License-Identifier: GPL-2.0-only
+/*
+ * POWER Data Stream Control Register (DSCR) explicit test
+ *
+ * This test modifies the DSCR value using mtspr instruction and
+ * verifies the change with mfspr instruction. It uses both the
+ * privilege state SPR and the problem state SPR for this purpose.
+ *
+ * When using the privilege state SPR, the instructions such as
+ * mfspr or mtspr are priviledged and the kernel emulates them
+ * for us. Instructions using problem state SPR can be exuecuted
+ * directly without any emulation if the HW supports them. Else
+ * they also get emulated by the kernel.
+ *
+ * Copyright 2012, Anton Blanchard, IBM Corporation.
+ * Copyright 2015, Anshuman Khandual, IBM Corporation.
+ */
+#include "dscr.h"
+
+int dscr_explicit(void)
+{
+	unsigned long i, dscr = 0;
+
+	SKIP_IF(!have_hwcap2(PPC_FEATURE2_DSCR));
+
+	srand(getpid());
+	set_dscr(dscr);
+
+	for (i = 0; i < COUNT; i++) {
+		unsigned long cur_dscr, cur_dscr_usr;
+		double ret = uniform_deviate(rand());
+
+		if (ret < 0.001) {
+			dscr++;
+			if (dscr > DSCR_MAX)
+				dscr = 0;
+
+			set_dscr(dscr);
+		}
+
+		cur_dscr = get_dscr();
+		if (cur_dscr != dscr) {
+			fprintf(stderr, "Kernel DSCR should be %ld but "
+					"is %ld\n", dscr, cur_dscr);
+			return 1;
+		}
+
+		ret = uniform_deviate(rand());
+		if (ret < 0.001) {
+			dscr++;
+			if (dscr > DSCR_MAX)
+				dscr = 0;
+
+			set_dscr_usr(dscr);
+		}
+
+		cur_dscr_usr = get_dscr_usr();
+		if (cur_dscr_usr != dscr) {
+			fprintf(stderr, "User DSCR should be %ld but "
+					"is %ld\n", dscr, cur_dscr_usr);
+			return 1;
+		}
+	}
+	return 0;
+}
+
+int main(int argc, char *argv[])
+{
+	return test_harness(dscr_explicit, "dscr_explicit_test");
+}
diff --git a/tools/testing/selftests/powerpc/dscr/dscr_inherit_exec_test.c b/tools/testing/selftests/powerpc/dscr/dscr_inherit_exec_test.c
new file mode 100644
index 000000000..c6a81b2d6
--- /dev/null
+++ b/tools/testing/selftests/powerpc/dscr/dscr_inherit_exec_test.c
@@ -0,0 +1,108 @@
+// SPDX-License-Identifier: GPL-2.0-only
+/*
+ * POWER Data Stream Control Register (DSCR) fork exec test
+ *
+ * This testcase modifies the DSCR using mtspr, forks & execs and
+ * verifies that the child is using the changed DSCR using mfspr.
+ *
+ * When using the privilege state SPR, the instructions such as
+ * mfspr or mtspr are privileged and the kernel emulates them
+ * for us. Instructions using problem state SPR can be executed
+ * directly without any emulation if the HW supports them. Else
+ * they also get emulated by the kernel.
+ *
+ * Copyright 2012, Anton Blanchard, IBM Corporation.
+ * Copyright 2015, Anshuman Khandual, IBM Corporation.
+ */
+#include "dscr.h"
+
+static char *prog;
+
+static void do_exec(unsigned long parent_dscr)
+{
+	unsigned long cur_dscr, cur_dscr_usr;
+
+	cur_dscr = get_dscr();
+	cur_dscr_usr = get_dscr_usr();
+
+	if (cur_dscr != parent_dscr) {
+		fprintf(stderr, "Parent DSCR %ld was not inherited "
+				"over exec (kernel value)\n", parent_dscr);
+		exit(1);
+	}
+
+	if (cur_dscr_usr != parent_dscr) {
+		fprintf(stderr, "Parent DSCR %ld was not inherited "
+				"over exec (user value)\n", parent_dscr);
+		exit(1);
+	}
+	exit(0);
+}
+
+int dscr_inherit_exec(void)
+{
+	unsigned long i, dscr = 0;
+	pid_t pid;
+
+	SKIP_IF(!have_hwcap2(PPC_FEATURE2_DSCR));
+
+	for (i = 0; i < COUNT; i++) {
+		dscr++;
+		if (dscr > DSCR_MAX)
+			dscr = 0;
+
+		if (dscr == get_default_dscr())
+			continue;
+
+		if (i % 2 == 0)
+			set_dscr_usr(dscr);
+		else
+			set_dscr(dscr);
+
+		pid = fork();
+		if (pid == -1) {
+			perror("fork() failed");
+			exit(1);
+		} else if (pid) {
+			int status;
+
+			if (waitpid(pid, &status, 0) == -1) {
+				perror("waitpid() failed");
+				exit(1);
+			}
+
+			if (!WIFEXITED(status)) {
+				fprintf(stderr, "Child didn't exit cleanly\n");
+				exit(1);
+			}
+
+			if (WEXITSTATUS(status) != 0) {
+				fprintf(stderr, "Child didn't exit cleanly\n");
+				return 1;
+			}
+		} else {
+			char dscr_str[16];
+
+			sprintf(dscr_str, "%ld", dscr);
+			execlp(prog, prog, "exec", dscr_str, NULL);
+			exit(1);
+		}
+	}
+	return 0;
+}
+
+int main(int argc, char *argv[])
+{
+	if (argc == 3 && !strcmp(argv[1], "exec")) {
+		unsigned long parent_dscr;
+
+		parent_dscr = atoi(argv[2]);
+		do_exec(parent_dscr);
+	} else if (argc != 1) {
+		fprintf(stderr, "Usage: %s\n", argv[0]);
+		exit(1);
+	}
+
+	prog = argv[0];
+	return test_harness(dscr_inherit_exec, "dscr_inherit_exec_test");
+}
diff --git a/tools/testing/selftests/powerpc/dscr/dscr_inherit_test.c b/tools/testing/selftests/powerpc/dscr/dscr_inherit_test.c
new file mode 100644
index 000000000..f9dfd3d3c
--- /dev/null
+++ b/tools/testing/selftests/powerpc/dscr/dscr_inherit_test.c
@@ -0,0 +1,86 @@
+// SPDX-License-Identifier: GPL-2.0-only
+/*
+ * POWER Data Stream Control Register (DSCR) fork test
+ *
+ * This testcase modifies the DSCR using mtspr, forks and then
+ * verifies that the child process has the correct changed DSCR
+ * value using mfspr.
+ *
+ * When using the privilege state SPR, the instructions such as
+ * mfspr or mtspr are priviledged and the kernel emulates them
+ * for us. Instructions using problem state SPR can be exuecuted
+ * directly without any emulation if the HW supports them. Else
+ * they also get emulated by the kernel.
+ *
+ * Copyright 2012, Anton Blanchard, IBM Corporation.
+ * Copyright 2015, Anshuman Khandual, IBM Corporation.
+ */
+#include "dscr.h"
+
+int dscr_inherit(void)
+{
+	unsigned long i, dscr = 0;
+	pid_t pid;
+
+	SKIP_IF(!have_hwcap2(PPC_FEATURE2_DSCR));
+
+	srand(getpid());
+	set_dscr(dscr);
+
+	for (i = 0; i < COUNT; i++) {
+		unsigned long cur_dscr, cur_dscr_usr;
+
+		dscr++;
+		if (dscr > DSCR_MAX)
+			dscr = 0;
+
+		if (i % 2 == 0)
+			set_dscr_usr(dscr);
+		else
+			set_dscr(dscr);
+
+		pid = fork();
+		if (pid == -1) {
+			perror("fork() failed");
+			exit(1);
+		} else if (pid) {
+			int status;
+
+			if (waitpid(pid, &status, 0) == -1) {
+				perror("waitpid() failed");
+				exit(1);
+			}
+
+			if (!WIFEXITED(status)) {
+				fprintf(stderr, "Child didn't exit cleanly\n");
+				exit(1);
+			}
+
+			if (WEXITSTATUS(status) != 0) {
+				fprintf(stderr, "Child didn't exit cleanly\n");
+				return 1;
+			}
+		} else {
+			cur_dscr = get_dscr();
+			if (cur_dscr != dscr) {
+				fprintf(stderr, "Kernel DSCR should be %ld "
+					"but is %ld\n", dscr, cur_dscr);
+				exit(1);
+			}
+
+			cur_dscr_usr = get_dscr_usr();
+			if (cur_dscr_usr != dscr) {
+				fprintf(stderr, "User DSCR should be %ld "
+					"but is %ld\n", dscr, cur_dscr_usr);
+				exit(1);
+			}
+			exit(0);
+		}
+	}
+	return 0;
+}
+
+int main(int argc, char *argv[])
+{
+	return test_harness(dscr_inherit, "dscr_inherit_test");
+}
diff --git a/tools/testing/selftests/powerpc/dscr/dscr_sysfs_test.c b/tools/testing/selftests/powerpc/dscr/dscr_sysfs_test.c
new file mode 100644
index 000000000..f20d1c166
--- /dev/null
+++ b/tools/testing/selftests/powerpc/dscr/dscr_sysfs_test.c
@@ -0,0 +1,103 @@
+// SPDX-License-Identifier: GPL-2.0-only
+/*
+ * POWER Data Stream Control Register (DSCR) sysfs interface test
+ *
+ * This test updates to system wide DSCR default through the sysfs interface
+ * and then verifies that all the CPU specific DSCR defaults are updated as
+ * well verified from their sysfs interfaces.
+ *
+ * Copyright 2015, Anshuman Khandual, IBM Corporation.
+ */
+#include "dscr.h"
+
+static int check_cpu_dscr_default(char *file, unsigned long val)
+{
+	char buf[10];
+	int fd, rc;
+
+	fd = open(file, O_RDWR);
+	if (fd == -1) {
+		perror("open() failed");
+		return 1;
+	}
+
+	rc = read(fd, buf, sizeof(buf));
+	if (rc == -1) {
+		perror("read() failed");
+		close(fd);
+		return 1;
+	}
+	close(fd);
+
+	buf[rc] = '\0';
+	if (strtol(buf, NULL, 16) != val) {
+		printf("DSCR match failed: %ld (system) %ld (cpu)\n",
+					val, strtol(buf, NULL, 16));
+		return 1;
+	}
+	return 0;
+}
+
+static int check_all_cpu_dscr_defaults(unsigned long val)
+{
+	DIR *sysfs;
+	struct dirent *dp;
+	char file[LEN_MAX];
+
+	sysfs = opendir(CPU_PATH);
+	if (!sysfs) {
+		perror("opendir() failed");
+		return 1;
+	}
+
+	while ((dp = readdir(sysfs))) {
+		int len;
+
+		if (!(dp->d_type & DT_DIR))
+			continue;
+		if (!strcmp(dp->d_name, "cpuidle"))
+			continue;
+		if (!strstr(dp->d_name, "cpu"))
+			continue;
+
+		len = snprintf(file, LEN_MAX, "%s%s/dscr", CPU_PATH, dp->d_name);
+		if (len >= LEN_MAX)
+			continue;
+		if (access(file, F_OK))
+			continue;
+
+		if (check_cpu_dscr_default(file, val)) {
+			closedir(sysfs);
+			return 1;
+		}
+	}
+	closedir(sysfs);
+	return 0;
+}
+
+int dscr_sysfs(void)
+{
+	unsigned long orig_dscr_default;
+	int i, j;
+
+	SKIP_IF(!have_hwcap2(PPC_FEATURE2_DSCR));
+
+	orig_dscr_default = get_default_dscr();
+	for (i = 0; i < COUNT; i++) {
+		for (j = 0; j < DSCR_MAX; j++) {
+			set_default_dscr(j);
+			if (check_all_cpu_dscr_defaults(j))
+				goto fail;
+		}
+	}
+	set_default_dscr(orig_dscr_default);
+	return 0;
+fail:
+	set_default_dscr(orig_dscr_default);
+	return 1;
+}
+
+int main(int argc, char *argv[])
+{
+	return test_harness(dscr_sysfs, "dscr_sysfs_test");
+}
diff --git a/tools/testing/selftests/powerpc/dscr/dscr_sysfs_thread_test.c b/tools/testing/selftests/powerpc/dscr/dscr_sysfs_thread_test.c
new file mode 100644
index 000000000..191ed126f
--- /dev/null
+++ b/tools/testing/selftests/powerpc/dscr/dscr_sysfs_thread_test.c
@@ -0,0 +1,79 @@
+// SPDX-License-Identifier: GPL-2.0-only
+/*
+ * POWER Data Stream Control Register (DSCR) sysfs thread test
+ *
+ * This test updates the system wide DSCR default value through
+ * sysfs interface which should then update all the CPU specific
+ * DSCR default values which must also be then visible to threads
+ * executing on individual CPUs on the system.
+ *
+ * Copyright 2015, Anshuman Khandual, IBM Corporation.
+ */
+#define _GNU_SOURCE
+#include "dscr.h"
+
+static int test_thread_dscr(unsigned long val)
+{
+	unsigned long cur_dscr, cur_dscr_usr;
+
+	cur_dscr = get_dscr();
+	cur_dscr_usr = get_dscr_usr();
+
+	if (val != cur_dscr) {
+		printf("[cpu %d] Kernel DSCR should be %ld but is %ld\n",
+					sched_getcpu(), val, cur_dscr);
+		return 1;
+	}
+
+	if (val != cur_dscr_usr) {
+		printf("[cpu %d] User DSCR should be %ld but is %ld\n",
+					sched_getcpu(), val, cur_dscr_usr);
+		return 1;
+	}
+	return 0;
+}
+
+static int check_cpu_dscr_thread(unsigned long val)
+{
+	cpu_set_t mask;
+	int cpu;
+
+	for (cpu = 0; cpu < CPU_SETSIZE; cpu++) {
+		CPU_ZERO(&mask);
+		CPU_SET(cpu, &mask);
+		if (sched_setaffinity(0, sizeof(mask), &mask))
+			continue;
+
+		if (test_thread_dscr(val))
+			return 1;
+	}
+	return 0;
+
+}
+
+int dscr_sysfs_thread(void)
+{
+	unsigned long orig_dscr_default;
+	int i, j;
+
+	SKIP_IF(!have_hwcap2(PPC_FEATURE2_DSCR));
+
+	orig_dscr_default = get_default_dscr();
+	for (i = 0; i < COUNT; i++) {
+		for (j = 0; j < DSCR_MAX; j++) {
+			set_default_dscr(j);
+			if (check_cpu_dscr_thread(j))
+				goto fail;
+		}
+	}
+	set_default_dscr(orig_dscr_default);
+	return 0;
+fail:
+	set_default_dscr(orig_dscr_default);
+	return 1;
+}
+
+int main(int argc, char *argv[])
+{
+	return test_harness(dscr_sysfs_thread, "dscr_sysfs_thread_test");
+}
diff --git a/tools/testing/selftests/powerpc/dscr/dscr_user_test.c b/tools/testing/selftests/powerpc/dscr/dscr_user_test.c
new file mode 100644
index 000000000..e09072446
--- /dev/null
+++ b/tools/testing/selftests/powerpc/dscr/dscr_user_test.c
@@ -0,0 +1,60 @@
+// SPDX-License-Identifier: GPL-2.0-only
+/*
+ * POWER Data Stream Control Register (DSCR) SPR test
+ *
+ * This test modifies the DSCR value through both the SPR number
+ * based mtspr instruction and then makes sure that the same is
+ * reflected through mfspr instruction using either of the SPR
+ * numbers.
+ *
+ * When using the privilege state SPR, the instructions such as
+ * mfspr or mtspr are priviledged and the kernel emulates them
+ * for us. Instructions using problem state SPR can be exuecuted
+ * directly without any emulation if the HW supports them. Else
+ * they also get emulated by the kernel.
+ *
+ * Copyright 2013, Anton Blanchard, IBM Corporation.
+ * Copyright 2015, Anshuman Khandual, IBM Corporation.
+ */
+#include "dscr.h"
+
+static int check_dscr(char *str)
+{
+	unsigned long cur_dscr, cur_dscr_usr;
+
+	cur_dscr = get_dscr();
+	cur_dscr_usr = get_dscr_usr();
+	if (cur_dscr != cur_dscr_usr) {
+		printf("%s set, kernel get %lx != user get %lx\n",
+					str, cur_dscr, cur_dscr_usr);
+		return 1;
+	}
+	return 0;
+}
+
+int dscr_user(void)
+{
+	int i;
+
+	SKIP_IF(!have_hwcap2(PPC_FEATURE2_DSCR));
+
+	check_dscr("");
+
+	for (i = 0; i < COUNT; i++) {
+		set_dscr(i);
+		if (check_dscr("kernel"))
+			return 1;
+	}
+
+	for (i = 0; i < COUNT; i++) {
+		set_dscr_usr(i);
+		if (check_dscr("user"))
+			return 1;
+	}
+	return 0;
+}
+
+int main(int argc, char *argv[])
+{
+	return test_harness(dscr_user, "dscr_user_test");
+}
diff --git a/tools/testing/selftests/powerpc/dscr/settings b/tools/testing/selftests/powerpc/dscr/settings
new file mode 100644
index 000000000..e7b941753
--- /dev/null
+++ b/tools/testing/selftests/powerpc/dscr/settings
@@ -0,0 +1 @@
+timeout=0
diff --git a/tools/testing/selftests/powerpc/eeh/Makefile b/tools/testing/selftests/powerpc/eeh/Makefile
new file mode 100644
index 000000000..b397babd5
--- /dev/null
+++ b/tools/testing/selftests/powerpc/eeh/Makefile
@@ -0,0 +1,9 @@
+# SPDX-License-Identifier: GPL-2.0
+noarg:
+	$(MAKE) -C ../
+
+TEST_PROGS := eeh-basic.sh
+TEST_FILES := eeh-functions.sh
+
+top_srcdir = ../../../../..
+include ../../lib.mk
diff --git a/tools/testing/selftests/powerpc/eeh/eeh-basic.sh b/tools/testing/selftests/powerpc/eeh/eeh-basic.sh
new file mode 100755
index 000000000..64779f073
--- /dev/null
+++ b/tools/testing/selftests/powerpc/eeh/eeh-basic.sh
@@ -0,0 +1,90 @@
+#!/bin/sh
+# SPDX-License-Identifier: GPL-2.0-only
+
+KSELFTESTS_SKIP=4
+
+. ./eeh-functions.sh
+
+if ! eeh_supported ; then
+	echo "EEH not supported on this system, skipping"
+	exit $KSELFTESTS_SKIP;
+fi
+
+if [ ! -e "/sys/kernel/debug/powerpc/eeh_dev_check" ] && \
+   [ ! -e "/sys/kernel/debug/powerpc/eeh_dev_break" ] ; then
+	echo "debugfs EEH testing files are missing. Is debugfs mounted?"
+	exit $KSELFTESTS_SKIP;
+fi
+
+pre_lspci=`mktemp`
+lspci > $pre_lspci
+
+# Bump the max freeze count to something absurd so we don't
+# trip over it while breaking things.
+echo 5000 > /sys/kernel/debug/powerpc/eeh_max_freezes
+
+# record the devices that we break in here. Assuming everything
+# goes to plan we should get them back once the recover process
+# is finished.
+devices=""
+
+# Build up a list of candidate devices.
+for dev in `ls -1 /sys/bus/pci/devices/ | grep '\.0$'` ; do
+	# skip bridges since we can't recover them (yet...)
+	if [ -e "/sys/bus/pci/devices/$dev/pci_bus" ] ; then
+		echo "$dev, Skipped: bridge"
+		continue;
+	fi
+
+	# Skip VFs for now since we don't have a reliable way
+	# to break them.
+	if [ -e "/sys/bus/pci/devices/$dev/physfn" ] ; then
+		echo "$dev, Skipped: virtfn"
+		continue;
+	fi
+
+	if [ "ahci" = "$(basename $(realpath /sys/bus/pci/devices/$dev/driver))" ] ; then
+		echo "$dev, Skipped: ahci doesn't support recovery"
+		continue
+	fi
+
+	# Don't inject errosr into an already-frozen PE. This happens with
+	# PEs that contain multiple PCI devices (e.g. multi-function cards)
+	# and injecting new errors during the recovery process will probably
+	# result in the recovery failing and the device being marked as
+	# failed.
+	if ! pe_ok $dev ; then
+		echo "$dev, Skipped: Bad initial PE state"
+		continue;
+	fi
+
+	echo "$dev, Added"
+
+	# Add to this list of device to check
+	devices="$devices $dev"
+done
+
+dev_count="$(echo $devices | wc -w)"
+echo "Found ${dev_count} breakable devices..."
+
+failed=0
+for dev in $devices ; do
+	echo "Breaking $dev..."
+
+	if ! pe_ok $dev ; then
+		echo "Skipping $dev, Initial PE state is not ok"
+		failed="$((failed + 1))"
+		continue;
+	fi
+
+	if ! eeh_one_dev $dev ; then
+		failed="$((failed + 1))"
+	fi
+done
+
+echo "$failed devices failed to recover ($dev_count tested)"
+lspci | diff -u $pre_lspci -
+rm -f $pre_lspci
+
+test "$failed" -eq 0
+exit $?
diff --git a/tools/testing/selftests/powerpc/eeh/eeh-functions.sh b/tools/testing/selftests/powerpc/eeh/eeh-functions.sh
new file mode 100755
index 000000000..00dc32c0e
--- /dev/null
+++ b/tools/testing/selftests/powerpc/eeh/eeh-functions.sh
@@ -0,0 +1,85 @@
+#!/bin/sh
+# SPDX-License-Identifier: GPL-2.0-only
+
+pe_ok() {
+	local dev="$1"
+	local path="/sys/bus/pci/devices/$dev/eeh_pe_state"
+
+	# if a driver doesn't support the error handling callbacks then the
+	# device is recovered by removing and re-probing it. This causes the
+	# sysfs directory to disappear so read the PE state once and squash
+	# any potential error messages
+	local eeh_state="$(cat $path 2>/dev/null)"
+	if [ -z "$eeh_state" ]; then
+		return 1;
+	fi
+
+	local fw_state="$(echo $eeh_state | cut -d' ' -f1)"
+	local sw_state="$(echo $eeh_state | cut -d' ' -f2)"
+
+	# If EEH_PE_ISOLATED or EEH_PE_RECOVERING are set then the PE is in an
+	# error state or being recovered. Either way, not ok.
+	if [ "$((sw_state & 0x3))" -ne 0 ] ; then
+		return 1
+	fi
+
+	# A functioning PE should have the EEH_STATE_MMIO_ACTIVE and
+	# EEH_STATE_DMA_ACTIVE flags set. For some goddamn stupid reason
+	# the platform backends set these when the PE is in reset. The
+	# RECOVERING check above should stop any false positives though.
+	if [ "$((fw_state & 0x18))" -ne "$((0x18))" ] ; then
+		return 1
+	fi
+
+	return 0;
+}
+
+eeh_supported() {
+	test -e /proc/powerpc/eeh && \
+	grep -q 'EEH Subsystem is enabled' /proc/powerpc/eeh
+}
+
+eeh_one_dev() {
+	local dev="$1"
+
+	# Using this function from the command line is sometimes useful for
+	# testing so check that the argument is a well-formed sysfs device
+	# name.
+	if ! test -e /sys/bus/pci/devices/$dev/ ; then
+		echo "Error: '$dev' must be a sysfs device name (DDDD:BB:DD.F)"
+		return 1;
+	fi
+
+	# Break it
+	echo $dev >/sys/kernel/debug/powerpc/eeh_dev_break
+
+	# Force an EEH device check. If the kernel has already
+	# noticed the EEH (due to a driver poll or whatever), this
+	# is a no-op.
+	echo $dev >/sys/kernel/debug/powerpc/eeh_dev_check
+
+	# Default to a 60s timeout when waiting for a device to recover. This
+	# is an arbitrary default which can be overridden by setting the
+	# EEH_MAX_WAIT environmental variable when required.
+
+	# The current record holder for longest recovery time is:
+	#  "Adaptec Series 8 12G SAS/PCIe 3" at 39 seconds
+	max_wait=${EEH_MAX_WAIT:=60}
+
+	for i in `seq 0 ${max_wait}` ; do
+		if pe_ok $dev ; then
+			break;
+		fi
+		echo "$dev, waited $i/${max_wait}"
+		sleep 1
+	done
+
+	if ! pe_ok $dev ; then
+		echo "$dev, Failed to recover!"
+		return 1;
+	fi
+
+	echo "$dev, Recovered after $i seconds"
+	return 0;
+}
+
diff --git a/tools/testing/selftests/powerpc/harness.c b/tools/testing/selftests/powerpc/harness.c
new file mode 100644
index 000000000..0ad4f12b3
--- /dev/null
+++ b/tools/testing/selftests/powerpc/harness.c
@@ -0,0 +1,133 @@
+// SPDX-License-Identifier: GPL-2.0-only
+/*
+ * Copyright 2013, Michael Ellerman, IBM Corp.
+ */
+
+#include <errno.h>
+#include <signal.h>
+#include <stdbool.h>
+#include <stdio.h>
+#include <stdlib.h>
+#include <sys/types.h>
+#include <sys/wait.h>
+#include <unistd.h>
+#include <elf.h>
+#include <fcntl.h>
+#include <link.h>
+#include <sys/stat.h>
+
+#include "subunit.h"
+#include "utils.h"
+
+#define KILL_TIMEOUT	5
+
+/* Setting timeout to -1 disables the alarm */
+static uint64_t timeout = 120;
+
+int run_test(int (test_function)(void), char *name)
+{
+	bool terminated;
+	int rc, status;
+	pid_t pid;
+
+	/* Make sure output is flushed before forking */
+	fflush(stdout);
+
+	pid = fork();
+	if (pid == 0) {
+		setpgid(0, 0);
+		exit(test_function());
+	} else if (pid == -1) {
+		perror("fork");
+		return 1;
+	}
+
+	setpgid(pid, pid);
+
+	if (timeout != -1)
+		/* Wake us up in timeout seconds */
+		alarm(timeout);
+	terminated = false;
+
+wait:
+	rc = waitpid(pid, &status, 0);
+	if (rc == -1) {
+		if (errno != EINTR) {
+			printf("unknown error from waitpid\n");
+			return 1;
+		}
+
+		if (terminated) {
+			printf("!! force killing %s\n", name);
+			kill(-pid, SIGKILL);
+			return 1;
+		} else {
+			printf("!! killing %s\n", name);
+			kill(-pid, SIGTERM);
+			terminated = true;
+			alarm(KILL_TIMEOUT);
+			goto wait;
+		}
+	}
+
+	/* Kill anything else in the process group that is still running */
+	kill(-pid, SIGTERM);
+
+	if (WIFEXITED(status))
+		status = WEXITSTATUS(status);
+	else {
+		if (WIFSIGNALED(status))
+			printf("!! child died by signal %d\n", WTERMSIG(status));
+		else
+			printf("!! child died by unknown cause\n");
+
+		status = 1; /* Signal or other */
+	}
+
+	return status;
+}
+
+static void sig_handler(int signum)
+{
+	/* Just wake us up from waitpid */
+}
+
+static struct sigaction sig_action = {
+	.sa_handler = sig_handler,
+};
+
+void test_harness_set_timeout(uint64_t time)
+{
+	timeout = time;
+}
+
+int test_harness(int (test_function)(void), char *name)
+{
+	int rc;
+
+	test_start(name);
+	test_set_git_version(GIT_VERSION);
+
+	if (sigaction(SIGINT, &sig_action, NULL)) {
+		perror("sigaction (sigint)");
+		test_error(name);
+		return 1;
+	}
+
+	if (sigaction(SIGALRM, &sig_action, NULL)) {
+		perror("sigaction (sigalrm)");
+		test_error(name);
+		return 1;
+	}
+
+	rc = run_test(test_function, name);
+
+	if (rc == MAGIC_SKIP_RETURN_VALUE) {
+		test_skip(name);
+		/* so that skipped test is not marked as failed */
+		rc = 0;
+	} else
+		test_finish(name, rc);
+
+	return rc;
+}
diff --git a/tools/testing/selftests/powerpc/include/basic_asm.h b/tools/testing/selftests/powerpc/include/basic_asm.h
new file mode 100644
index 000000000..886dc026f
--- /dev/null
+++ b/tools/testing/selftests/powerpc/include/basic_asm.h
@@ -0,0 +1,74 @@
+/* SPDX-License-Identifier: GPL-2.0 */
+#ifndef _SELFTESTS_POWERPC_BASIC_ASM_H
+#define _SELFTESTS_POWERPC_BASIC_ASM_H
+
+#include <ppc-asm.h>
+#include <asm/unistd.h>
+
+#define LOAD_REG_IMMEDIATE(reg, expr) \
+	lis	reg, (expr)@highest;	\
+	ori	reg, reg, (expr)@higher;	\
+	rldicr	reg, reg, 32, 31;	\
+	oris	reg, reg, (expr)@high;	\
+	ori	reg, reg, (expr)@l;
+
+/*
+ * Note: These macros assume that variables being stored on the stack are
+ * doublewords, while this is usually the case it may not always be the
+ * case for each use case.
+ */
+#if defined(_CALL_ELF) && _CALL_ELF == 2
+#define STACK_FRAME_MIN_SIZE 32
+#define STACK_FRAME_TOC_POS  24
+#define __STACK_FRAME_PARAM(_param)  (32 + ((_param)*8))
+#define __STACK_FRAME_LOCAL(_num_params, _var_num)  \
+	((STACK_FRAME_PARAM(_num_params)) + ((_var_num)*8))
+#else
+#define STACK_FRAME_MIN_SIZE 112
+#define STACK_FRAME_TOC_POS  40
+#define __STACK_FRAME_PARAM(i)  (48 + ((i)*8))
+
+/*
+ * Caveat: if a function passed more than 8 doublewords, the caller will have
+ * made more space... which would render the 112 incorrect.
+ */
+#define __STACK_FRAME_LOCAL(_num_params, _var_num)  \
+	(112 + ((_var_num)*8))
+#endif
+
+/* Parameter x saved to the stack */
+#define STACK_FRAME_PARAM(var)    __STACK_FRAME_PARAM(var)
+
+/* Local variable x saved to the stack after x parameters */
+#define STACK_FRAME_LOCAL(num_params, var)    \
+	__STACK_FRAME_LOCAL(num_params, var)
+#define STACK_FRAME_LR_POS   16
+#define STACK_FRAME_CR_POS   8
+
+/*
+ * It is very important to note here that _extra is the extra amount of
+ * stack space needed. This space can be accessed using STACK_FRAME_PARAM()
+ * or STACK_FRAME_LOCAL() macros.
+ *
+ * r1 and r2 are not defined in ppc-asm.h (instead they are defined as sp
+ * and toc). Kernel programmers tend to prefer rX even for r1 and r2, hence
+ * %1 and %r2. r0 is defined in ppc-asm.h and therefore %r0 gets
+ * preprocessed incorrectly, hence r0.
+ */
+#define PUSH_BASIC_STACK(_extra) \
+	mflr	r0; \
+	std	r0, STACK_FRAME_LR_POS(%r1); \
+	stdu	%r1, -(_extra + STACK_FRAME_MIN_SIZE)(%r1); \
+	mfcr	r0; \
+	stw	r0, STACK_FRAME_CR_POS(%r1); \
+	std	%r2, STACK_FRAME_TOC_POS(%r1);
+
+#define POP_BASIC_STACK(_extra) \
+	ld	%r2, STACK_FRAME_TOC_POS(%r1); \
+	lwz	r0, STACK_FRAME_CR_POS(%r1); \
+	mtcr	r0; \
+	addi	%r1, %r1, (_extra + STACK_FRAME_MIN_SIZE); \
+	ld	r0, STACK_FRAME_LR_POS(%r1); \
+	mtlr	r0;
+
+#endif /* _SELFTESTS_POWERPC_BASIC_ASM_H */
diff --git a/tools/testing/selftests/powerpc/include/fpu_asm.h b/tools/testing/selftests/powerpc/include/fpu_asm.h
new file mode 100644
index 000000000..58ac2ce33
--- /dev/null
+++ b/tools/testing/selftests/powerpc/include/fpu_asm.h
@@ -0,0 +1,76 @@
+/* SPDX-License-Identifier: GPL-2.0-or-later */
+/*
+ * Copyright 2016, Cyril Bur, IBM Corp.
+ */
+
+#ifndef _SELFTESTS_POWERPC_FPU_ASM_H
+#define _SELFTESTS_POWERPC_FPU_ASM_H
+#include "basic_asm.h"
+
+#define PUSH_FPU(stack_size) \
+	stfd	f31,(stack_size + STACK_FRAME_MIN_SIZE)(%r1); \
+	stfd	f30,(stack_size + STACK_FRAME_MIN_SIZE - 8)(%r1); \
+	stfd	f29,(stack_size + STACK_FRAME_MIN_SIZE - 16)(%r1); \
+	stfd	f28,(stack_size + STACK_FRAME_MIN_SIZE - 24)(%r1); \
+	stfd	f27,(stack_size + STACK_FRAME_MIN_SIZE - 32)(%r1); \
+	stfd	f26,(stack_size + STACK_FRAME_MIN_SIZE - 40)(%r1); \
+	stfd	f25,(stack_size + STACK_FRAME_MIN_SIZE - 48)(%r1); \
+	stfd	f24,(stack_size + STACK_FRAME_MIN_SIZE - 56)(%r1); \
+	stfd	f23,(stack_size + STACK_FRAME_MIN_SIZE - 64)(%r1); \
+	stfd	f22,(stack_size + STACK_FRAME_MIN_SIZE - 72)(%r1); \
+	stfd	f21,(stack_size + STACK_FRAME_MIN_SIZE - 80)(%r1); \
+	stfd	f20,(stack_size + STACK_FRAME_MIN_SIZE - 88)(%r1); \
+	stfd	f19,(stack_size + STACK_FRAME_MIN_SIZE - 96)(%r1); \
+	stfd	f18,(stack_size + STACK_FRAME_MIN_SIZE - 104)(%r1); \
+	stfd	f17,(stack_size + STACK_FRAME_MIN_SIZE - 112)(%r1); \
+	stfd	f16,(stack_size + STACK_FRAME_MIN_SIZE - 120)(%r1); \
+	stfd	f15,(stack_size + STACK_FRAME_MIN_SIZE - 128)(%r1); \
+	stfd	f14,(stack_size + STACK_FRAME_MIN_SIZE - 136)(%r1);
+
+#define POP_FPU(stack_size) \
+	lfd	f31,(stack_size + STACK_FRAME_MIN_SIZE)(%r1); \
+	lfd	f30,(stack_size + STACK_FRAME_MIN_SIZE - 8)(%r1); \
+	lfd	f29,(stack_size + STACK_FRAME_MIN_SIZE - 16)(%r1); \
+	lfd	f28,(stack_size + STACK_FRAME_MIN_SIZE - 24)(%r1); \
+	lfd	f27,(stack_size + STACK_FRAME_MIN_SIZE - 32)(%r1); \
+	lfd	f26,(stack_size + STACK_FRAME_MIN_SIZE - 40)(%r1); \
+	lfd	f25,(stack_size + STACK_FRAME_MIN_SIZE - 48)(%r1); \
+	lfd	f24,(stack_size + STACK_FRAME_MIN_SIZE - 56)(%r1); \
+	lfd	f23,(stack_size + STACK_FRAME_MIN_SIZE - 64)(%r1); \
+	lfd	f22,(stack_size + STACK_FRAME_MIN_SIZE - 72)(%r1); \
+	lfd	f21,(stack_size + STACK_FRAME_MIN_SIZE - 80)(%r1); \
+	lfd	f20,(stack_size + STACK_FRAME_MIN_SIZE - 88)(%r1); \
+	lfd	f19,(stack_size + STACK_FRAME_MIN_SIZE - 96)(%r1); \
+	lfd	f18,(stack_size + STACK_FRAME_MIN_SIZE - 104)(%r1); \
+	lfd	f17,(stack_size + STACK_FRAME_MIN_SIZE - 112)(%r1); \
+	lfd	f16,(stack_size + STACK_FRAME_MIN_SIZE - 120)(%r1); \
+	lfd	f15,(stack_size + STACK_FRAME_MIN_SIZE - 128)(%r1); \
+	lfd	f14,(stack_size + STACK_FRAME_MIN_SIZE - 136)(%r1);
+
+/*
+ * Careful calling this, it will 'clobber' fpu (by design)
+ * Don't call this from C
+ */
+FUNC_START(load_fpu)
+	lfd	f14,0(r3)
+	lfd	f15,8(r3)
+	lfd	f16,16(r3)
+	lfd	f17,24(r3)
+	lfd	f18,32(r3)
+	lfd	f19,40(r3)
+	lfd	f20,48(r3)
+	lfd	f21,56(r3)
+	lfd	f22,64(r3)
+	lfd	f23,72(r3)
+	lfd	f24,80(r3)
+	lfd	f25,88(r3)
+	lfd	f26,96(r3)
+	lfd	f27,104(r3)
+	lfd	f28,112(r3)
+	lfd	f29,120(r3)
+	lfd	f30,128(r3)
+	lfd	f31,136(r3)
+	blr
+FUNC_END(load_fpu)
+
+#endif /* _SELFTESTS_POWERPC_FPU_ASM_H */
diff --git a/tools/testing/selftests/powerpc/include/gpr_asm.h b/tools/testing/selftests/powerpc/include/gpr_asm.h
new file mode 100644
index 000000000..5db74f5c6
--- /dev/null
+++ b/tools/testing/selftests/powerpc/include/gpr_asm.h
@@ -0,0 +1,92 @@
+/* SPDX-License-Identifier: GPL-2.0-or-later */
+/*
+ * Copyright 2016, Cyril Bur, IBM Corp.
+ */
+
+#ifndef _SELFTESTS_POWERPC_GPR_ASM_H
+#define _SELFTESTS_POWERPC_GPR_ASM_H
+
+#include "basic_asm.h"
+
+#define __PUSH_NVREGS(top_pos); \
+	std r31,(top_pos)(%r1); \
+	std r30,(top_pos - 8)(%r1); \
+	std r29,(top_pos - 16)(%r1); \
+	std r28,(top_pos - 24)(%r1); \
+	std r27,(top_pos - 32)(%r1); \
+	std r26,(top_pos - 40)(%r1); \
+	std r25,(top_pos - 48)(%r1); \
+	std r24,(top_pos - 56)(%r1); \
+	std r23,(top_pos - 64)(%r1); \
+	std r22,(top_pos - 72)(%r1); \
+	std r21,(top_pos - 80)(%r1); \
+	std r20,(top_pos - 88)(%r1); \
+	std r19,(top_pos - 96)(%r1); \
+	std r18,(top_pos - 104)(%r1); \
+	std r17,(top_pos - 112)(%r1); \
+	std r16,(top_pos - 120)(%r1); \
+	std r15,(top_pos - 128)(%r1); \
+	std r14,(top_pos - 136)(%r1)
+
+#define __POP_NVREGS(top_pos); \
+	ld r31,(top_pos)(%r1); \
+	ld r30,(top_pos - 8)(%r1); \
+	ld r29,(top_pos - 16)(%r1); \
+	ld r28,(top_pos - 24)(%r1); \
+	ld r27,(top_pos - 32)(%r1); \
+	ld r26,(top_pos - 40)(%r1); \
+	ld r25,(top_pos - 48)(%r1); \
+	ld r24,(top_pos - 56)(%r1); \
+	ld r23,(top_pos - 64)(%r1); \
+	ld r22,(top_pos - 72)(%r1); \
+	ld r21,(top_pos - 80)(%r1); \
+	ld r20,(top_pos - 88)(%r1); \
+	ld r19,(top_pos - 96)(%r1); \
+	ld r18,(top_pos - 104)(%r1); \
+	ld r17,(top_pos - 112)(%r1); \
+	ld r16,(top_pos - 120)(%r1); \
+	ld r15,(top_pos - 128)(%r1); \
+	ld r14,(top_pos - 136)(%r1)
+
+#define PUSH_NVREGS(stack_size) \
+	__PUSH_NVREGS(stack_size + STACK_FRAME_MIN_SIZE)
+
+/* 18 NV FPU REGS */
+#define PUSH_NVREGS_BELOW_FPU(stack_size) \
+	__PUSH_NVREGS(stack_size + STACK_FRAME_MIN_SIZE - (18 * 8))
+
+#define POP_NVREGS(stack_size) \
+	__POP_NVREGS(stack_size + STACK_FRAME_MIN_SIZE)
+
+/* 18 NV FPU REGS */
+#define POP_NVREGS_BELOW_FPU(stack_size) \
+	__POP_NVREGS(stack_size + STACK_FRAME_MIN_SIZE - (18 * 8))
+
+/*
+ * Careful calling this, it will 'clobber' NVGPRs (by design)
+ * Don't call this from C
+ */
+FUNC_START(load_gpr)
+	ld	r14,0(r3)
+	ld	r15,8(r3)
+	ld	r16,16(r3)
+	ld	r17,24(r3)
+	ld	r18,32(r3)
+	ld	r19,40(r3)
+	ld	r20,48(r3)
+	ld	r21,56(r3)
+	ld	r22,64(r3)
+	ld	r23,72(r3)
+	ld	r24,80(r3)
+	ld	r25,88(r3)
+	ld	r26,96(r3)
+	ld	r27,104(r3)
+	ld	r28,112(r3)
+	ld	r29,120(r3)
+	ld	r30,128(r3)
+	ld	r31,136(r3)
+	blr
+FUNC_END(load_gpr)
+
+
+#endif /* _SELFTESTS_POWERPC_GPR_ASM_H */
diff --git a/tools/testing/selftests/powerpc/include/instructions.h b/tools/testing/selftests/powerpc/include/instructions.h
new file mode 100644
index 000000000..4efa6314b
--- /dev/null
+++ b/tools/testing/selftests/powerpc/include/instructions.h
@@ -0,0 +1,146 @@
+/* SPDX-License-Identifier: GPL-2.0 */
+#ifndef _SELFTESTS_POWERPC_INSTRUCTIONS_H
+#define _SELFTESTS_POWERPC_INSTRUCTIONS_H
+
+#include <stdio.h>
+#include <stdlib.h>
+
+/* This defines the "copy" instruction from Power ISA 3.0 Book II, section 4.4. */
+#define __COPY(RA, RB, L) \
+	(0x7c00060c | (RA) << (31-15) | (RB) << (31-20) | (L) << (31-10))
+#define COPY(RA, RB, L) \
+	.long __COPY((RA), (RB), (L))
+
+static inline void copy(void *i)
+{
+	asm volatile(str(COPY(0, %0, 0))";"
+			:
+			: "b" (i)
+			: "memory"
+		    );
+}
+
+static inline void copy_first(void *i)
+{
+	asm volatile(str(COPY(0, %0, 1))";"
+			:
+			: "b" (i)
+			: "memory"
+		    );
+}
+
+/* This defines the "paste" instruction from Power ISA 3.0 Book II, section 4.4. */
+#define __PASTE(RA, RB, L, RC) \
+	(0x7c00070c | (RA) << (31-15) | (RB) << (31-20) | (L) << (31-10) | (RC) << (31-31))
+#define PASTE(RA, RB, L, RC) \
+	.long __PASTE((RA), (RB), (L), (RC))
+
+static inline int paste(void *i)
+{
+	int cr;
+
+	asm volatile(str(PASTE(0, %1, 0, 0))";"
+			"mfcr %0;"
+			: "=r" (cr)
+			: "b" (i)
+			: "memory"
+		    );
+	return cr;
+}
+
+static inline int paste_last(void *i)
+{
+	int cr;
+
+	asm volatile(str(PASTE(0, %1, 1, 1))";"
+			"mfcr %0;"
+			: "=r" (cr)
+			: "b" (i)
+			: "memory"
+		    );
+	return cr;
+}
+
+#define PPC_INST_COPY                  __COPY(0, 0, 0)
+#define PPC_INST_COPY_FIRST            __COPY(0, 0, 1)
+#define PPC_INST_PASTE                 __PASTE(0, 0, 0, 0)
+#define PPC_INST_PASTE_LAST            __PASTE(0, 0, 1, 1)
+
+/* This defines the prefixed load/store instructions */
+#ifdef __ASSEMBLY__
+#  define stringify_in_c(...)	__VA_ARGS__
+#else
+#  define __stringify_in_c(...)	#__VA_ARGS__
+#  define stringify_in_c(...)	__stringify_in_c(__VA_ARGS__) " "
+#endif
+
+#define __PPC_RA(a)	(((a) & 0x1f) << 16)
+#define __PPC_RS(s)	(((s) & 0x1f) << 21)
+#define __PPC_RT(t)	__PPC_RS(t)
+#define __PPC_PREFIX_R(r)	(((r) & 0x1) << 20)
+
+#define PPC_PREFIX_MLS			0x06000000
+#define PPC_PREFIX_8LS			0x04000000
+
+#define PPC_INST_LBZ			0x88000000
+#define PPC_INST_LHZ			0xa0000000
+#define PPC_INST_LHA			0xa8000000
+#define PPC_INST_LWZ			0x80000000
+#define PPC_INST_STB			0x98000000
+#define PPC_INST_STH			0xb0000000
+#define PPC_INST_STW			0x90000000
+#define PPC_INST_STD			0xf8000000
+#define PPC_INST_LFS			0xc0000000
+#define PPC_INST_LFD			0xc8000000
+#define PPC_INST_STFS			0xd0000000
+#define PPC_INST_STFD			0xd8000000
+
+#define PREFIX_MLS(instr, t, a, r, d)	stringify_in_c(.balign 64, , 4;)		\
+					stringify_in_c(.long PPC_PREFIX_MLS |		\
+						       __PPC_PREFIX_R(r) |		\
+						       (((d) >> 16) & 0x3ffff);)	\
+					stringify_in_c(.long (instr)  |			\
+						       __PPC_RT(t) |			\
+						       __PPC_RA(a) |			\
+						       ((d) & 0xffff);\n)
+
+#define PREFIX_8LS(instr, t, a, r, d)	stringify_in_c(.balign 64, , 4;)		\
+					stringify_in_c(.long PPC_PREFIX_8LS |		\
+						       __PPC_PREFIX_R(r) |		\
+						       (((d) >> 16) & 0x3ffff);)	\
+					stringify_in_c(.long (instr)  |			\
+						       __PPC_RT(t) |			\
+						       __PPC_RA(a) |			\
+						       ((d) & 0xffff);\n)
+
+/* Prefixed Integer Load/Store instructions */
+#define PLBZ(t, a, r, d)		PREFIX_MLS(PPC_INST_LBZ, t, a, r, d)
+#define PLHZ(t, a, r, d)		PREFIX_MLS(PPC_INST_LHZ, t, a, r, d)
+#define PLHA(t, a, r, d)		PREFIX_MLS(PPC_INST_LHA, t, a, r, d)
+#define PLWZ(t, a, r, d)		PREFIX_MLS(PPC_INST_LWZ, t, a, r, d)
+#define PLWA(t, a, r, d)		PREFIX_8LS(0xa4000000, t, a, r, d)
+#define PLD(t, a, r, d)			PREFIX_8LS(0xe4000000, t, a, r, d)
+#define PLQ(t, a, r, d)			PREFIX_8LS(0xe0000000, t, a, r, d)
+#define PSTB(s, a, r, d)		PREFIX_MLS(PPC_INST_STB, s, a, r, d)
+#define PSTH(s, a, r, d)		PREFIX_MLS(PPC_INST_STH, s, a, r, d)
+#define PSTW(s, a, r, d)		PREFIX_MLS(PPC_INST_STW, s, a, r, d)
+#define PSTD(s, a, r, d)		PREFIX_8LS(0xf4000000, s, a, r, d)
+#define PSTQ(s, a, r, d)		PREFIX_8LS(0xf0000000, s, a, r, d)
+
+/* Prefixed Floating-Point Load/Store Instructions */
+#define PLFS(frt, a, r, d)		PREFIX_MLS(PPC_INST_LFS, frt, a, r, d)
+#define PLFD(frt, a, r, d)		PREFIX_MLS(PPC_INST_LFD, frt, a, r, d)
+#define PSTFS(frs, a, r, d)		PREFIX_MLS(PPC_INST_STFS, frs, a, r, d)
+#define PSTFD(frs, a, r, d)		PREFIX_MLS(PPC_INST_STFD, frs, a, r, d)
+
+/* Prefixed VSX Load/Store Instructions */
+#define PLXSD(vrt, a, r, d)		PREFIX_8LS(0xa8000000, vrt, a, r, d)
+#define PLXSSP(vrt, a, r, d)		PREFIX_8LS(0xac000000, vrt, a, r, d)
+#define PLXV0(s, a, r, d)		PREFIX_8LS(0xc8000000, s, a, r, d)
+#define PLXV1(s, a, r, d)		PREFIX_8LS(0xcc000000, s, a, r, d)
+#define PSTXSD(vrs, a, r, d)		PREFIX_8LS(0xb8000000, vrs, a, r, d)
+#define PSTXSSP(vrs, a, r, d)		PREFIX_8LS(0xbc000000, vrs, a, r, d)
+#define PSTXV0(s, a, r, d)		PREFIX_8LS(0xd8000000, s, a, r, d)
+#define PSTXV1(s, a, r, d)		PREFIX_8LS(0xdc000000, s, a, r, d)
+
+#endif /* _SELFTESTS_POWERPC_INSTRUCTIONS_H */
diff --git a/tools/testing/selftests/powerpc/include/pkeys.h b/tools/testing/selftests/powerpc/include/pkeys.h
new file mode 100644
index 000000000..3312cb1b0
--- /dev/null
+++ b/tools/testing/selftests/powerpc/include/pkeys.h
@@ -0,0 +1,136 @@
+/* SPDX-License-Identifier: GPL-2.0 */
+/*
+ * Copyright 2020, Sandipan Das, IBM Corp.
+ */
+
+#ifndef _SELFTESTS_POWERPC_PKEYS_H
+#define _SELFTESTS_POWERPC_PKEYS_H
+
+#include <sys/mman.h>
+
+#include "reg.h"
+#include "utils.h"
+
+/*
+ * Older versions of libc use the Intel-specific access rights.
+ * Hence, override the definitions as they might be incorrect.
+ */
+#undef PKEY_DISABLE_ACCESS
+#define PKEY_DISABLE_ACCESS	0x3
+
+#undef PKEY_DISABLE_WRITE
+#define PKEY_DISABLE_WRITE	0x2
+
+#undef PKEY_DISABLE_EXECUTE
+#define PKEY_DISABLE_EXECUTE	0x4
+
+/* Older versions of libc do not not define this */
+#ifndef SEGV_PKUERR
+#define SEGV_PKUERR	4
+#endif
+
+#define SI_PKEY_OFFSET	0x20
+
+#define __NR_pkey_mprotect	386
+#define __NR_pkey_alloc		384
+#define __NR_pkey_free		385
+
+#define PKEY_BITS_PER_PKEY	2
+#define NR_PKEYS		32
+#define PKEY_BITS_MASK		((1UL << PKEY_BITS_PER_PKEY) - 1)
+
+inline unsigned long pkeyreg_get(void)
+{
+	return mfspr(SPRN_AMR);
+}
+
+inline void pkeyreg_set(unsigned long amr)
+{
+	set_amr(amr);
+}
+
+void pkey_set_rights(int pkey, unsigned long rights)
+{
+	unsigned long amr, shift;
+
+	shift = (NR_PKEYS - pkey - 1) * PKEY_BITS_PER_PKEY;
+	amr = pkeyreg_get();
+	amr &= ~(PKEY_BITS_MASK << shift);
+	amr |= (rights & PKEY_BITS_MASK) << shift;
+	pkeyreg_set(amr);
+}
+
+int sys_pkey_mprotect(void *addr, size_t len, int prot, int pkey)
+{
+	return syscall(__NR_pkey_mprotect, addr, len, prot, pkey);
+}
+
+int sys_pkey_alloc(unsigned long flags, unsigned long rights)
+{
+	return syscall(__NR_pkey_alloc, flags, rights);
+}
+
+int sys_pkey_free(int pkey)
+{
+	return syscall(__NR_pkey_free, pkey);
+}
+
+int pkeys_unsupported(void)
+{
+	bool hash_mmu = false;
+	int pkey;
+
+	/* Protection keys are currently supported on Hash MMU only */
+	FAIL_IF(using_hash_mmu(&hash_mmu));
+	SKIP_IF(!hash_mmu);
+
+	/* Check if the system call is supported */
+	pkey = sys_pkey_alloc(0, 0);
+	SKIP_IF(pkey < 0);
+	sys_pkey_free(pkey);
+
+	return 0;
+}
+
+int siginfo_pkey(siginfo_t *si)
+{
+	/*
+	 * In older versions of libc, siginfo_t does not have si_pkey as
+	 * a member.
+	 */
+#ifdef si_pkey
+	return si->si_pkey;
+#else
+	return *((int *)(((char *) si) + SI_PKEY_OFFSET));
+#endif
+}
+
+#define pkey_rights(r) ({						\
+	static char buf[4] = "rwx";					\
+	unsigned int amr_bits;						\
+	if ((r) & PKEY_DISABLE_EXECUTE)					\
+		buf[2] = '-';						\
+	amr_bits = (r) & PKEY_BITS_MASK;				\
+	if (amr_bits & PKEY_DISABLE_WRITE)				\
+		buf[1] = '-';						\
+	if (amr_bits & PKEY_DISABLE_ACCESS & ~PKEY_DISABLE_WRITE)	\
+		buf[0] = '-';						\
+	buf;								\
+})
+
+unsigned long next_pkey_rights(unsigned long rights)
+{
+	if (rights == PKEY_DISABLE_ACCESS)
+		return PKEY_DISABLE_EXECUTE;
+	else if (rights == (PKEY_DISABLE_ACCESS | PKEY_DISABLE_EXECUTE))
+		return 0;
+
+	if ((rights & PKEY_BITS_MASK) == 0)
+		rights |= PKEY_DISABLE_WRITE;
+	else if ((rights & PKEY_BITS_MASK) == PKEY_DISABLE_WRITE)
+		rights |= PKEY_DISABLE_ACCESS;
+
+	return rights;
+}
+
+#endif /* _SELFTESTS_POWERPC_PKEYS_H */
diff --git a/tools/testing/selftests/powerpc/include/reg.h b/tools/testing/selftests/powerpc/include/reg.h
new file mode 100644
index 000000000..c0f2742a3
--- /dev/null
+++ b/tools/testing/selftests/powerpc/include/reg.h
@@ -0,0 +1,163 @@
+/* SPDX-License-Identifier: GPL-2.0-only */
+/*
+ * Copyright 2014, Michael Ellerman, IBM Corp.
+ */
+
+#ifndef _SELFTESTS_POWERPC_REG_H
+#define _SELFTESTS_POWERPC_REG_H
+
+#define __stringify_1(x)        #x
+#define __stringify(x)          __stringify_1(x)
+
+#define mfspr(rn)	({unsigned long rval; \
+			 asm volatile("mfspr %0," _str(rn) \
+				    : "=r" (rval)); rval; })
+#define mtspr(rn, v)	asm volatile("mtspr " _str(rn) ",%0" : \
+				    : "r" ((unsigned long)(v)) \
+				    : "memory")
+
+#define mb()		asm volatile("sync" : : : "memory");
+#define barrier()	asm volatile("" : : : "memory");
+
+#define SPRN_MMCR2     769
+#define SPRN_MMCRA     770
+#define SPRN_MMCR0     779
+#define   MMCR0_PMAO   0x00000080
+#define   MMCR0_PMAE   0x04000000
+#define   MMCR0_FC     0x80000000
+#define SPRN_EBBHR     804
+#define SPRN_EBBRR     805
+#define SPRN_BESCR     806     /* Branch event status & control register */
+#define SPRN_BESCRS    800     /* Branch event status & control set (1 bits set to 1) */
+#define SPRN_BESCRSU   801     /* Branch event status & control set upper */
+#define SPRN_BESCRR    802     /* Branch event status & control REset (1 bits set to 0) */
+#define SPRN_BESCRRU   803     /* Branch event status & control REset upper */
+
+#define BESCR_PMEO     0x1     /* PMU Event-based exception Occurred */
+#define BESCR_PME      (0x1ul << 32) /* PMU Event-based exception Enable */
+
+#define SPRN_PMC1      771
+#define SPRN_PMC2      772
+#define SPRN_PMC3      773
+#define SPRN_PMC4      774
+#define SPRN_PMC5      775
+#define SPRN_PMC6      776
+
+#define SPRN_SIAR      780
+#define SPRN_SDAR      781
+#define SPRN_SIER      768
+
+#define SPRN_TEXASR     0x82    /* Transaction Exception and Status Register */
+#define SPRN_TFIAR      0x81    /* Transaction Failure Inst Addr    */
+#define SPRN_TFHAR      0x80    /* Transaction Failure Handler Addr */
+#define SPRN_TAR        0x32f	/* Target Address Register */
+
+#define SPRN_DSCR_PRIV 0x11	/* Privilege State DSCR */
+#define SPRN_DSCR      0x03	/* Data Stream Control Register */
+#define SPRN_PPR       896	/* Program Priority Register */
+#define SPRN_AMR       13	/* Authority Mask Register - problem state */
+
+#define set_amr(v)	asm volatile("isync;" \
+				     "mtspr " __stringify(SPRN_AMR) ",%0;" \
+				     "isync" : \
+				    : "r" ((unsigned long)(v)) \
+				    : "memory")
+
+/* TEXASR register bits */
+#define TEXASR_FC	0xFE00000000000000
+#define TEXASR_FP	0x0100000000000000
+#define TEXASR_DA	0x0080000000000000
+#define TEXASR_NO	0x0040000000000000
+#define TEXASR_FO	0x0020000000000000
+#define TEXASR_SIC	0x0010000000000000
+#define TEXASR_NTC	0x0008000000000000
+#define TEXASR_TC	0x0004000000000000
+#define TEXASR_TIC	0x0002000000000000
+#define TEXASR_IC	0x0001000000000000
+#define TEXASR_IFC	0x0000800000000000
+#define TEXASR_ABT	0x0000000100000000
+#define TEXASR_SPD	0x0000000080000000
+#define TEXASR_HV	0x0000000020000000
+#define TEXASR_PR	0x0000000010000000
+#define TEXASR_FS	0x0000000008000000
+#define TEXASR_TE	0x0000000004000000
+#define TEXASR_ROT	0x0000000002000000
+
+/* MSR register bits */
+#define MSR_TS_S_LG     33              /* Trans Mem state: Suspended */
+#define MSR_TS_T_LG	34              /* Trans Mem state: Active */
+
+#define __MASK(X)       (1UL<<(X))
+
+/* macro to check TM MSR bits */
+#define MSR_TS_S        __MASK(MSR_TS_S_LG)   /* Transaction Suspended */
+#define MSR_TS_T	__MASK(MSR_TS_T_LG)   /* Transaction Transactional */
+
+/* Vector Instructions */
+#define VSX_XX1(xs, ra, rb)	(((xs) & 0x1f) << 21 | ((ra) << 16) |  \
+				 ((rb) << 11) | (((xs) >> 5)))
+#define STXVD2X(xs, ra, rb)	.long (0x7c000798 | VSX_XX1((xs), (ra), (rb)))
+#define LXVD2X(xs, ra, rb)	.long (0x7c000698 | VSX_XX1((xs), (ra), (rb)))
+
+#define ASM_LOAD_GPR_IMMED(_asm_symbol_name_immed) \
+		"li 14, %[" #_asm_symbol_name_immed "];" \
+		"li 15, %[" #_asm_symbol_name_immed "];" \
+		"li 16, %[" #_asm_symbol_name_immed "];" \
+		"li 17, %[" #_asm_symbol_name_immed "];" \
+		"li 18, %[" #_asm_symbol_name_immed "];" \
+		"li 19, %[" #_asm_symbol_name_immed "];" \
+		"li 20, %[" #_asm_symbol_name_immed "];" \
+		"li 21, %[" #_asm_symbol_name_immed "];" \
+		"li 22, %[" #_asm_symbol_name_immed "];" \
+		"li 23, %[" #_asm_symbol_name_immed "];" \
+		"li 24, %[" #_asm_symbol_name_immed "];" \
+		"li 25, %[" #_asm_symbol_name_immed "];" \
+		"li 26, %[" #_asm_symbol_name_immed "];" \
+		"li 27, %[" #_asm_symbol_name_immed "];" \
+		"li 28, %[" #_asm_symbol_name_immed "];" \
+		"li 29, %[" #_asm_symbol_name_immed "];" \
+		"li 30, %[" #_asm_symbol_name_immed "];" \
+		"li 31, %[" #_asm_symbol_name_immed "];"
+
+#define ASM_LOAD_FPR_SINGLE_PRECISION(_asm_symbol_name_addr) \
+		"lfs 0, 0(%[" #_asm_symbol_name_addr "]);" \
+		"lfs 1, 0(%[" #_asm_symbol_name_addr "]);" \
+		"lfs 2, 0(%[" #_asm_symbol_name_addr "]);" \
+		"lfs 3, 0(%[" #_asm_symbol_name_addr "]);" \
+		"lfs 4, 0(%[" #_asm_symbol_name_addr "]);" \
+		"lfs 5, 0(%[" #_asm_symbol_name_addr "]);" \
+		"lfs 6, 0(%[" #_asm_symbol_name_addr "]);" \
+		"lfs 7, 0(%[" #_asm_symbol_name_addr "]);" \
+		"lfs 8, 0(%[" #_asm_symbol_name_addr "]);" \
+		"lfs 9, 0(%[" #_asm_symbol_name_addr "]);" \
+		"lfs 10, 0(%[" #_asm_symbol_name_addr "]);" \
+		"lfs 11, 0(%[" #_asm_symbol_name_addr "]);" \
+		"lfs 12, 0(%[" #_asm_symbol_name_addr "]);" \
+		"lfs 13, 0(%[" #_asm_symbol_name_addr "]);" \
+		"lfs 14, 0(%[" #_asm_symbol_name_addr "]);" \
+		"lfs 15, 0(%[" #_asm_symbol_name_addr "]);" \
+		"lfs 16, 0(%[" #_asm_symbol_name_addr "]);" \
+		"lfs 17, 0(%[" #_asm_symbol_name_addr "]);" \
+		"lfs 18, 0(%[" #_asm_symbol_name_addr "]);" \
+		"lfs 19, 0(%[" #_asm_symbol_name_addr "]);" \
+		"lfs 20, 0(%[" #_asm_symbol_name_addr "]);" \
+		"lfs 21, 0(%[" #_asm_symbol_name_addr "]);" \
+		"lfs 22, 0(%[" #_asm_symbol_name_addr "]);" \
+		"lfs 23, 0(%[" #_asm_symbol_name_addr "]);" \
+		"lfs 24, 0(%[" #_asm_symbol_name_addr "]);" \
+		"lfs 25, 0(%[" #_asm_symbol_name_addr "]);" \
+		"lfs 26, 0(%[" #_asm_symbol_name_addr "]);" \
+		"lfs 27, 0(%[" #_asm_symbol_name_addr "]);" \
+		"lfs 28, 0(%[" #_asm_symbol_name_addr "]);" \
+		"lfs 29, 0(%[" #_asm_symbol_name_addr "]);" \
+		"lfs 30, 0(%[" #_asm_symbol_name_addr "]);" \
+		"lfs 31, 0(%[" #_asm_symbol_name_addr "]);"
+
+#ifndef __ASSEMBLER__
+void store_gpr(unsigned long *addr);
+void load_gpr(unsigned long *addr);
+void load_fpr_single_precision(float *addr);
+void store_fpr_single_precision(float *addr);
+#endif /* end of __ASSEMBLER__ */
+
+#endif /* _SELFTESTS_POWERPC_REG_H */
diff --git a/tools/testing/selftests/powerpc/include/subunit.h b/tools/testing/selftests/powerpc/include/subunit.h
new file mode 100644
index 000000000..068d55fdf
--- /dev/null
+++ b/tools/testing/selftests/powerpc/include/subunit.h
@@ -0,0 +1,52 @@
+/* SPDX-License-Identifier: GPL-2.0-only */
+/*
+ * Copyright 2013, Michael Ellerman, IBM Corp.
+ */
+
+#ifndef _SELFTESTS_POWERPC_SUBUNIT_H
+#define _SELFTESTS_POWERPC_SUBUNIT_H
+
+static inline void test_start(char *name)
+{
+	printf("test: %s\n", name);
+}
+
+static inline void test_failure_detail(char *name, char *detail)
+{
+	printf("failure: %s [%s]\n", name, detail);
+}
+
+static inline void test_failure(char *name)
+{
+	printf("failure: %s\n", name);
+}
+
+static inline void test_error(char *name)
+{
+	printf("error: %s\n", name);
+}
+
+static inline void test_skip(char *name)
+{
+	printf("skip: %s\n", name);
+}
+
+static inline void test_success(char *name)
+{
+	printf("success: %s\n", name);
+}
+
+static inline void test_finish(char *name, int status)
+{
+	if (status)
+		test_failure(name);
+	else
+		test_success(name);
+}
+
+static inline void test_set_git_version(char *value)
+{
+	printf("tags: git_version:%s\n", value);
+}
+
+#endif /* _SELFTESTS_POWERPC_SUBUNIT_H */
diff --git a/tools/testing/selftests/powerpc/include/utils.h b/tools/testing/selftests/powerpc/include/utils.h
new file mode 100644
index 000000000..b7d188fc8
--- /dev/null
+++ b/tools/testing/selftests/powerpc/include/utils.h
@@ -0,0 +1,148 @@
+/* SPDX-License-Identifier: GPL-2.0-only */
+/*
+ * Copyright 2013, Michael Ellerman, IBM Corp.
+ */
+
+#ifndef _SELFTESTS_POWERPC_UTILS_H
+#define _SELFTESTS_POWERPC_UTILS_H
+
+#define __cacheline_aligned __attribute__((aligned(128)))
+
+#include <stdint.h>
+#include <stdbool.h>
+#include <linux/auxvec.h>
+#include <linux/perf_event.h>
+#include <asm/cputable.h>
+#include "reg.h"
+
+/* Avoid headaches with PRI?64 - just use %ll? always */
+typedef unsigned long long u64;
+typedef   signed long long s64;
+
+/* Just for familiarity */
+typedef uint32_t u32;
+typedef uint16_t u16;
+typedef uint8_t u8;
+
+void test_harness_set_timeout(uint64_t time);
+int test_harness(int (test_function)(void), char *name);
+
+int read_auxv(char *buf, ssize_t buf_size);
+void *find_auxv_entry(int type, char *auxv);
+void *get_auxv_entry(int type);
+
+int pick_online_cpu(void);
+
+int read_debugfs_file(char *debugfs_file, int *result);
+int write_debugfs_file(char *debugfs_file, int result);
+int read_sysfs_file(char *debugfs_file, char *result, size_t result_size);
+int perf_event_open_counter(unsigned int type,
+			    unsigned long config, int group_fd);
+int perf_event_enable(int fd);
+int perf_event_disable(int fd);
+int perf_event_reset(int fd);
+
+struct perf_event_read {
+	__u64 nr;
+	__u64 l1d_misses;
+};
+
+#if !defined(__GLIBC_PREREQ) || !__GLIBC_PREREQ(2, 30)
+#include <unistd.h>
+#include <sys/syscall.h>
+
+static inline pid_t gettid(void)
+{
+	return syscall(SYS_gettid);
+}
+#endif
+
+static inline bool have_hwcap(unsigned long ftr)
+{
+	return ((unsigned long)get_auxv_entry(AT_HWCAP) & ftr) == ftr;
+}
+
+#ifdef AT_HWCAP2
+static inline bool have_hwcap2(unsigned long ftr2)
+{
+	return ((unsigned long)get_auxv_entry(AT_HWCAP2) & ftr2) == ftr2;
+}
+#else
+static inline bool have_hwcap2(unsigned long ftr2)
+{
+	return false;
+}
+#endif
+
+bool is_ppc64le(void);
+int using_hash_mmu(bool *using_hash);
+
+/* Yes, this is evil */
+#define FAIL_IF(x)						\
+do {								\
+	if ((x)) {						\
+		fprintf(stderr,					\
+		"[FAIL] Test FAILED on line %d\n", __LINE__);	\
+		return 1;					\
+	}							\
+} while (0)
+
+#define FAIL_IF_EXIT(x)						\
+do {								\
+	if ((x)) {						\
+		fprintf(stderr,					\
+		"[FAIL] Test FAILED on line %d\n", __LINE__);	\
+		_exit(1);					\
+	}							\
+} while (0)
+
+/* The test harness uses this, yes it's gross */
+#define MAGIC_SKIP_RETURN_VALUE	99
+
+#define SKIP_IF(x)						\
+do {								\
+	if ((x)) {						\
+		fprintf(stderr,					\
+		"[SKIP] Test skipped on line %d\n", __LINE__);	\
+		return MAGIC_SKIP_RETURN_VALUE;			\
+	}							\
+} while (0)
+
+#define SKIP_IF_MSG(x, msg)					\
+do {								\
+	if ((x)) {						\
+		fprintf(stderr,					\
+		"[SKIP] Test skipped on line %d: %s\n",		\
+		 __LINE__, msg);				\
+		return MAGIC_SKIP_RETURN_VALUE;			\
+	}							\
+} while (0)
+
+#define _str(s) #s
+#define str(s) _str(s)
+
+#define sigsafe_err(msg)	({ \
+		ssize_t nbytes __attribute__((unused)); \
+		nbytes = write(STDERR_FILENO, msg, strlen(msg)); })
+
+/* POWER9 feature */
+#ifndef PPC_FEATURE2_ARCH_3_00
+#define PPC_FEATURE2_ARCH_3_00 0x00800000
+#endif
+
+/* POWER10 feature */
+#ifndef PPC_FEATURE2_ARCH_3_1
+#define PPC_FEATURE2_ARCH_3_1 0x00040000
+#endif
+
+#if defined(__powerpc64__)
+#define UCONTEXT_NIA(UC)	(UC)->uc_mcontext.gp_regs[PT_NIP]
+#define UCONTEXT_MSR(UC)	(UC)->uc_mcontext.gp_regs[PT_MSR]
+#elif defined(__powerpc__)
+#define UCONTEXT_NIA(UC)	(UC)->uc_mcontext.uc_regs->gregs[PT_NIP]
+#define UCONTEXT_MSR(UC)	(UC)->uc_mcontext.uc_regs->gregs[PT_MSR]
+#else
+#error implement UCONTEXT_NIA
+#endif
+
+#endif /* _SELFTESTS_POWERPC_UTILS_H */
diff --git a/tools/testing/selftests/powerpc/include/vmx_asm.h b/tools/testing/selftests/powerpc/include/vmx_asm.h
new file mode 100644
index 000000000..ad9fb1b40
--- /dev/null
+++ b/tools/testing/selftests/powerpc/include/vmx_asm.h
@@ -0,0 +1,92 @@
+/* SPDX-License-Identifier: GPL-2.0-or-later */
+/*
+ * Copyright 2015, Cyril Bur, IBM Corp.
+ */
+
+#include "basic_asm.h"
+
+/* POS MUST BE 16 ALIGNED! */
+#define PUSH_VMX(pos,reg) \
+	li	reg,pos; \
+	stvx	v20,reg,%r1; \
+	addi	reg,reg,16; \
+	stvx	v21,reg,%r1; \
+	addi	reg,reg,16; \
+	stvx	v22,reg,%r1; \
+	addi	reg,reg,16; \
+	stvx	v23,reg,%r1; \
+	addi	reg,reg,16; \
+	stvx	v24,reg,%r1; \
+	addi	reg,reg,16; \
+	stvx	v25,reg,%r1; \
+	addi	reg,reg,16; \
+	stvx	v26,reg,%r1; \
+	addi	reg,reg,16; \
+	stvx	v27,reg,%r1; \
+	addi	reg,reg,16; \
+	stvx	v28,reg,%r1; \
+	addi	reg,reg,16; \
+	stvx	v29,reg,%r1; \
+	addi	reg,reg,16; \
+	stvx	v30,reg,%r1; \
+	addi	reg,reg,16; \
+	stvx	v31,reg,%r1;
+
+/* POS MUST BE 16 ALIGNED! */
+#define POP_VMX(pos,reg) \
+	li	reg,pos; \
+	lvx	v20,reg,%r1; \
+	addi	reg,reg,16; \
+	lvx	v21,reg,%r1; \
+	addi	reg,reg,16; \
+	lvx	v22,reg,%r1; \
+	addi	reg,reg,16; \
+	lvx	v23,reg,%r1; \
+	addi	reg,reg,16; \
+	lvx	v24,reg,%r1; \
+	addi	reg,reg,16; \
+	lvx	v25,reg,%r1; \
+	addi	reg,reg,16; \
+	lvx	v26,reg,%r1; \
+	addi	reg,reg,16; \
+	lvx	v27,reg,%r1; \
+	addi	reg,reg,16; \
+	lvx	v28,reg,%r1; \
+	addi	reg,reg,16; \
+	lvx	v29,reg,%r1; \
+	addi	reg,reg,16; \
+	lvx	v30,reg,%r1; \
+	addi	reg,reg,16; \
+	lvx	v31,reg,%r1;
+
+/*
+ * Careful this will 'clobber' vmx (by design)
+ * Don't call this from C
+ */
+FUNC_START(load_vmx)
+	li	r5,0
+	lvx	v20,r5,r3
+	addi	r5,r5,16
+	lvx	v21,r5,r3
+	addi	r5,r5,16
+	lvx	v22,r5,r3
+	addi	r5,r5,16
+	lvx	v23,r5,r3
+	addi	r5,r5,16
+	lvx	v24,r5,r3
+	addi	r5,r5,16
+	lvx	v25,r5,r3
+	addi	r5,r5,16
+	lvx	v26,r5,r3
+	addi	r5,r5,16
+	lvx	v27,r5,r3
+	addi	r5,r5,16
+	lvx	v28,r5,r3
+	addi	r5,r5,16
+	lvx	v29,r5,r3
+	addi	r5,r5,16
+	lvx	v30,r5,r3
+	addi	r5,r5,16
+	lvx	v31,r5,r3
+	blr
+FUNC_END(load_vmx)
diff --git a/tools/testing/selftests/powerpc/include/vsx_asm.h b/tools/testing/selftests/powerpc/include/vsx_asm.h
new file mode 100644
index 000000000..434ca2f9b
--- /dev/null
+++ b/tools/testing/selftests/powerpc/include/vsx_asm.h
@@ -0,0 +1,67 @@
+/* SPDX-License-Identifier: GPL-2.0-or-later */
+/*
+ * Copyright 2015, Cyril Bur, IBM Corp.
+ */
+
+#include "basic_asm.h"
+
+/*
+ * Careful this will 'clobber' vsx (by design), VSX are always
+ * volatile though so unlike vmx this isn't so much of an issue
+ * Still should avoid calling from C
+ */
+FUNC_START(load_vsx)
+	li	r5,0
+	lxvd2x	vs20,r5,r3
+	addi	r5,r5,16
+	lxvd2x	vs21,r5,r3
+	addi	r5,r5,16
+	lxvd2x	vs22,r5,r3
+	addi	r5,r5,16
+	lxvd2x	vs23,r5,r3
+	addi	r5,r5,16
+	lxvd2x	vs24,r5,r3
+	addi	r5,r5,16
+	lxvd2x	vs25,r5,r3
+	addi	r5,r5,16
+	lxvd2x	vs26,r5,r3
+	addi	r5,r5,16
+	lxvd2x	vs27,r5,r3
+	addi	r5,r5,16
+	lxvd2x	vs28,r5,r3
+	addi	r5,r5,16
+	lxvd2x	vs29,r5,r3
+	addi	r5,r5,16
+	lxvd2x	vs30,r5,r3
+	addi	r5,r5,16
+	lxvd2x	vs31,r5,r3
+	blr
+FUNC_END(load_vsx)
+
+FUNC_START(store_vsx)
+	li	r5,0
+	stxvd2x	vs20,r5,r3
+	addi	r5,r5,16
+	stxvd2x	vs21,r5,r3
+	addi	r5,r5,16
+	stxvd2x	vs22,r5,r3
+	addi	r5,r5,16
+	stxvd2x	vs23,r5,r3
+	addi	r5,r5,16
+	stxvd2x	vs24,r5,r3
+	addi	r5,r5,16
+	stxvd2x	vs25,r5,r3
+	addi	r5,r5,16
+	stxvd2x	vs26,r5,r3
+	addi	r5,r5,16
+	stxvd2x	vs27,r5,r3
+	addi	r5,r5,16
+	stxvd2x	vs28,r5,r3
+	addi	r5,r5,16
+	stxvd2x	vs29,r5,r3
+	addi	r5,r5,16
+	stxvd2x	vs30,r5,r3
+	addi	r5,r5,16
+	stxvd2x	vs31,r5,r3
+	blr
+FUNC_END(store_vsx)
diff --git a/tools/testing/selftests/powerpc/lib/reg.S b/tools/testing/selftests/powerpc/lib/reg.S
new file mode 100644
index 000000000..9304ea7d5
--- /dev/null
+++ b/tools/testing/selftests/powerpc/lib/reg.S
@@ -0,0 +1,393 @@
+/* SPDX-License-Identifier: GPL-2.0-or-later */
+/*
+ * test helper assembly functions
+ *
+ * Copyright (C) 2016 Simon Guo, IBM Corporation.
+ */
+#include <ppc-asm.h>
+#include "reg.h"
+
+
+/* Non volatile GPR - unsigned long buf[18] */
+FUNC_START(load_gpr)
+	ld	14, 0*8(3)
+	ld	15, 1*8(3)
+	ld	16, 2*8(3)
+	ld	17, 3*8(3)
+	ld	18, 4*8(3)
+	ld	19, 5*8(3)
+	ld	20, 6*8(3)
+	ld	21, 7*8(3)
+	ld	22, 8*8(3)
+	ld	23, 9*8(3)
+	ld	24, 10*8(3)
+	ld	25, 11*8(3)
+	ld	26, 12*8(3)
+	ld	27, 13*8(3)
+	ld	28, 14*8(3)
+	ld	29, 15*8(3)
+	ld	30, 16*8(3)
+	ld	31, 17*8(3)
+	blr
+FUNC_END(load_gpr)
+
+FUNC_START(store_gpr)
+	std	14, 0*8(3)
+	std	15, 1*8(3)
+	std	16, 2*8(3)
+	std	17, 3*8(3)
+	std	18, 4*8(3)
+	std	19, 5*8(3)
+	std	20, 6*8(3)
+	std	21, 7*8(3)
+	std	22, 8*8(3)
+	std	23, 9*8(3)
+	std	24, 10*8(3)
+	std	25, 11*8(3)
+	std	26, 12*8(3)
+	std	27, 13*8(3)
+	std	28, 14*8(3)
+	std	29, 15*8(3)
+	std	30, 16*8(3)
+	std	31, 17*8(3)
+	blr
+FUNC_END(store_gpr)
+
+/* Single Precision Float - float buf[32] */
+FUNC_START(load_fpr_single_precision)
+	lfs 0, 0*4(3)
+	lfs 1, 1*4(3)
+	lfs 2, 2*4(3)
+	lfs 3, 3*4(3)
+	lfs 4, 4*4(3)
+	lfs 5, 5*4(3)
+	lfs 6, 6*4(3)
+	lfs 7, 7*4(3)
+	lfs 8, 8*4(3)
+	lfs 9, 9*4(3)
+	lfs 10, 10*4(3)
+	lfs 11, 11*4(3)
+	lfs 12, 12*4(3)
+	lfs 13, 13*4(3)
+	lfs 14, 14*4(3)
+	lfs 15, 15*4(3)
+	lfs 16, 16*4(3)
+	lfs 17, 17*4(3)
+	lfs 18, 18*4(3)
+	lfs 19, 19*4(3)
+	lfs 20, 20*4(3)
+	lfs 21, 21*4(3)
+	lfs 22, 22*4(3)
+	lfs 23, 23*4(3)
+	lfs 24, 24*4(3)
+	lfs 25, 25*4(3)
+	lfs 26, 26*4(3)
+	lfs 27, 27*4(3)
+	lfs 28, 28*4(3)
+	lfs 29, 29*4(3)
+	lfs 30, 30*4(3)
+	lfs 31, 31*4(3)
+	blr
+FUNC_END(load_fpr_single_precision)
+
+/* Single Precision Float - float buf[32] */
+FUNC_START(store_fpr_single_precision)
+	stfs 0, 0*4(3)
+	stfs 1, 1*4(3)
+	stfs 2, 2*4(3)
+	stfs 3, 3*4(3)
+	stfs 4, 4*4(3)
+	stfs 5, 5*4(3)
+	stfs 6, 6*4(3)
+	stfs 7, 7*4(3)
+	stfs 8, 8*4(3)
+	stfs 9, 9*4(3)
+	stfs 10, 10*4(3)
+	stfs 11, 11*4(3)
+	stfs 12, 12*4(3)
+	stfs 13, 13*4(3)
+	stfs 14, 14*4(3)
+	stfs 15, 15*4(3)
+	stfs 16, 16*4(3)
+	stfs 17, 17*4(3)
+	stfs 18, 18*4(3)
+	stfs 19, 19*4(3)
+	stfs 20, 20*4(3)
+	stfs 21, 21*4(3)
+	stfs 22, 22*4(3)
+	stfs 23, 23*4(3)
+	stfs 24, 24*4(3)
+	stfs 25, 25*4(3)
+	stfs 26, 26*4(3)
+	stfs 27, 27*4(3)
+	stfs 28, 28*4(3)
+	stfs 29, 29*4(3)
+	stfs 30, 30*4(3)
+	stfs 31, 31*4(3)
+	blr
+FUNC_END(store_fpr_single_precision)
+
+/* VMX/VSX registers - unsigned long buf[128] */
+FUNC_START(loadvsx)
+	lis	4, 0
+	LXVD2X	(0,(4),(3))
+	addi	4, 4, 16
+	LXVD2X	(1,(4),(3))
+	addi	4, 4, 16
+	LXVD2X	(2,(4),(3))
+	addi	4, 4, 16
+	LXVD2X	(3,(4),(3))
+	addi	4, 4, 16
+	LXVD2X	(4,(4),(3))
+	addi	4, 4, 16
+	LXVD2X	(5,(4),(3))
+	addi	4, 4, 16
+	LXVD2X	(6,(4),(3))
+	addi	4, 4, 16
+	LXVD2X	(7,(4),(3))
+	addi	4, 4, 16
+	LXVD2X	(8,(4),(3))
+	addi	4, 4, 16
+	LXVD2X	(9,(4),(3))
+	addi	4, 4, 16
+	LXVD2X	(10,(4),(3))
+	addi	4, 4, 16
+	LXVD2X	(11,(4),(3))
+	addi	4, 4, 16
+	LXVD2X	(12,(4),(3))
+	addi	4, 4, 16
+	LXVD2X	(13,(4),(3))
+	addi	4, 4, 16
+	LXVD2X	(14,(4),(3))
+	addi	4, 4, 16
+	LXVD2X	(15,(4),(3))
+	addi	4, 4, 16
+	LXVD2X	(16,(4),(3))
+	addi	4, 4, 16
+	LXVD2X	(17,(4),(3))
+	addi	4, 4, 16
+	LXVD2X	(18,(4),(3))
+	addi	4, 4, 16
+	LXVD2X	(19,(4),(3))
+	addi	4, 4, 16
+	LXVD2X	(20,(4),(3))
+	addi	4, 4, 16
+	LXVD2X	(21,(4),(3))
+	addi	4, 4, 16
+	LXVD2X	(22,(4),(3))
+	addi	4, 4, 16
+	LXVD2X	(23,(4),(3))
+	addi	4, 4, 16
+	LXVD2X	(24,(4),(3))
+	addi	4, 4, 16
+	LXVD2X	(25,(4),(3))
+	addi	4, 4, 16
+	LXVD2X	(26,(4),(3))
+	addi	4, 4, 16
+	LXVD2X	(27,(4),(3))
+	addi	4, 4, 16
+	LXVD2X	(28,(4),(3))
+	addi	4, 4, 16
+	LXVD2X	(29,(4),(3))
+	addi	4, 4, 16
+	LXVD2X	(30,(4),(3))
+	addi	4, 4, 16
+	LXVD2X	(31,(4),(3))
+	addi	4, 4, 16
+	LXVD2X	(32,(4),(3))
+	addi	4, 4, 16
+	LXVD2X	(33,(4),(3))
+	addi	4, 4, 16
+	LXVD2X	(34,(4),(3))
+	addi	4, 4, 16
+	LXVD2X	(35,(4),(3))
+	addi	4, 4, 16
+	LXVD2X	(36,(4),(3))
+	addi	4, 4, 16
+	LXVD2X	(37,(4),(3))
+	addi	4, 4, 16
+	LXVD2X	(38,(4),(3))
+	addi	4, 4, 16
+	LXVD2X	(39,(4),(3))
+	addi	4, 4, 16
+	LXVD2X	(40,(4),(3))
+	addi	4, 4, 16
+	LXVD2X	(41,(4),(3))
+	addi	4, 4, 16
+	LXVD2X	(42,(4),(3))
+	addi	4, 4, 16
+	LXVD2X	(43,(4),(3))
+	addi	4, 4, 16
+	LXVD2X	(44,(4),(3))
+	addi	4, 4, 16
+	LXVD2X	(45,(4),(3))
+	addi	4, 4, 16
+	LXVD2X	(46,(4),(3))
+	addi	4, 4, 16
+	LXVD2X	(47,(4),(3))
+	addi	4, 4, 16
+	LXVD2X	(48,(4),(3))
+	addi	4, 4, 16
+	LXVD2X	(49,(4),(3))
+	addi	4, 4, 16
+	LXVD2X	(50,(4),(3))
+	addi	4, 4, 16
+	LXVD2X	(51,(4),(3))
+	addi	4, 4, 16
+	LXVD2X	(52,(4),(3))
+	addi	4, 4, 16
+	LXVD2X	(53,(4),(3))
+	addi	4, 4, 16
+	LXVD2X	(54,(4),(3))
+	addi	4, 4, 16
+	LXVD2X	(55,(4),(3))
+	addi	4, 4, 16
+	LXVD2X	(56,(4),(3))
+	addi	4, 4, 16
+	LXVD2X	(57,(4),(3))
+	addi	4, 4, 16
+	LXVD2X	(58,(4),(3))
+	addi	4, 4, 16
+	LXVD2X	(59,(4),(3))
+	addi	4, 4, 16
+	LXVD2X	(60,(4),(3))
+	addi	4, 4, 16
+	LXVD2X	(61,(4),(3))
+	addi	4, 4, 16
+	LXVD2X	(62,(4),(3))
+	addi	4, 4, 16
+	LXVD2X	(63,(4),(3))
+	blr
+FUNC_END(loadvsx)
+
+FUNC_START(storevsx)
+	lis	4, 0
+	STXVD2X	(0,(4),(3))
+	addi	4, 4, 16
+	STXVD2X	(1,(4),(3))
+	addi	4, 4, 16
+	STXVD2X	(2,(4),(3))
+	addi	4, 4, 16
+	STXVD2X	(3,(4),(3))
+	addi	4, 4, 16
+	STXVD2X	(4,(4),(3))
+	addi	4, 4, 16
+	STXVD2X	(5,(4),(3))
+	addi	4, 4, 16
+	STXVD2X	(6,(4),(3))
+	addi	4, 4, 16
+	STXVD2X	(7,(4),(3))
+	addi	4, 4, 16
+	STXVD2X	(8,(4),(3))
+	addi	4, 4, 16
+	STXVD2X	(9,(4),(3))
+	addi	4, 4, 16
+	STXVD2X	(10,(4),(3))
+	addi	4, 4, 16
+	STXVD2X	(11,(4),(3))
+	addi	4, 4, 16
+	STXVD2X	(12,(4),(3))
+	addi	4, 4, 16
+	STXVD2X	(13,(4),(3))
+	addi	4, 4, 16
+	STXVD2X	(14,(4),(3))
+	addi	4, 4, 16
+	STXVD2X	(15,(4),(3))
+	addi	4, 4, 16
+	STXVD2X	(16,(4),(3))
+	addi	4, 4, 16
+	STXVD2X	(17,(4),(3))
+	addi	4, 4, 16
+	STXVD2X	(18,(4),(3))
+	addi	4, 4, 16
+	STXVD2X	(19,(4),(3))
+	addi	4, 4, 16
+	STXVD2X	(20,(4),(3))
+	addi	4, 4, 16
+	STXVD2X	(21,(4),(3))
+	addi	4, 4, 16
+	STXVD2X	(22,(4),(3))
+	addi	4, 4, 16
+	STXVD2X	(23,(4),(3))
+	addi	4, 4, 16
+	STXVD2X	(24,(4),(3))
+	addi	4, 4, 16
+	STXVD2X	(25,(4),(3))
+	addi	4, 4, 16
+	STXVD2X	(26,(4),(3))
+	addi	4, 4, 16
+	STXVD2X	(27,(4),(3))
+	addi	4, 4, 16
+	STXVD2X	(28,(4),(3))
+	addi	4, 4, 16
+	STXVD2X	(29,(4),(3))
+	addi	4, 4, 16
+	STXVD2X	(30,(4),(3))
+	addi	4, 4, 16
+	STXVD2X	(31,(4),(3))
+	addi	4, 4, 16
+	STXVD2X	(32,(4),(3))
+	addi	4, 4, 16
+	STXVD2X	(33,(4),(3))
+	addi	4, 4, 16
+	STXVD2X	(34,(4),(3))
+	addi	4, 4, 16
+	STXVD2X	(35,(4),(3))
+	addi	4, 4, 16
+	STXVD2X	(36,(4),(3))
+	addi	4, 4, 16
+	STXVD2X	(37,(4),(3))
+	addi	4, 4, 16
+	STXVD2X	(38,(4),(3))
+	addi	4, 4, 16
+	STXVD2X	(39,(4),(3))
+	addi	4, 4, 16
+	STXVD2X	(40,(4),(3))
+	addi	4, 4, 16
+	STXVD2X	(41,(4),(3))
+	addi	4, 4, 16
+	STXVD2X	(42,(4),(3))
+	addi	4, 4, 16
+	STXVD2X	(43,(4),(3))
+	addi	4, 4, 16
+	STXVD2X	(44,(4),(3))
+	addi	4, 4, 16
+	STXVD2X	(45,(4),(3))
+	addi	4, 4, 16
+	STXVD2X	(46,(4),(3))
+	addi	4, 4, 16
+	STXVD2X	(47,(4),(3))
+	addi	4, 4, 16
+	STXVD2X	(48,(4),(3))
+	addi	4, 4, 16
+	STXVD2X	(49,(4),(3))
+	addi	4, 4, 16
+	STXVD2X	(50,(4),(3))
+	addi	4, 4, 16
+	STXVD2X	(51,(4),(3))
+	addi	4, 4, 16
+	STXVD2X	(52,(4),(3))
+	addi	4, 4, 16
+	STXVD2X	(53,(4),(3))
+	addi	4, 4, 16
+	STXVD2X	(54,(4),(3))
+	addi	4, 4, 16
+	STXVD2X	(55,(4),(3))
+	addi	4, 4, 16
+	STXVD2X	(56,(4),(3))
+	addi	4, 4, 16
+	STXVD2X	(57,(4),(3))
+	addi	4, 4, 16
+	STXVD2X	(58,(4),(3))
+	addi	4, 4, 16
+	STXVD2X	(59,(4),(3))
+	addi	4, 4, 16
+	STXVD2X	(60,(4),(3))
+	addi	4, 4, 16
+	STXVD2X	(61,(4),(3))
+	addi	4, 4, 16
+	STXVD2X	(62,(4),(3))
+	addi	4, 4, 16
+	STXVD2X	(63,(4),(3))
+	blr
+FUNC_END(storevsx)
diff --git a/tools/testing/selftests/powerpc/math/.gitignore b/tools/testing/selftests/powerpc/math/.gitignore
new file mode 100644
index 000000000..d0c23b2e4
--- /dev/null
+++ b/tools/testing/selftests/powerpc/math/.gitignore
@@ -0,0 +1,9 @@
+# SPDX-License-Identifier: GPL-2.0-only
+fpu_syscall
+vmx_syscall
+fpu_preempt
+vmx_preempt
+fpu_signal
+vmx_signal
+vsx_preempt
+fpu_denormal
diff --git a/tools/testing/selftests/powerpc/math/Makefile b/tools/testing/selftests/powerpc/math/Makefile
new file mode 100644
index 000000000..fcc91c205
--- /dev/null
+++ b/tools/testing/selftests/powerpc/math/Makefile
@@ -0,0 +1,19 @@
+# SPDX-License-Identifier: GPL-2.0
+TEST_GEN_PROGS := fpu_syscall fpu_preempt fpu_signal fpu_denormal vmx_syscall vmx_preempt vmx_signal vsx_preempt
+
+top_srcdir = ../../../../..
+include ../../lib.mk
+
+$(TEST_GEN_PROGS): ../harness.c
+$(TEST_GEN_PROGS): CFLAGS += -O2 -g -pthread -m64 -maltivec
+
+$(OUTPUT)/fpu_syscall: fpu_asm.S
+$(OUTPUT)/fpu_preempt: fpu_asm.S
+$(OUTPUT)/fpu_signal:  fpu_asm.S
+
+$(OUTPUT)/vmx_syscall: vmx_asm.S ../utils.c
+$(OUTPUT)/vmx_preempt: vmx_asm.S ../utils.c
+$(OUTPUT)/vmx_signal: vmx_asm.S ../utils.c
+
+$(OUTPUT)/vsx_preempt: CFLAGS += -mvsx
+$(OUTPUT)/vsx_preempt: vsx_asm.S ../utils.c
diff --git a/tools/testing/selftests/powerpc/math/fpu_asm.S b/tools/testing/selftests/powerpc/math/fpu_asm.S
new file mode 100644
index 000000000..9dc0c158f
--- /dev/null
+++ b/tools/testing/selftests/powerpc/math/fpu_asm.S
@@ -0,0 +1,131 @@
+/* SPDX-License-Identifier: GPL-2.0-or-later */
+/*
+ * Copyright 2015, Cyril Bur, IBM Corp.
+ */
+
+#include "basic_asm.h"
+#include "fpu_asm.h"
+
+FUNC_START(check_fpu)
+	mr r4,r3
+	li	r3,1 # assume a bad result
+	lfd	f0,0(r4)
+	fcmpu	cr1,f0,f14
+	bne	cr1,1f
+	lfd	f0,8(r4)
+	fcmpu	cr1,f0,f15
+	bne	cr1,1f
+	lfd	f0,16(r4)
+	fcmpu	cr1,f0,f16
+	bne	cr1,1f
+	lfd	f0,24(r4)
+	fcmpu	cr1,f0,f17
+	bne	cr1,1f
+	lfd	f0,32(r4)
+	fcmpu	cr1,f0,f18
+	bne	cr1,1f
+	lfd	f0,40(r4)
+	fcmpu	cr1,f0,f19
+	bne	cr1,1f
+	lfd	f0,48(r4)
+	fcmpu	cr1,f0,f20
+	bne	cr1,1f
+	lfd	f0,56(r4)
+	fcmpu	cr1,f0,f21
+	bne	cr1,1f
+	lfd	f0,64(r4)
+	fcmpu	cr1,f0,f22
+	bne	cr1,1f
+	lfd	f0,72(r4)
+	fcmpu	cr1,f0,f23
+	bne	cr1,1f
+	lfd	f0,80(r4)
+	fcmpu	cr1,f0,f24
+	bne	cr1,1f
+	lfd	f0,88(r4)
+	fcmpu	cr1,f0,f25
+	bne	cr1,1f
+	lfd	f0,96(r4)
+	fcmpu	cr1,f0,f26
+	bne	cr1,1f
+	lfd	f0,104(r4)
+	fcmpu	cr1,f0,f27
+	bne	cr1,1f
+	lfd	f0,112(r4)
+	fcmpu	cr1,f0,f28
+	bne	cr1,1f
+	lfd	f0,120(r4)
+	fcmpu	cr1,f0,f29
+	bne	cr1,1f
+	lfd	f0,128(r4)
+	fcmpu	cr1,f0,f30
+	bne	cr1,1f
+	lfd	f0,136(r4)
+	fcmpu	cr1,f0,f31
+	bne	cr1,1f
+	li	r3,0 # Success!!!
+1:	blr
+
+FUNC_START(test_fpu)
+	# r3 holds pointer to where to put the result of fork
+	# r4 holds pointer to the pid
+	# f14-f31 are non volatiles
+	PUSH_BASIC_STACK(256)
+	PUSH_FPU(256)
+	std	r3,STACK_FRAME_PARAM(0)(sp) # Address of darray
+	std r4,STACK_FRAME_PARAM(1)(sp) # Address of pid
+
+	bl load_fpu
+	nop
+	li	r0,__NR_fork
+	sc
+
+	# pass the result of the fork to the caller
+	ld	r9,STACK_FRAME_PARAM(1)(sp)
+	std	r3,0(r9)
+
+	ld r3,STACK_FRAME_PARAM(0)(sp)
+	bl check_fpu
+	nop
+
+	POP_FPU(256)
+	POP_BASIC_STACK(256)
+	blr
+FUNC_END(test_fpu)
+
+# int preempt_fpu(double *darray, int *threads_running, int *running)
+# On starting will (atomically) decrement not_ready as a signal that the FPU
+# has been loaded with darray. Will proceed to check the validity of the FPU
+# registers while running is not zero.
+FUNC_START(preempt_fpu)
+	PUSH_BASIC_STACK(256)
+	PUSH_FPU(256)
+	std r3,STACK_FRAME_PARAM(0)(sp) # double *darray
+	std r4,STACK_FRAME_PARAM(1)(sp) # int *threads_starting
+	std r5,STACK_FRAME_PARAM(2)(sp) # int *running
+
+	bl load_fpu
+	nop
+
+	sync
+	# Atomic DEC
+	ld r3,STACK_FRAME_PARAM(1)(sp)
+1:	lwarx r4,0,r3
+	addi r4,r4,-1
+	stwcx. r4,0,r3
+	bne- 1b
+
+2:	ld r3,STACK_FRAME_PARAM(0)(sp)
+	bl check_fpu
+	nop
+	cmpdi r3,0
+	bne 3f
+	ld r4,STACK_FRAME_PARAM(2)(sp)
+	ld r5,0(r4)
+	cmpwi r5,0
+	bne 2b
+
+3:	POP_FPU(256)
+	POP_BASIC_STACK(256)
+	blr
+FUNC_END(preempt_fpu)
diff --git a/tools/testing/selftests/powerpc/math/fpu_denormal.c b/tools/testing/selftests/powerpc/math/fpu_denormal.c
new file mode 100644
index 000000000..5f96682ab
--- /dev/null
+++ b/tools/testing/selftests/powerpc/math/fpu_denormal.c
@@ -0,0 +1,38 @@
+// SPDX-License-Identifier: GPL-2.0-or-later
+/*
+ * Copyright IBM Corp. 2020
+ *
+ * This test attempts to cause a FP denormal exception on POWER8 CPUs. Unfortunately
+ * if the denormal handler is not configured or working properly, this can cause a bad
+ * crash in kernel mode when the kernel tries to save FP registers when the process
+ * exits.
+ */
+
+#include <stdio.h>
+#include <string.h>
+
+#include "utils.h"
+
+static int test_denormal_fpu(void)
+{
+	unsigned int m32;
+	unsigned long m64;
+	volatile float f;
+	volatile double d;
+
+	/* try to induce lfs <denormal> ; stfd */
+
+	m32 = 0x00715fcf; /* random denormal */
+	memcpy((float *)&f, &m32, sizeof(f));
+	d = f;
+	memcpy(&m64, (double *)&d, sizeof(d));
+
+	FAIL_IF((long)(m64 != 0x380c57f3c0000000)); /* renormalised value */
+
+	return 0;
+}
+
+int main(int argc, char *argv[])
+{
+	return test_harness(test_denormal_fpu, "fpu_denormal");
+}
diff --git a/tools/testing/selftests/powerpc/math/fpu_preempt.c b/tools/testing/selftests/powerpc/math/fpu_preempt.c
new file mode 100644
index 000000000..3e5b5663d
--- /dev/null
+++ b/tools/testing/selftests/powerpc/math/fpu_preempt.c
@@ -0,0 +1,110 @@
+// SPDX-License-Identifier: GPL-2.0-or-later
+/*
+ * Copyright 2015, Cyril Bur, IBM Corp.
+ *
+ * This test attempts to see if the FPU registers change across preemption.
+ * Two things should be noted here a) The check_fpu function in asm only checks
+ * the non volatile registers as it is reused from the syscall test b) There is
+ * no way to be sure preemption happened so this test just uses many threads
+ * and a long wait. As such, a successful test doesn't mean much but a failure
+ * is bad.
+ */
+
+#include <stdio.h>
+#include <unistd.h>
+#include <sys/syscall.h>
+#include <sys/time.h>
+#include <sys/types.h>
+#include <sys/wait.h>
+#include <stdlib.h>
+#include <pthread.h>
+
+#include "utils.h"
+
+/* Time to wait for workers to get preempted (seconds) */
+#define PREEMPT_TIME 20
+/*
+ * Factor by which to multiply number of online CPUs for total number of
+ * worker threads
+ */
+#define THREAD_FACTOR 8
+
+
+__thread double darray[] = {0.1, 0.2, 0.3, 0.4, 0.5, 0.6, 0.7, 0.8, 0.9, 1.0,
+		     1.1, 1.2, 1.3, 1.4, 1.5, 1.6, 1.7, 1.8, 1.9, 2.0,
+		     2.1};
+
+int threads_starting;
+int running;
+
+extern int preempt_fpu(double *darray, int *threads_starting, int *running);
+
+void *preempt_fpu_c(void *p)
+{
+	long rc;
+	int i;
+
+	srand(pthread_self());
+	for (i = 0; i < 21; i++)
+		darray[i] = rand();
+
+	rc = preempt_fpu(darray, &threads_starting, &running);
+
+	return (void *)rc;
+}
+
+int test_preempt_fpu(void)
+{
+	int i, rc, threads;
+	pthread_t *tids;
+
+	threads = sysconf(_SC_NPROCESSORS_ONLN) * THREAD_FACTOR;
+	tids = malloc((threads) * sizeof(pthread_t));
+	FAIL_IF(!tids);
+
+	running = true;
+	threads_starting = threads;
+	for (i = 0; i < threads; i++) {
+		rc = pthread_create(&tids[i], NULL, preempt_fpu_c, NULL);
+		FAIL_IF(rc);
+	}
+
+	setbuf(stdout, NULL);
+	/* Not really necessary but nice to wait for every thread to start */
+	printf("\tWaiting for all workers to start...");
+	while(threads_starting)
+		asm volatile("": : :"memory");
+	printf("done\n");
+
+	printf("\tWaiting for %d seconds to let some workers get preempted...", PREEMPT_TIME);
+	sleep(PREEMPT_TIME);
+	printf("done\n");
+
+	printf("\tStopping workers...");
+	/*
+	 * Working are checking this value every loop. In preempt_fpu 'cmpwi r5,0; bne 2b'.
+	 * r5 will have loaded the value of running.
+	 */
+	running = 0;
+	for (i = 0; i < threads; i++) {
+		void *rc_p;
+		pthread_join(tids[i], &rc_p);
+
+		/*
+		 * Harness will say the fail was here, look at why preempt_fpu
+		 * returned
+		 */
+		if ((long) rc_p)
+			printf("oops\n");
+		FAIL_IF((long) rc_p);
+	}
+	printf("done\n");
+
+	free(tids);
+	return 0;
+}
+
+int main(int argc, char *argv[])
+{
+	return test_harness(test_preempt_fpu, "fpu_preempt");
+}
diff --git a/tools/testing/selftests/powerpc/math/fpu_signal.c b/tools/testing/selftests/powerpc/math/fpu_signal.c
new file mode 100644
index 000000000..7b1addd50
--- /dev/null
+++ b/tools/testing/selftests/powerpc/math/fpu_signal.c
@@ -0,0 +1,131 @@
+// SPDX-License-Identifier: GPL-2.0-or-later
+/*
+ * Copyright 2015, Cyril Bur, IBM Corp.
+ *
+ * This test attempts to see if the FPU registers are correctly reported in a
+ * signal context. Each worker just spins checking its FPU registers, at some
+ * point a signal will interrupt it and C code will check the signal context
+ * ensuring it is also the same.
+ */
+
+#include <stdio.h>
+#include <unistd.h>
+#include <sys/syscall.h>
+#include <sys/time.h>
+#include <sys/types.h>
+#include <sys/wait.h>
+#include <stdlib.h>
+#include <pthread.h>
+
+#include "utils.h"
+
+/* Number of times each thread should receive the signal */
+#define ITERATIONS 10
+/*
+ * Factor by which to multiply number of online CPUs for total number of
+ * worker threads
+ */
+#define THREAD_FACTOR 8
+
+__thread double darray[] = {0.1, 0.2, 0.3, 0.4, 0.5, 0.6, 0.7, 0.8, 0.9, 1.0,
+		     1.1, 1.2, 1.3, 1.4, 1.5, 1.6, 1.7, 1.8, 1.9, 2.0,
+		     2.1};
+
+bool bad_context;
+int threads_starting;
+int running;
+
+extern long preempt_fpu(double *darray, int *threads_starting, int *running);
+
+void signal_fpu_sig(int sig, siginfo_t *info, void *context)
+{
+	int i;
+	ucontext_t *uc = context;
+	mcontext_t *mc = &uc->uc_mcontext;
+
+	/* Only the non volatiles were loaded up */
+	for (i = 14; i < 32; i++) {
+		if (mc->fp_regs[i] != darray[i - 14]) {
+			bad_context = true;
+			break;
+		}
+	}
+}
+
+void *signal_fpu_c(void *p)
+{
+	int i;
+	long rc;
+	struct sigaction act;
+	act.sa_sigaction = signal_fpu_sig;
+	act.sa_flags = SA_SIGINFO;
+	rc = sigaction(SIGUSR1, &act, NULL);
+	if (rc)
+		return p;
+
+	srand(pthread_self());
+	for (i = 0; i < 21; i++)
+		darray[i] = rand();
+
+	rc = preempt_fpu(darray, &threads_starting, &running);
+
+	return (void *) rc;
+}
+
+int test_signal_fpu(void)
+{
+	int i, j, rc, threads;
+	void *rc_p;
+	pthread_t *tids;
+
+	threads = sysconf(_SC_NPROCESSORS_ONLN) * THREAD_FACTOR;
+	tids = malloc(threads * sizeof(pthread_t));
+	FAIL_IF(!tids);
+
+	running = true;
+	threads_starting = threads;
+	for (i = 0; i < threads; i++) {
+		rc = pthread_create(&tids[i], NULL, signal_fpu_c, NULL);
+		FAIL_IF(rc);
+	}
+
+	setbuf(stdout, NULL);
+	printf("\tWaiting for all workers to start...");
+	while (threads_starting)
+		asm volatile("": : :"memory");
+	printf("done\n");
+
+	printf("\tSending signals to all threads %d times...", ITERATIONS);
+	for (i = 0; i < ITERATIONS; i++) {
+		for (j = 0; j < threads; j++) {
+			pthread_kill(tids[j], SIGUSR1);
+		}
+		sleep(1);
+	}
+	printf("done\n");
+
+	printf("\tStopping workers...");
+	running = 0;
+	for (i = 0; i < threads; i++) {
+		pthread_join(tids[i], &rc_p);
+
+		/*
+		 * Harness will say the fail was here, look at why signal_fpu
+		 * returned
+		 */
+		if ((long) rc_p || bad_context)
+			printf("oops\n");
+		if (bad_context)
+			fprintf(stderr, "\t!! bad_context is true\n");
+		FAIL_IF((long) rc_p || bad_context);
+	}
+	printf("done\n");
+
+	free(tids);
+	return 0;
+}
+
+int main(int argc, char *argv[])
+{
+	return test_harness(test_signal_fpu, "fpu_signal");
+}
diff --git a/tools/testing/selftests/powerpc/math/fpu_syscall.c b/tools/testing/selftests/powerpc/math/fpu_syscall.c
new file mode 100644
index 000000000..694f225c7
--- /dev/null
+++ b/tools/testing/selftests/powerpc/math/fpu_syscall.c
@@ -0,0 +1,86 @@
+// SPDX-License-Identifier: GPL-2.0-or-later
+/*
+ * Copyright 2015, Cyril Bur, IBM Corp.
+ *
+ * This test attempts to see if the FPU registers change across a syscall (fork).
+ */
+
+#include <stdio.h>
+#include <unistd.h>
+#include <sys/syscall.h>
+#include <sys/time.h>
+#include <sys/types.h>
+#include <sys/wait.h>
+#include <stdlib.h>
+
+#include "utils.h"
+
+extern int test_fpu(double *darray, pid_t *pid);
+
+double darray[] = {0.1, 0.2, 0.3, 0.4, 0.5, 0.6, 0.7, 0.8, 0.9, 1.0,
+		     1.1, 1.2, 1.3, 1.4, 1.5, 1.6, 1.7, 1.8, 1.9, 2.0,
+		     2.1};
+
+int syscall_fpu(void)
+{
+	pid_t fork_pid;
+	int i;
+	int ret;
+	int child_ret;
+	for (i = 0; i < 1000; i++) {
+		/* test_fpu will fork() */
+		ret = test_fpu(darray, &fork_pid);
+		if (fork_pid == -1)
+			return -1;
+		if (fork_pid == 0)
+			exit(ret);
+		waitpid(fork_pid, &child_ret, 0);
+		if (ret || child_ret)
+			return 1;
+	}
+
+	return 0;
+}
+
+int test_syscall_fpu(void)
+{
+	/*
+	 * Setup an environment with much context switching
+	 */
+	pid_t pid2;
+	pid_t pid = fork();
+	int ret;
+	int child_ret;
+	FAIL_IF(pid == -1);
+
+	pid2 = fork();
+	/* Can't FAIL_IF(pid2 == -1); because already forked once */
+	if (pid2 == -1) {
+		/*
+		 * Couldn't fork, ensure test is a fail
+		 */
+		child_ret = ret = 1;
+	} else {
+		ret = syscall_fpu();
+		if (pid2)
+			waitpid(pid2, &child_ret, 0);
+		else
+			exit(ret);
+	}
+
+	ret |= child_ret;
+
+	if (pid)
+		waitpid(pid, &child_ret, 0);
+	else
+		exit(ret);
+
+	FAIL_IF(ret || child_ret);
+	return 0;
+}
+
+int main(int argc, char *argv[])
+{
+	return test_harness(test_syscall_fpu, "syscall_fpu");
+
+}
diff --git a/tools/testing/selftests/powerpc/math/vmx_asm.S b/tools/testing/selftests/powerpc/math/vmx_asm.S
new file mode 100644
index 000000000..11b0704c5
--- /dev/null
+++ b/tools/testing/selftests/powerpc/math/vmx_asm.S
@@ -0,0 +1,148 @@
+/* SPDX-License-Identifier: GPL-2.0-or-later */
+/*
+ * Copyright 2015, Cyril Bur, IBM Corp.
+ */
+
+#include "basic_asm.h"
+#include "vmx_asm.h"
+
+# Should be safe from C, only touches r4, r5 and v0,v1,v2
+FUNC_START(check_vmx)
+	PUSH_BASIC_STACK(32)
+	mr r4,r3
+	li	r3,1 # assume a bad result
+	li	r5,0
+	lvx	v0,r5,r4
+	vcmpequd.	v1,v0,v20
+	vmr	v2,v1
+
+	addi	r5,r5,16
+	lvx	v0,r5,r4
+	vcmpequd.	v1,v0,v21
+	vand	v2,v2,v1
+
+	addi	r5,r5,16
+	lvx	v0,r5,r4
+	vcmpequd.	v1,v0,v22
+	vand	v2,v2,v1
+
+	addi	r5,r5,16
+	lvx	v0,r5,r4
+	vcmpequd.	v1,v0,v23
+	vand	v2,v2,v1
+
+	addi	r5,r5,16
+	lvx	v0,r5,r4
+	vcmpequd.	v1,v0,v24
+	vand	v2,v2,v1
+
+	addi	r5,r5,16
+	lvx	v0,r5,r4
+	vcmpequd.	v1,v0,v25
+	vand	v2,v2,v1
+
+	addi	r5,r5,16
+	lvx	v0,r5,r4
+	vcmpequd.	v1,v0,v26
+	vand	v2,v2,v1
+
+	addi	r5,r5,16
+	lvx	v0,r5,r4
+	vcmpequd.	v1,v0,v27
+	vand	v2,v2,v1
+
+	addi	r5,r5,16
+	lvx	v0,r5,r4
+	vcmpequd.	v1,v0,v28
+	vand	v2,v2,v1
+
+	addi	r5,r5,16
+	lvx	v0,r5,r4
+	vcmpequd.	v1,v0,v29
+	vand	v2,v2,v1
+
+	addi	r5,r5,16
+	lvx	v0,r5,r4
+	vcmpequd.	v1,v0,v30
+	vand	v2,v2,v1
+
+	addi	r5,r5,16
+	lvx	v0,r5,r4
+	vcmpequd.	v1,v0,v31
+	vand	v2,v2,v1
+
+	li	r5,STACK_FRAME_LOCAL(0,0)
+	stvx	v2,r5,sp
+	ldx	r0,r5,sp
+	cmpdi	r0,0xffffffffffffffff
+	bne	1f
+	li	r3,0
+1:	POP_BASIC_STACK(32)
+	blr
+FUNC_END(check_vmx)
+
+# Safe from C
+FUNC_START(test_vmx)
+	# r3 holds pointer to where to put the result of fork
+	# r4 holds pointer to the pid
+	# v20-v31 are non-volatile
+	PUSH_BASIC_STACK(512)
+	std	r3,STACK_FRAME_PARAM(0)(sp) # Address of varray
+	std r4,STACK_FRAME_PARAM(1)(sp) # address of pid
+	PUSH_VMX(STACK_FRAME_LOCAL(2,0),r4)
+
+	bl load_vmx
+	nop
+
+	li	r0,__NR_fork
+	sc
+	# Pass the result of fork back to the caller
+	ld	r9,STACK_FRAME_PARAM(1)(sp)
+	std	r3,0(r9)
+
+	ld r3,STACK_FRAME_PARAM(0)(sp)
+	bl check_vmx
+	nop
+
+	POP_VMX(STACK_FRAME_LOCAL(2,0),r4)
+	POP_BASIC_STACK(512)
+	blr
+FUNC_END(test_vmx)
+
+# int preempt_vmx(vector int *varray, int *threads_starting, int *running)
+# On starting will (atomically) decrement threads_starting as a signal that
+# the VMX have been loaded with varray. Will proceed to check the validity of
+# the VMX registers while running is not zero.
+FUNC_START(preempt_vmx)
+	PUSH_BASIC_STACK(512)
+	std r3,STACK_FRAME_PARAM(0)(sp) # vector int *varray
+	std r4,STACK_FRAME_PARAM(1)(sp) # int *threads_starting
+	std r5,STACK_FRAME_PARAM(2)(sp) # int *running
+	# VMX need to write to 16 byte aligned addresses, skip STACK_FRAME_LOCAL(3,0)
+	PUSH_VMX(STACK_FRAME_LOCAL(4,0),r4)
+
+	bl load_vmx
+	nop
+
+	sync
+	# Atomic DEC
+	ld r3,STACK_FRAME_PARAM(1)(sp)
+1:	lwarx r4,0,r3
+	addi r4,r4,-1
+	stwcx. r4,0,r3
+	bne- 1b
+
+2:	ld r3,STACK_FRAME_PARAM(0)(sp)
+	bl check_vmx
+	nop
+	cmpdi r3,0
+	bne 3f
+	ld r4,STACK_FRAME_PARAM(2)(sp)
+	ld r5,0(r4)
+	cmpwi r5,0
+	bne 2b
+
+3:	POP_VMX(STACK_FRAME_LOCAL(4,0),r4)
+	POP_BASIC_STACK(512)
+	blr
+FUNC_END(preempt_vmx)
diff --git a/tools/testing/selftests/powerpc/math/vmx_preempt.c b/tools/testing/selftests/powerpc/math/vmx_preempt.c
new file mode 100644
index 000000000..6f7cf400c
--- /dev/null
+++ b/tools/testing/selftests/powerpc/math/vmx_preempt.c
@@ -0,0 +1,113 @@
+// SPDX-License-Identifier: GPL-2.0-or-later
+/*
+ * Copyright 2015, Cyril Bur, IBM Corp.
+ *
+ * This test attempts to see if the VMX registers change across preemption.
+ * Two things should be noted here a) The check_vmx function in asm only checks
+ * the non volatile registers as it is reused from the syscall test b) There is
+ * no way to be sure preemption happened so this test just uses many threads
+ * and a long wait. As such, a successful test doesn't mean much but a failure
+ * is bad.
+ */
+
+#include <stdio.h>
+#include <unistd.h>
+#include <sys/syscall.h>
+#include <sys/time.h>
+#include <sys/types.h>
+#include <sys/wait.h>
+#include <stdlib.h>
+#include <pthread.h>
+
+#include "utils.h"
+
+/* Time to wait for workers to get preempted (seconds) */
+#define PREEMPT_TIME 20
+/*
+ * Factor by which to multiply number of online CPUs for total number of
+ * worker threads
+ */
+#define THREAD_FACTOR 8
+
+__thread vector int varray[] = {{1, 2, 3, 4}, {5, 6, 7, 8}, {9, 10,11,12},
+	{13,14,15,16},{17,18,19,20},{21,22,23,24},
+	{25,26,27,28},{29,30,31,32},{33,34,35,36},
+	{37,38,39,40},{41,42,43,44},{45,46,47,48}};
+
+int threads_starting;
+int running;
+
+extern int preempt_vmx(vector int *varray, int *threads_starting, int *running);
+
+void *preempt_vmx_c(void *p)
+{
+	int i, j;
+	long rc;
+
+	srand(pthread_self());
+	for (i = 0; i < 12; i++)
+		for (j = 0; j < 4; j++)
+			varray[i][j] = rand();
+
+	rc = preempt_vmx(varray, &threads_starting, &running);
+
+	return (void *)rc;
+}
+
+int test_preempt_vmx(void)
+{
+	int i, rc, threads;
+	pthread_t *tids;
+
+	// vcmpequd used in vmx_asm.S is v2.07
+	SKIP_IF(!have_hwcap2(PPC_FEATURE2_ARCH_2_07));
+
+	threads = sysconf(_SC_NPROCESSORS_ONLN) * THREAD_FACTOR;
+	tids = malloc(threads * sizeof(pthread_t));
+	FAIL_IF(!tids);
+
+	running = true;
+	threads_starting = threads;
+	for (i = 0; i < threads; i++) {
+		rc = pthread_create(&tids[i], NULL, preempt_vmx_c, NULL);
+		FAIL_IF(rc);
+	}
+
+	setbuf(stdout, NULL);
+	/* Not really nessesary but nice to wait for every thread to start */
+	printf("\tWaiting for all workers to start...");
+	while(threads_starting)
+		asm volatile("": : :"memory");
+	printf("done\n");
+
+	printf("\tWaiting for %d seconds to let some workers get preempted...", PREEMPT_TIME);
+	sleep(PREEMPT_TIME);
+	printf("done\n");
+
+	printf("\tStopping workers...");
+	/*
+	 * Working are checking this value every loop. In preempt_vmx 'cmpwi r5,0; bne 2b'.
+	 * r5 will have loaded the value of running.
+	 */
+	running = 0;
+	for (i = 0; i < threads; i++) {
+		void *rc_p;
+		pthread_join(tids[i], &rc_p);
+
+		/*
+		 * Harness will say the fail was here, look at why preempt_vmx
+		 * returned
+		 */
+		if ((long) rc_p)
+			printf("oops\n");
+		FAIL_IF((long) rc_p);
+	}
+	printf("done\n");
+
+	return 0;
+}
+
+int main(int argc, char *argv[])
+{
+	return test_harness(test_preempt_vmx, "vmx_preempt");
+}
diff --git a/tools/testing/selftests/powerpc/math/vmx_signal.c b/tools/testing/selftests/powerpc/math/vmx_signal.c
new file mode 100644
index 000000000..b340a5c4e
--- /dev/null
+++ b/tools/testing/selftests/powerpc/math/vmx_signal.c
@@ -0,0 +1,155 @@
+// SPDX-License-Identifier: GPL-2.0-or-later
+/*
+ * Copyright 2015, Cyril Bur, IBM Corp.
+ *
+ * This test attempts to see if the VMX registers are correctly reported in a
+ * signal context. Each worker just spins checking its VMX registers, at some
+ * point a signal will interrupt it and C code will check the signal context
+ * ensuring it is also the same.
+ */
+
+#include <stdio.h>
+#include <unistd.h>
+#include <sys/syscall.h>
+#include <sys/time.h>
+#include <sys/types.h>
+#include <sys/wait.h>
+#include <stdlib.h>
+#include <string.h>
+#include <pthread.h>
+#include <altivec.h>
+
+#include "utils.h"
+
+/* Number of times each thread should receive the signal */
+#define ITERATIONS 10
+/*
+ * Factor by which to multiply number of online CPUs for total number of
+ * worker threads
+ */
+#define THREAD_FACTOR 8
+
+__thread vector int varray[] = {{1, 2, 3, 4}, {5, 6, 7, 8}, {9, 10,11,12},
+	{13,14,15,16},{17,18,19,20},{21,22,23,24},
+	{25,26,27,28},{29,30,31,32},{33,34,35,36},
+	{37,38,39,40},{41,42,43,44},{45,46,47,48}};
+
+bool bad_context;
+int running;
+int threads_starting;
+
+extern int preempt_vmx(vector int *varray, int *threads_starting, int *sentinal);
+
+void signal_vmx_sig(int sig, siginfo_t *info, void *context)
+{
+	int i;
+	ucontext_t *uc = context;
+	mcontext_t *mc = &uc->uc_mcontext;
+
+	/* Only the non volatiles were loaded up */
+	for (i = 20; i < 32; i++) {
+		if (memcmp(mc->v_regs->vrregs[i], &varray[i - 20], 16)) {
+			int j;
+			/*
+			 * Shouldn't printf() in a signal handler, however, this is a
+			 * test and we've detected failure. Understanding what failed
+			 * is paramount. All that happens after this is tests exit with
+			 * failure.
+			 */
+			printf("VMX mismatch at reg %d!\n", i);
+			printf("Reg | Actual                  | Expected\n");
+			for (j = 20; j < 32; j++) {
+				printf("%d  | 0x%04x%04x%04x%04x      | 0x%04x%04x%04x%04x\n", j, mc->v_regs->vrregs[j][0],
+					   mc->v_regs->vrregs[j][1], mc->v_regs->vrregs[j][2], mc->v_regs->vrregs[j][3],
+					   varray[j - 20][0], varray[j - 20][1], varray[j - 20][2], varray[j - 20][3]);
+			}
+			bad_context = true;
+			break;
+		}
+	}
+}
+
+void *signal_vmx_c(void *p)
+{
+	int i, j;
+	long rc;
+	struct sigaction act;
+	act.sa_sigaction = signal_vmx_sig;
+	act.sa_flags = SA_SIGINFO;
+	rc = sigaction(SIGUSR1, &act, NULL);
+	if (rc)
+		return p;
+
+	srand(pthread_self());
+	for (i = 0; i < 12; i++)
+		for (j = 0; j < 4; j++)
+			varray[i][j] = rand();
+
+	rc = preempt_vmx(varray, &threads_starting, &running);
+
+	return (void *) rc;
+}
+
+int test_signal_vmx(void)
+{
+	int i, j, rc, threads;
+	void *rc_p;
+	pthread_t *tids;
+
+	// vcmpequd used in vmx_asm.S is v2.07
+	SKIP_IF(!have_hwcap2(PPC_FEATURE2_ARCH_2_07));
+
+	threads = sysconf(_SC_NPROCESSORS_ONLN) * THREAD_FACTOR;
+	tids = malloc(threads * sizeof(pthread_t));
+	FAIL_IF(!tids);
+
+	running = true;
+	threads_starting = threads;
+	for (i = 0; i < threads; i++) {
+		rc = pthread_create(&tids[i], NULL, signal_vmx_c, NULL);
+		FAIL_IF(rc);
+	}
+
+	setbuf(stdout, NULL);
+	printf("\tWaiting for %d workers to start... %d", threads, threads_starting);
+	while (threads_starting) {
+		asm volatile("": : :"memory");
+		usleep(1000);
+		printf(", %d", threads_starting);
+	}
+	printf(" ...done\n");
+
+	printf("\tSending signals to all threads %d times...", ITERATIONS);
+	for (i = 0; i < ITERATIONS; i++) {
+		for (j = 0; j < threads; j++) {
+			pthread_kill(tids[j], SIGUSR1);
+		}
+		sleep(1);
+	}
+	printf("done\n");
+
+	printf("\tKilling workers...");
+	running = 0;
+	for (i = 0; i < threads; i++) {
+		pthread_join(tids[i], &rc_p);
+
+		/*
+		 * Harness will say the fail was here, look at why signal_vmx
+		 * returned
+		 */
+		if ((long) rc_p || bad_context)
+			printf("oops\n");
+		if (bad_context)
+			fprintf(stderr, "\t!! bad_context is true\n");
+		FAIL_IF((long) rc_p || bad_context);
+	}
+	printf("done\n");
+
+	free(tids);
+	return 0;
+}
+
+int main(int argc, char *argv[])
+{
+	return test_harness(test_signal_vmx, "vmx_signal");
+}
diff --git a/tools/testing/selftests/powerpc/math/vmx_syscall.c b/tools/testing/selftests/powerpc/math/vmx_syscall.c
new file mode 100644
index 000000000..03c78dfe3
--- /dev/null
+++ b/tools/testing/selftests/powerpc/math/vmx_syscall.c
@@ -0,0 +1,92 @@
+// SPDX-License-Identifier: GPL-2.0-or-later
+/*
+ * Copyright 2015, Cyril Bur, IBM Corp.
+ *
+ * This test attempts to see if the VMX registers change across a syscall (fork).
+ */
+
+#include <altivec.h>
+#include <stdio.h>
+#include <unistd.h>
+#include <sys/syscall.h>
+#include <sys/time.h>
+#include <stdlib.h>
+#include <sys/types.h>
+#include <sys/wait.h>
+#include "utils.h"
+
+vector int varray[] = {{1, 2, 3, 4}, {5, 6, 7, 8}, {9, 10,11,12},
+	{13,14,15,16},{17,18,19,20},{21,22,23,24},
+	{25,26,27,28},{29,30,31,32},{33,34,35,36},
+	{37,38,39,40},{41,42,43,44},{45,46,47,48}};
+
+extern int test_vmx(vector int *varray, pid_t *pid);
+
+int vmx_syscall(void)
+{
+	pid_t fork_pid;
+	int i;
+	int ret;
+	int child_ret;
+	for (i = 0; i < 1000; i++) {
+		/* test_vmx will fork() */
+		ret = test_vmx(varray, &fork_pid);
+		if (fork_pid == -1)
+			return -1;
+		if (fork_pid == 0)
+			exit(ret);
+		waitpid(fork_pid, &child_ret, 0);
+		if (ret || child_ret)
+			return 1;
+	}
+
+	return 0;
+}
+
+int test_vmx_syscall(void)
+{
+	/*
+	 * Setup an environment with much context switching
+	 */
+	pid_t pid2;
+	pid_t pid;
+	int ret;
+	int child_ret;
+
+	// vcmpequd used in vmx_asm.S is v2.07
+	SKIP_IF(!have_hwcap2(PPC_FEATURE2_ARCH_2_07));
+
+	pid = fork();
+	FAIL_IF(pid == -1);
+
+	pid2 = fork();
+	ret = vmx_syscall();
+	/* Can't FAIL_IF(pid2 == -1); because we've already forked */
+	if (pid2 == -1) {
+		/*
+		 * Couldn't fork, ensure child_ret is set and is a fail
+		 */
+		ret = child_ret = 1;
+	} else {
+		if (pid2)
+			waitpid(pid2, &child_ret, 0);
+		else
+			exit(ret);
+	}
+
+	ret |= child_ret;
+
+	if (pid)
+		waitpid(pid, &child_ret, 0);
+	else
+		exit(ret);
+
+	FAIL_IF(ret || child_ret);
+	return 0;
+}
+
+int main(int argc, char *argv[])
+{
+	return test_harness(test_vmx_syscall, "vmx_syscall");
+
+}
diff --git a/tools/testing/selftests/powerpc/math/vsx_asm.S b/tools/testing/selftests/powerpc/math/vsx_asm.S
new file mode 100644
index 000000000..ffc165d98
--- /dev/null
+++ b/tools/testing/selftests/powerpc/math/vsx_asm.S
@@ -0,0 +1,57 @@
+/* SPDX-License-Identifier: GPL-2.0-or-later */
+/*
+ * Copyright 2015, Cyril Bur, IBM Corp.
+ */
+
+#include "basic_asm.h"
+#include "vsx_asm.h"
+
+#long check_vsx(vector int *r3);
+#This function wraps storeing VSX regs to the end of an array and a
+#call to a comparison function in C which boils down to a memcmp()
+FUNC_START(check_vsx)
+	PUSH_BASIC_STACK(32)
+	std	r3,STACK_FRAME_PARAM(0)(sp)
+	addi r3, r3, 16 * 12 #Second half of array
+	bl store_vsx
+	ld r3,STACK_FRAME_PARAM(0)(sp)
+	bl vsx_memcmp
+	POP_BASIC_STACK(32)
+	blr
+FUNC_END(check_vsx)
+
+# int preempt_vmx(vector int *varray, int *threads_starting,
+#                 int *running);
+# On starting will (atomically) decrement threads_starting as a signal
+# that the VMX have been loaded with varray. Will proceed to check the
+# validity of the VMX registers while running is not zero.
+FUNC_START(preempt_vsx)
+	PUSH_BASIC_STACK(512)
+	std r3,STACK_FRAME_PARAM(0)(sp) # vector int *varray
+	std r4,STACK_FRAME_PARAM(1)(sp) # int *threads_starting
+	std r5,STACK_FRAME_PARAM(2)(sp) # int *running
+
+	bl load_vsx
+	nop
+
+	sync
+	# Atomic DEC
+	ld r3,STACK_FRAME_PARAM(1)(sp)
+1:	lwarx r4,0,r3
+	addi r4,r4,-1
+	stwcx. r4,0,r3
+	bne- 1b
+
+2:	ld r3,STACK_FRAME_PARAM(0)(sp)
+	bl check_vsx
+	nop
+	cmpdi r3,0
+	bne 3f
+	ld r4,STACK_FRAME_PARAM(2)(sp)
+	ld r5,0(r4)
+	cmpwi r5,0
+	bne 2b
+
+3:	POP_BASIC_STACK(512)
+	blr
+FUNC_END(preempt_vsx)
diff --git a/tools/testing/selftests/powerpc/math/vsx_preempt.c b/tools/testing/selftests/powerpc/math/vsx_preempt.c
new file mode 100644
index 000000000..d1601bb88
--- /dev/null
+++ b/tools/testing/selftests/powerpc/math/vsx_preempt.c
@@ -0,0 +1,145 @@
+// SPDX-License-Identifier: GPL-2.0-or-later
+/*
+ * Copyright 2015, Cyril Bur, IBM Corp.
+ *
+ * This test attempts to see if the VSX registers change across preemption.
+ * There is no way to be sure preemption happened so this test just
+ * uses many threads and a long wait. As such, a successful test
+ * doesn't mean much but a failure is bad.
+ */
+
+#include <stdio.h>
+#include <string.h>
+#include <unistd.h>
+#include <sys/syscall.h>
+#include <sys/time.h>
+#include <sys/types.h>
+#include <sys/wait.h>
+#include <stdlib.h>
+#include <pthread.h>
+
+#include "utils.h"
+
+/* Time to wait for workers to get preempted (seconds) */
+#define PREEMPT_TIME 20
+/*
+ * Factor by which to multiply number of online CPUs for total number of
+ * worker threads
+ */
+#define THREAD_FACTOR 8
+
+/*
+ * Ensure there is twice the number of non-volatile VMX regs!
+ * check_vmx() is going to use the other half as space to put the live
+ * registers before calling vsx_memcmp()
+ */
+__thread vector int varray[24] = {
+	{1, 2, 3, 4 }, {5, 6, 7, 8 }, {9, 10,11,12},
+	{13,14,15,16}, {17,18,19,20}, {21,22,23,24},
+	{25,26,27,28}, {29,30,31,32}, {33,34,35,36},
+	{37,38,39,40}, {41,42,43,44}, {45,46,47,48}
+};
+
+int threads_starting;
+int running;
+
+extern long preempt_vsx(vector int *varray, int *threads_starting, int *running);
+
+long vsx_memcmp(vector int *a) {
+	vector int zero = {0, 0, 0, 0};
+	int i;
+
+	FAIL_IF(a != varray);
+
+	for(i = 0; i < 12; i++) {
+		if (memcmp(&a[i + 12], &zero, sizeof(vector int)) == 0) {
+			fprintf(stderr, "Detected zero from the VSX reg %d\n", i + 12);
+			return 2;
+		}
+	}
+
+	if (memcmp(a, &a[12], 12 * sizeof(vector int))) {
+		long *p = (long *)a;
+		fprintf(stderr, "VSX mismatch\n");
+		for (i = 0; i < 24; i=i+2)
+			fprintf(stderr, "%d: 0x%08lx%08lx | 0x%08lx%08lx\n",
+					i/2 + i%2 + 20, p[i], p[i + 1], p[i + 24], p[i + 25]);
+		return 1;
+	}
+	return 0;
+}
+
+void *preempt_vsx_c(void *p)
+{
+	int i, j;
+	long rc;
+	srand(pthread_self());
+	for (i = 0; i < 12; i++)
+		for (j = 0; j < 4; j++) {
+			varray[i][j] = rand();
+			/* Don't want zero because it hides kernel problems */
+			if (varray[i][j] == 0)
+				j--;
+		}
+	rc = preempt_vsx(varray, &threads_starting, &running);
+	if (rc == 2)
+		fprintf(stderr, "Caught zeros in VSX compares\n");
+	return (void *)rc;
+}
+
+int test_preempt_vsx(void)
+{
+	int i, rc, threads;
+	pthread_t *tids;
+
+	SKIP_IF(!have_hwcap(PPC_FEATURE_HAS_VSX));
+
+	threads = sysconf(_SC_NPROCESSORS_ONLN) * THREAD_FACTOR;
+	tids = malloc(threads * sizeof(pthread_t));
+	FAIL_IF(!tids);
+
+	running = true;
+	threads_starting = threads;
+	for (i = 0; i < threads; i++) {
+		rc = pthread_create(&tids[i], NULL, preempt_vsx_c, NULL);
+		FAIL_IF(rc);
+	}
+
+	setbuf(stdout, NULL);
+	/* Not really nessesary but nice to wait for every thread to start */
+	printf("\tWaiting for %d workers to start...", threads_starting);
+	while(threads_starting)
+		asm volatile("": : :"memory");
+	printf("done\n");
+
+	printf("\tWaiting for %d seconds to let some workers get preempted...", PREEMPT_TIME);
+	sleep(PREEMPT_TIME);
+	printf("done\n");
+
+	printf("\tStopping workers...");
+	/*
+	 * Working are checking this value every loop. In preempt_vsx 'cmpwi r5,0; bne 2b'.
+	 * r5 will have loaded the value of running.
+	 */
+	running = 0;
+	for (i = 0; i < threads; i++) {
+		void *rc_p;
+		pthread_join(tids[i], &rc_p);
+
+		/*
+		 * Harness will say the fail was here, look at why preempt_vsx
+		 * returned
+		 */
+		if ((long) rc_p)
+			printf("oops\n");
+		FAIL_IF((long) rc_p);
+	}
+	printf("done\n");
+
+	return 0;
+}
+
+int main(int argc, char *argv[])
+{
+	return test_harness(test_preempt_vsx, "vsx_preempt");
+}
diff --git a/tools/testing/selftests/powerpc/mm/.gitignore b/tools/testing/selftests/powerpc/mm/.gitignore
new file mode 100644
index 000000000..aac4a59f9
--- /dev/null
+++ b/tools/testing/selftests/powerpc/mm/.gitignore
@@ -0,0 +1,14 @@
+# SPDX-License-Identifier: GPL-2.0-only
+hugetlb_vs_thp_test
+subpage_prot
+tempfile
+prot_sao
+segv_errors
+wild_bctr
+large_vm_fork_separation
+bad_accesses
+tlbie_test
+pkey_exec_prot
+pkey_siginfo
+stack_expansion_ldst
+stack_expansion_signal
diff --git a/tools/testing/selftests/powerpc/mm/Makefile b/tools/testing/selftests/powerpc/mm/Makefile
new file mode 100644
index 000000000..defe488d6
--- /dev/null
+++ b/tools/testing/selftests/powerpc/mm/Makefile
@@ -0,0 +1,34 @@
+# SPDX-License-Identifier: GPL-2.0
+noarg:
+	$(MAKE) -C ../
+
+TEST_GEN_PROGS := hugetlb_vs_thp_test subpage_prot prot_sao segv_errors wild_bctr \
+		  large_vm_fork_separation bad_accesses pkey_exec_prot \
+		  pkey_siginfo stack_expansion_signal stack_expansion_ldst
+
+TEST_GEN_PROGS_EXTENDED := tlbie_test
+TEST_GEN_FILES := tempfile
+
+top_srcdir = ../../../../..
+include ../../lib.mk
+
+$(TEST_GEN_PROGS): ../harness.c ../utils.c
+
+$(OUTPUT)/prot_sao: ../utils.c
+
+$(OUTPUT)/wild_bctr: CFLAGS += -m64
+$(OUTPUT)/large_vm_fork_separation: CFLAGS += -m64
+$(OUTPUT)/bad_accesses: CFLAGS += -m64
+$(OUTPUT)/pkey_exec_prot: CFLAGS += -m64
+$(OUTPUT)/pkey_siginfo: CFLAGS += -m64
+
+$(OUTPUT)/stack_expansion_signal: ../utils.c ../pmu/lib.c
+
+$(OUTPUT)/stack_expansion_ldst: CFLAGS += -fno-stack-protector
+$(OUTPUT)/stack_expansion_ldst: ../utils.c
+
+$(OUTPUT)/tempfile:
+	dd if=/dev/zero of=$@ bs=64k count=1
+
+$(OUTPUT)/tlbie_test: LDLIBS += -lpthread
+$(OUTPUT)/pkey_siginfo: LDLIBS += -lpthread
diff --git a/tools/testing/selftests/powerpc/mm/bad_accesses.c b/tools/testing/selftests/powerpc/mm/bad_accesses.c
new file mode 100644
index 000000000..fd747b2ff
--- /dev/null
+++ b/tools/testing/selftests/powerpc/mm/bad_accesses.c
@@ -0,0 +1,144 @@
+// SPDX-License-Identifier: GPL-2.0+
+//
+// Copyright 2019, Michael Ellerman, IBM Corp.
+//
+// Test that out-of-bounds reads/writes behave as expected.
+
+#include <setjmp.h>
+#include <stdbool.h>
+#include <stdio.h>
+#include <stdlib.h>
+#include <string.h>
+#include <sys/types.h>
+#include <sys/wait.h>
+#include <unistd.h>
+
+#include "utils.h"
+
+// Old distros (Ubuntu 16.04 at least) don't define this
+#ifndef SEGV_BNDERR
+#define SEGV_BNDERR	3
+#endif
+
+// 64-bit kernel is always here
+#define PAGE_OFFSET	(0xcul << 60)
+
+static unsigned long kernel_virt_end;
+
+static volatile int fault_code;
+static volatile unsigned long fault_addr;
+static jmp_buf setjmp_env;
+
+static void segv_handler(int n, siginfo_t *info, void *ctxt_v)
+{
+	fault_code = info->si_code;
+	fault_addr = (unsigned long)info->si_addr;
+	siglongjmp(setjmp_env, 1);
+}
+
+int bad_access(char *p, bool write)
+{
+	char x;
+
+	fault_code = 0;
+	fault_addr = 0;
+
+	if (sigsetjmp(setjmp_env, 1) == 0) {
+		if (write)
+			*p = 1;
+		else
+			x = *p;
+
+		printf("Bad - no SEGV! (%c)\n", x);
+		return 1;
+	}
+
+	// If we see MAPERR that means we took a page fault rather than an SLB
+	// miss. We only expect to take page faults for addresses within the
+	// valid kernel range.
+	FAIL_IF(fault_code == SEGV_MAPERR && \
+		(fault_addr < PAGE_OFFSET || fault_addr >= kernel_virt_end));
+
+	FAIL_IF(fault_code != SEGV_MAPERR && fault_code != SEGV_BNDERR);
+
+	return 0;
+}
+
+static int test(void)
+{
+	unsigned long i, j, addr, region_shift, page_shift, page_size;
+	struct sigaction sig;
+	bool hash_mmu;
+
+	sig = (struct sigaction) {
+		.sa_sigaction = segv_handler,
+		.sa_flags = SA_SIGINFO,
+	};
+
+	FAIL_IF(sigaction(SIGSEGV, &sig, NULL) != 0);
+
+	FAIL_IF(using_hash_mmu(&hash_mmu));
+
+	page_size = sysconf(_SC_PAGESIZE);
+	if (page_size == (64 * 1024))
+		page_shift = 16;
+	else
+		page_shift = 12;
+
+	if (page_size == (64 * 1024) || !hash_mmu) {
+		region_shift = 52;
+
+		// We have 7 512T regions (4 kernel linear, vmalloc, io, vmemmap)
+		kernel_virt_end = PAGE_OFFSET + (7 * (512ul << 40));
+	} else if (page_size == (4 * 1024) && hash_mmu) {
+		region_shift = 46;
+
+		// We have 7 64T regions (4 kernel linear, vmalloc, io, vmemmap)
+		kernel_virt_end = PAGE_OFFSET + (7 * (64ul << 40));
+	} else
+		FAIL_IF(true);
+
+	printf("Using %s MMU, PAGE_SIZE = %dKB start address 0x%016lx\n",
+	       hash_mmu ? "hash" : "radix",
+	       (1 << page_shift) >> 10,
+	       1ul << region_shift);
+
+	// This generates access patterns like:
+	//   0x0010000000000000
+	//   0x0010000000010000
+	//   0x0010000000020000
+	//   ...
+	//   0x0014000000000000
+	//   0x0018000000000000
+	//   0x0020000000000000
+	//   0x0020000000010000
+	//   0x0020000000020000
+	//   ...
+	//   0xf400000000000000
+	//   0xf800000000000000
+
+	for (i = 1; i <= ((0xful << 60) >> region_shift); i++) {
+		for (j = page_shift - 1; j < 60; j++) {
+			unsigned long base, delta;
+
+			base  = i << region_shift;
+			delta = 1ul << j;
+
+			if (delta >= base)
+				break;
+
+			addr = (base | delta) & ~((1 << page_shift) - 1);
+
+			FAIL_IF(bad_access((char *)addr, false));
+			FAIL_IF(bad_access((char *)addr, true));
+		}
+	}
+
+	return 0;
+}
+
+int main(void)
+{
+	test_harness_set_timeout(300);
+	return test_harness(test, "bad_accesses");
+}
diff --git a/tools/testing/selftests/powerpc/mm/hugetlb_vs_thp_test.c b/tools/testing/selftests/powerpc/mm/hugetlb_vs_thp_test.c
new file mode 100644
index 000000000..9932359ce
--- /dev/null
+++ b/tools/testing/selftests/powerpc/mm/hugetlb_vs_thp_test.c
@@ -0,0 +1,77 @@
+// SPDX-License-Identifier: GPL-2.0
+#include <stdio.h>
+#include <sys/mman.h>
+#include <unistd.h>
+
+#include "utils.h"
+
+/* This must match the huge page & THP size */
+#define SIZE	(16 * 1024 * 1024)
+
+static int test_body(void)
+{
+	void *addr;
+	char *p;
+
+	addr = (void *)0xa0000000;
+
+	p = mmap(addr, SIZE, PROT_READ | PROT_WRITE,
+		 MAP_HUGETLB | MAP_ANONYMOUS | MAP_PRIVATE, -1, 0);
+	if (p != MAP_FAILED) {
+		/*
+		 * Typically the mmap will fail because no huge pages are
+		 * allocated on the system. But if there are huge pages
+		 * allocated the mmap will succeed. That's fine too, we just
+		 * munmap here before continuing.  munmap() length of
+		 * MAP_HUGETLB memory must be hugepage aligned.
+		 */
+		if (munmap(addr, SIZE)) {
+			perror("munmap");
+			return 1;
+		}
+	}
+
+	p = mmap(addr, SIZE, PROT_READ | PROT_WRITE,
+		 MAP_ANONYMOUS | MAP_PRIVATE, -1, 0);
+	if (p == MAP_FAILED) {
+		printf("Mapping failed @ %p\n", addr);
+		perror("mmap");
+		return 1;
+	}
+
+	/*
+	 * Either a user or kernel access is sufficient to trigger the bug.
+	 * A kernel access is easier to spot & debug, as it will trigger the
+	 * softlockup or RCU stall detectors, and when the system is kicked
+	 * into xmon we get a backtrace in the kernel.
+	 *
+	 * A good option is:
+	 *  getcwd(p, SIZE);
+	 *
+	 * For the purposes of this testcase it's preferable to spin in
+	 * userspace, so the harness can kill us if we get stuck. That way we
+	 * see a test failure rather than a dead system.
+	 */
+	*p = 0xf;
+
+	munmap(addr, SIZE);
+
+	return 0;
+}
+
+static int test_main(void)
+{
+	int i;
+
+	/* 10,000 because it's a "bunch", and completes reasonably quickly */
+	for (i = 0; i < 10000; i++)
+		if (test_body())
+			return 1;
+
+	return 0;
+}
+
+int main(void)
+{
+	return test_harness(test_main, "hugetlb_vs_thp");
+}
diff --git a/tools/testing/selftests/powerpc/mm/large_vm_fork_separation.c b/tools/testing/selftests/powerpc/mm/large_vm_fork_separation.c
new file mode 100644
index 000000000..2363a7f3a
--- /dev/null
+++ b/tools/testing/selftests/powerpc/mm/large_vm_fork_separation.c
@@ -0,0 +1,87 @@
+// SPDX-License-Identifier: GPL-2.0+
+//
+// Copyright 2019, Michael Ellerman, IBM Corp.
+//
+// Test that allocating memory beyond the memory limit and then forking is
+// handled correctly, ie. the child is able to access the mappings beyond the
+// memory limit and the child's writes are not visible to the parent.
+
+#include <stdio.h>
+#include <stdlib.h>
+#include <sys/mman.h>
+#include <sys/types.h>
+#include <sys/wait.h>
+#include <unistd.h>
+
+#include "utils.h"
+
+
+#ifndef MAP_FIXED_NOREPLACE
+#define MAP_FIXED_NOREPLACE	MAP_FIXED	// "Should be safe" above 512TB
+#endif
+
+
+static int test(void)
+{
+	int p2c[2], c2p[2], rc, status, c, *p;
+	unsigned long page_size;
+	pid_t pid;
+
+	page_size = sysconf(_SC_PAGESIZE);
+	SKIP_IF(page_size != 65536);
+
+	// Create a mapping at 512TB to allocate an extended_id
+	p = mmap((void *)(512ul << 40), page_size, PROT_READ | PROT_WRITE,
+		MAP_PRIVATE | MAP_ANONYMOUS | MAP_FIXED_NOREPLACE, -1, 0);
+	if (p == MAP_FAILED) {
+		perror("mmap");
+		printf("Error: couldn't mmap(), confirm kernel has 4TB support?\n");
+		return 1;
+	}
+
+	printf("parent writing %p = 1\n", p);
+	*p = 1;
+
+	FAIL_IF(pipe(p2c) == -1 || pipe(c2p) == -1);
+
+	pid = fork();
+	if (pid == 0) {
+		FAIL_IF(read(p2c[0], &c, 1) != 1);
+
+		pid = getpid();
+		printf("child writing  %p = %d\n", p, pid);
+		*p = pid;
+
+		FAIL_IF(write(c2p[1], &c, 1) != 1);
+		FAIL_IF(read(p2c[0], &c, 1) != 1);
+		exit(0);
+	}
+
+	c = 0;
+	FAIL_IF(write(p2c[1], &c, 1) != 1);
+	FAIL_IF(read(c2p[0], &c, 1) != 1);
+
+	// Prevent compiler optimisation
+	barrier();
+
+	rc = 0;
+	printf("parent reading %p = %d\n", p, *p);
+	if (*p != 1) {
+		printf("Error: BUG! parent saw child's write! *p = %d\n", *p);
+		rc = 1;
+	}
+
+	FAIL_IF(write(p2c[1], &c, 1) != 1);
+	FAIL_IF(waitpid(pid, &status, 0) == -1);
+	FAIL_IF(!WIFEXITED(status) || WEXITSTATUS(status));
+
+	if (rc == 0)
+		printf("success: test completed OK\n");
+
+	return rc;
+}
+
+int main(void)
+{
+	return test_harness(test, "large_vm_fork_separation");
+}
diff --git a/tools/testing/selftests/powerpc/mm/pkey_exec_prot.c b/tools/testing/selftests/powerpc/mm/pkey_exec_prot.c
new file mode 100644
index 000000000..0af4f0266
--- /dev/null
+++ b/tools/testing/selftests/powerpc/mm/pkey_exec_prot.c
@@ -0,0 +1,294 @@
+// SPDX-License-Identifier: GPL-2.0+
+
+/*
+ * Copyright 2020, Sandipan Das, IBM Corp.
+ *
+ * Test if applying execute protection on pages using memory
+ * protection keys works as expected.
+ */
+
+#define _GNU_SOURCE
+#include <stdio.h>
+#include <stdlib.h>
+#include <string.h>
+#include <signal.h>
+
+#include <unistd.h>
+
+#include "pkeys.h"
+
+#define PPC_INST_NOP	0x60000000
+#define PPC_INST_TRAP	0x7fe00008
+#define PPC_INST_BLR	0x4e800020
+
+static volatile sig_atomic_t fault_pkey, fault_code, fault_type;
+static volatile sig_atomic_t remaining_faults;
+static volatile unsigned int *fault_addr;
+static unsigned long pgsize, numinsns;
+static unsigned int *insns;
+
+static void trap_handler(int signum, siginfo_t *sinfo, void *ctx)
+{
+	/* Check if this fault originated from the expected address */
+	if (sinfo->si_addr != (void *) fault_addr)
+		sigsafe_err("got a fault for an unexpected address\n");
+
+	_exit(1);
+}
+
+static void segv_handler(int signum, siginfo_t *sinfo, void *ctx)
+{
+	int signal_pkey;
+
+	signal_pkey = siginfo_pkey(sinfo);
+	fault_code = sinfo->si_code;
+
+	/* Check if this fault originated from the expected address */
+	if (sinfo->si_addr != (void *) fault_addr) {
+		sigsafe_err("got a fault for an unexpected address\n");
+		_exit(1);
+	}
+
+	/* Check if too many faults have occurred for a single test case */
+	if (!remaining_faults) {
+		sigsafe_err("got too many faults for the same address\n");
+		_exit(1);
+	}
+
+
+	/* Restore permissions in order to continue */
+	switch (fault_code) {
+	case SEGV_ACCERR:
+		if (mprotect(insns, pgsize, PROT_READ | PROT_WRITE)) {
+			sigsafe_err("failed to set access permissions\n");
+			_exit(1);
+		}
+		break;
+	case SEGV_PKUERR:
+		if (signal_pkey != fault_pkey) {
+			sigsafe_err("got a fault for an unexpected pkey\n");
+			_exit(1);
+		}
+
+		switch (fault_type) {
+		case PKEY_DISABLE_ACCESS:
+			pkey_set_rights(fault_pkey, 0);
+			break;
+		case PKEY_DISABLE_EXECUTE:
+			/*
+			 * Reassociate the exec-only pkey with the region
+			 * to be able to continue. Unlike AMR, we cannot
+			 * set IAMR directly from userspace to restore the
+			 * permissions.
+			 */
+			if (mprotect(insns, pgsize, PROT_EXEC)) {
+				sigsafe_err("failed to set execute permissions\n");
+				_exit(1);
+			}
+			break;
+		default:
+			sigsafe_err("got a fault with an unexpected type\n");
+			_exit(1);
+		}
+		break;
+	default:
+		sigsafe_err("got a fault with an unexpected code\n");
+		_exit(1);
+	}
+
+	remaining_faults--;
+}
+
+static int test(void)
+{
+	struct sigaction segv_act, trap_act;
+	unsigned long rights;
+	int pkey, ret, i;
+
+	ret = pkeys_unsupported();
+	if (ret)
+		return ret;
+
+	/* Setup SIGSEGV handler */
+	segv_act.sa_handler = 0;
+	segv_act.sa_sigaction = segv_handler;
+	FAIL_IF(sigprocmask(SIG_SETMASK, 0, &segv_act.sa_mask) != 0);
+	segv_act.sa_flags = SA_SIGINFO;
+	segv_act.sa_restorer = 0;
+	FAIL_IF(sigaction(SIGSEGV, &segv_act, NULL) != 0);
+
+	/* Setup SIGTRAP handler */
+	trap_act.sa_handler = 0;
+	trap_act.sa_sigaction = trap_handler;
+	FAIL_IF(sigprocmask(SIG_SETMASK, 0, &trap_act.sa_mask) != 0);
+	trap_act.sa_flags = SA_SIGINFO;
+	trap_act.sa_restorer = 0;
+	FAIL_IF(sigaction(SIGTRAP, &trap_act, NULL) != 0);
+
+	/* Setup executable region */
+	pgsize = getpagesize();
+	numinsns = pgsize / sizeof(unsigned int);
+	insns = (unsigned int *) mmap(NULL, pgsize, PROT_READ | PROT_WRITE,
+				      MAP_PRIVATE | MAP_ANONYMOUS, -1, 0);
+	FAIL_IF(insns == MAP_FAILED);
+
+	/* Write the instruction words */
+	for (i = 1; i < numinsns - 1; i++)
+		insns[i] = PPC_INST_NOP;
+
+	/*
+	 * Set the first instruction as an unconditional trap. If
+	 * the last write to this address succeeds, this should
+	 * get overwritten by a no-op.
+	 */
+	insns[0] = PPC_INST_TRAP;
+
+	/*
+	 * Later, to jump to the executable region, we use a branch
+	 * and link instruction (bctrl) which sets the return address
+	 * automatically in LR. Use that to return back.
+	 */
+	insns[numinsns - 1] = PPC_INST_BLR;
+
+	/* Allocate a pkey that restricts execution */
+	rights = PKEY_DISABLE_EXECUTE;
+	pkey = sys_pkey_alloc(0, rights);
+	FAIL_IF(pkey < 0);
+
+	/*
+	 * Pick the first instruction's address from the executable
+	 * region.
+	 */
+	fault_addr = insns;
+
+	/* The following two cases will avoid SEGV_PKUERR */
+	fault_type = -1;
+	fault_pkey = -1;
+
+	/*
+	 * Read an instruction word from the address when AMR bits
+	 * are not set i.e. the pkey permits both read and write
+	 * access.
+	 *
+	 * This should not generate a fault as having PROT_EXEC
+	 * implies PROT_READ on GNU systems. The pkey currently
+	 * restricts execution only based on the IAMR bits. The
+	 * AMR bits are cleared.
+	 */
+	remaining_faults = 0;
+	FAIL_IF(sys_pkey_mprotect(insns, pgsize, PROT_EXEC, pkey) != 0);
+	printf("read from %p, pkey permissions are %s\n", fault_addr,
+	       pkey_rights(rights));
+	i = *fault_addr;
+	FAIL_IF(remaining_faults != 0);
+
+	/*
+	 * Write an instruction word to the address when AMR bits
+	 * are not set i.e. the pkey permits both read and write
+	 * access.
+	 *
+	 * This should generate an access fault as having just
+	 * PROT_EXEC also restricts writes. The pkey currently
+	 * restricts execution only based on the IAMR bits. The
+	 * AMR bits are cleared.
+	 */
+	remaining_faults = 1;
+	FAIL_IF(sys_pkey_mprotect(insns, pgsize, PROT_EXEC, pkey) != 0);
+	printf("write to %p, pkey permissions are %s\n", fault_addr,
+	       pkey_rights(rights));
+	*fault_addr = PPC_INST_TRAP;
+	FAIL_IF(remaining_faults != 0 || fault_code != SEGV_ACCERR);
+
+	/* The following three cases will generate SEGV_PKUERR */
+	rights |= PKEY_DISABLE_ACCESS;
+	fault_type = PKEY_DISABLE_ACCESS;
+	fault_pkey = pkey;
+
+	/*
+	 * Read an instruction word from the address when AMR bits
+	 * are set i.e. the pkey permits neither read nor write
+	 * access.
+	 *
+	 * This should generate a pkey fault based on AMR bits only
+	 * as having PROT_EXEC implicitly allows reads.
+	 */
+	remaining_faults = 1;
+	FAIL_IF(sys_pkey_mprotect(insns, pgsize, PROT_EXEC, pkey) != 0);
+	pkey_set_rights(pkey, rights);
+	printf("read from %p, pkey permissions are %s\n", fault_addr,
+	       pkey_rights(rights));
+	i = *fault_addr;
+	FAIL_IF(remaining_faults != 0 || fault_code != SEGV_PKUERR);
+
+	/*
+	 * Write an instruction word to the address when AMR bits
+	 * are set i.e. the pkey permits neither read nor write
+	 * access.
+	 *
+	 * This should generate two faults. First, a pkey fault
+	 * based on AMR bits and then an access fault since
+	 * PROT_EXEC does not allow writes.
+	 */
+	remaining_faults = 2;
+	FAIL_IF(sys_pkey_mprotect(insns, pgsize, PROT_EXEC, pkey) != 0);
+	pkey_set_rights(pkey, rights);
+	printf("write to %p, pkey permissions are %s\n", fault_addr,
+	       pkey_rights(rights));
+	*fault_addr = PPC_INST_NOP;
+	FAIL_IF(remaining_faults != 0 || fault_code != SEGV_ACCERR);
+
+	/* Free the current pkey */
+	sys_pkey_free(pkey);
+
+	rights = 0;
+	do {
+		/*
+		 * Allocate pkeys with all valid combinations of read,
+		 * write and execute restrictions.
+		 */
+		pkey = sys_pkey_alloc(0, rights);
+		FAIL_IF(pkey < 0);
+
+		/*
+		 * Jump to the executable region. AMR bits may or may not
+		 * be set but they should not affect execution.
+		 *
+		 * This should generate pkey faults based on IAMR bits which
+		 * may be set to restrict execution.
+		 *
+		 * The first iteration also checks if the overwrite of the
+		 * first instruction word from a trap to a no-op succeeded.
+		 */
+		fault_pkey = pkey;
+		fault_type = -1;
+		remaining_faults = 0;
+		if (rights & PKEY_DISABLE_EXECUTE) {
+			fault_type = PKEY_DISABLE_EXECUTE;
+			remaining_faults = 1;
+		}
+
+		FAIL_IF(sys_pkey_mprotect(insns, pgsize, PROT_EXEC, pkey) != 0);
+		printf("execute at %p, pkey permissions are %s\n", fault_addr,
+		       pkey_rights(rights));
+		asm volatile("mtctr	%0; bctrl" : : "r"(insns));
+		FAIL_IF(remaining_faults != 0);
+		if (rights & PKEY_DISABLE_EXECUTE)
+			FAIL_IF(fault_code != SEGV_PKUERR);
+
+		/* Free the current pkey */
+		sys_pkey_free(pkey);
+
+		/* Find next valid combination of pkey rights */
+		rights = next_pkey_rights(rights);
+	} while (rights);
+
+	/* Cleanup */
+	munmap((void *) insns, pgsize);
+
+	return 0;
+}
+
+int main(void)
+{
+	return test_harness(test, "pkey_exec_prot");
+}
diff --git a/tools/testing/selftests/powerpc/mm/pkey_siginfo.c b/tools/testing/selftests/powerpc/mm/pkey_siginfo.c
new file mode 100644
index 000000000..2db76e56d
--- /dev/null
+++ b/tools/testing/selftests/powerpc/mm/pkey_siginfo.c
@@ -0,0 +1,333 @@
+// SPDX-License-Identifier: GPL-2.0
+
+/*
+ * Copyright 2020, Sandipan Das, IBM Corp.
+ *
+ * Test if the signal information reports the correct memory protection
+ * key upon getting a key access violation fault for a page that was
+ * attempted to be protected by two different keys from two competing
+ * threads at the same time.
+ */
+
+#define _GNU_SOURCE
+#include <stdio.h>
+#include <stdlib.h>
+#include <string.h>
+#include <signal.h>
+
+#include <unistd.h>
+#include <pthread.h>
+#include <sys/mman.h>
+
+#include "pkeys.h"
+
+#define PPC_INST_NOP	0x60000000
+#define PPC_INST_BLR	0x4e800020
+#define PROT_RWX	(PROT_READ | PROT_WRITE | PROT_EXEC)
+
+#define NUM_ITERATIONS	1000000
+
+static volatile sig_atomic_t perm_pkey, rest_pkey;
+static volatile sig_atomic_t rights, fault_count;
+static volatile unsigned int *volatile fault_addr;
+static pthread_barrier_t iteration_barrier;
+
+static void segv_handler(int signum, siginfo_t *sinfo, void *ctx)
+{
+	void *pgstart;
+	size_t pgsize;
+	int pkey;
+
+	pkey = siginfo_pkey(sinfo);
+
+	/* Check if this fault originated from a pkey access violation */
+	if (sinfo->si_code != SEGV_PKUERR) {
+		sigsafe_err("got a fault for an unexpected reason\n");
+		_exit(1);
+	}
+
+	/* Check if this fault originated from the expected address */
+	if (sinfo->si_addr != (void *) fault_addr) {
+		sigsafe_err("got a fault for an unexpected address\n");
+		_exit(1);
+	}
+
+	/* Check if this fault originated from the restrictive pkey */
+	if (pkey != rest_pkey) {
+		sigsafe_err("got a fault for an unexpected pkey\n");
+		_exit(1);
+	}
+
+	/* Check if too many faults have occurred for the same iteration */
+	if (fault_count > 0) {
+		sigsafe_err("got too many faults for the same address\n");
+		_exit(1);
+	}
+
+	pgsize = getpagesize();
+	pgstart = (void *) ((unsigned long) fault_addr & ~(pgsize - 1));
+
+	/*
+	 * If the current fault occurred due to lack of execute rights,
+	 * reassociate the page with the exec-only pkey since execute
+	 * rights cannot be changed directly for the faulting pkey as
+	 * IAMR is inaccessible from userspace.
+	 *
+	 * Otherwise, if the current fault occurred due to lack of
+	 * read-write rights, change the AMR permission bits for the
+	 * pkey.
+	 *
+	 * This will let the test continue.
+	 */
+	if (rights == PKEY_DISABLE_EXECUTE &&
+	    mprotect(pgstart, pgsize, PROT_EXEC))
+		_exit(1);
+	else
+		pkey_set_rights(pkey, 0);
+
+	fault_count++;
+}
+
+struct region {
+	unsigned long rights;
+	unsigned int *base;
+	size_t size;
+};
+
+static void *protect(void *p)
+{
+	unsigned long rights;
+	unsigned int *base;
+	size_t size;
+	int tid, i;
+
+	tid = gettid();
+	base = ((struct region *) p)->base;
+	size = ((struct region *) p)->size;
+	FAIL_IF_EXIT(!base);
+
+	/* No read, write and execute restrictions */
+	rights = 0;
+
+	printf("tid %d, pkey permissions are %s\n", tid, pkey_rights(rights));
+
+	/* Allocate the permissive pkey */
+	perm_pkey = sys_pkey_alloc(0, rights);
+	FAIL_IF_EXIT(perm_pkey < 0);
+
+	/*
+	 * Repeatedly try to protect the common region with a permissive
+	 * pkey
+	 */
+	for (i = 0; i < NUM_ITERATIONS; i++) {
+		/*
+		 * Wait until the other thread has finished allocating the
+		 * restrictive pkey or until the next iteration has begun
+		 */
+		pthread_barrier_wait(&iteration_barrier);
+
+		/* Try to associate the permissive pkey with the region */
+		FAIL_IF_EXIT(sys_pkey_mprotect(base, size, PROT_RWX,
+					       perm_pkey));
+	}
+
+	/* Free the permissive pkey */
+	sys_pkey_free(perm_pkey);
+
+	return NULL;
+}
+
+static void *protect_access(void *p)
+{
+	size_t size, numinsns;
+	unsigned int *base;
+	int tid, i;
+
+	tid = gettid();
+	base = ((struct region *) p)->base;
+	size = ((struct region *) p)->size;
+	rights = ((struct region *) p)->rights;
+	numinsns = size / sizeof(base[0]);
+	FAIL_IF_EXIT(!base);
+
+	/* Allocate the restrictive pkey */
+	rest_pkey = sys_pkey_alloc(0, rights);
+	FAIL_IF_EXIT(rest_pkey < 0);
+
+	printf("tid %d, pkey permissions are %s\n", tid, pkey_rights(rights));
+	printf("tid %d, %s randomly in range [%p, %p]\n", tid,
+	       (rights == PKEY_DISABLE_EXECUTE) ? "execute" :
+	       (rights == PKEY_DISABLE_WRITE)  ? "write" : "read",
+	       base, base + numinsns);
+
+	/*
+	 * Repeatedly try to protect the common region with a restrictive
+	 * pkey and read, write or execute from it
+	 */
+	for (i = 0; i < NUM_ITERATIONS; i++) {
+		/*
+		 * Wait until the other thread has finished allocating the
+		 * permissive pkey or until the next iteration has begun
+		 */
+		pthread_barrier_wait(&iteration_barrier);
+
+		/* Try to associate the restrictive pkey with the region */
+		FAIL_IF_EXIT(sys_pkey_mprotect(base, size, PROT_RWX,
+					       rest_pkey));
+
+		/* Choose a random instruction word address from the region */
+		fault_addr = base + (rand() % numinsns);
+		fault_count = 0;
+
+		switch (rights) {
+		/* Read protection test */
+		case PKEY_DISABLE_ACCESS:
+			/*
+			 * Read an instruction word from the region and
+			 * verify if it has not been overwritten to
+			 * something unexpected
+			 */
+			FAIL_IF_EXIT(*fault_addr != PPC_INST_NOP &&
+				     *fault_addr != PPC_INST_BLR);
+			break;
+
+		/* Write protection test */
+		case PKEY_DISABLE_WRITE:
+			/*
+			 * Write an instruction word to the region and
+			 * verify if the overwrite has succeeded
+			 */
+			*fault_addr = PPC_INST_BLR;
+			FAIL_IF_EXIT(*fault_addr != PPC_INST_BLR);
+			break;
+
+		/* Execute protection test */
+		case PKEY_DISABLE_EXECUTE:
+			/* Jump to the region and execute instructions */
+			asm volatile(
+				"mtctr	%0; bctrl"
+				: : "r"(fault_addr) : "ctr", "lr");
+			break;
+		}
+
+		/*
+		 * Restore the restrictions originally imposed by the
+		 * restrictive pkey as the signal handler would have
+		 * cleared out the corresponding AMR bits
+		 */
+		pkey_set_rights(rest_pkey, rights);
+	}
+
+	/* Free restrictive pkey */
+	sys_pkey_free(rest_pkey);
+
+	return NULL;
+}
+
+static void reset_pkeys(unsigned long rights)
+{
+	int pkeys[NR_PKEYS], i;
+
+	/* Exhaustively allocate all available pkeys */
+	for (i = 0; i < NR_PKEYS; i++)
+		pkeys[i] = sys_pkey_alloc(0, rights);
+
+	/* Free all allocated pkeys */
+	for (i = 0; i < NR_PKEYS; i++)
+		sys_pkey_free(pkeys[i]);
+}
+
+static int test(void)
+{
+	pthread_t prot_thread, pacc_thread;
+	struct sigaction act;
+	pthread_attr_t attr;
+	size_t numinsns;
+	struct region r;
+	int ret, i;
+
+	srand(time(NULL));
+	ret = pkeys_unsupported();
+	if (ret)
+		return ret;
+
+	/* Allocate the region */
+	r.size = getpagesize();
+	r.base = mmap(NULL, r.size, PROT_RWX,
+		      MAP_PRIVATE | MAP_ANONYMOUS, -1, 0);
+	FAIL_IF(r.base == MAP_FAILED);
+
+	/*
+	 * Fill the region with no-ops with a branch at the end
+	 * for returning to the caller
+	 */
+	numinsns = r.size / sizeof(r.base[0]);
+	for (i = 0; i < numinsns - 1; i++)
+		r.base[i] = PPC_INST_NOP;
+	r.base[i] = PPC_INST_BLR;
+
+	/* Setup SIGSEGV handler */
+	act.sa_handler = 0;
+	act.sa_sigaction = segv_handler;
+	FAIL_IF(sigprocmask(SIG_SETMASK, 0, &act.sa_mask) != 0);
+	act.sa_flags = SA_SIGINFO;
+	act.sa_restorer = 0;
+	FAIL_IF(sigaction(SIGSEGV, &act, NULL) != 0);
+
+	/*
+	 * For these tests, the parent process should clear all bits of
+	 * AMR and IAMR, i.e. impose no restrictions, for all available
+	 * pkeys. This will be the base for the initial AMR and IAMR
+	 * values for all the test thread pairs.
+	 *
+	 * If the AMR and IAMR bits of all available pkeys are cleared
+	 * before running the tests and a fault is generated when
+	 * attempting to read, write or execute instructions from a
+	 * pkey protected region, the pkey responsible for this must be
+	 * the one from the protect-and-access thread since the other
+	 * one is fully permissive. Despite that, if the pkey reported
+	 * by siginfo is not the restrictive pkey, then there must be a
+	 * kernel bug.
+	 */
+	reset_pkeys(0);
+
+	/* Setup barrier for protect and protect-and-access threads */
+	FAIL_IF(pthread_attr_init(&attr) != 0);
+	FAIL_IF(pthread_barrier_init(&iteration_barrier, NULL, 2) != 0);
+
+	/* Setup and start protect and protect-and-read threads */
+	puts("starting thread pair (protect, protect-and-read)");
+	r.rights = PKEY_DISABLE_ACCESS;
+	FAIL_IF(pthread_create(&prot_thread, &attr, &protect, &r) != 0);
+	FAIL_IF(pthread_create(&pacc_thread, &attr, &protect_access, &r) != 0);
+	FAIL_IF(pthread_join(prot_thread, NULL) != 0);
+	FAIL_IF(pthread_join(pacc_thread, NULL) != 0);
+
+	/* Setup and start protect and protect-and-write threads */
+	puts("starting thread pair (protect, protect-and-write)");
+	r.rights = PKEY_DISABLE_WRITE;
+	FAIL_IF(pthread_create(&prot_thread, &attr, &protect, &r) != 0);
+	FAIL_IF(pthread_create(&pacc_thread, &attr, &protect_access, &r) != 0);
+	FAIL_IF(pthread_join(prot_thread, NULL) != 0);
+	FAIL_IF(pthread_join(pacc_thread, NULL) != 0);
+
+	/* Setup and start protect and protect-and-execute threads */
+	puts("starting thread pair (protect, protect-and-execute)");
+	r.rights = PKEY_DISABLE_EXECUTE;
+	FAIL_IF(pthread_create(&prot_thread, &attr, &protect, &r) != 0);
+	FAIL_IF(pthread_create(&pacc_thread, &attr, &protect_access, &r) != 0);
+	FAIL_IF(pthread_join(prot_thread, NULL) != 0);
+	FAIL_IF(pthread_join(pacc_thread, NULL) != 0);
+
+	/* Cleanup */
+	FAIL_IF(pthread_attr_destroy(&attr) != 0);
+	FAIL_IF(pthread_barrier_destroy(&iteration_barrier) != 0);
+	munmap(r.base, r.size);
+
+	return 0;
+}
+
+int main(void)
+{
+	return test_harness(test, "pkey_siginfo");
+}
diff --git a/tools/testing/selftests/powerpc/mm/prot_sao.c b/tools/testing/selftests/powerpc/mm/prot_sao.c
new file mode 100644
index 000000000..30b71b1d7
--- /dev/null
+++ b/tools/testing/selftests/powerpc/mm/prot_sao.c
@@ -0,0 +1,48 @@
+// SPDX-License-Identifier: GPL-2.0-only
+/*
+ * Copyright 2016, Michael Ellerman, IBM Corp.
+ */
+
+#include <stdio.h>
+#include <stdlib.h>
+#include <string.h>
+#include <sys/mman.h>
+#include <unistd.h>
+
+#include <asm/cputable.h>
+
+#include "utils.h"
+
+#define SIZE (64 * 1024)
+
+int test_prot_sao(void)
+{
+	char *p;
+
+	/*
+	 * SAO was introduced in 2.06 and removed in 3.1. It's disabled in
+	 * guests/LPARs by default, so also skip if we are running in a guest.
+	 */
+	SKIP_IF(!have_hwcap(PPC_FEATURE_ARCH_2_06) ||
+		have_hwcap2(PPC_FEATURE2_ARCH_3_1) ||
+		access("/proc/device-tree/rtas/ibm,hypertas-functions", F_OK) == 0);
+
+	/*
+	 * Ensure we can ask for PROT_SAO.
+	 * We can't really verify that it does the right thing, but at least we
+	 * confirm the kernel will accept it.
+	 */
+	p = mmap(NULL, SIZE, PROT_READ | PROT_WRITE | PROT_SAO,
+		 MAP_ANONYMOUS | MAP_PRIVATE, -1, 0);
+	FAIL_IF(p == MAP_FAILED);
+
+	/* Write to the mapping, to at least cause a fault */
+	memset(p, 0xaa, SIZE);
+
+	return 0;
+}
+
+int main(void)
+{
+	return test_harness(test_prot_sao, "prot-sao");
+}
diff --git a/tools/testing/selftests/powerpc/mm/segv_errors.c b/tools/testing/selftests/powerpc/mm/segv_errors.c
new file mode 100644
index 000000000..06ae76ee3
--- /dev/null
+++ b/tools/testing/selftests/powerpc/mm/segv_errors.c
@@ -0,0 +1,78 @@
+// SPDX-License-Identifier: GPL-2.0
+
+/*
+ * Copyright 2017 John Sperbeck
+ *
+ * Test that an access to a mapped but inaccessible area causes a SEGV and
+ * reports si_code == SEGV_ACCERR.
+ */
+
+#include <stdbool.h>
+#include <stdio.h>
+#include <stdlib.h>
+#include <string.h>
+#include <unistd.h>
+#include <signal.h>
+#include <sys/mman.h>
+#include <assert.h>
+#include <ucontext.h>
+
+#include "utils.h"
+
+static bool faulted;
+static int si_code;
+
+static void segv_handler(int n, siginfo_t *info, void *ctxt_v)
+{
+	ucontext_t *ctxt = (ucontext_t *)ctxt_v;
+	struct pt_regs *regs = ctxt->uc_mcontext.regs;
+
+	faulted = true;
+	si_code = info->si_code;
+	regs->nip += 4;
+}
+
+int test_segv_errors(void)
+{
+	struct sigaction act = {
+		.sa_sigaction = segv_handler,
+		.sa_flags = SA_SIGINFO,
+	};
+	char c, *p = NULL;
+
+	p = mmap(NULL, getpagesize(), 0, MAP_PRIVATE|MAP_ANONYMOUS, -1, 0);
+	FAIL_IF(p == MAP_FAILED);
+
+	FAIL_IF(sigaction(SIGSEGV, &act, NULL) != 0);
+
+	faulted = false;
+	si_code = 0;
+
+	/*
+	 * We just need a compiler barrier, but mb() works and has the nice
+	 * property of being easy to spot in the disassembly.
+	 */
+	mb();
+	c = *p;
+	mb();
+
+	FAIL_IF(!faulted);
+	FAIL_IF(si_code != SEGV_ACCERR);
+
+	faulted = false;
+	si_code = 0;
+
+	mb();
+	*p = c;
+	mb();
+
+	FAIL_IF(!faulted);
+	FAIL_IF(si_code != SEGV_ACCERR);
+
+	return 0;
+}
+
+int main(void)
+{
+	return test_harness(test_segv_errors, "segv_errors");
+}
diff --git a/tools/testing/selftests/powerpc/mm/stack_expansion_ldst.c b/tools/testing/selftests/powerpc/mm/stack_expansion_ldst.c
new file mode 100644
index 000000000..ed9143990
--- /dev/null
+++ b/tools/testing/selftests/powerpc/mm/stack_expansion_ldst.c
@@ -0,0 +1,202 @@
+// SPDX-License-Identifier: GPL-2.0
+/*
+ * Test that loads/stores expand the stack segment, or trigger a SEGV, in
+ * various conditions.
+ *
+ * Based on test code by Tom Lane.
+ */
+
+#undef NDEBUG
+#include <assert.h>
+
+#include <err.h>
+#include <errno.h>
+#include <stdio.h>
+#include <signal.h>
+#include <stdlib.h>
+#include <string.h>
+#include <sys/resource.h>
+#include <sys/time.h>
+#include <sys/types.h>
+#include <sys/wait.h>
+#include <unistd.h>
+
+#define _KB (1024)
+#define _MB (1024 * 1024)
+
+volatile char *stack_top_ptr;
+volatile unsigned long stack_top_sp;
+volatile char c;
+
+enum access_type {
+	LOAD,
+	STORE,
+};
+
+/*
+ * Consume stack until the stack pointer is below @target_sp, then do an access
+ * (load or store) at offset @delta from either the base of the stack or the
+ * current stack pointer.
+ */
+__attribute__ ((noinline))
+int consume_stack(unsigned long target_sp, unsigned long stack_high, int delta, enum access_type type)
+{
+	unsigned long target;
+	char stack_cur;
+
+	if ((unsigned long)&stack_cur > target_sp)
+		return consume_stack(target_sp, stack_high, delta, type);
+	else {
+		// We don't really need this, but without it GCC might not
+		// generate a recursive call above.
+		stack_top_ptr = &stack_cur;
+
+#ifdef __powerpc__
+		asm volatile ("mr %[sp], %%r1" : [sp] "=r" (stack_top_sp));
+#else
+		asm volatile ("mov %%rsp, %[sp]" : [sp] "=r" (stack_top_sp));
+#endif
+		target = stack_high - delta + 1;
+		volatile char *p = (char *)target;
+
+		if (type == STORE)
+			*p = c;
+		else
+			c = *p;
+
+		// Do something to prevent the stack frame being popped prior to
+		// our access above.
+		getpid();
+	}
+
+	return 0;
+}
+
+static int search_proc_maps(char *needle, unsigned long *low, unsigned long *high)
+{
+	unsigned long start, end;
+	static char buf[4096];
+	char name[128];
+	FILE *f;
+	int rc;
+
+	f = fopen("/proc/self/maps", "r");
+	if (!f) {
+		perror("fopen");
+		return -1;
+	}
+
+	while (fgets(buf, sizeof(buf), f)) {
+		rc = sscanf(buf, "%lx-%lx %*c%*c%*c%*c %*x %*d:%*d %*d %127s\n",
+			    &start, &end, name);
+		if (rc == 2)
+			continue;
+
+		if (rc != 3) {
+			printf("sscanf errored\n");
+			rc = -1;
+			break;
+		}
+
+		if (strstr(name, needle)) {
+			*low = start;
+			*high = end - 1;
+			rc = 0;
+			break;
+		}
+	}
+
+	fclose(f);
+
+	return rc;
+}
+
+int child(unsigned int stack_used, int delta, enum access_type type)
+{
+	unsigned long low, stack_high;
+
+	assert(search_proc_maps("[stack]", &low, &stack_high) == 0);
+
+	assert(consume_stack(stack_high - stack_used, stack_high, delta, type) == 0);
+
+	printf("Access OK: %s delta %-7d used size 0x%06x stack high 0x%lx top_ptr %p top sp 0x%lx actual used 0x%lx\n",
+	       type == LOAD ? "load" : "store", delta, stack_used, stack_high,
+	       stack_top_ptr, stack_top_sp, stack_high - stack_top_sp + 1);
+
+	return 0;
+}
+
+static int test_one(unsigned int stack_used, int delta, enum access_type type)
+{
+	pid_t pid;
+	int rc;
+
+	pid = fork();
+	if (pid == 0)
+		exit(child(stack_used, delta, type));
+
+	assert(waitpid(pid, &rc, 0) != -1);
+
+	if (WIFEXITED(rc) && WEXITSTATUS(rc) == 0)
+		return 0;
+
+	// We don't expect a non-zero exit that's not a signal
+	assert(!WIFEXITED(rc));
+
+	printf("Faulted:   %s delta %-7d used size 0x%06x signal %d\n",
+	       type == LOAD ? "load" : "store", delta, stack_used,
+	       WTERMSIG(rc));
+
+	return 1;
+}
+
+// This is fairly arbitrary but is well below any of the targets below,
+// so that the delta between the stack pointer and the target is large.
+#define DEFAULT_SIZE	(32 * _KB)
+
+static void test_one_type(enum access_type type, unsigned long page_size, unsigned long rlim_cur)
+{
+	unsigned long delta;
+
+	// We should be able to access anywhere within the rlimit
+	for (delta = page_size; delta <= rlim_cur; delta += page_size)
+		assert(test_one(DEFAULT_SIZE, delta, type) == 0);
+
+	assert(test_one(DEFAULT_SIZE, rlim_cur, type) == 0);
+
+	// But if we go past the rlimit it should fail
+	assert(test_one(DEFAULT_SIZE, rlim_cur + 1, type) != 0);
+}
+
+static int test(void)
+{
+	unsigned long page_size;
+	struct rlimit rlimit;
+
+	page_size = getpagesize();
+	getrlimit(RLIMIT_STACK, &rlimit);
+	printf("Stack rlimit is 0x%lx\n", rlimit.rlim_cur);
+
+	printf("Testing loads ...\n");
+	test_one_type(LOAD, page_size, rlimit.rlim_cur);
+	printf("Testing stores ...\n");
+	test_one_type(STORE, page_size, rlimit.rlim_cur);
+
+	printf("All OK\n");
+
+	return 0;
+}
+
+#ifdef __powerpc__
+#include "utils.h"
+
+int main(void)
+{
+	return test_harness(test, "stack_expansion_ldst");
+}
+#else
+int main(void)
+{
+	return test();
+}
+#endif
diff --git a/tools/testing/selftests/powerpc/mm/stack_expansion_signal.c b/tools/testing/selftests/powerpc/mm/stack_expansion_signal.c
new file mode 100644
index 000000000..c8b32a29e
--- /dev/null
+++ b/tools/testing/selftests/powerpc/mm/stack_expansion_signal.c
@@ -0,0 +1,118 @@
+// SPDX-License-Identifier: GPL-2.0
+/*
+ * Test that signal delivery is able to expand the stack segment without
+ * triggering a SEGV.
+ *
+ * Based on test code by Tom Lane.
+ */
+
+#include <err.h>
+#include <stdio.h>
+#include <stdlib.h>
+#include <signal.h>
+#include <sys/types.h>
+#include <unistd.h>
+
+#include "../pmu/lib.h"
+#include "utils.h"
+
+#define _KB (1024)
+#define _MB (1024 * 1024)
+
+static char *stack_base_ptr;
+static char *stack_top_ptr;
+
+static volatile sig_atomic_t sig_occurred = 0;
+
+static void sigusr1_handler(int signal_arg)
+{
+	sig_occurred = 1;
+}
+
+static int consume_stack(unsigned int stack_size, union pipe write_pipe)
+{
+	char stack_cur;
+
+	if ((stack_base_ptr - &stack_cur) < stack_size)
+		return consume_stack(stack_size, write_pipe);
+	else {
+		stack_top_ptr = &stack_cur;
+
+		FAIL_IF(notify_parent(write_pipe));
+
+		while (!sig_occurred)
+			barrier();
+	}
+
+	return 0;
+}
+
+static int child(unsigned int stack_size, union pipe write_pipe)
+{
+	struct sigaction act;
+	char stack_base;
+
+	act.sa_handler = sigusr1_handler;
+	sigemptyset(&act.sa_mask);
+	act.sa_flags = 0;
+	if (sigaction(SIGUSR1, &act, NULL) < 0)
+		err(1, "sigaction");
+
+	stack_base_ptr = (char *) (((size_t) &stack_base + 65535) & ~65535UL);
+
+	FAIL_IF(consume_stack(stack_size, write_pipe));
+
+	printf("size 0x%06x: OK, stack base %p top %p (%zx used)\n",
+		stack_size, stack_base_ptr, stack_top_ptr,
+		stack_base_ptr - stack_top_ptr);
+
+	return 0;
+}
+
+static int test_one_size(unsigned int stack_size)
+{
+	union pipe read_pipe, write_pipe;
+	pid_t pid;
+
+	FAIL_IF(pipe(read_pipe.fds) == -1);
+	FAIL_IF(pipe(write_pipe.fds) == -1);
+
+	pid = fork();
+	if (pid == 0) {
+		close(read_pipe.read_fd);
+		close(write_pipe.write_fd);
+		exit(child(stack_size, read_pipe));
+	}
+
+	close(read_pipe.write_fd);
+	close(write_pipe.read_fd);
+	FAIL_IF(sync_with_child(read_pipe, write_pipe));
+
+	kill(pid, SIGUSR1);
+
+	FAIL_IF(wait_for_child(pid));
+
+	close(read_pipe.read_fd);
+	close(write_pipe.write_fd);
+
+	return 0;
+}
+
+int test(void)
+{
+	unsigned int i, size;
+
+	// Test with used stack from 1MB - 64K to 1MB + 64K
+	// Increment by 64 to get more coverage of odd sizes
+	for (i = 0; i < (128 * _KB); i += 64) {
+		size = i + (1 * _MB) - (64 * _KB);
+		FAIL_IF(test_one_size(size));
+	}
+
+	return 0;
+}
+
+int main(void)
+{
+	return test_harness(test, "stack_expansion_signal");
+}
diff --git a/tools/testing/selftests/powerpc/mm/subpage_prot.c b/tools/testing/selftests/powerpc/mm/subpage_prot.c
new file mode 100644
index 000000000..3ae77ba93
--- /dev/null
+++ b/tools/testing/selftests/powerpc/mm/subpage_prot.c
@@ -0,0 +1,236 @@
+/*
+ * Copyright IBM Corp.
+ *
+ * This program is free software; you can redistribute it and/or modify it
+ * under the terms of version 2.1 of the GNU Lesser General Public License
+ * as published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it would be useful, but
+ * WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.
+ *
+ */
+
+#include <assert.h>
+#include <errno.h>
+#include <fcntl.h>
+#include <signal.h>
+#include <stdarg.h>
+#include <stdio.h>
+#include <stdlib.h>
+#include <string.h>
+#include <sys/mman.h>
+#include <sys/ptrace.h>
+#include <sys/syscall.h>
+#include <ucontext.h>
+#include <unistd.h>
+
+#include "utils.h"
+
+char *file_name;
+
+int in_test;
+volatile int faulted;
+volatile void *dar;
+int errors;
+
+static void segv(int signum, siginfo_t *info, void *ctxt_v)
+{
+	ucontext_t *ctxt = (ucontext_t *)ctxt_v;
+	struct pt_regs *regs = ctxt->uc_mcontext.regs;
+
+	if (!in_test) {
+		fprintf(stderr, "Segfault outside of test !\n");
+		exit(1);
+	}
+
+	faulted = 1;
+	dar = (void *)regs->dar;
+	regs->nip += 4;
+}
+
+static inline void do_read(const volatile void *addr)
+{
+	int ret;
+
+	asm volatile("lwz %0,0(%1); twi 0,%0,0; isync;\n"
+		     : "=r" (ret) : "r" (addr) : "memory");
+}
+
+static inline void do_write(const volatile void *addr)
+{
+	int val = 0x1234567;
+
+	asm volatile("stw %0,0(%1); sync; \n"
+		     : : "r" (val), "r" (addr) : "memory");
+}
+
+static inline void check_faulted(void *addr, long page, long subpage, int write)
+{
+	int want_fault = (subpage == ((page + 3) % 16));
+
+	if (write)
+		want_fault |= (subpage == ((page + 1) % 16));
+
+	if (faulted != want_fault) {
+		printf("Failed at %p (p=%ld,sp=%ld,w=%d), want=%s, got=%s !\n",
+		       addr, page, subpage, write,
+		       want_fault ? "fault" : "pass",
+		       faulted ? "fault" : "pass");
+		++errors;
+	}
+
+	if (faulted) {
+		if (dar != addr) {
+			printf("Fault expected at %p and happened at %p !\n",
+			       addr, dar);
+		}
+		faulted = 0;
+		asm volatile("sync" : : : "memory");
+	}
+}
+
+static int run_test(void *addr, unsigned long size)
+{
+	unsigned int *map;
+	long i, j, pages, err;
+
+	pages = size / 0x10000;
+	map = malloc(pages * 4);
+	assert(map);
+
+	/*
+	 * for each page, mark subpage i % 16 read only and subpage
+	 * (i + 3) % 16 inaccessible
+	 */
+	for (i = 0; i < pages; i++) {
+		map[i] = (0x40000000 >> (((i + 1) * 2) % 32)) |
+			(0xc0000000 >> (((i + 3) * 2) % 32));
+	}
+
+	err = syscall(__NR_subpage_prot, addr, size, map);
+	if (err) {
+		perror("subpage_perm");
+		return 1;
+	}
+	free(map);
+
+	in_test = 1;
+	errors = 0;
+	for (i = 0; i < pages; i++) {
+		for (j = 0; j < 16; j++, addr += 0x1000) {
+			do_read(addr);
+			check_faulted(addr, i, j, 0);
+			do_write(addr);
+			check_faulted(addr, i, j, 1);
+		}
+	}
+
+	in_test = 0;
+	if (errors) {
+		printf("%d errors detected\n", errors);
+		return 1;
+	}
+
+	return 0;
+}
+
+static int syscall_available(void)
+{
+	int rc;
+
+	errno = 0;
+	rc = syscall(__NR_subpage_prot, 0, 0, 0);
+
+	return rc == 0 || (errno != ENOENT && errno != ENOSYS);
+}
+
+int test_anon(void)
+{
+	unsigned long align;
+	struct sigaction act = {
+		.sa_sigaction = segv,
+		.sa_flags = SA_SIGINFO
+	};
+	void *mallocblock;
+	unsigned long mallocsize;
+
+	SKIP_IF(!syscall_available());
+
+	if (getpagesize() != 0x10000) {
+		fprintf(stderr, "Kernel page size must be 64K!\n");
+		return 1;
+	}
+
+	sigaction(SIGSEGV, &act, NULL);
+
+	mallocsize = 4 * 16 * 1024 * 1024;
+
+	FAIL_IF(posix_memalign(&mallocblock, 64 * 1024, mallocsize));
+
+	align = (unsigned long)mallocblock;
+	if (align & 0xffff)
+		align = (align | 0xffff) + 1;
+
+	mallocblock = (void *)align;
+
+	printf("allocated malloc block of 0x%lx bytes at %p\n",
+	       mallocsize, mallocblock);
+
+	printf("testing malloc block...\n");
+
+	return run_test(mallocblock, mallocsize);
+}
+
+int test_file(void)
+{
+	struct sigaction act = {
+		.sa_sigaction = segv,
+		.sa_flags = SA_SIGINFO
+	};
+	void *fileblock;
+	off_t filesize;
+	int fd;
+
+	SKIP_IF(!syscall_available());
+
+	fd = open(file_name, O_RDWR);
+	if (fd == -1) {
+		perror("failed to open file");
+		return 1;
+	}
+	sigaction(SIGSEGV, &act, NULL);
+
+	filesize = lseek(fd, 0, SEEK_END);
+	if (filesize & 0xffff)
+		filesize &= ~0xfffful;
+
+	fileblock = mmap(NULL, filesize, PROT_READ | PROT_WRITE,
+			 MAP_SHARED, fd, 0);
+	if (fileblock == MAP_FAILED) {
+		perror("failed to map file");
+		return 1;
+	}
+	printf("allocated %s for 0x%lx bytes at %p\n",
+	       file_name, filesize, fileblock);
+
+	printf("testing file map...\n");
+
+	return run_test(fileblock, filesize);
+}
+
+int main(int argc, char *argv[])
+{
+	int rc;
+
+	rc = test_harness(test_anon, "subpage_prot_anon");
+	if (rc)
+		return rc;
+
+	if (argc > 1)
+		file_name = argv[1];
+	else
+		file_name = "tempfile";
+
+	return test_harness(test_file, "subpage_prot_file");
+}
diff --git a/tools/testing/selftests/powerpc/mm/tlbie_test.c b/tools/testing/selftests/powerpc/mm/tlbie_test.c
new file mode 100644
index 000000000..f85a0938a
--- /dev/null
+++ b/tools/testing/selftests/powerpc/mm/tlbie_test.c
@@ -0,0 +1,734 @@
+// SPDX-License-Identifier: GPL-2.0
+
+/*
+ * Copyright 2019, Nick Piggin, Gautham R. Shenoy, Aneesh Kumar K.V, IBM Corp.
+ */
+
+/*
+ *
+ * Test tlbie/mtpidr race. We have 4 threads doing flush/load/compare/store
+ * sequence in a loop. The same threads also rung a context switch task
+ * that does sched_yield() in loop.
+ *
+ * The snapshot thread mark the mmap area PROT_READ in between, make a copy
+ * and copy it back to the original area. This helps us to detect if any
+ * store continued to happen after we marked the memory PROT_READ.
+ */
+
+#define _GNU_SOURCE
+#include <stdio.h>
+#include <sys/mman.h>
+#include <sys/types.h>
+#include <sys/wait.h>
+#include <sys/ipc.h>
+#include <sys/shm.h>
+#include <sys/stat.h>
+#include <sys/time.h>
+#include <linux/futex.h>
+#include <unistd.h>
+#include <asm/unistd.h>
+#include <string.h>
+#include <stdlib.h>
+#include <fcntl.h>
+#include <sched.h>
+#include <time.h>
+#include <stdarg.h>
+#include <sched.h>
+#include <pthread.h>
+#include <signal.h>
+#include <sys/prctl.h>
+
+static inline void dcbf(volatile unsigned int *addr)
+{
+	__asm__ __volatile__ ("dcbf %y0; sync" : : "Z"(*(unsigned char *)addr) : "memory");
+}
+
+static void err_msg(char *msg)
+{
+
+	time_t now;
+	time(&now);
+	printf("=================================\n");
+	printf("    Error: %s\n", msg);
+	printf("    %s", ctime(&now));
+	printf("=================================\n");
+	exit(1);
+}
+
+static char *map1;
+static char *map2;
+static pid_t rim_process_pid;
+
+/*
+ * A "rim-sequence" is defined to be the sequence of the following
+ * operations performed on a memory word:
+ *	1) FLUSH the contents of that word.
+ *	2) LOAD the contents of that word.
+ *	3) COMPARE the contents of that word with the content that was
+ *	           previously stored at that word
+ *	4) STORE new content into that word.
+ *
+ * The threads in this test that perform the rim-sequence are termed
+ * as rim_threads.
+ */
+
+/*
+ * A "corruption" is defined to be the failed COMPARE operation in a
+ * rim-sequence.
+ *
+ * A rim_thread that detects a corruption informs about it to all the
+ * other rim_threads, and the mem_snapshot thread.
+ */
+static volatile unsigned int corruption_found;
+
+/*
+ * This defines the maximum number of rim_threads in this test.
+ *
+ * The THREAD_ID_BITS denote the number of bits required
+ * to represent the thread_ids [0..MAX_THREADS - 1].
+ * We are being a bit paranoid here and set it to 8 bits,
+ * though 6 bits suffice.
+ *
+ */
+#define MAX_THREADS 		64
+#define THREAD_ID_BITS		8
+#define THREAD_ID_MASK		((1 << THREAD_ID_BITS) - 1)
+static unsigned int rim_thread_ids[MAX_THREADS];
+static pthread_t rim_threads[MAX_THREADS];
+
+
+/*
+ * Each rim_thread works on an exclusive "chunk" of size
+ * RIM_CHUNK_SIZE.
+ *
+ * The ith rim_thread works on the ith chunk.
+ *
+ * The ith chunk begins at
+ * map1 + (i * RIM_CHUNK_SIZE)
+ */
+#define RIM_CHUNK_SIZE  	1024
+#define BITS_PER_BYTE 		8
+#define WORD_SIZE     		(sizeof(unsigned int))
+#define WORD_BITS		(WORD_SIZE * BITS_PER_BYTE)
+#define WORDS_PER_CHUNK		(RIM_CHUNK_SIZE/WORD_SIZE)
+
+static inline char *compute_chunk_start_addr(unsigned int thread_id)
+{
+	char *chunk_start;
+
+	chunk_start = (char *)((unsigned long)map1 +
+			       (thread_id * RIM_CHUNK_SIZE));
+
+	return chunk_start;
+}
+
+/*
+ * The "word-offset" of a word-aligned address inside a chunk, is
+ * defined to be the number of words that precede the address in that
+ * chunk.
+ *
+ * WORD_OFFSET_BITS denote the number of bits required to represent
+ * the word-offsets of all the word-aligned addresses of a chunk.
+ */
+#define WORD_OFFSET_BITS	(__builtin_ctz(WORDS_PER_CHUNK))
+#define WORD_OFFSET_MASK	((1 << WORD_OFFSET_BITS) - 1)
+
+static inline unsigned int compute_word_offset(char *start, unsigned int *addr)
+{
+	unsigned int delta_bytes, ret;
+	delta_bytes = (unsigned long)addr - (unsigned long)start;
+
+	ret = delta_bytes/WORD_SIZE;
+
+	return ret;
+}
+
+/*
+ * A "sweep" is defined to be the sequential execution of the
+ * rim-sequence by a rim_thread on its chunk one word at a time,
+ * starting from the first word of its chunk and ending with the last
+ * word of its chunk.
+ *
+ * Each sweep of a rim_thread is uniquely identified by a sweep_id.
+ * SWEEP_ID_BITS denote the number of bits required to represent
+ * the sweep_ids of rim_threads.
+ *
+ * As to why SWEEP_ID_BITS are computed as a function of THREAD_ID_BITS,
+ * WORD_OFFSET_BITS, and WORD_BITS, see the "store-pattern" below.
+ */
+#define SWEEP_ID_BITS		(WORD_BITS - (THREAD_ID_BITS + WORD_OFFSET_BITS))
+#define SWEEP_ID_MASK		((1 << SWEEP_ID_BITS) - 1)
+
+/*
+ * A "store-pattern" is the word-pattern that is stored into a word
+ * location in the 4)STORE step of the rim-sequence.
+ *
+ * In the store-pattern, we shall encode:
+ *
+ *      - The thread-id of the rim_thread performing the store
+ *        (The most significant THREAD_ID_BITS)
+ *
+ *      - The word-offset of the address into which the store is being
+ *        performed (The next WORD_OFFSET_BITS)
+ *
+ *      - The sweep_id of the current sweep in which the store is
+ *        being performed. (The lower SWEEP_ID_BITS)
+ *
+ * Store Pattern: 32 bits
+ * |------------------|--------------------|---------------------------------|
+ * |    Thread id     |  Word offset       |         sweep_id                |
+ * |------------------|--------------------|---------------------------------|
+ *    THREAD_ID_BITS     WORD_OFFSET_BITS          SWEEP_ID_BITS
+ *
+ * In the store pattern, the (Thread-id + Word-offset) uniquely identify the
+ * address to which the store is being performed i.e,
+ *    address == map1 +
+ *              (Thread-id * RIM_CHUNK_SIZE) + (Word-offset * WORD_SIZE)
+ *
+ * And the sweep_id in the store pattern identifies the time when the
+ * store was performed by the rim_thread.
+ *
+ * We shall use this property in the 3)COMPARE step of the
+ * rim-sequence.
+ */
+#define SWEEP_ID_SHIFT	0
+#define WORD_OFFSET_SHIFT	(SWEEP_ID_BITS)
+#define THREAD_ID_SHIFT		(WORD_OFFSET_BITS + SWEEP_ID_BITS)
+
+/*
+ * Compute the store pattern for a given thread with id @tid, at
+ * location @addr in the sweep identified by @sweep_id
+ */
+static inline unsigned int compute_store_pattern(unsigned int tid,
+						 unsigned int *addr,
+						 unsigned int sweep_id)
+{
+	unsigned int ret = 0;
+	char *start = compute_chunk_start_addr(tid);
+	unsigned int word_offset = compute_word_offset(start, addr);
+
+	ret += (tid & THREAD_ID_MASK) << THREAD_ID_SHIFT;
+	ret += (word_offset & WORD_OFFSET_MASK) << WORD_OFFSET_SHIFT;
+	ret += (sweep_id & SWEEP_ID_MASK) << SWEEP_ID_SHIFT;
+	return ret;
+}
+
+/* Extract the thread-id from the given store-pattern */
+static inline unsigned int extract_tid(unsigned int pattern)
+{
+	unsigned int ret;
+
+	ret = (pattern >> THREAD_ID_SHIFT) & THREAD_ID_MASK;
+	return ret;
+}
+
+/* Extract the word-offset from the given store-pattern */
+static inline unsigned int extract_word_offset(unsigned int pattern)
+{
+	unsigned int ret;
+
+	ret = (pattern >> WORD_OFFSET_SHIFT) & WORD_OFFSET_MASK;
+
+	return ret;
+}
+
+/* Extract the sweep-id from the given store-pattern */
+static inline unsigned int extract_sweep_id(unsigned int pattern)
+
+{
+	unsigned int ret;
+
+	ret = (pattern >> SWEEP_ID_SHIFT) & SWEEP_ID_MASK;
+
+	return ret;
+}
+
+/************************************************************
+ *                                                          *
+ *          Logging the output of the verification          *
+ *                                                          *
+ ************************************************************/
+#define LOGDIR_NAME_SIZE 100
+static char logdir[LOGDIR_NAME_SIZE];
+
+static FILE *fp[MAX_THREADS];
+static const char logfilename[] ="Thread-%02d-Chunk";
+
+static inline void start_verification_log(unsigned int tid,
+					  unsigned int *addr,
+					  unsigned int cur_sweep_id,
+					  unsigned int prev_sweep_id)
+{
+	FILE *f;
+	char logfile[30];
+	char path[LOGDIR_NAME_SIZE + 30];
+	char separator[2] = "/";
+	char *chunk_start = compute_chunk_start_addr(tid);
+	unsigned int size = RIM_CHUNK_SIZE;
+
+	sprintf(logfile, logfilename, tid);
+	strcpy(path, logdir);
+	strcat(path, separator);
+	strcat(path, logfile);
+	f = fopen(path, "w");
+
+	if (!f) {
+		err_msg("Unable to create logfile\n");
+	}
+
+	fp[tid] = f;
+
+	fprintf(f, "----------------------------------------------------------\n");
+	fprintf(f, "PID                = %d\n", rim_process_pid);
+	fprintf(f, "Thread id          = %02d\n", tid);
+	fprintf(f, "Chunk Start Addr   = 0x%016lx\n", (unsigned long)chunk_start);
+	fprintf(f, "Chunk Size         = %d\n", size);
+	fprintf(f, "Next Store Addr    = 0x%016lx\n", (unsigned long)addr);
+	fprintf(f, "Current sweep-id   = 0x%08x\n", cur_sweep_id);
+	fprintf(f, "Previous sweep-id  = 0x%08x\n", prev_sweep_id);
+	fprintf(f, "----------------------------------------------------------\n");
+}
+
+static inline void log_anamoly(unsigned int tid, unsigned int *addr,
+			       unsigned int expected, unsigned int observed)
+{
+	FILE *f = fp[tid];
+
+	fprintf(f, "Thread %02d: Addr 0x%lx: Expected 0x%x, Observed 0x%x\n",
+	        tid, (unsigned long)addr, expected, observed);
+	fprintf(f, "Thread %02d: Expected Thread id   = %02d\n", tid, extract_tid(expected));
+	fprintf(f, "Thread %02d: Observed Thread id   = %02d\n", tid, extract_tid(observed));
+	fprintf(f, "Thread %02d: Expected Word offset = %03d\n", tid, extract_word_offset(expected));
+	fprintf(f, "Thread %02d: Observed Word offset = %03d\n", tid, extract_word_offset(observed));
+	fprintf(f, "Thread %02d: Expected sweep-id    = 0x%x\n", tid, extract_sweep_id(expected));
+	fprintf(f, "Thread %02d: Observed sweep-id    = 0x%x\n", tid, extract_sweep_id(observed));
+	fprintf(f, "----------------------------------------------------------\n");
+}
+
+static inline void end_verification_log(unsigned int tid, unsigned nr_anamolies)
+{
+	FILE *f = fp[tid];
+	char logfile[30];
+	char path[LOGDIR_NAME_SIZE + 30];
+	char separator[] = "/";
+
+	fclose(f);
+
+	if (nr_anamolies == 0) {
+		remove(path);
+		return;
+	}
+
+	sprintf(logfile, logfilename, tid);
+	strcpy(path, logdir);
+	strcat(path, separator);
+	strcat(path, logfile);
+
+	printf("Thread %02d chunk has %d corrupted words. For details check %s\n",
+		tid, nr_anamolies, path);
+}
+
+/*
+ * When a COMPARE step of a rim-sequence fails, the rim_thread informs
+ * everyone else via the shared_memory pointed to by
+ * corruption_found variable. On seeing this, every thread verifies the
+ * content of its chunk as follows.
+ *
+ * Suppose a thread identified with @tid was about to store (but not
+ * yet stored) to @next_store_addr in its current sweep identified
+ * @cur_sweep_id. Let @prev_sweep_id indicate the previous sweep_id.
+ *
+ * This implies that for all the addresses @addr < @next_store_addr,
+ * Thread @tid has already performed a store as part of its current
+ * sweep. Hence we expect the content of such @addr to be:
+ *    |-------------------------------------------------|
+ *    | tid   | word_offset(addr) |    cur_sweep_id     |
+ *    |-------------------------------------------------|
+ *
+ * Since Thread @tid is yet to perform stores on address
+ * @next_store_addr and above, we expect the content of such an
+ * address @addr to be:
+ *    |-------------------------------------------------|
+ *    | tid   | word_offset(addr) |    prev_sweep_id    |
+ *    |-------------------------------------------------|
+ *
+ * The verifier function @verify_chunk does this verification and logs
+ * any anamolies that it finds.
+ */
+static void verify_chunk(unsigned int tid, unsigned int *next_store_addr,
+		  unsigned int cur_sweep_id,
+		  unsigned int prev_sweep_id)
+{
+	unsigned int *iter_ptr;
+	unsigned int size = RIM_CHUNK_SIZE;
+	unsigned int expected;
+	unsigned int observed;
+	char *chunk_start = compute_chunk_start_addr(tid);
+
+	int nr_anamolies = 0;
+
+	start_verification_log(tid, next_store_addr,
+			       cur_sweep_id, prev_sweep_id);
+
+	for (iter_ptr = (unsigned int *)chunk_start;
+	     (unsigned long)iter_ptr < (unsigned long)chunk_start + size;
+	     iter_ptr++) {
+		unsigned int expected_sweep_id;
+
+		if (iter_ptr < next_store_addr) {
+			expected_sweep_id = cur_sweep_id;
+		} else {
+			expected_sweep_id = prev_sweep_id;
+		}
+
+		expected = compute_store_pattern(tid, iter_ptr, expected_sweep_id);
+
+		dcbf((volatile unsigned int*)iter_ptr); //Flush before reading
+		observed = *iter_ptr;
+
+	        if (observed != expected) {
+			nr_anamolies++;
+			log_anamoly(tid, iter_ptr, expected, observed);
+		}
+	}
+
+	end_verification_log(tid, nr_anamolies);
+}
+
+static void set_pthread_cpu(pthread_t th, int cpu)
+{
+	cpu_set_t run_cpu_mask;
+	struct sched_param param;
+
+	CPU_ZERO(&run_cpu_mask);
+	CPU_SET(cpu, &run_cpu_mask);
+	pthread_setaffinity_np(th, sizeof(cpu_set_t), &run_cpu_mask);
+
+	param.sched_priority = 1;
+	if (0 && sched_setscheduler(0, SCHED_FIFO, &param) == -1) {
+		/* haven't reproduced with this setting, it kills random preemption which may be a factor */
+		fprintf(stderr, "could not set SCHED_FIFO, run as root?\n");
+	}
+}
+
+static void set_mycpu(int cpu)
+{
+	cpu_set_t run_cpu_mask;
+	struct sched_param param;
+
+	CPU_ZERO(&run_cpu_mask);
+	CPU_SET(cpu, &run_cpu_mask);
+	sched_setaffinity(0, sizeof(cpu_set_t), &run_cpu_mask);
+
+	param.sched_priority = 1;
+	if (0 && sched_setscheduler(0, SCHED_FIFO, &param) == -1) {
+		fprintf(stderr, "could not set SCHED_FIFO, run as root?\n");
+	}
+}
+
+static volatile int segv_wait;
+
+static void segv_handler(int signo, siginfo_t *info, void *extra)
+{
+	while (segv_wait) {
+		sched_yield();
+	}
+
+}
+
+static void set_segv_handler(void)
+{
+	struct sigaction sa;
+
+	sa.sa_flags = SA_SIGINFO;
+	sa.sa_sigaction = segv_handler;
+
+	if (sigaction(SIGSEGV, &sa, NULL) == -1) {
+		perror("sigaction");
+		exit(EXIT_FAILURE);
+	}
+}
+
+int timeout = 0;
+/*
+ * This function is executed by every rim_thread.
+ *
+ * This function performs sweeps over the exclusive chunks of the
+ * rim_threads executing the rim-sequence one word at a time.
+ */
+static void *rim_fn(void *arg)
+{
+	unsigned int tid = *((unsigned int *)arg);
+
+	int size = RIM_CHUNK_SIZE;
+	char *chunk_start = compute_chunk_start_addr(tid);
+
+	unsigned int prev_sweep_id;
+	unsigned int cur_sweep_id = 0;
+
+	/* word access */
+	unsigned int pattern = cur_sweep_id;
+	unsigned int *pattern_ptr = &pattern;
+	unsigned int *w_ptr, read_data;
+
+	set_segv_handler();
+
+	/*
+	 * Let us initialize the chunk:
+	 *
+	 * Each word-aligned address addr in the chunk,
+	 * is initialized to :
+	 *    |-------------------------------------------------|
+	 *    | tid   | word_offset(addr) |         0           |
+	 *    |-------------------------------------------------|
+	 */
+	for (w_ptr = (unsigned int *)chunk_start;
+	     (unsigned long)w_ptr < (unsigned long)(chunk_start) + size;
+	     w_ptr++) {
+
+		*pattern_ptr = compute_store_pattern(tid, w_ptr, cur_sweep_id);
+		*w_ptr = *pattern_ptr;
+	}
+
+	while (!corruption_found && !timeout) {
+		prev_sweep_id = cur_sweep_id;
+		cur_sweep_id = cur_sweep_id + 1;
+
+		for (w_ptr = (unsigned int *)chunk_start;
+		     (unsigned long)w_ptr < (unsigned long)(chunk_start) + size;
+		     w_ptr++)  {
+			unsigned int old_pattern;
+
+			/*
+			 * Compute the pattern that we would have
+			 * stored at this location in the previous
+			 * sweep.
+			 */
+			old_pattern = compute_store_pattern(tid, w_ptr, prev_sweep_id);
+
+			/*
+			 * FLUSH:Ensure that we flush the contents of
+			 *       the cache before loading
+			 */
+			dcbf((volatile unsigned int*)w_ptr); //Flush
+
+			/* LOAD: Read the value */
+			read_data = *w_ptr; //Load
+
+			/*
+			 * COMPARE: Is it the same as what we had stored
+			 *          in the previous sweep ? It better be!
+			 */
+			if (read_data != old_pattern) {
+				/* No it isn't! Tell everyone */
+				corruption_found = 1;
+			}
+
+			/*
+			 * Before performing a store, let us check if
+			 * any rim_thread has found a corruption.
+			 */
+			if (corruption_found || timeout) {
+				/*
+				 * Yes. Someone (including us!) has found
+				 * a corruption :(
+				 *
+				 * Let us verify that our chunk is
+				 * correct.
+				 */
+				/* But first, let us allow the dust to settle down! */
+				verify_chunk(tid, w_ptr, cur_sweep_id, prev_sweep_id);
+
+				return 0;
+			}
+
+			/*
+			 * Compute the new pattern that we are going
+			 * to write to this location
+			 */
+			*pattern_ptr = compute_store_pattern(tid, w_ptr, cur_sweep_id);
+
+			/*
+			 * STORE: Now let us write this pattern into
+			 *        the location
+			 */
+			*w_ptr = *pattern_ptr;
+		}
+	}
+
+	return NULL;
+}
+
+
+static unsigned long start_cpu = 0;
+static unsigned long nrthreads = 4;
+
+static pthread_t mem_snapshot_thread;
+
+static void *mem_snapshot_fn(void *arg)
+{
+	int page_size = getpagesize();
+	size_t size = page_size;
+	void *tmp = malloc(size);
+
+	while (!corruption_found && !timeout) {
+		/* Stop memory migration once corruption is found */
+		segv_wait = 1;
+
+		mprotect(map1, size, PROT_READ);
+
+		/*
+		 * Load from the working alias (map1). Loading from map2
+		 * also fails.
+		 */
+		memcpy(tmp, map1, size);
+
+		/*
+		 * Stores must go via map2 which has write permissions, but
+		 * the corrupted data tends to be seen in the snapshot buffer,
+		 * so corruption does not appear to be introduced at the
+		 * copy-back via map2 alias here.
+		 */
+		memcpy(map2, tmp, size);
+		/*
+		 * Before releasing other threads, must ensure the copy
+		 * back to
+		 */
+		asm volatile("sync" ::: "memory");
+		mprotect(map1, size, PROT_READ|PROT_WRITE);
+		asm volatile("sync" ::: "memory");
+		segv_wait = 0;
+
+		usleep(1); /* This value makes a big difference */
+	}
+
+	return 0;
+}
+
+void alrm_sighandler(int sig)
+{
+	timeout = 1;
+}
+
+int main(int argc, char *argv[])
+{
+	int c;
+	int page_size = getpagesize();
+	time_t now;
+	int i, dir_error;
+	pthread_attr_t attr;
+	key_t shm_key = (key_t) getpid();
+	int shmid, run_time = 20 * 60;
+	struct sigaction sa_alrm;
+
+	snprintf(logdir, LOGDIR_NAME_SIZE,
+		 "/tmp/logdir-%u", (unsigned int)getpid());
+	while ((c = getopt(argc, argv, "r:hn:l:t:")) != -1) {
+		switch(c) {
+		case 'r':
+			start_cpu = strtoul(optarg, NULL, 10);
+			break;
+		case 'h':
+			printf("%s [-r <start_cpu>] [-n <nrthreads>] [-l <logdir>] [-t <timeout>]\n", argv[0]);
+			exit(0);
+			break;
+		case 'n':
+			nrthreads = strtoul(optarg, NULL, 10);
+			break;
+		case 'l':
+			strncpy(logdir, optarg, LOGDIR_NAME_SIZE - 1);
+			break;
+		case 't':
+			run_time = strtoul(optarg, NULL, 10);
+			break;
+		default:
+			printf("invalid option\n");
+			exit(0);
+			break;
+		}
+	}
+
+	if (nrthreads > MAX_THREADS)
+		nrthreads = MAX_THREADS;
+
+	shmid = shmget(shm_key, page_size, IPC_CREAT|0666);
+	if (shmid < 0) {
+		err_msg("Failed shmget\n");
+	}
+
+	map1 = shmat(shmid, NULL, 0);
+	if (map1 == (void *) -1) {
+		err_msg("Failed shmat");
+	}
+
+	map2 = shmat(shmid, NULL, 0);
+	if (map2 == (void *) -1) {
+		err_msg("Failed shmat");
+	}
+
+	dir_error = mkdir(logdir, 0755);
+
+	if (dir_error) {
+		err_msg("Failed mkdir");
+	}
+
+	printf("start_cpu list:%lu\n", start_cpu);
+	printf("number of worker threads:%lu + 1 snapshot thread\n", nrthreads);
+	printf("Allocated address:0x%016lx + secondary map:0x%016lx\n", (unsigned long)map1, (unsigned long)map2);
+	printf("logdir at : %s\n", logdir);
+	printf("Timeout: %d seconds\n", run_time);
+
+	time(&now);
+	printf("=================================\n");
+	printf("     Starting Test\n");
+	printf("     %s", ctime(&now));
+	printf("=================================\n");
+
+	for (i = 0; i < nrthreads; i++) {
+		if (1 && !fork()) {
+			prctl(PR_SET_PDEATHSIG, SIGKILL);
+			set_mycpu(start_cpu + i);
+			for (;;)
+				sched_yield();
+			exit(0);
+		}
+	}
+
+
+	sa_alrm.sa_handler = &alrm_sighandler;
+	sigemptyset(&sa_alrm.sa_mask);
+	sa_alrm.sa_flags = 0;
+
+	if (sigaction(SIGALRM, &sa_alrm, 0) == -1) {
+		err_msg("Failed signal handler registration\n");
+	}
+
+	alarm(run_time);
+
+	pthread_attr_init(&attr);
+	for (i = 0; i < nrthreads; i++) {
+		rim_thread_ids[i] = i;
+		pthread_create(&rim_threads[i], &attr, rim_fn, &rim_thread_ids[i]);
+		set_pthread_cpu(rim_threads[i], start_cpu + i);
+	}
+
+	pthread_create(&mem_snapshot_thread, &attr, mem_snapshot_fn, map1);
+	set_pthread_cpu(mem_snapshot_thread, start_cpu + i);
+
+
+	pthread_join(mem_snapshot_thread, NULL);
+	for (i = 0; i < nrthreads; i++) {
+		pthread_join(rim_threads[i], NULL);
+	}
+
+	if (!timeout) {
+		time(&now);
+		printf("=================================\n");
+		printf("      Data Corruption Detected\n");
+		printf("      %s", ctime(&now));
+		printf("      See logfiles in %s\n", logdir);
+		printf("=================================\n");
+		return 1;
+	}
+	return 0;
+}
diff --git a/tools/testing/selftests/powerpc/mm/wild_bctr.c b/tools/testing/selftests/powerpc/mm/wild_bctr.c
new file mode 100644
index 000000000..f2fa101c5
--- /dev/null
+++ b/tools/testing/selftests/powerpc/mm/wild_bctr.c
@@ -0,0 +1,170 @@
+// SPDX-License-Identifier: GPL-2.0+
+/*
+ * Copyright 2018, Michael Ellerman, IBM Corp.
+ *
+ * Test that an out-of-bounds branch to counter behaves as expected.
+ */
+
+#include <setjmp.h>
+#include <stdio.h>
+#include <stdlib.h>
+#include <string.h>
+#include <sys/mman.h>
+#include <sys/types.h>
+#include <sys/wait.h>
+#include <ucontext.h>
+#include <unistd.h>
+
+#include "utils.h"
+
+
+#define BAD_NIP	0x788c545a18000000ull
+
+static struct pt_regs signal_regs;
+static jmp_buf setjmp_env;
+
+static void save_regs(ucontext_t *ctxt)
+{
+	struct pt_regs *regs = ctxt->uc_mcontext.regs;
+
+	memcpy(&signal_regs, regs, sizeof(signal_regs));
+}
+
+static void segv_handler(int signum, siginfo_t *info, void *ctxt_v)
+{
+	save_regs(ctxt_v);
+	longjmp(setjmp_env, 1);
+}
+
+static void usr2_handler(int signum, siginfo_t *info, void *ctxt_v)
+{
+	save_regs(ctxt_v);
+}
+
+static int ok(void)
+{
+	printf("Everything is OK in here.\n");
+	return 0;
+}
+
+#define REG_POISON	0x5a5a
+#define POISONED_REG(n)	((((unsigned long)REG_POISON) << 48) | ((n) << 32) | \
+			 (((unsigned long)REG_POISON) << 16) | (n))
+
+static inline void poison_regs(void)
+{
+	#define POISON_REG(n)	\
+	  "lis  " __stringify(n) "," __stringify(REG_POISON) ";" \
+	  "addi " __stringify(n) "," __stringify(n) "," __stringify(n) ";" \
+	  "sldi " __stringify(n) "," __stringify(n) ", 32 ;" \
+	  "oris " __stringify(n) "," __stringify(n) "," __stringify(REG_POISON) ";" \
+	  "addi " __stringify(n) "," __stringify(n) "," __stringify(n) ";"
+
+	asm (POISON_REG(15)
+	     POISON_REG(16)
+	     POISON_REG(17)
+	     POISON_REG(18)
+	     POISON_REG(19)
+	     POISON_REG(20)
+	     POISON_REG(21)
+	     POISON_REG(22)
+	     POISON_REG(23)
+	     POISON_REG(24)
+	     POISON_REG(25)
+	     POISON_REG(26)
+	     POISON_REG(27)
+	     POISON_REG(28)
+	     POISON_REG(29)
+	     : // inputs
+	     : // outputs
+	     : "15", "16", "17", "18", "19", "20", "21", "22", "23", "24", "25",
+	       "26", "27", "28", "29"
+	);
+	#undef POISON_REG
+}
+
+static int check_regs(void)
+{
+	unsigned long i;
+
+	for (i = 15; i <= 29; i++)
+		FAIL_IF(signal_regs.gpr[i] != POISONED_REG(i));
+
+	printf("Regs OK\n");
+	return 0;
+}
+
+static void dump_regs(void)
+{
+	for (int i = 0; i < 32; i += 4) {
+		printf("r%02d 0x%016lx  r%02d 0x%016lx  " \
+		       "r%02d 0x%016lx  r%02d 0x%016lx\n",
+		       i, signal_regs.gpr[i],
+		       i+1, signal_regs.gpr[i+1],
+		       i+2, signal_regs.gpr[i+2],
+		       i+3, signal_regs.gpr[i+3]);
+	}
+}
+
+#ifdef _CALL_AIXDESC
+struct opd {
+	unsigned long ip;
+	unsigned long toc;
+	unsigned long env;
+};
+static struct opd bad_opd = {
+	.ip = BAD_NIP,
+};
+#define BAD_FUNC (&bad_opd)
+#else
+#define BAD_FUNC BAD_NIP
+#endif
+
+int test_wild_bctr(void)
+{
+	int (*func_ptr)(void);
+	struct sigaction segv = {
+		.sa_sigaction = segv_handler,
+		.sa_flags = SA_SIGINFO
+	};
+	struct sigaction usr2 = {
+		.sa_sigaction = usr2_handler,
+		.sa_flags = SA_SIGINFO
+	};
+
+	FAIL_IF(sigaction(SIGSEGV, &segv, NULL));
+	FAIL_IF(sigaction(SIGUSR2, &usr2, NULL));
+
+	bzero(&signal_regs, sizeof(signal_regs));
+
+	if (setjmp(setjmp_env) == 0) {
+		func_ptr = ok;
+		func_ptr();
+
+		kill(getpid(), SIGUSR2);
+		printf("Regs before:\n");
+		dump_regs();
+		bzero(&signal_regs, sizeof(signal_regs));
+
+		poison_regs();
+
+		func_ptr = (int (*)(void))BAD_FUNC;
+		func_ptr();
+
+		FAIL_IF(1); /* we didn't segv? */
+	}
+
+	FAIL_IF(signal_regs.nip != BAD_NIP);
+
+	printf("All good - took SEGV as expected branching to 0x%llx\n", BAD_NIP);
+
+	dump_regs();
+	FAIL_IF(check_regs());
+
+	return 0;
+}
+
+int main(void)
+{
+	return test_harness(test_wild_bctr, "wild_bctr");
+}
diff --git a/tools/testing/selftests/powerpc/nx-gzip/99-nx-gzip.rules b/tools/testing/selftests/powerpc/nx-gzip/99-nx-gzip.rules
new file mode 100644
index 000000000..5a7118495
--- /dev/null
+++ b/tools/testing/selftests/powerpc/nx-gzip/99-nx-gzip.rules
@@ -0,0 +1 @@
+SUBSYSTEM=="nxgzip", KERNEL=="nx-gzip", MODE="0666"
diff --git a/tools/testing/selftests/powerpc/nx-gzip/Makefile b/tools/testing/selftests/powerpc/nx-gzip/Makefile
new file mode 100644
index 000000000..640fad6cc
--- /dev/null
+++ b/tools/testing/selftests/powerpc/nx-gzip/Makefile
@@ -0,0 +1,8 @@
+CFLAGS = -O3 -m64 -I./include
+
+TEST_GEN_FILES := gzfht_test gunz_test
+TEST_PROGS := nx-gzip-test.sh
+
+include ../../lib.mk
+
+$(TEST_GEN_FILES): gzip_vas.c
diff --git a/tools/testing/selftests/powerpc/nx-gzip/README b/tools/testing/selftests/powerpc/nx-gzip/README
new file mode 100644
index 000000000..9809dbaa1
--- /dev/null
+++ b/tools/testing/selftests/powerpc/nx-gzip/README
@@ -0,0 +1,45 @@
+Test the nx-gzip function:
+=========================
+
+Verify that following device exists:
+  /dev/crypto/nx-gzip
+If you get a permission error run as sudo or set the device permissions:
+   sudo chmod go+rw /dev/crypto/nx-gzip
+However, chmod may not survive across boots. You may create a udev file such
+as:
+   /etc/udev/rules.d/99-nx-gzip.rules
+
+
+To manually build and run:
+$ gcc -O3 -I./include -o gzfht_test gzfht_test.c gzip_vas.c
+$ gcc -O3 -I./include -o gunz_test gunz_test.c gzip_vas.c
+
+
+Compress any file using Fixed Huffman mode. Output will have a .nx.gz suffix:
+$ ./gzfht_test gzip_vas.c
+file gzip_vas.c read, 6413 bytes
+compressed 6413 to 3124 bytes total, crc32 checksum = abd15e8a
+
+
+Uncompress the previous output. Output will have a .nx.gunzip suffix:
+./gunz_test gzip_vas.c.nx.gz
+gzHeader FLG 0
+00 00 00 00 04 03
+gzHeader MTIME, XFL, OS ignored
+computed checksum abd15e8a isize 0000190d
+stored   checksum abd15e8a isize 0000190d
+decomp is complete: fclose
+
+
+Compare two files:
+$ sha1sum gzip_vas.c.nx.gz.nx.gunzip gzip_vas.c
+bf43e3c0c3651f5f22b6f9784cd9b1eeab4120b6  gzip_vas.c.nx.gz.nx.gunzip
+bf43e3c0c3651f5f22b6f9784cd9b1eeab4120b6  gzip_vas.c
+
+
+Note that the code here are intended for testing the nx-gzip hardware function.
+They are not intended for demonstrating performance or compression ratio.
+By being simplistic these selftests expect to allocate the entire set of source
+and target pages in the memory so it needs enough memory to work.
+For more information and source code consider using:
+https://github.com/libnxz/power-gzip
diff --git a/tools/testing/selftests/powerpc/nx-gzip/gunz_test.c b/tools/testing/selftests/powerpc/nx-gzip/gunz_test.c
new file mode 100644
index 000000000..7c23d3dd7
--- /dev/null
+++ b/tools/testing/selftests/powerpc/nx-gzip/gunz_test.c
@@ -0,0 +1,1028 @@
+// SPDX-License-Identifier: GPL-2.0-or-later
+
+/* P9 gunzip sample code for demonstrating the P9 NX hardware
+ * interface.  Not intended for productive uses or for performance or
+ * compression ratio measurements.  Note also that /dev/crypto/gzip,
+ * VAS and skiboot support are required
+ *
+ * Copyright 2020 IBM Corp.
+ *
+ * Author: Bulent Abali <abali@us.ibm.com>
+ *
+ * https://github.com/libnxz/power-gzip for zlib api and other utils
+ * Definitions of acronyms used here.  See
+ * P9 NX Gzip Accelerator User's Manual for details:
+ * https://github.com/libnxz/power-gzip/blob/develop/doc/power_nx_gzip_um.pdf
+ *
+ * adler/crc: 32 bit checksums appended to stream tail
+ * ce:       completion extension
+ * cpb:      coprocessor parameter block (metadata)
+ * crb:      coprocessor request block (command)
+ * csb:      coprocessor status block (status)
+ * dht:      dynamic huffman table
+ * dde:      data descriptor element (address, length)
+ * ddl:      list of ddes
+ * dh/fh:    dynamic and fixed huffman types
+ * fc:       coprocessor function code
+ * histlen:  history/dictionary length
+ * history:  sliding window of up to 32KB of data
+ * lzcount:  Deflate LZ symbol counts
+ * rembytecnt: remaining byte count
+ * sfbt:     source final block type; last block's type during decomp
+ * spbc:     source processed byte count
+ * subc:     source unprocessed bit count
+ * tebc:     target ending bit count; valid bits in the last byte
+ * tpbc:     target processed byte count
+ * vas:      virtual accelerator switch; the user mode interface
+ */
+
+#define _ISOC11_SOURCE	// For aligned_alloc()
+#define _DEFAULT_SOURCE	// For endian.h
+
+#include <stdio.h>
+#include <stdlib.h>
+#include <string.h>
+#include <unistd.h>
+#include <stdint.h>
+#include <sys/types.h>
+#include <sys/stat.h>
+#include <sys/time.h>
+#include <sys/fcntl.h>
+#include <sys/mman.h>
+#include <endian.h>
+#include <bits/endian.h>
+#include <sys/ioctl.h>
+#include <assert.h>
+#include <errno.h>
+#include <signal.h>
+#include "nxu.h"
+#include "nx.h"
+#include "crb.h"
+
+int nx_dbg;
+FILE *nx_gzip_log;
+
+#define NX_MIN(X, Y) (((X) < (Y))?(X):(Y))
+#define NX_MAX(X, Y) (((X) > (Y))?(X):(Y))
+
+#define GETINPC(X) fgetc(X)
+#define FNAME_MAX 1024
+
+/* fifo queue management */
+#define fifo_used_bytes(used) (used)
+#define fifo_free_bytes(used, len) ((len)-(used))
+/* amount of free bytes in the first and last parts */
+#define fifo_free_first_bytes(cur, used, len)  ((((cur)+(used)) <= (len)) \
+						  ? (len)-((cur)+(used)) : 0)
+#define fifo_free_last_bytes(cur, used, len)   ((((cur)+(used)) <= (len)) \
+						  ? (cur) : (len)-(used))
+/* amount of used bytes in the first and last parts */
+#define fifo_used_first_bytes(cur, used, len)  ((((cur)+(used)) <= (len)) \
+						  ? (used) : (len)-(cur))
+#define fifo_used_last_bytes(cur, used, len)   ((((cur)+(used)) <= (len)) \
+						  ? 0 : ((used)+(cur))-(len))
+/* first and last free parts start here */
+#define fifo_free_first_offset(cur, used)      ((cur)+(used))
+#define fifo_free_last_offset(cur, used, len)  \
+					   fifo_used_last_bytes(cur, used, len)
+/* first and last used parts start here */
+#define fifo_used_first_offset(cur)            (cur)
+#define fifo_used_last_offset(cur)             (0)
+
+const int fifo_in_len = 1<<24;
+const int fifo_out_len = 1<<24;
+const int page_sz = 1<<16;
+const int line_sz = 1<<7;
+const int window_max = 1<<15;
+
+/*
+ * Adds an (address, len) pair to the list of ddes (ddl) and updates
+ * the base dde.  ddl[0] is the only dde in a direct dde which
+ * contains a single (addr,len) pair.  For more pairs, ddl[0] becomes
+ * the indirect (base) dde that points to a list of direct ddes.
+ * See Section 6.4 of the NX-gzip user manual for DDE description.
+ * Addr=NULL, len=0 clears the ddl[0].  Returns the total number of
+ * bytes in ddl.  Caller is responsible for allocting the array of
+ * nx_dde_t *ddl.  If N addresses are required in the scatter-gather
+ * list, the ddl array must have N+1 entries minimum.
+ */
+static inline uint32_t nx_append_dde(struct nx_dde_t *ddl, void *addr,
+					uint32_t len)
+{
+	uint32_t ddecnt;
+	uint32_t bytes;
+
+	if (addr == NULL && len == 0) {
+		clearp_dde(ddl);
+		return 0;
+	}
+
+	NXPRT(fprintf(stderr, "%d: %s addr %p len %x\n", __LINE__, addr,
+			__func__, len));
+
+	/* Number of ddes in the dde list ; == 0 when it is a direct dde */
+	ddecnt = getpnn(ddl, dde_count);
+	bytes = getp32(ddl, ddebc);
+
+	if (ddecnt == 0 && bytes == 0) {
+		/* First dde is unused; make it a direct dde */
+		bytes = len;
+		putp32(ddl, ddebc, bytes);
+		putp64(ddl, ddead, (uint64_t) addr);
+	} else if (ddecnt == 0) {
+		/* Converting direct to indirect dde
+		 * ddl[0] becomes head dde of ddl
+		 * copy direct to indirect first.
+		 */
+		ddl[1] = ddl[0];
+
+		/* Add the new dde next */
+		clear_dde(ddl[2]);
+		put32(ddl[2], ddebc, len);
+		put64(ddl[2], ddead, (uint64_t) addr);
+
+		/* Ddl head points to 2 direct ddes */
+		ddecnt = 2;
+		putpnn(ddl, dde_count, ddecnt);
+		bytes = bytes + len;
+		putp32(ddl, ddebc, bytes);
+		/* Pointer to the first direct dde */
+		putp64(ddl, ddead, (uint64_t) &ddl[1]);
+	} else {
+		/* Append a dde to an existing indirect ddl */
+		++ddecnt;
+		clear_dde(ddl[ddecnt]);
+		put64(ddl[ddecnt], ddead, (uint64_t) addr);
+		put32(ddl[ddecnt], ddebc, len);
+
+		putpnn(ddl, dde_count, ddecnt);
+		bytes = bytes + len;
+		putp32(ddl, ddebc, bytes); /* byte sum of all dde */
+	}
+	return bytes;
+}
+
+/*
+ * Touch specified number of pages represented in number bytes
+ * beginning from the first buffer in a dde list.
+ * Do not touch the pages past buf_sz-th byte's page.
+ *
+ * Set buf_sz = 0 to touch all pages described by the ddep.
+ */
+static int nx_touch_pages_dde(struct nx_dde_t *ddep, long buf_sz, long page_sz,
+				int wr)
+{
+	uint32_t indirect_count;
+	uint32_t buf_len;
+	long total;
+	uint64_t buf_addr;
+	struct nx_dde_t *dde_list;
+	int i;
+
+	assert(!!ddep);
+
+	indirect_count = getpnn(ddep, dde_count);
+
+	NXPRT(fprintf(stderr, "%s dde_count %d request len ", __func__,
+			indirect_count));
+	NXPRT(fprintf(stderr, "0x%lx\n", buf_sz));
+
+	if (indirect_count == 0) {
+		/* Direct dde */
+		buf_len = getp32(ddep, ddebc);
+		buf_addr = getp64(ddep, ddead);
+
+		NXPRT(fprintf(stderr, "touch direct ddebc 0x%x ddead %p\n",
+				buf_len, (void *)buf_addr));
+
+		if (buf_sz == 0)
+			nxu_touch_pages((void *)buf_addr, buf_len, page_sz, wr);
+		else
+			nxu_touch_pages((void *)buf_addr, NX_MIN(buf_len,
+					buf_sz), page_sz, wr);
+
+		return ERR_NX_OK;
+	}
+
+	/* Indirect dde */
+	if (indirect_count > MAX_DDE_COUNT)
+		return ERR_NX_EXCESSIVE_DDE;
+
+	/* First address of the list */
+	dde_list = (struct nx_dde_t *) getp64(ddep, ddead);
+
+	if (buf_sz == 0)
+		buf_sz = getp32(ddep, ddebc);
+
+	total = 0;
+	for (i = 0; i < indirect_count; i++) {
+		buf_len = get32(dde_list[i], ddebc);
+		buf_addr = get64(dde_list[i], ddead);
+		total += buf_len;
+
+		NXPRT(fprintf(stderr, "touch loop len 0x%x ddead %p total ",
+				buf_len, (void *)buf_addr));
+		NXPRT(fprintf(stderr, "0x%lx\n", total));
+
+		/* Touching fewer pages than encoded in the ddebc */
+		if (total > buf_sz) {
+			buf_len = NX_MIN(buf_len, total - buf_sz);
+			nxu_touch_pages((void *)buf_addr, buf_len, page_sz, wr);
+			NXPRT(fprintf(stderr, "touch loop break len 0x%x ",
+				      buf_len));
+			NXPRT(fprintf(stderr, "ddead %p\n", (void *)buf_addr));
+			break;
+		}
+		nxu_touch_pages((void *)buf_addr, buf_len, page_sz, wr);
+	}
+	return ERR_NX_OK;
+}
+
+/*
+ * Src and dst buffers are supplied in scatter gather lists.
+ * NX function code and other parameters supplied in cmdp.
+ */
+static int nx_submit_job(struct nx_dde_t *src, struct nx_dde_t *dst,
+			 struct nx_gzip_crb_cpb_t *cmdp, void *handle)
+{
+	uint64_t csbaddr;
+
+	memset((void *)&cmdp->crb.csb, 0, sizeof(cmdp->crb.csb));
+
+	cmdp->crb.source_dde = *src;
+	cmdp->crb.target_dde = *dst;
+
+	/* Status, output byte count in tpbc */
+	csbaddr = ((uint64_t) &cmdp->crb.csb) & csb_address_mask;
+	put64(cmdp->crb, csb_address, csbaddr);
+
+	/* NX reports input bytes in spbc; cleared */
+	cmdp->cpb.out_spbc_comp_wrap = 0;
+	cmdp->cpb.out_spbc_comp_with_count = 0;
+	cmdp->cpb.out_spbc_decomp = 0;
+
+	/* Clear output */
+	put32(cmdp->cpb, out_crc, INIT_CRC);
+	put32(cmdp->cpb, out_adler, INIT_ADLER);
+
+	/* Submit the crb, the job descriptor, to the accelerator. */
+	return nxu_submit_job(cmdp, handle);
+}
+
+int decompress_file(int argc, char **argv, void *devhandle)
+{
+	FILE *inpf = NULL;
+	FILE *outf = NULL;
+
+	int c, expect, i, cc, rc = 0;
+	char gzfname[FNAME_MAX];
+
+	/* Queuing, file ops, byte counting */
+	char *fifo_in, *fifo_out;
+	int used_in, cur_in, used_out, cur_out, read_sz, n;
+	int first_free, last_free, first_used, last_used;
+	int first_offset, last_offset;
+	int write_sz, free_space, source_sz;
+	int source_sz_estimate, target_sz_estimate;
+	uint64_t last_comp_ratio = 0; /* 1000 max */
+	uint64_t total_out = 0;
+	int is_final, is_eof;
+
+	/* nx hardware */
+	int sfbt, subc, spbc, tpbc, nx_ce, fc, resuming = 0;
+	int history_len = 0;
+	struct nx_gzip_crb_cpb_t cmd, *cmdp;
+	struct nx_dde_t *ddl_in;
+	struct nx_dde_t dde_in[6] __aligned(128);
+	struct nx_dde_t *ddl_out;
+	struct nx_dde_t dde_out[6] __aligned(128);
+	int pgfault_retries;
+
+	/* when using mmap'ed files */
+	off_t input_file_offset;
+
+	if (argc > 2) {
+		fprintf(stderr, "usage: %s <fname> or stdin\n", argv[0]);
+		fprintf(stderr, "    writes to stdout or <fname>.nx.gunzip\n");
+		return -1;
+	}
+
+	if (argc == 1) {
+		inpf = stdin;
+		outf = stdout;
+	} else if (argc == 2) {
+		char w[1024];
+		char *wp;
+
+		inpf = fopen(argv[1], "r");
+		if (inpf == NULL) {
+			perror(argv[1]);
+			return -1;
+		}
+
+		/* Make a new file name to write to.  Ignoring '.gz' */
+		wp = (NULL != (wp = strrchr(argv[1], '/'))) ? (wp+1) : argv[1];
+		strcpy(w, wp);
+		strcat(w, ".nx.gunzip");
+
+		outf = fopen(w, "w");
+		if (outf == NULL) {
+			perror(w);
+			return -1;
+		}
+	}
+
+	/* Decode the gzip header */
+	c = GETINPC(inpf); expect = 0x1f; /* ID1 */
+	if (c != expect)
+		goto err1;
+
+	c = GETINPC(inpf); expect = 0x8b; /* ID2 */
+	if (c != expect)
+		goto err1;
+
+	c = GETINPC(inpf); expect = 0x08; /* CM */
+	if (c != expect)
+		goto err1;
+
+	int flg = GETINPC(inpf); /* FLG */
+
+	if (flg & 0xE0 || flg & 0x4 || flg == EOF)
+		goto err2;
+
+	fprintf(stderr, "gzHeader FLG %x\n", flg);
+
+	/* Read 6 bytes; ignoring the MTIME, XFL, OS fields in this
+	 * sample code.
+	 */
+	for (i = 0; i < 6; i++) {
+		char tmp[10];
+
+		tmp[i] = GETINPC(inpf);
+		if (tmp[i] == EOF)
+			goto err3;
+		fprintf(stderr, "%02x ", tmp[i]);
+		if (i == 5)
+			fprintf(stderr, "\n");
+	}
+	fprintf(stderr, "gzHeader MTIME, XFL, OS ignored\n");
+
+	/* FNAME */
+	if (flg & 0x8) {
+		int k = 0;
+
+		do {
+			c = GETINPC(inpf);
+			if (c == EOF || k >= FNAME_MAX)
+				goto err3;
+			gzfname[k++] = c;
+		} while (c);
+		fprintf(stderr, "gzHeader FNAME: %s\n", gzfname);
+	}
+
+	/* FHCRC */
+	if (flg & 0x2) {
+		c = GETINPC(inpf);
+		if (c == EOF)
+			goto err3;
+		c = GETINPC(inpf);
+		if (c == EOF)
+			goto err3;
+		fprintf(stderr, "gzHeader FHCRC: ignored\n");
+	}
+
+	used_in = cur_in = used_out = cur_out = 0;
+	is_final = is_eof = 0;
+
+	/* Allocate one page larger to prevent page faults due to NX
+	 * overfetching.
+	 * Either do this (char*)(uintptr_t)aligned_alloc or use
+	 * -std=c11 flag to make the int-to-pointer warning go away.
+	 */
+	assert((fifo_in  = (char *)(uintptr_t)aligned_alloc(line_sz,
+				   fifo_in_len + page_sz)) != NULL);
+	assert((fifo_out = (char *)(uintptr_t)aligned_alloc(line_sz,
+				   fifo_out_len + page_sz + line_sz)) != NULL);
+	/* Leave unused space due to history rounding rules */
+	fifo_out = fifo_out + line_sz;
+	nxu_touch_pages(fifo_out, fifo_out_len, page_sz, 1);
+
+	ddl_in  = &dde_in[0];
+	ddl_out = &dde_out[0];
+	cmdp = &cmd;
+	memset(&cmdp->crb, 0, sizeof(cmdp->crb));
+
+read_state:
+
+	/* Read from .gz file */
+
+	NXPRT(fprintf(stderr, "read_state:\n"));
+
+	if (is_eof != 0)
+		goto write_state;
+
+	/* We read in to fifo_in in two steps: first: read in to from
+	 * cur_in to the end of the buffer.  last: if free space wrapped
+	 * around, read from fifo_in offset 0 to offset cur_in.
+	 */
+
+	/* Reset fifo head to reduce unnecessary wrap arounds */
+	cur_in = (used_in == 0) ? 0 : cur_in;
+
+	/* Free space total is reduced by a gap */
+	free_space = NX_MAX(0, fifo_free_bytes(used_in, fifo_in_len)
+			    - line_sz);
+
+	/* Free space may wrap around as first and last */
+	first_free = fifo_free_first_bytes(cur_in, used_in, fifo_in_len);
+	last_free  = fifo_free_last_bytes(cur_in, used_in, fifo_in_len);
+
+	/* Start offsets of the free memory */
+	first_offset = fifo_free_first_offset(cur_in, used_in);
+	last_offset  = fifo_free_last_offset(cur_in, used_in, fifo_in_len);
+
+	/* Reduce read_sz because of the line_sz gap */
+	read_sz = NX_MIN(free_space, first_free);
+	n = 0;
+	if (read_sz > 0) {
+		/* Read in to offset cur_in + used_in */
+		n = fread(fifo_in + first_offset, 1, read_sz, inpf);
+		used_in = used_in + n;
+		free_space = free_space - n;
+		assert(n <= read_sz);
+		if (n != read_sz) {
+			/* Either EOF or error; exit the read loop */
+			is_eof = 1;
+			goto write_state;
+		}
+	}
+
+	/* If free space wrapped around */
+	if (last_free > 0) {
+		/* Reduce read_sz because of the line_sz gap */
+		read_sz = NX_MIN(free_space, last_free);
+		n = 0;
+		if (read_sz > 0) {
+			n = fread(fifo_in + last_offset, 1, read_sz, inpf);
+			used_in = used_in + n;       /* Increase used space */
+			free_space = free_space - n; /* Decrease free space */
+			assert(n <= read_sz);
+			if (n != read_sz) {
+				/* Either EOF or error; exit the read loop */
+				is_eof = 1;
+				goto write_state;
+			}
+		}
+	}
+
+	/* At this point we have used_in bytes in fifo_in with the
+	 * data head starting at cur_in and possibly wrapping around.
+	 */
+
+write_state:
+
+	/* Write decompressed data to output file */
+
+	NXPRT(fprintf(stderr, "write_state:\n"));
+
+	if (used_out == 0)
+		goto decomp_state;
+
+	/* If fifo_out has data waiting, write it out to the file to
+	 * make free target space for the accelerator used bytes in
+	 * the first and last parts of fifo_out.
+	 */
+
+	first_used = fifo_used_first_bytes(cur_out, used_out, fifo_out_len);
+	last_used  = fifo_used_last_bytes(cur_out, used_out, fifo_out_len);
+
+	write_sz = first_used;
+
+	n = 0;
+	if (write_sz > 0) {
+		n = fwrite(fifo_out + cur_out, 1, write_sz, outf);
+		used_out = used_out - n;
+		/* Move head of the fifo */
+		cur_out = (cur_out + n) % fifo_out_len;
+		assert(n <= write_sz);
+		if (n != write_sz) {
+			fprintf(stderr, "error: write\n");
+			rc = -1;
+			goto err5;
+		}
+	}
+
+	if (last_used > 0) { /* If more data available in the last part */
+		write_sz = last_used; /* Keep it here for later */
+		n = 0;
+		if (write_sz > 0) {
+			n = fwrite(fifo_out, 1, write_sz, outf);
+			used_out = used_out - n;
+			cur_out = (cur_out + n) % fifo_out_len;
+			assert(n <= write_sz);
+			if (n != write_sz) {
+				fprintf(stderr, "error: write\n");
+				rc = -1;
+				goto err5;
+			}
+		}
+	}
+
+decomp_state:
+
+	/* NX decompresses input data */
+
+	NXPRT(fprintf(stderr, "decomp_state:\n"));
+
+	if (is_final)
+		goto finish_state;
+
+	/* Address/len lists */
+	clearp_dde(ddl_in);
+	clearp_dde(ddl_out);
+
+	/* FC, CRC, HistLen, Table 6-6 */
+	if (resuming) {
+		/* Resuming a partially decompressed input.
+		 * The key to resume is supplying the 32KB
+		 * dictionary (history) to NX, which is basically
+		 * the last 32KB of output produced.
+		 */
+		fc = GZIP_FC_DECOMPRESS_RESUME;
+
+		cmdp->cpb.in_crc   = cmdp->cpb.out_crc;
+		cmdp->cpb.in_adler = cmdp->cpb.out_adler;
+
+		/* Round up the history size to quadword.  Section 2.10 */
+		history_len = (history_len + 15) / 16;
+		putnn(cmdp->cpb, in_histlen, history_len);
+		history_len = history_len * 16; /* bytes */
+
+		if (history_len > 0) {
+			/* Chain in the history buffer to the DDE list */
+			if (cur_out >= history_len) {
+				nx_append_dde(ddl_in, fifo_out
+					      + (cur_out - history_len),
+					      history_len);
+			} else {
+				nx_append_dde(ddl_in, fifo_out
+					      + ((fifo_out_len + cur_out)
+					      - history_len),
+					      history_len - cur_out);
+				/* Up to 32KB history wraps around fifo_out */
+				nx_append_dde(ddl_in, fifo_out, cur_out);
+			}
+
+		}
+	} else {
+		/* First decompress job */
+		fc = GZIP_FC_DECOMPRESS;
+
+		history_len = 0;
+		/* Writing 0 clears out subc as well */
+		cmdp->cpb.in_histlen = 0;
+		total_out = 0;
+
+		put32(cmdp->cpb, in_crc, INIT_CRC);
+		put32(cmdp->cpb, in_adler, INIT_ADLER);
+		put32(cmdp->cpb, out_crc, INIT_CRC);
+		put32(cmdp->cpb, out_adler, INIT_ADLER);
+
+		/* Assuming 10% compression ratio initially; use the
+		 * most recently measured compression ratio as a
+		 * heuristic to estimate the input and output
+		 * sizes.  If we give too much input, the target buffer
+		 * overflows and NX cycles are wasted, and then we
+		 * must retry with smaller input size.  1000 is 100%.
+		 */
+		last_comp_ratio = 100UL;
+	}
+	cmdp->crb.gzip_fc = 0;
+	putnn(cmdp->crb, gzip_fc, fc);
+
+	/*
+	 * NX source buffers
+	 */
+	first_used = fifo_used_first_bytes(cur_in, used_in, fifo_in_len);
+	last_used = fifo_used_last_bytes(cur_in, used_in, fifo_in_len);
+
+	if (first_used > 0)
+		nx_append_dde(ddl_in, fifo_in + cur_in, first_used);
+
+	if (last_used > 0)
+		nx_append_dde(ddl_in, fifo_in, last_used);
+
+	/*
+	 * NX target buffers
+	 */
+	first_free = fifo_free_first_bytes(cur_out, used_out, fifo_out_len);
+	last_free = fifo_free_last_bytes(cur_out, used_out, fifo_out_len);
+
+	/* Reduce output free space amount not to overwrite the history */
+	int target_max = NX_MAX(0, fifo_free_bytes(used_out, fifo_out_len)
+				- (1<<16));
+
+	NXPRT(fprintf(stderr, "target_max %d (0x%x)\n", target_max,
+		      target_max));
+
+	first_free = NX_MIN(target_max, first_free);
+	if (first_free > 0) {
+		first_offset = fifo_free_first_offset(cur_out, used_out);
+		nx_append_dde(ddl_out, fifo_out + first_offset, first_free);
+	}
+
+	if (last_free > 0) {
+		last_free = NX_MIN(target_max - first_free, last_free);
+		if (last_free > 0) {
+			last_offset = fifo_free_last_offset(cur_out, used_out,
+							    fifo_out_len);
+			nx_append_dde(ddl_out, fifo_out + last_offset,
+				      last_free);
+		}
+	}
+
+	/* Target buffer size is used to limit the source data size
+	 * based on previous measurements of compression ratio.
+	 */
+
+	/* source_sz includes history */
+	source_sz = getp32(ddl_in, ddebc);
+	assert(source_sz > history_len);
+	source_sz = source_sz - history_len;
+
+	/* Estimating how much source is needed to 3/4 fill a
+	 * target_max size target buffer.  If we overshoot, then NX
+	 * must repeat the job with smaller input and we waste
+	 * bandwidth.  If we undershoot then we use more NX calls than
+	 * necessary.
+	 */
+
+	source_sz_estimate = ((uint64_t)target_max * last_comp_ratio * 3UL)
+				/ 4000;
+
+	if (source_sz_estimate < source_sz) {
+		/* Target might be small, therefore limiting the
+		 * source data.
+		 */
+		source_sz = source_sz_estimate;
+		target_sz_estimate = target_max;
+	} else {
+		/* Source file might be small, therefore limiting target
+		 * touch pages to a smaller value to save processor cycles.
+		 */
+		target_sz_estimate = ((uint64_t)source_sz * 1000UL)
+					/ (last_comp_ratio + 1);
+		target_sz_estimate = NX_MIN(2 * target_sz_estimate,
+					    target_max);
+	}
+
+	source_sz = source_sz + history_len;
+
+	/* Some NX condition codes require submitting the NX job again.
+	 * Kernel doesn't handle NX page faults. Expects user code to
+	 * touch pages.
+	 */
+	pgfault_retries = NX_MAX_FAULTS;
+
+restart_nx:
+
+	putp32(ddl_in, ddebc, source_sz);
+
+	/* Fault in pages */
+	nxu_touch_pages(cmdp, sizeof(struct nx_gzip_crb_cpb_t), page_sz, 1);
+	nx_touch_pages_dde(ddl_in, 0, page_sz, 0);
+	nx_touch_pages_dde(ddl_out, target_sz_estimate, page_sz, 1);
+
+	/* Send job to NX */
+	cc = nx_submit_job(ddl_in, ddl_out, cmdp, devhandle);
+
+	switch (cc) {
+
+	case ERR_NX_AT_FAULT:
+
+		/* We touched the pages ahead of time.  In the most common case
+		 * we shouldn't be here.  But may be some pages were paged out.
+		 * Kernel should have placed the faulting address to fsaddr.
+		 */
+		NXPRT(fprintf(stderr, "ERR_NX_AT_FAULT %p\n",
+			      (void *)cmdp->crb.csb.fsaddr));
+
+		if (pgfault_retries == NX_MAX_FAULTS) {
+			/* Try once with exact number of pages */
+			--pgfault_retries;
+			goto restart_nx;
+		} else if (pgfault_retries > 0) {
+			/* If still faulting try fewer input pages
+			 * assuming memory outage
+			 */
+			if (source_sz > page_sz)
+				source_sz = NX_MAX(source_sz / 2, page_sz);
+			--pgfault_retries;
+			goto restart_nx;
+		} else {
+			fprintf(stderr, "cannot make progress; too many ");
+			fprintf(stderr, "page fault retries cc= %d\n", cc);
+			rc = -1;
+			goto err5;
+		}
+
+	case ERR_NX_DATA_LENGTH:
+
+		NXPRT(fprintf(stderr, "ERR_NX_DATA_LENGTH; "));
+		NXPRT(fprintf(stderr, "stream may have trailing data\n"));
+
+		/* Not an error in the most common case; it just says
+		 * there is trailing data that we must examine.
+		 *
+		 * CC=3 CE(1)=0 CE(0)=1 indicates partial completion
+		 * Fig.6-7 and Table 6-8.
+		 */
+		nx_ce = get_csb_ce_ms3b(cmdp->crb.csb);
+
+		if (!csb_ce_termination(nx_ce) &&
+		    csb_ce_partial_completion(nx_ce)) {
+			/* Check CPB for more information
+			 * spbc and tpbc are valid
+			 */
+			sfbt = getnn(cmdp->cpb, out_sfbt); /* Table 6-4 */
+			subc = getnn(cmdp->cpb, out_subc); /* Table 6-4 */
+			spbc = get32(cmdp->cpb, out_spbc_decomp);
+			tpbc = get32(cmdp->crb.csb, tpbc);
+			assert(target_max >= tpbc);
+
+			goto ok_cc3; /* not an error */
+		} else {
+			/* History length error when CE(1)=1 CE(0)=0. */
+			rc = -1;
+			fprintf(stderr, "history length error cc= %d\n", cc);
+			goto err5;
+		}
+
+	case ERR_NX_TARGET_SPACE:
+
+		/* Target buffer not large enough; retry smaller input
+		 * data; give at least 1 byte.  SPBC/TPBC are not valid.
+		 */
+		assert(source_sz > history_len);
+		source_sz = ((source_sz - history_len + 2) / 2) + history_len;
+		NXPRT(fprintf(stderr, "ERR_NX_TARGET_SPACE; retry with "));
+		NXPRT(fprintf(stderr, "smaller input data src %d hist %d\n",
+			      source_sz, history_len));
+		goto restart_nx;
+
+	case ERR_NX_OK:
+
+		/* This should not happen for gzip formatted data;
+		 * we need trailing crc and isize
+		 */
+		fprintf(stderr, "ERR_NX_OK\n");
+		spbc = get32(cmdp->cpb, out_spbc_decomp);
+		tpbc = get32(cmdp->crb.csb, tpbc);
+		assert(target_max >= tpbc);
+		assert(spbc >= history_len);
+		source_sz = spbc - history_len;
+		goto offsets_state;
+
+	default:
+		fprintf(stderr, "error: cc= %d\n", cc);
+		rc = -1;
+		goto err5;
+	}
+
+ok_cc3:
+
+	NXPRT(fprintf(stderr, "cc3: sfbt: %x\n", sfbt));
+
+	assert(spbc > history_len);
+	source_sz = spbc - history_len;
+
+	/* Table 6-4: Source Final Block Type (SFBT) describes the
+	 * last processed deflate block and clues the software how to
+	 * resume the next job.  SUBC indicates how many input bits NX
+	 * consumed but did not process.  SPBC indicates how many
+	 * bytes of source were given to the accelerator including
+	 * history bytes.
+	 */
+
+	switch (sfbt) {
+		int dhtlen;
+
+	case 0x0: /* Deflate final EOB received */
+
+		/* Calculating the checksum start position. */
+
+		source_sz = source_sz - subc / 8;
+		is_final = 1;
+		break;
+
+		/* Resume decompression cases are below. Basically
+		 * indicates where NX has suspended and how to resume
+		 * the input stream.
+		 */
+
+	case 0x8: /* Within a literal block; use rembytecount */
+	case 0x9: /* Within a literal block; use rembytecount; bfinal=1 */
+
+		/* Supply the partially processed source byte again */
+		source_sz = source_sz - ((subc + 7) / 8);
+
+		/* SUBC LS 3bits: number of bits in the first source byte need
+		 * to be processed.
+		 * 000 means all 8 bits;  Table 6-3
+		 * Clear subc, histlen, sfbt, rembytecnt, dhtlen
+		 */
+		cmdp->cpb.in_subc = 0;
+		cmdp->cpb.in_sfbt = 0;
+		putnn(cmdp->cpb, in_subc, subc % 8);
+		putnn(cmdp->cpb, in_sfbt, sfbt);
+		putnn(cmdp->cpb, in_rembytecnt, getnn(cmdp->cpb,
+						      out_rembytecnt));
+		break;
+
+	case 0xA: /* Within a FH block; */
+	case 0xB: /* Within a FH block; bfinal=1 */
+
+		source_sz = source_sz - ((subc + 7) / 8);
+
+		/* Clear subc, histlen, sfbt, rembytecnt, dhtlen */
+		cmdp->cpb.in_subc = 0;
+		cmdp->cpb.in_sfbt = 0;
+		putnn(cmdp->cpb, in_subc, subc % 8);
+		putnn(cmdp->cpb, in_sfbt, sfbt);
+		break;
+
+	case 0xC: /* Within a DH block; */
+	case 0xD: /* Within a DH block; bfinal=1 */
+
+		source_sz = source_sz - ((subc + 7) / 8);
+
+		/* Clear subc, histlen, sfbt, rembytecnt, dhtlen */
+		cmdp->cpb.in_subc = 0;
+		cmdp->cpb.in_sfbt = 0;
+		putnn(cmdp->cpb, in_subc, subc % 8);
+		putnn(cmdp->cpb, in_sfbt, sfbt);
+
+		dhtlen = getnn(cmdp->cpb, out_dhtlen);
+		putnn(cmdp->cpb, in_dhtlen, dhtlen);
+		assert(dhtlen >= 42);
+
+		/* Round up to a qword */
+		dhtlen = (dhtlen + 127) / 128;
+
+		while (dhtlen > 0) { /* Copy dht from cpb.out to cpb.in */
+			--dhtlen;
+			cmdp->cpb.in_dht[dhtlen] = cmdp->cpb.out_dht[dhtlen];
+		}
+		break;
+
+	case 0xE: /* Within a block header; bfinal=0; */
+		     /* Also given if source data exactly ends (SUBC=0) with
+		      * EOB code with BFINAL=0.  Means the next byte will
+		      * contain a block header.
+		      */
+	case 0xF: /* within a block header with BFINAL=1. */
+
+		source_sz = source_sz - ((subc + 7) / 8);
+
+		/* Clear subc, histlen, sfbt, rembytecnt, dhtlen */
+		cmdp->cpb.in_subc = 0;
+		cmdp->cpb.in_sfbt = 0;
+		putnn(cmdp->cpb, in_subc, subc % 8);
+		putnn(cmdp->cpb, in_sfbt, sfbt);
+
+		/* Engine did not process any data */
+		if (is_eof && (source_sz == 0))
+			is_final = 1;
+	}
+
+offsets_state:
+
+	/* Adjust the source and target buffer offsets and lengths  */
+
+	NXPRT(fprintf(stderr, "offsets_state:\n"));
+
+	/* Delete input data from fifo_in */
+	used_in = used_in - source_sz;
+	cur_in = (cur_in + source_sz) % fifo_in_len;
+	input_file_offset = input_file_offset + source_sz;
+
+	/* Add output data to fifo_out */
+	used_out = used_out + tpbc;
+
+	assert(used_out <= fifo_out_len);
+
+	total_out = total_out + tpbc;
+
+	/* Deflate history is 32KB max.  No need to supply more
+	 * than 32KB on a resume.
+	 */
+	history_len = (total_out > window_max) ? window_max : total_out;
+
+	/* To estimate expected expansion in the next NX job; 500 means 50%.
+	 * Deflate best case is around 1 to 1000.
+	 */
+	last_comp_ratio = (1000UL * ((uint64_t)source_sz + 1))
+			  / ((uint64_t)tpbc + 1);
+	last_comp_ratio = NX_MAX(NX_MIN(1000UL, last_comp_ratio), 1);
+	NXPRT(fprintf(stderr, "comp_ratio %ld source_sz %d spbc %d tpbc %d\n",
+		      last_comp_ratio, source_sz, spbc, tpbc));
+
+	resuming = 1;
+
+finish_state:
+
+	NXPRT(fprintf(stderr, "finish_state:\n"));
+
+	if (is_final) {
+		if (used_out)
+			goto write_state; /* More data to write out */
+		else if (used_in < 8) {
+			/* Need at least 8 more bytes containing gzip crc
+			 * and isize.
+			 */
+			rc = -1;
+			goto err4;
+		} else {
+			/* Compare checksums and exit */
+			int i;
+			unsigned char tail[8];
+			uint32_t cksum, isize;
+
+			for (i = 0; i < 8; i++)
+				tail[i] = fifo_in[(cur_in + i) % fifo_in_len];
+			fprintf(stderr, "computed checksum %08x isize %08x\n",
+				cmdp->cpb.out_crc, (uint32_t) (total_out
+				% (1ULL<<32)));
+			cksum = ((uint32_t) tail[0] | (uint32_t) tail[1]<<8
+				 | (uint32_t) tail[2]<<16
+				 | (uint32_t) tail[3]<<24);
+			isize = ((uint32_t) tail[4] | (uint32_t) tail[5]<<8
+				 | (uint32_t) tail[6]<<16
+				 | (uint32_t) tail[7]<<24);
+			fprintf(stderr, "stored   checksum %08x isize %08x\n",
+				cksum, isize);
+
+			if (cksum == cmdp->cpb.out_crc && isize == (uint32_t)
+			    (total_out % (1ULL<<32))) {
+				rc = 0;	goto ok1;
+			} else {
+				rc = -1; goto err4;
+			}
+		}
+	} else
+		goto read_state;
+
+	return -1;
+
+err1:
+	fprintf(stderr, "error: not a gzip file, expect %x, read %x\n",
+		expect, c);
+	return -1;
+
+err2:
+	fprintf(stderr, "error: the FLG byte is wrong or not being handled\n");
+	return -1;
+
+err3:
+	fprintf(stderr, "error: gzip header\n");
+	return -1;
+
+err4:
+	fprintf(stderr, "error: checksum missing or mismatch\n");
+
+err5:
+ok1:
+	fprintf(stderr, "decomp is complete: fclose\n");
+	fclose(outf);
+
+	return rc;
+}
+
+
+int main(int argc, char **argv)
+{
+	int rc;
+	struct sigaction act;
+	void *handle;
+
+	nx_dbg = 0;
+	nx_gzip_log = NULL;
+	act.sa_handler = 0;
+	act.sa_sigaction = nxu_sigsegv_handler;
+	act.sa_flags = SA_SIGINFO;
+	act.sa_restorer = 0;
+	sigemptyset(&act.sa_mask);
+	sigaction(SIGSEGV, &act, NULL);
+
+	handle = nx_function_begin(NX_FUNC_COMP_GZIP, 0);
+	if (!handle) {
+		fprintf(stderr, "Unable to init NX, errno %d\n", errno);
+		exit(-1);
+	}
+
+	rc = decompress_file(argc, argv, handle);
+
+	nx_function_end(handle);
+
+	return rc;
+}
diff --git a/tools/testing/selftests/powerpc/nx-gzip/gzfht_test.c b/tools/testing/selftests/powerpc/nx-gzip/gzfht_test.c
new file mode 100644
index 000000000..02dffb65d
--- /dev/null
+++ b/tools/testing/selftests/powerpc/nx-gzip/gzfht_test.c
@@ -0,0 +1,433 @@
+// SPDX-License-Identifier: GPL-2.0-or-later
+
+/* P9 gzip sample code for demonstrating the P9 NX hardware interface.
+ * Not intended for productive uses or for performance or compression
+ * ratio measurements.  For simplicity of demonstration, this sample
+ * code compresses in to fixed Huffman blocks only (Deflate btype=1)
+ * and has very simple memory management.  Dynamic Huffman blocks
+ * (Deflate btype=2) are more involved as detailed in the user guide.
+ * Note also that /dev/crypto/gzip, VAS and skiboot support are
+ * required.
+ *
+ * Copyright 2020 IBM Corp.
+ *
+ * https://github.com/libnxz/power-gzip for zlib api and other utils
+ *
+ * Author: Bulent Abali <abali@us.ibm.com>
+ *
+ * Definitions of acronyms used here. See
+ * P9 NX Gzip Accelerator User's Manual for details:
+ * https://github.com/libnxz/power-gzip/blob/develop/doc/power_nx_gzip_um.pdf
+ *
+ * adler/crc: 32 bit checksums appended to stream tail
+ * ce:       completion extension
+ * cpb:      coprocessor parameter block (metadata)
+ * crb:      coprocessor request block (command)
+ * csb:      coprocessor status block (status)
+ * dht:      dynamic huffman table
+ * dde:      data descriptor element (address, length)
+ * ddl:      list of ddes
+ * dh/fh:    dynamic and fixed huffman types
+ * fc:       coprocessor function code
+ * histlen:  history/dictionary length
+ * history:  sliding window of up to 32KB of data
+ * lzcount:  Deflate LZ symbol counts
+ * rembytecnt: remaining byte count
+ * sfbt:     source final block type; last block's type during decomp
+ * spbc:     source processed byte count
+ * subc:     source unprocessed bit count
+ * tebc:     target ending bit count; valid bits in the last byte
+ * tpbc:     target processed byte count
+ * vas:      virtual accelerator switch; the user mode interface
+ */
+
+#define _ISOC11_SOURCE	// For aligned_alloc()
+#define _DEFAULT_SOURCE	// For endian.h
+
+#include <stdio.h>
+#include <stdlib.h>
+#include <string.h>
+#include <unistd.h>
+#include <stdint.h>
+#include <sys/types.h>
+#include <sys/stat.h>
+#include <sys/time.h>
+#include <sys/fcntl.h>
+#include <sys/mman.h>
+#include <endian.h>
+#include <bits/endian.h>
+#include <sys/ioctl.h>
+#include <assert.h>
+#include <errno.h>
+#include <signal.h>
+#include "nxu.h"
+#include "nx.h"
+
+int nx_dbg;
+FILE *nx_gzip_log;
+
+#define NX_MIN(X, Y) (((X) < (Y)) ? (X) : (Y))
+#define FNAME_MAX 1024
+#define FEXT ".nx.gz"
+
+/*
+ * LZ counts returned in the user supplied nx_gzip_crb_cpb_t structure.
+ */
+static int compress_fht_sample(char *src, uint32_t srclen, char *dst,
+				uint32_t dstlen, int with_count,
+				struct nx_gzip_crb_cpb_t *cmdp, void *handle)
+{
+	uint32_t fc;
+
+	assert(!!cmdp);
+
+	put32(cmdp->crb, gzip_fc, 0);  /* clear */
+	fc = (with_count) ? GZIP_FC_COMPRESS_RESUME_FHT_COUNT :
+			    GZIP_FC_COMPRESS_RESUME_FHT;
+	putnn(cmdp->crb, gzip_fc, fc);
+	putnn(cmdp->cpb, in_histlen, 0); /* resuming with no history */
+	memset((void *) &cmdp->crb.csb, 0, sizeof(cmdp->crb.csb));
+
+	/* Section 6.6 programming notes; spbc may be in two different
+	 * places depending on FC.
+	 */
+	if (!with_count)
+		put32(cmdp->cpb, out_spbc_comp, 0);
+	else
+		put32(cmdp->cpb, out_spbc_comp_with_count, 0);
+
+	/* Figure 6-3 6-4; CSB location */
+	put64(cmdp->crb, csb_address, 0);
+	put64(cmdp->crb, csb_address,
+	      (uint64_t) &cmdp->crb.csb & csb_address_mask);
+
+	/* Source direct dde (scatter-gather list) */
+	clear_dde(cmdp->crb.source_dde);
+	putnn(cmdp->crb.source_dde, dde_count, 0);
+	put32(cmdp->crb.source_dde, ddebc, srclen);
+	put64(cmdp->crb.source_dde, ddead, (uint64_t) src);
+
+	/* Target direct dde (scatter-gather list) */
+	clear_dde(cmdp->crb.target_dde);
+	putnn(cmdp->crb.target_dde, dde_count, 0);
+	put32(cmdp->crb.target_dde, ddebc, dstlen);
+	put64(cmdp->crb.target_dde, ddead, (uint64_t) dst);
+
+	/* Submit the crb, the job descriptor, to the accelerator */
+	return nxu_submit_job(cmdp, handle);
+}
+
+/*
+ * Prepares a blank no filename no timestamp gzip header and returns
+ * the number of bytes written to buf.
+ * Gzip specification at https://tools.ietf.org/html/rfc1952
+ */
+int gzip_header_blank(char *buf)
+{
+	int i = 0;
+
+	buf[i++] = 0x1f; /* ID1 */
+	buf[i++] = 0x8b; /* ID2 */
+	buf[i++] = 0x08; /* CM  */
+	buf[i++] = 0x00; /* FLG */
+	buf[i++] = 0x00; /* MTIME */
+	buf[i++] = 0x00; /* MTIME */
+	buf[i++] = 0x00; /* MTIME */
+	buf[i++] = 0x00; /* MTIME */
+	buf[i++] = 0x04; /* XFL 4=fastest */
+	buf[i++] = 0x03; /* OS UNIX */
+
+	return i;
+}
+
+/* Caller must free the allocated buffer return nonzero on error. */
+int read_alloc_input_file(char *fname, char **buf, size_t *bufsize)
+{
+	struct stat statbuf;
+	FILE *fp;
+	char *p;
+	size_t num_bytes;
+
+	if (stat(fname, &statbuf)) {
+		perror(fname);
+		return(-1);
+	}
+	fp = fopen(fname, "r");
+	if (fp == NULL) {
+		perror(fname);
+		return(-1);
+	}
+	assert(NULL != (p = (char *) malloc(statbuf.st_size)));
+	num_bytes = fread(p, 1, statbuf.st_size, fp);
+	if (ferror(fp) || (num_bytes != statbuf.st_size)) {
+		perror(fname);
+		return(-1);
+	}
+	*buf = p;
+	*bufsize = num_bytes;
+	return 0;
+}
+
+/* Returns nonzero on error */
+int write_output_file(char *fname, char *buf, size_t bufsize)
+{
+	FILE *fp;
+	size_t num_bytes;
+
+	fp = fopen(fname, "w");
+	if (fp == NULL) {
+		perror(fname);
+		return(-1);
+	}
+	num_bytes = fwrite(buf, 1, bufsize, fp);
+	if (ferror(fp) || (num_bytes != bufsize)) {
+		perror(fname);
+		return(-1);
+	}
+	fclose(fp);
+	return 0;
+}
+
+/*
+ * Z_SYNC_FLUSH as described in zlib.h.
+ * Returns number of appended bytes
+ */
+int append_sync_flush(char *buf, int tebc, int final)
+{
+	uint64_t flush;
+	int shift = (tebc & 0x7);
+
+	if (tebc > 0) {
+		/* Last byte is partially full */
+		buf = buf - 1;
+		*buf = *buf & (unsigned char) ((1<<tebc)-1);
+	} else
+		*buf = 0;
+	flush = ((0x1ULL & final) << shift) | *buf;
+	shift = shift + 3; /* BFINAL and BTYPE written */
+	shift = (shift <= 8) ? 8 : 16;
+	flush |= (0xFFFF0000ULL) << shift; /* Zero length block */
+	shift = shift + 32;
+	while (shift > 0) {
+		*buf++ = (unsigned char) (flush & 0xffULL);
+		flush = flush >> 8;
+		shift = shift - 8;
+	}
+	return(((tebc > 5) || (tebc == 0)) ? 5 : 4);
+}
+
+/*
+ * Final deflate block bit.  This call assumes the block
+ * beginning is byte aligned.
+ */
+static void set_bfinal(void *buf, int bfinal)
+{
+	char *b = buf;
+
+	if (bfinal)
+		*b = *b | (unsigned char) 0x01;
+	else
+		*b = *b & (unsigned char) 0xfe;
+}
+
+int compress_file(int argc, char **argv, void *handle)
+{
+	char *inbuf, *outbuf, *srcbuf, *dstbuf;
+	char outname[FNAME_MAX];
+	uint32_t srclen, dstlen;
+	uint32_t flushlen, chunk;
+	size_t inlen, outlen, dsttotlen, srctotlen;
+	uint32_t crc, spbc, tpbc, tebc;
+	int lzcounts = 0;
+	int cc;
+	int num_hdr_bytes;
+	struct nx_gzip_crb_cpb_t *cmdp;
+	uint32_t pagelen = 65536;
+	int fault_tries = NX_MAX_FAULTS;
+
+	cmdp = (void *)(uintptr_t)
+		aligned_alloc(sizeof(struct nx_gzip_crb_cpb_t),
+			      sizeof(struct nx_gzip_crb_cpb_t));
+
+	if (argc != 2) {
+		fprintf(stderr, "usage: %s <fname>\n", argv[0]);
+		exit(-1);
+	}
+	if (read_alloc_input_file(argv[1], &inbuf, &inlen))
+		exit(-1);
+	fprintf(stderr, "file %s read, %ld bytes\n", argv[1], inlen);
+
+	/* Generous output buffer for header/trailer */
+	outlen = 2 * inlen + 1024;
+
+	assert(NULL != (outbuf = (char *)malloc(outlen)));
+	nxu_touch_pages(outbuf, outlen, pagelen, 1);
+
+	/* Compress piecemeal in smallish chunks */
+	chunk = 1<<22;
+
+	/* Write the gzip header to the stream */
+	num_hdr_bytes = gzip_header_blank(outbuf);
+	dstbuf    = outbuf + num_hdr_bytes;
+	outlen    = outlen - num_hdr_bytes;
+	dsttotlen = num_hdr_bytes;
+
+	srcbuf    = inbuf;
+	srctotlen = 0;
+
+	/* Init the CRB, the coprocessor request block */
+	memset(&cmdp->crb, 0, sizeof(cmdp->crb));
+
+	/* Initial gzip crc32 */
+	put32(cmdp->cpb, in_crc, 0);
+
+	while (inlen > 0) {
+
+		/* Submit chunk size source data per job */
+		srclen = NX_MIN(chunk, inlen);
+		/* Supply large target in case data expands */
+		dstlen = NX_MIN(2*srclen, outlen);
+
+		/* Page faults are handled by the user code */
+
+		/* Fault-in pages; an improved code wouldn't touch so
+		 * many pages but would try to estimate the
+		 * compression ratio and adjust both the src and dst
+		 * touch amounts.
+		 */
+		nxu_touch_pages(cmdp, sizeof(struct nx_gzip_crb_cpb_t), pagelen,
+				1);
+		nxu_touch_pages(srcbuf, srclen, pagelen, 0);
+		nxu_touch_pages(dstbuf, dstlen, pagelen, 1);
+
+		cc = compress_fht_sample(
+			srcbuf, srclen,
+			dstbuf, dstlen,
+			lzcounts, cmdp, handle);
+
+		if (cc != ERR_NX_OK && cc != ERR_NX_TPBC_GT_SPBC &&
+		    cc != ERR_NX_AT_FAULT) {
+			fprintf(stderr, "nx error: cc= %d\n", cc);
+			exit(-1);
+		}
+
+		/* Page faults are handled by the user code */
+		if (cc == ERR_NX_AT_FAULT) {
+			NXPRT(fprintf(stderr, "page fault: cc= %d, ", cc));
+			NXPRT(fprintf(stderr, "try= %d, fsa= %08llx\n",
+				  fault_tries,
+				  (unsigned long long) cmdp->crb.csb.fsaddr));
+			fault_tries--;
+			if (fault_tries > 0) {
+				continue;
+			} else {
+				fprintf(stderr, "error: cannot progress; ");
+				fprintf(stderr, "too many faults\n");
+				exit(-1);
+			};
+		}
+
+		fault_tries = NX_MAX_FAULTS; /* Reset for the next chunk */
+
+		inlen     = inlen - srclen;
+		srcbuf    = srcbuf + srclen;
+		srctotlen = srctotlen + srclen;
+
+		/* Two possible locations for spbc depending on the function
+		 * code.
+		 */
+		spbc = (!lzcounts) ? get32(cmdp->cpb, out_spbc_comp) :
+			get32(cmdp->cpb, out_spbc_comp_with_count);
+		assert(spbc == srclen);
+
+		/* Target byte count */
+		tpbc = get32(cmdp->crb.csb, tpbc);
+		/* Target ending bit count */
+		tebc = getnn(cmdp->cpb, out_tebc);
+		NXPRT(fprintf(stderr, "compressed chunk %d ", spbc));
+		NXPRT(fprintf(stderr, "to %d bytes, tebc= %d\n", tpbc, tebc));
+
+		if (inlen > 0) { /* More chunks to go */
+			set_bfinal(dstbuf, 0);
+			dstbuf    = dstbuf + tpbc;
+			dsttotlen = dsttotlen + tpbc;
+			outlen    = outlen - tpbc;
+			/* Round up to the next byte with a flush
+			 * block; do not set the BFINAqL bit.
+			 */
+			flushlen  = append_sync_flush(dstbuf, tebc, 0);
+			dsttotlen = dsttotlen + flushlen;
+			outlen    = outlen - flushlen;
+			dstbuf    = dstbuf + flushlen;
+			NXPRT(fprintf(stderr, "added sync_flush %d bytes\n",
+					flushlen));
+		} else {  /* Done */
+			/* Set the BFINAL bit of the last block per Deflate
+			 * specification.
+			 */
+			set_bfinal(dstbuf, 1);
+			dstbuf    = dstbuf + tpbc;
+			dsttotlen = dsttotlen + tpbc;
+			outlen    = outlen - tpbc;
+		}
+
+		/* Resuming crc32 for the next chunk */
+		crc = get32(cmdp->cpb, out_crc);
+		put32(cmdp->cpb, in_crc, crc);
+		crc = be32toh(crc);
+	}
+
+	/* Append crc32 and ISIZE to the end */
+	memcpy(dstbuf, &crc, 4);
+	memcpy(dstbuf+4, &srctotlen, 4);
+	dsttotlen = dsttotlen + 8;
+	outlen    = outlen - 8;
+
+	assert(FNAME_MAX > (strlen(argv[1]) + strlen(FEXT)));
+	strcpy(outname, argv[1]);
+	strcat(outname, FEXT);
+	if (write_output_file(outname, outbuf, dsttotlen)) {
+		fprintf(stderr, "write error: %s\n", outname);
+		exit(-1);
+	}
+
+	fprintf(stderr, "compressed %ld to %ld bytes total, ", srctotlen,
+		dsttotlen);
+	fprintf(stderr, "crc32 checksum = %08x\n", crc);
+
+	if (inbuf != NULL)
+		free(inbuf);
+
+	if (outbuf != NULL)
+		free(outbuf);
+
+	return 0;
+}
+
+int main(int argc, char **argv)
+{
+	int rc;
+	struct sigaction act;
+	void *handle;
+
+	nx_dbg = 0;
+	nx_gzip_log = NULL;
+	act.sa_handler = 0;
+	act.sa_sigaction = nxu_sigsegv_handler;
+	act.sa_flags = SA_SIGINFO;
+	act.sa_restorer = 0;
+	sigemptyset(&act.sa_mask);
+	sigaction(SIGSEGV, &act, NULL);
+
+	handle = nx_function_begin(NX_FUNC_COMP_GZIP, 0);
+	if (!handle) {
+		fprintf(stderr, "Unable to init NX, errno %d\n", errno);
+		exit(-1);
+	}
+
+	rc = compress_file(argc, argv, handle);
+
+	nx_function_end(handle);
+
+	return rc;
+}
diff --git a/tools/testing/selftests/powerpc/nx-gzip/gzip_vas.c b/tools/testing/selftests/powerpc/nx-gzip/gzip_vas.c
new file mode 100644
index 000000000..c055885da
--- /dev/null
+++ b/tools/testing/selftests/powerpc/nx-gzip/gzip_vas.c
@@ -0,0 +1,316 @@
+// SPDX-License-Identifier: GPL-2.0-or-later
+
+/*
+ * Copyright 2020 IBM Corp.
+ *
+ * Author: Bulent Abali <abali@us.ibm.com>
+ *
+ */
+#include <stdio.h>
+#include <stdlib.h>
+#include <string.h>
+#include <unistd.h>
+#include <stdint.h>
+#include <sys/types.h>
+#include <sys/stat.h>
+#include <sys/time.h>
+#include <sys/fcntl.h>
+#include <sys/mman.h>
+#include <endian.h>
+#include <bits/endian.h>
+#include <sys/ioctl.h>
+#include <assert.h>
+#include <errno.h>
+#include <signal.h>
+#include "vas-api.h"
+#include "nx.h"
+#include "copy-paste.h"
+#include "nxu.h"
+#include "nx_dbg.h"
+#include <sys/platform/ppc.h>
+
+#define barrier()
+#define hwsync()    ({ asm volatile("sync" ::: "memory"); })
+
+#ifndef NX_NO_CPU_PRI
+#define cpu_pri_default()  ({ asm volatile ("or 2, 2, 2"); })
+#define cpu_pri_low()      ({ asm volatile ("or 31, 31, 31"); })
+#else
+#define cpu_pri_default()
+#define cpu_pri_low()
+#endif
+
+void *nx_fault_storage_address;
+
+struct nx_handle {
+	int fd;
+	int function;
+	void *paste_addr;
+};
+
+static int open_device_nodes(char *devname, int pri, struct nx_handle *handle)
+{
+	int rc, fd;
+	void *addr;
+	struct vas_tx_win_open_attr txattr;
+
+	fd = open(devname, O_RDWR);
+	if (fd < 0) {
+		fprintf(stderr, " open device name %s\n", devname);
+		return -errno;
+	}
+
+	memset(&txattr, 0, sizeof(txattr));
+	txattr.version = 1;
+	txattr.vas_id = pri;
+	rc = ioctl(fd, VAS_TX_WIN_OPEN, (unsigned long)&txattr);
+	if (rc < 0) {
+		fprintf(stderr, "ioctl() n %d, error %d\n", rc, errno);
+		rc = -errno;
+		goto out;
+	}
+
+	addr = mmap(NULL, 4096, PROT_READ|PROT_WRITE, MAP_SHARED, fd, 0ULL);
+	if (addr == MAP_FAILED) {
+		fprintf(stderr, "mmap() failed, errno %d\n", errno);
+		rc = -errno;
+		goto out;
+	}
+	handle->fd = fd;
+	handle->paste_addr = (void *)((char *)addr + 0x400);
+
+	rc = 0;
+out:
+	close(fd);
+	return rc;
+}
+
+void *nx_function_begin(int function, int pri)
+{
+	int rc;
+	char *devname = "/dev/crypto/nx-gzip";
+	struct nx_handle *nxhandle;
+
+	if (function != NX_FUNC_COMP_GZIP) {
+		errno = EINVAL;
+		fprintf(stderr, " NX_FUNC_COMP_GZIP not found\n");
+		return NULL;
+	}
+
+
+	nxhandle = malloc(sizeof(*nxhandle));
+	if (!nxhandle) {
+		errno = ENOMEM;
+		fprintf(stderr, " No memory\n");
+		return NULL;
+	}
+
+	nxhandle->function = function;
+	rc = open_device_nodes(devname, pri, nxhandle);
+	if (rc < 0) {
+		errno = -rc;
+		fprintf(stderr, " open_device_nodes failed\n");
+		return NULL;
+	}
+
+	return nxhandle;
+}
+
+int nx_function_end(void *handle)
+{
+	int rc = 0;
+	struct nx_handle *nxhandle = handle;
+
+	rc = munmap(nxhandle->paste_addr - 0x400, 4096);
+	if (rc < 0) {
+		fprintf(stderr, "munmap() failed, errno %d\n", errno);
+		return rc;
+	}
+	close(nxhandle->fd);
+	free(nxhandle);
+
+	return rc;
+}
+
+static int nx_wait_for_csb(struct nx_gzip_crb_cpb_t *cmdp)
+{
+	long poll = 0;
+	uint64_t t;
+
+	/* Save power and let other threads use the h/w. top may show
+	 * 100% but only because OS doesn't know we slowed the this
+	 * h/w thread while polling. We're letting other threads have
+	 * higher throughput on the core.
+	 */
+	cpu_pri_low();
+
+#define CSB_MAX_POLL 200000000UL
+#define USLEEP_TH     300000UL
+
+	t = __ppc_get_timebase();
+
+	while (getnn(cmdp->crb.csb, csb_v) == 0) {
+		++poll;
+		hwsync();
+
+		cpu_pri_low();
+
+		/* usleep(0) takes around 29000 ticks ~60 us.
+		 * 300000 is spinning for about 600 us then
+		 * start sleeping.
+		 */
+		if ((__ppc_get_timebase() - t) > USLEEP_TH) {
+			cpu_pri_default();
+			usleep(1);
+		}
+
+		if (poll > CSB_MAX_POLL)
+			break;
+
+		/* Fault address from signal handler */
+		if (nx_fault_storage_address) {
+			cpu_pri_default();
+			return -EAGAIN;
+		}
+
+	}
+
+	cpu_pri_default();
+
+	/* hw has updated csb and output buffer */
+	hwsync();
+
+	/* Check CSB flags. */
+	if (getnn(cmdp->crb.csb, csb_v) == 0) {
+		fprintf(stderr, "CSB still not valid after %d polls.\n",
+			(int) poll);
+		prt_err("CSB still not valid after %d polls, giving up.\n",
+			(int) poll);
+		return -ETIMEDOUT;
+	}
+
+	return 0;
+}
+
+static int nxu_run_job(struct nx_gzip_crb_cpb_t *cmdp, void *handle)
+{
+	int i, ret, retries;
+	struct nx_handle *nxhandle = handle;
+
+	assert(handle != NULL);
+	i = 0;
+	retries = 5000;
+	while (i++ < retries) {
+		hwsync();
+		vas_copy(&cmdp->crb, 0);
+		ret = vas_paste(nxhandle->paste_addr, 0);
+		hwsync();
+
+		NXPRT(fprintf(stderr, "Paste attempt %d/%d returns 0x%x\n",
+				i, retries, ret));
+
+		if ((ret == 2) || (ret == 3)) {
+
+			ret = nx_wait_for_csb(cmdp);
+			if (!ret) {
+				goto out;
+			} else if (ret == -EAGAIN) {
+				long x;
+
+				prt_err("Touching address %p, 0x%lx\n",
+					 nx_fault_storage_address,
+					 *(long *) nx_fault_storage_address);
+				x = *(long *) nx_fault_storage_address;
+				*(long *) nx_fault_storage_address = x;
+				nx_fault_storage_address = 0;
+				continue;
+			} else {
+				prt_err("wait_for_csb() returns %d\n", ret);
+				break;
+			}
+		} else {
+			if (i < 10) {
+				/* spin for few ticks */
+#define SPIN_TH 500UL
+				uint64_t fail_spin;
+
+				fail_spin = __ppc_get_timebase();
+				while ((__ppc_get_timebase() - fail_spin) <
+					 SPIN_TH)
+					;
+			} else {
+				/* sleep */
+				unsigned int pr = 0;
+
+				if (pr++ % 100 == 0) {
+					prt_err("Paste attempt %d/", i);
+					prt_err("%d, failed pid= %d\n", retries,
+						getpid());
+				}
+				usleep(1);
+			}
+			continue;
+		}
+	}
+
+out:
+	cpu_pri_default();
+
+	return ret;
+}
+
+int nxu_submit_job(struct nx_gzip_crb_cpb_t *cmdp, void *handle)
+{
+	int cc;
+
+	cc = nxu_run_job(cmdp, handle);
+
+	if (!cc)
+		cc = getnn(cmdp->crb.csb, csb_cc);      /* CC Table 6-8 */
+
+	return cc;
+}
+
+
+void nxu_sigsegv_handler(int sig, siginfo_t *info, void *ctx)
+{
+	fprintf(stderr, "%d: Got signal %d si_code %d, si_addr %p\n", getpid(),
+		sig, info->si_code, info->si_addr);
+
+	nx_fault_storage_address = info->si_addr;
+}
+
+/*
+ * Fault in pages prior to NX job submission.  wr=1 may be required to
+ * touch writeable pages.  System zero pages do not fault-in the page as
+ * intended.  Typically set wr=1 for NX target pages and set wr=0 for NX
+ * source pages.
+ */
+int nxu_touch_pages(void *buf, long buf_len, long page_len, int wr)
+{
+	char *begin = buf;
+	char *end = (char *) buf + buf_len - 1;
+	volatile char t;
+
+	assert(buf_len >= 0 && !!buf);
+
+	NXPRT(fprintf(stderr, "touch %p %p len 0x%lx wr=%d\n", buf,
+			(buf + buf_len), buf_len, wr));
+
+	if (buf_len <= 0 || buf == NULL)
+		return -1;
+
+	do {
+		t = *begin;
+		if (wr)
+			*begin = t;
+		begin = begin + page_len;
+	} while (begin < end);
+
+	/* When buf_sz is small or buf tail is in another page */
+	t = *end;
+	if (wr)
+		*end = t;
+
+	return 0;
+}
diff --git a/tools/testing/selftests/powerpc/nx-gzip/include/copy-paste.h b/tools/testing/selftests/powerpc/nx-gzip/include/copy-paste.h
new file mode 100644
index 000000000..0db2d6485
--- /dev/null
+++ b/tools/testing/selftests/powerpc/nx-gzip/include/copy-paste.h
@@ -0,0 +1,56 @@
+/* SPDX-License-Identifier: GPL-2.0-or-later */
+
+/* From asm-compat.h */
+#define __stringify_in_c(...)	#__VA_ARGS__
+#define stringify_in_c(...)	__stringify_in_c(__VA_ARGS__) " "
+
+/*
+ * Macros taken from arch/powerpc/include/asm/ppc-opcode.h and other
+ * header files.
+ */
+#define ___PPC_RA(a)    (((a) & 0x1f) << 16)
+#define ___PPC_RB(b)    (((b) & 0x1f) << 11)
+
+#define PPC_INST_COPY                   0x7c20060c
+#define PPC_INST_PASTE                  0x7c20070d
+
+#define PPC_COPY(a, b)          stringify_in_c(.long PPC_INST_COPY | \
+						___PPC_RA(a) | ___PPC_RB(b))
+#define PPC_PASTE(a, b)         stringify_in_c(.long PPC_INST_PASTE | \
+						___PPC_RA(a) | ___PPC_RB(b))
+#define CR0_SHIFT	28
+#define CR0_MASK	0xF
+/*
+ * Copy/paste instructions:
+ *
+ *	copy RA,RB
+ *		Copy contents of address (RA) + effective_address(RB)
+ *		to internal copy-buffer.
+ *
+ *	paste RA,RB
+ *		Paste contents of internal copy-buffer to the address
+ *		(RA) + effective_address(RB)
+ */
+static inline int vas_copy(void *crb, int offset)
+{
+	asm volatile(PPC_COPY(%0, %1)";"
+		:
+		: "b" (offset), "b" (crb)
+		: "memory");
+
+	return 0;
+}
+
+static inline int vas_paste(void *paste_address, int offset)
+{
+	__u32 cr;
+
+	cr = 0;
+	asm volatile(PPC_PASTE(%1, %2)";"
+		"mfocrf %0, 0x80;"
+		: "=r" (cr)
+		: "b" (offset), "b" (paste_address)
+		: "memory", "cr0");
+
+	return (cr >> CR0_SHIFT) & CR0_MASK;
+}
diff --git a/tools/testing/selftests/powerpc/nx-gzip/include/crb.h b/tools/testing/selftests/powerpc/nx-gzip/include/crb.h
new file mode 100644
index 000000000..ab101085f
--- /dev/null
+++ b/tools/testing/selftests/powerpc/nx-gzip/include/crb.h
@@ -0,0 +1,155 @@
+/* SPDX-License-Identifier: GPL-2.0-or-later */
+#ifndef __CRB_H
+#define __CRB_H
+#include <linux/types.h>
+#include "nx.h"
+
+/* CCW 842 CI/FC masks
+ * NX P8 workbook, section 4.3.1, figure 4-6
+ * "CI/FC Boundary by NX CT type"
+ */
+#define CCW_CI_842              (0x00003ff8)
+#define CCW_FC_842              (0x00000007)
+
+/* Chapter 6.5.8 Coprocessor-Completion Block (CCB) */
+
+#define CCB_VALUE		(0x3fffffffffffffff)
+#define CCB_ADDRESS		(0xfffffffffffffff8)
+#define CCB_CM			(0x0000000000000007)
+#define CCB_CM0			(0x0000000000000004)
+#define CCB_CM12		(0x0000000000000003)
+
+#define CCB_CM0_ALL_COMPLETIONS	(0x0)
+#define CCB_CM0_LAST_IN_CHAIN	(0x4)
+#define CCB_CM12_STORE		(0x0)
+#define CCB_CM12_INTERRUPT	(0x1)
+
+#define CCB_SIZE		(0x10)
+#define CCB_ALIGN		CCB_SIZE
+
+struct coprocessor_completion_block {
+	__be64 value;
+	__be64 address;
+} __aligned(CCB_ALIGN);
+
+
+/* Chapter 6.5.7 Coprocessor-Status Block (CSB) */
+
+#define CSB_V			(0x80)
+#define CSB_F			(0x04)
+#define CSB_CH			(0x03)
+#define CSB_CE_INCOMPLETE	(0x80)
+#define CSB_CE_TERMINATION	(0x40)
+#define CSB_CE_TPBC		(0x20)
+
+#define CSB_CC_SUCCESS		(0)
+#define CSB_CC_INVALID_ALIGN	(1)
+#define CSB_CC_OPERAND_OVERLAP	(2)
+#define CSB_CC_DATA_LENGTH	(3)
+#define CSB_CC_TRANSLATION	(5)
+#define CSB_CC_PROTECTION	(6)
+#define CSB_CC_RD_EXTERNAL	(7)
+#define CSB_CC_INVALID_OPERAND	(8)
+#define CSB_CC_PRIVILEGE	(9)
+#define CSB_CC_INTERNAL		(10)
+#define CSB_CC_WR_EXTERNAL	(12)
+#define CSB_CC_NOSPC		(13)
+#define CSB_CC_EXCESSIVE_DDE	(14)
+#define CSB_CC_WR_TRANSLATION	(15)
+#define CSB_CC_WR_PROTECTION	(16)
+#define CSB_CC_UNKNOWN_CODE	(17)
+#define CSB_CC_ABORT		(18)
+#define CSB_CC_TRANSPORT	(20)
+#define CSB_CC_SEGMENTED_DDL	(31)
+#define CSB_CC_PROGRESS_POINT	(32)
+#define CSB_CC_DDE_OVERFLOW	(33)
+#define CSB_CC_SESSION		(34)
+#define CSB_CC_PROVISION	(36)
+#define CSB_CC_CHAIN		(37)
+#define CSB_CC_SEQUENCE		(38)
+#define CSB_CC_HW		(39)
+
+#define CSB_SIZE		(0x10)
+#define CSB_ALIGN		CSB_SIZE
+
+struct coprocessor_status_block {
+	__u8 flags;
+	__u8 cs;
+	__u8 cc;
+	__u8 ce;
+	__be32 count;
+	__be64 address;
+} __aligned(CSB_ALIGN);
+
+
+/* Chapter 6.5.10 Data-Descriptor List (DDL)
+ * each list contains one or more Data-Descriptor Entries (DDE)
+ */
+
+#define DDE_P			(0x8000)
+
+#define DDE_SIZE		(0x10)
+#define DDE_ALIGN		DDE_SIZE
+
+struct data_descriptor_entry {
+	__be16 flags;
+	__u8 count;
+	__u8 index;
+	__be32 length;
+	__be64 address;
+} __aligned(DDE_ALIGN);
+
+
+/* Chapter 6.5.2 Coprocessor-Request Block (CRB) */
+
+#define CRB_SIZE		(0x80)
+#define CRB_ALIGN		(0x100) /* Errata: requires 256 alignment */
+
+
+/* Coprocessor Status Block field
+ *   ADDRESS	address of CSB
+ *   C		CCB is valid
+ *   AT		0 = addrs are virtual, 1 = addrs are phys
+ *   M		enable perf monitor
+ */
+#define CRB_CSB_ADDRESS		(0xfffffffffffffff0)
+#define CRB_CSB_C		(0x0000000000000008)
+#define CRB_CSB_AT		(0x0000000000000002)
+#define CRB_CSB_M		(0x0000000000000001)
+
+struct coprocessor_request_block {
+	__be32 ccw;
+	__be32 flags;
+	__be64 csb_addr;
+
+	struct data_descriptor_entry source;
+	struct data_descriptor_entry target;
+
+	struct coprocessor_completion_block ccb;
+
+	__u8 reserved[48];
+
+	struct coprocessor_status_block csb;
+} __aligned(CRB_ALIGN);
+
+#define crb_csb_addr(c)         __be64_to_cpu(c->csb_addr)
+#define crb_nx_fault_addr(c)    __be64_to_cpu(c->stamp.nx.fault_storage_addr)
+#define crb_nx_flags(c)         c->stamp.nx.flags
+#define crb_nx_fault_status(c)  c->stamp.nx.fault_status
+#define crb_nx_pswid(c)		c->stamp.nx.pswid
+
+
+/* RFC02167 Initiate Coprocessor Instructions document
+ * Chapter 8.2.1.1.1 RS
+ * Chapter 8.2.3 Coprocessor Directive
+ * Chapter 8.2.4 Execution
+ *
+ * The CCW must be converted to BE before passing to icswx()
+ */
+
+#define CCW_PS                  (0xff000000)
+#define CCW_CT                  (0x00ff0000)
+#define CCW_CD                  (0x0000ffff)
+#define CCW_CL                  (0x0000c000)
+
+#endif
diff --git a/tools/testing/selftests/powerpc/nx-gzip/include/nx.h b/tools/testing/selftests/powerpc/nx-gzip/include/nx.h
new file mode 100644
index 000000000..1abe23fc2
--- /dev/null
+++ b/tools/testing/selftests/powerpc/nx-gzip/include/nx.h
@@ -0,0 +1,38 @@
+/* SPDX-License-Identifier: GPL-2.0-or-later */
+/*
+ * Copyright 2020 IBM Corp.
+ *
+ */
+#ifndef _NX_H
+#define _NX_H
+
+#include <stdbool.h>
+
+#define	NX_FUNC_COMP_842	1
+#define NX_FUNC_COMP_GZIP	2
+
+#ifndef __aligned
+#define __aligned(x)	__attribute__((aligned(x)))
+#endif
+
+struct nx842_func_args {
+	bool use_crc;
+	bool decompress;		/* true decompress; false compress */
+	bool move_data;
+	int timeout;			/* seconds */
+};
+
+struct nxbuf_t {
+	int len;
+	char *buf;
+};
+
+/* @function should be EFT (aka 842), GZIP etc */
+void *nx_function_begin(int function, int pri);
+
+int nx_function(void *handle, struct nxbuf_t *in, struct nxbuf_t *out,
+		void *arg);
+
+int nx_function_end(void *handle);
+
+#endif	/* _NX_H */
diff --git a/tools/testing/selftests/powerpc/nx-gzip/include/nx_dbg.h b/tools/testing/selftests/powerpc/nx-gzip/include/nx_dbg.h
new file mode 100644
index 000000000..16464e19c
--- /dev/null
+++ b/tools/testing/selftests/powerpc/nx-gzip/include/nx_dbg.h
@@ -0,0 +1,95 @@
+/* SPDX-License-Identifier: GPL-2.0-or-later */
+/*
+ * Copyright 2020 IBM Corporation
+ *
+ */
+
+#ifndef _NXU_DBG_H_
+#define _NXU_DBG_H_
+
+#include <sys/file.h>
+#include <stdint.h>
+#include <stdio.h>
+#include <time.h>
+#include <pthread.h>
+
+extern FILE * nx_gzip_log;
+extern int nx_gzip_trace;
+extern unsigned int nx_gzip_inflate_impl;
+extern unsigned int nx_gzip_deflate_impl;
+extern unsigned int nx_gzip_inflate_flags;
+extern unsigned int nx_gzip_deflate_flags;
+
+extern int nx_dbg;
+pthread_mutex_t mutex_log;
+
+#define nx_gzip_trace_enabled()       (nx_gzip_trace & 0x1)
+#define nx_gzip_hw_trace_enabled()    (nx_gzip_trace & 0x2)
+#define nx_gzip_sw_trace_enabled()    (nx_gzip_trace & 0x4)
+#define nx_gzip_gather_statistics()   (nx_gzip_trace & 0x8)
+#define nx_gzip_per_stream_stat()     (nx_gzip_trace & 0x10)
+
+#define prt(fmt, ...) do { \
+	pthread_mutex_lock(&mutex_log);					\
+	flock(nx_gzip_log->_fileno, LOCK_EX);				\
+	time_t t; struct tm *m; time(&t); m = localtime(&t);		\
+	fprintf(nx_gzip_log, "[%04d/%02d/%02d %02d:%02d:%02d] "		\
+		"pid %d: " fmt,	\
+		(int)m->tm_year + 1900, (int)m->tm_mon+1, (int)m->tm_mday, \
+		(int)m->tm_hour, (int)m->tm_min, (int)m->tm_sec,	\
+		(int)getpid(), ## __VA_ARGS__);				\
+	fflush(nx_gzip_log);						\
+	flock(nx_gzip_log->_fileno, LOCK_UN);				\
+	pthread_mutex_unlock(&mutex_log);				\
+} while (0)
+
+/* Use in case of an error */
+#define prt_err(fmt, ...) do { if (nx_dbg >= 0) {			\
+	prt("%s:%u: Error: "fmt,					\
+		__FILE__, __LINE__, ## __VA_ARGS__);			\
+}} while (0)
+
+/* Use in case of an warning */
+#define prt_warn(fmt, ...) do {	if (nx_dbg >= 1) {			\
+	prt("%s:%u: Warning: "fmt,					\
+		__FILE__, __LINE__, ## __VA_ARGS__);			\
+}} while (0)
+
+/* Informational printouts */
+#define prt_info(fmt, ...) do {	if (nx_dbg >= 2) {			\
+	prt("Info: "fmt, ## __VA_ARGS__);				\
+}} while (0)
+
+/* Trace zlib wrapper code */
+#define prt_trace(fmt, ...) do { if (nx_gzip_trace_enabled()) {		\
+	prt("### "fmt, ## __VA_ARGS__);					\
+}} while (0)
+
+/* Trace statistics */
+#define prt_stat(fmt, ...) do {	if (nx_gzip_gather_statistics()) {	\
+	prt("### "fmt, ## __VA_ARGS__);					\
+}} while (0)
+
+/* Trace zlib hardware implementation */
+#define hw_trace(fmt, ...) do {						\
+		if (nx_gzip_hw_trace_enabled())				\
+			fprintf(nx_gzip_log, "hhh " fmt, ## __VA_ARGS__); \
+	} while (0)
+
+/* Trace zlib software implementation */
+#define sw_trace(fmt, ...) do {						\
+		if (nx_gzip_sw_trace_enabled())				\
+			fprintf(nx_gzip_log, "sss " fmt, ## __VA_ARGS__); \
+	} while (0)
+
+
+/**
+ * str_to_num - Convert string into number and copy with endings like
+ *              KiB for kilobyte
+ *              MiB for megabyte
+ *              GiB for gigabyte
+ */
+uint64_t str_to_num(char *str);
+void nx_lib_debug(int onoff);
+
+#endif	/* _NXU_DBG_H_ */
diff --git a/tools/testing/selftests/powerpc/nx-gzip/include/nxu.h b/tools/testing/selftests/powerpc/nx-gzip/include/nxu.h
new file mode 100644
index 000000000..20a4e883e
--- /dev/null
+++ b/tools/testing/selftests/powerpc/nx-gzip/include/nxu.h
@@ -0,0 +1,650 @@
+/* SPDX-License-Identifier: GPL-2.0-or-later */
+/*
+ * Hardware interface of the NX-GZIP compression accelerator
+ *
+ * Copyright (C) IBM Corporation, 2020
+ *
+ * Author: Bulent Abali <abali@us.ibm.com>
+ *
+ */
+
+#ifndef _NXU_H
+#define _NXU_H
+
+#include <stdint.h>
+#include <endian.h>
+#include "nx.h"
+
+/* deflate */
+#define LLSZ   286
+#define DSZ    30
+
+/* nx */
+#define DHTSZ  18
+#define DHT_MAXSZ 288
+#define MAX_DDE_COUNT 256
+
+/* util */
+#ifdef NXDBG
+#define NXPRT(X)	X
+#else
+#define NXPRT(X)
+#endif
+
+#ifdef NXTIMER
+#include <sys/platform/ppc.h>
+#define NX_CLK(X)	X
+#define nx_get_time()	__ppc_get_timebase()
+#define nx_get_freq()	__ppc_get_timebase_freq()
+#else
+#define NX_CLK(X)
+#define nx_get_time()  (-1)
+#define nx_get_freq()  (-1)
+#endif
+
+#define NX_MAX_FAULTS  500
+
+/*
+ * Definitions of acronyms used here. See
+ * P9 NX Gzip Accelerator User's Manual for details:
+ * https://github.com/libnxz/power-gzip/blob/develop/doc/power_nx_gzip_um.pdf
+ *
+ * adler/crc: 32 bit checksums appended to stream tail
+ * ce:       completion extension
+ * cpb:      coprocessor parameter block (metadata)
+ * crb:      coprocessor request block (command)
+ * csb:      coprocessor status block (status)
+ * dht:      dynamic huffman table
+ * dde:      data descriptor element (address, length)
+ * ddl:      list of ddes
+ * dh/fh:    dynamic and fixed huffman types
+ * fc:       coprocessor function code
+ * histlen:  history/dictionary length
+ * history:  sliding window of up to 32KB of data
+ * lzcount:  Deflate LZ symbol counts
+ * rembytecnt: remaining byte count
+ * sfbt:     source final block type; last block's type during decomp
+ * spbc:     source processed byte count
+ * subc:     source unprocessed bit count
+ * tebc:     target ending bit count; valid bits in the last byte
+ * tpbc:     target processed byte count
+ * vas:      virtual accelerator switch; the user mode interface
+ */
+
+union nx_qw_t {
+	uint32_t word[4];
+	uint64_t dword[2];
+} __aligned(16);
+
+/*
+ * Note: NX registers with fewer than 32 bits are declared by
+ * convention as uint32_t variables in unions. If *_offset and *_mask
+ * are defined for a variable, then use get_ put_ macros to
+ * conveniently access the register fields for endian conversions.
+ */
+
+struct nx_dde_t {
+	/* Data Descriptor Element, Section 6.4 */
+	union {
+		uint32_t dde_count;
+		/* When dde_count == 0 ddead is a pointer to a data buffer;
+		 * ddebc is the buffer length bytes.
+		 * When dde_count > 0 dde is an indirect dde; ddead is a
+		 * pointer to a contiguous list of direct ddes; ddebc is the
+		 * total length of all data pointed to by the list of direct
+		 * ddes. Note that only one level of indirection is permitted.
+		 * See Section 6.4 of the user manual for additional details.
+		 */
+	};
+	uint32_t ddebc; /* dde byte count */
+	uint64_t ddead; /* dde address */
+} __aligned(16);
+
+struct nx_csb_t {
+	/* Coprocessor Status Block, Section 6.6  */
+	union {
+		uint32_t csb_v;
+		/* Valid bit. v must be set to 0 by the program
+		 * before submitting the coprocessor command.
+		 * Software can poll for the v bit
+		 */
+
+		uint32_t csb_f;
+		/* 16B CSB size. Written to 0 by DMA when it writes the CPB */
+
+		uint32_t csb_cs;
+		/* cs completion sequence; unused */
+
+		uint32_t csb_cc;
+		/* cc completion code; cc != 0 exception occurred */
+
+		uint32_t csb_ce;
+		/* ce completion extension */
+
+	};
+	uint32_t tpbc;
+	/* target processed byte count TPBC */
+
+	uint64_t fsaddr;
+	/* Section 6.12.1 CSB NonZero error summary.  FSA Failing storage
+	 * address.  Address where error occurred. When available, written
+	 * to A field of CSB
+	 */
+} __aligned(16);
+
+struct nx_ccb_t {
+	/* Coprocessor Completion Block, Section 6.7 */
+
+	uint32_t reserved[3];
+	union {
+		/* When crb.c==0 (no ccb defined) it is reserved;
+		 * When crb.c==1 (ccb defined) it is cm
+		 */
+
+		uint32_t ccb_cm;
+		/* Signal interrupt of crb.c==1 and cm==1 */
+
+		uint32_t word;
+		/* generic access to the 32bit word */
+	};
+} __aligned(16);
+
+struct vas_stamped_crb_t {
+	/*
+	 * CRB operand of the paste coprocessor instruction is stamped
+	 * in quadword 4 with the information shown here as its written
+	 * in to the receive FIFO of the coprocessor
+	 */
+
+	union {
+		uint32_t vas_buf_num;
+		/* Verification only vas buffer number which correlates to
+		 * the low order bits of the atag in the paste command
+		 */
+
+		uint32_t send_wc_id;
+		/* Pointer to Send Window Context that provides for NX address
+		 * translation information, such as MSR and LPCR bits, job
+		 * completion interrupt RA, PSWID, and job utilization counter.
+		 */
+
+	};
+	union {
+		uint32_t recv_wc_id;
+		/* Pointer to Receive Window Context. NX uses this to return
+		 * credits to a Receive FIFO as entries are dequeued.
+		 */
+
+	};
+	uint32_t reserved2;
+	union {
+		uint32_t vas_invalid;
+		/* Invalid bit. If this bit is 1 the CRB is discarded by
+		 * NX upon fetching from the receive FIFO. If this bit is 0
+		 * the CRB is processed normally. The bit is stamped to 0
+		 * by VAS and may be written to 1 by hypervisor while
+		 * the CRB is in the receive FIFO (in memory).
+		 */
+
+	};
+};
+
+struct nx_stamped_fault_crb_t {
+	/*
+	 * A CRB that has a translation fault is stamped by NX in quadword 4
+	 * and pasted to the Fault Send Window in VAS.
+	 */
+	uint64_t fsa;
+	union {
+		uint32_t nxsf_t;
+		uint32_t nxsf_fs;
+	};
+	uint32_t pswid;
+};
+
+union stamped_crb_t {
+	struct vas_stamped_crb_t      vas;
+	struct nx_stamped_fault_crb_t nx;
+};
+
+struct nx_gzip_cpb_t {
+	/*
+	 * Coprocessor Parameter Block In/Out are used to pass metadata
+	 * to/from accelerator.  Tables 6.5 and 6.6 of the user manual.
+	 */
+
+	/* CPBInput */
+
+	struct {
+		union {
+		union nx_qw_t qw0;
+			struct {
+				uint32_t in_adler;            /* bits 0:31  */
+				uint32_t in_crc;              /* bits 32:63 */
+				union {
+					uint32_t in_histlen;  /* bits 64:75 */
+					uint32_t in_subc;     /* bits 93:95 */
+				};
+				union {
+					/* bits 108:111 */
+					uint32_t in_sfbt;
+					/* bits 112:127 */
+					uint32_t in_rembytecnt;
+					/* bits 116:127 */
+					uint32_t in_dhtlen;
+				};
+			};
+		};
+		union {
+			union nx_qw_t  in_dht[DHTSZ];	/* qw[1:18]     */
+			char in_dht_char[DHT_MAXSZ];	/* byte access  */
+		};
+		union nx_qw_t  reserved[5];		/* qw[19:23]    */
+	};
+
+	/* CPBOutput */
+
+	volatile struct {
+		union {
+			union nx_qw_t qw24;
+			struct {
+				uint32_t out_adler;    /* bits 0:31  qw[24] */
+				uint32_t out_crc;      /* bits 32:63 qw[24] */
+				union {
+					/* bits 77:79 qw[24] */
+					uint32_t out_tebc;
+					/* bits 80:95 qw[24] */
+					uint32_t out_subc;
+				};
+				union {
+					/* bits 108:111 qw[24] */
+					uint32_t out_sfbt;
+					/* bits 112:127 qw[24] */
+					uint32_t out_rembytecnt;
+					/* bits 116:127 qw[24] */
+					uint32_t out_dhtlen;
+				};
+			};
+		};
+		union {
+			union nx_qw_t  qw25[79];        /* qw[25:103] */
+			/* qw[25] compress no lzcounts or wrap */
+			uint32_t out_spbc_comp_wrap;
+			uint32_t out_spbc_wrap;         /* qw[25] wrap */
+			/* qw[25] compress no lzcounts */
+			uint32_t out_spbc_comp;
+			 /* 286 LL and 30 D symbol counts */
+			uint32_t out_lzcount[LLSZ+DSZ];
+			struct {
+				union nx_qw_t  out_dht[DHTSZ];  /* qw[25:42] */
+				/* qw[43] decompress */
+				uint32_t out_spbc_decomp;
+			};
+		};
+		/* qw[104] compress with lzcounts */
+		uint32_t out_spbc_comp_with_count;
+	};
+} __aligned(128);
+
+struct nx_gzip_crb_t {
+	union {                   /* byte[0:3]   */
+		uint32_t gzip_fc;     /* bits[24-31] */
+	};
+	uint32_t reserved1;       /* byte[4:7]   */
+	union {
+		uint64_t csb_address; /* byte[8:15]  */
+		struct {
+			uint32_t reserved2;
+			union {
+				uint32_t crb_c;
+				/* c==0 no ccb defined */
+
+				uint32_t crb_at;
+				/* at==0 address type is ignored;
+				 * all addrs effective assumed.
+				 */
+
+			};
+		};
+	};
+	struct nx_dde_t source_dde;           /* byte[16:31] */
+	struct nx_dde_t target_dde;           /* byte[32:47] */
+	volatile struct nx_ccb_t ccb;         /* byte[48:63] */
+	volatile union {
+		/* byte[64:239] shift csb by 128 bytes out of the crb; csb was
+		 * in crb earlier; JReilly says csb written with partial inject
+		 */
+		union nx_qw_t reserved64[11];
+		union stamped_crb_t stamp;       /* byte[64:79] */
+	};
+	volatile struct nx_csb_t csb;
+} __aligned(128);
+
+struct nx_gzip_crb_cpb_t {
+	struct nx_gzip_crb_t crb;
+	struct nx_gzip_cpb_t cpb;
+} __aligned(2048);
+
+
+/*
+ * NX hardware convention has the msb bit on the left numbered 0.
+ * The defines below has *_offset defined as the right most bit
+ * position of a field.  x of size_mask(x) is the field width in bits.
+ */
+
+#define size_mask(x)          ((1U<<(x))-1)
+
+/*
+ * Offsets and Widths within the containing 32 bits of the various NX
+ * gzip hardware registers.  Use the getnn/putnn macros to access
+ * these regs
+ */
+
+#define dde_count_mask        size_mask(8)
+#define dde_count_offset      23
+
+/* CSB */
+
+#define csb_v_mask            size_mask(1)
+#define csb_v_offset          0
+#define csb_f_mask            size_mask(1)
+#define csb_f_offset          6
+#define csb_cs_mask           size_mask(8)
+#define csb_cs_offset         15
+#define csb_cc_mask           size_mask(8)
+#define csb_cc_offset         23
+#define csb_ce_mask           size_mask(8)
+#define csb_ce_offset         31
+
+/* CCB */
+
+#define ccb_cm_mask           size_mask(3)
+#define ccb_cm_offset         31
+
+/* VAS stamped CRB fields */
+
+#define vas_buf_num_mask      size_mask(6)
+#define vas_buf_num_offset    5
+#define send_wc_id_mask       size_mask(16)
+#define send_wc_id_offset     31
+#define recv_wc_id_mask       size_mask(16)
+#define recv_wc_id_offset     31
+#define vas_invalid_mask      size_mask(1)
+#define vas_invalid_offset    31
+
+/* NX stamped fault CRB fields */
+
+#define nxsf_t_mask           size_mask(1)
+#define nxsf_t_offset         23
+#define nxsf_fs_mask          size_mask(8)
+#define nxsf_fs_offset        31
+
+/* CPB input */
+
+#define in_histlen_mask       size_mask(12)
+#define in_histlen_offset     11
+#define in_dhtlen_mask        size_mask(12)
+#define in_dhtlen_offset      31
+#define in_subc_mask          size_mask(3)
+#define in_subc_offset        31
+#define in_sfbt_mask          size_mask(4)
+#define in_sfbt_offset        15
+#define in_rembytecnt_mask    size_mask(16)
+#define in_rembytecnt_offset  31
+
+/* CPB output */
+
+#define out_tebc_mask         size_mask(3)
+#define out_tebc_offset       15
+#define out_subc_mask         size_mask(16)
+#define out_subc_offset       31
+#define out_sfbt_mask         size_mask(4)
+#define out_sfbt_offset       15
+#define out_rembytecnt_mask   size_mask(16)
+#define out_rembytecnt_offset 31
+#define out_dhtlen_mask       size_mask(12)
+#define out_dhtlen_offset     31
+
+/* CRB */
+
+#define gzip_fc_mask          size_mask(8)
+#define gzip_fc_offset        31
+#define crb_c_mask            size_mask(1)
+#define crb_c_offset          28
+#define crb_at_mask           size_mask(1)
+#define crb_at_offset         30
+#define csb_address_mask      ~(15UL) /* mask off bottom 4b */
+
+/*
+ * Access macros for the registers.  Do not access registers directly
+ * because of the endian conversion.  P9 processor may run either as
+ * Little or Big endian. However the NX coprocessor regs are always
+ * big endian.
+ * Use the 32 and 64b macros to access respective
+ * register sizes.
+ * Use nn forms for the register fields shorter than 32 bits.
+ */
+
+#define getnn(ST, REG)      ((be32toh(ST.REG) >> (31-REG##_offset)) \
+				 & REG##_mask)
+#define getpnn(ST, REG)     ((be32toh((ST)->REG) >> (31-REG##_offset)) \
+				 & REG##_mask)
+#define get32(ST, REG)      (be32toh(ST.REG))
+#define getp32(ST, REG)     (be32toh((ST)->REG))
+#define get64(ST, REG)      (be64toh(ST.REG))
+#define getp64(ST, REG)     (be64toh((ST)->REG))
+
+#define unget32(ST, REG)    (get32(ST, REG) & ~((REG##_mask) \
+				<< (31-REG##_offset)))
+/* get 32bits less the REG field */
+
+#define ungetp32(ST, REG)   (getp32(ST, REG) & ~((REG##_mask) \
+				<< (31-REG##_offset)))
+/* get 32bits less the REG field */
+
+#define clear_regs(ST)      memset((void *)(&(ST)), 0, sizeof(ST))
+#define clear_dde(ST)       do { ST.dde_count = ST.ddebc = 0; ST.ddead = 0; \
+				} while (0)
+#define clearp_dde(ST)      do { (ST)->dde_count = (ST)->ddebc = 0; \
+				 (ST)->ddead = 0; \
+				} while (0)
+#define clear_struct(ST)    memset((void *)(&(ST)), 0, sizeof(ST))
+#define putnn(ST, REG, X)   (ST.REG = htobe32(unget32(ST, REG) | (((X) \
+				 & REG##_mask) << (31-REG##_offset))))
+#define putpnn(ST, REG, X)  ((ST)->REG = htobe32(ungetp32(ST, REG) \
+				| (((X) & REG##_mask) << (31-REG##_offset))))
+
+#define put32(ST, REG, X)   (ST.REG = htobe32(X))
+#define putp32(ST, REG, X)  ((ST)->REG = htobe32(X))
+#define put64(ST, REG, X)   (ST.REG = htobe64(X))
+#define putp64(ST, REG, X)  ((ST)->REG = htobe64(X))
+
+/*
+ * Completion extension ce(0) ce(1) ce(2).  Bits ce(3-7)
+ * unused.  Section 6.6 Figure 6.7.
+ */
+
+#define get_csb_ce(ST) ((uint32_t)getnn(ST, csb_ce))
+#define get_csb_ce_ms3b(ST) (get_csb_ce(ST) >> 5)
+#define put_csb_ce_ms3b(ST, X) putnn(ST, csb_ce, ((uint32_t)(X) << 5))
+
+#define CSB_CE_PARTIAL         0x4
+#define CSB_CE_TERMINATE       0x2
+#define CSB_CE_TPBC_VALID      0x1
+
+#define csb_ce_termination(X)         (!!((X) & CSB_CE_TERMINATE))
+/* termination, output buffers may be modified, SPBC/TPBC invalid Fig.6-7 */
+
+#define csb_ce_check_completion(X)    (!csb_ce_termination(X))
+/* if not terminated then check full or partial completion */
+
+#define csb_ce_partial_completion(X)  (!!((X) & CSB_CE_PARTIAL))
+#define csb_ce_full_completion(X)     (!csb_ce_partial_completion(X))
+#define csb_ce_tpbc_valid(X)          (!!((X) & CSB_CE_TPBC_VALID))
+/* TPBC indicates successfully stored data count */
+
+#define csb_ce_default_err(X)         csb_ce_termination(X)
+/* most error CEs have CE(0)=0 and CE(1)=1 */
+
+#define csb_ce_cc3_partial(X)         csb_ce_partial_completion(X)
+/* some CC=3 are partially completed, Table 6-8 */
+
+#define csb_ce_cc64(X)                ((X)&(CSB_CE_PARTIAL \
+					| CSB_CE_TERMINATE) == 0)
+/* Compression: when TPBC>SPBC then CC=64 Table 6-8; target didn't
+ * compress smaller than source.
+ */
+
+/* Decompress SFBT combinations Tables 5-3, 6-4, 6-6 */
+
+#define SFBT_BFINAL 0x1
+#define SFBT_LIT    0x4
+#define SFBT_FHT    0x5
+#define SFBT_DHT    0x6
+#define SFBT_HDR    0x7
+
+/*
+ * NX gzip function codes. Table 6.2.
+ * Bits 0:4 are the FC. Bit 5 is used by the DMA controller to
+ * select one of the two Byte Count Limits.
+ */
+
+#define GZIP_FC_LIMIT_MASK                               0x01
+#define GZIP_FC_COMPRESS_FHT                             0x00
+#define GZIP_FC_COMPRESS_DHT                             0x02
+#define GZIP_FC_COMPRESS_FHT_COUNT                       0x04
+#define GZIP_FC_COMPRESS_DHT_COUNT                       0x06
+#define GZIP_FC_COMPRESS_RESUME_FHT                      0x08
+#define GZIP_FC_COMPRESS_RESUME_DHT                      0x0a
+#define GZIP_FC_COMPRESS_RESUME_FHT_COUNT                0x0c
+#define GZIP_FC_COMPRESS_RESUME_DHT_COUNT                0x0e
+#define GZIP_FC_DECOMPRESS                               0x10
+#define GZIP_FC_DECOMPRESS_SINGLE_BLK_N_SUSPEND          0x12
+#define GZIP_FC_DECOMPRESS_RESUME                        0x14
+#define GZIP_FC_DECOMPRESS_RESUME_SINGLE_BLK_N_SUSPEND   0x16
+#define GZIP_FC_WRAP                                     0x1e
+
+#define fc_is_compress(fc)  (((fc) & 0x10) == 0)
+#define fc_has_count(fc)    (fc_is_compress(fc) && (((fc) & 0x4) != 0))
+
+/* CSB.CC Error codes */
+
+#define ERR_NX_OK             0
+#define ERR_NX_ALIGNMENT      1
+#define ERR_NX_OPOVERLAP      2
+#define ERR_NX_DATA_LENGTH    3
+#define ERR_NX_TRANSLATION    5
+#define ERR_NX_PROTECTION     6
+#define ERR_NX_EXTERNAL_UE7   7
+#define ERR_NX_INVALID_OP     8
+#define ERR_NX_PRIVILEGE      9
+#define ERR_NX_INTERNAL_UE   10
+#define ERR_NX_EXTERN_UE_WR  12
+#define ERR_NX_TARGET_SPACE  13
+#define ERR_NX_EXCESSIVE_DDE 14
+#define ERR_NX_TRANSL_WR     15
+#define ERR_NX_PROTECT_WR    16
+#define ERR_NX_SUBFUNCTION   17
+#define ERR_NX_FUNC_ABORT    18
+#define ERR_NX_BYTE_MAX      19
+#define ERR_NX_CORRUPT_CRB   20
+#define ERR_NX_INVALID_CRB   21
+#define ERR_NX_INVALID_DDE   30
+#define ERR_NX_SEGMENTED_DDL 31
+#define ERR_NX_DDE_OVERFLOW  33
+#define ERR_NX_TPBC_GT_SPBC  64
+#define ERR_NX_MISSING_CODE  66
+#define ERR_NX_INVALID_DIST  67
+#define ERR_NX_INVALID_DHT   68
+#define ERR_NX_EXTERNAL_UE90 90
+#define ERR_NX_WDOG_TIMER   224
+#define ERR_NX_AT_FAULT     250
+#define ERR_NX_INTR_SERVER  252
+#define ERR_NX_UE253        253
+#define ERR_NX_NO_HW        254
+#define ERR_NX_HUNG_OP      255
+#define ERR_NX_END          256
+
+/* initial values for non-resume operations */
+#define INIT_CRC   0  /* crc32(0L, Z_NULL, 0) */
+#define INIT_ADLER 1  /* adler32(0L, Z_NULL, 0)  adler is initialized to 1 */
+
+/* prototypes */
+int nxu_submit_job(struct nx_gzip_crb_cpb_t *c, void *handle);
+
+extern void nxu_sigsegv_handler(int sig, siginfo_t *info, void *ctx);
+extern int nxu_touch_pages(void *buf, long buf_len, long page_len, int wr);
+
+/* caller supplies a print buffer 4*sizeof(crb) */
+
+char *nx_crb_str(struct nx_gzip_crb_t *crb, char *prbuf);
+char *nx_cpb_str(struct nx_gzip_cpb_t *cpb, char *prbuf);
+char *nx_prt_hex(void *cp, int sz, char *prbuf);
+char *nx_lzcount_str(struct nx_gzip_cpb_t *cpb, char *prbuf);
+char *nx_strerror(int e);
+
+#ifdef NX_SIM
+#include <stdio.h>
+int nx_sim_init(void *ctx);
+int nx_sim_end(void *ctx);
+int nxu_run_sim_job(struct nx_gzip_crb_cpb_t *c, void *ctx);
+#endif /* NX_SIM */
+
+/* Deflate stream manipulation */
+
+#define set_final_bit(x)	(x |= (unsigned char)1)
+#define clr_final_bit(x)	(x &= ~(unsigned char)1)
+
+#define append_empty_fh_blk(p, b) do { *(p) = (2 | (1&(b))); *((p)+1) = 0; \
+					} while (0)
+/* append 10 bits 0000001b 00...... ;
+ * assumes appending starts on a byte boundary; b is the final bit.
+ */
+
+
+#ifdef NX_842
+
+/* 842 Engine */
+
+struct nx_eft_crb_t {
+	union {                   /* byte[0:3]   */
+		uint32_t eft_fc;      /* bits[29-31] */
+	};
+	uint32_t reserved1;       /* byte[4:7]   */
+	union {
+		uint64_t csb_address; /* byte[8:15]  */
+		struct {
+			uint32_t reserved2;
+			union {
+				uint32_t crb_c;
+				/* c==0 no ccb defined */
+
+				uint32_t crb_at;
+				/* at==0 address type is ignored;
+				 * all addrs effective assumed.
+				 */
+
+			};
+		};
+	};
+	struct nx_dde_t source_dde;           /* byte[16:31] */
+	struct nx_dde_t target_dde;           /* byte[32:47] */
+	struct nx_ccb_t ccb;                  /* byte[48:63] */
+	union {
+		union nx_qw_t reserved64[3];     /* byte[64:96] */
+	};
+	struct nx_csb_t csb;
+} __aligned(128);
+
+/* 842 CRB */
+
+#define EFT_FC_MASK                 size_mask(3)
+#define EFT_FC_OFFSET               31
+#define EFT_FC_COMPRESS             0x0
+#define EFT_FC_COMPRESS_WITH_CRC    0x1
+#define EFT_FC_DECOMPRESS           0x2
+#define EFT_FC_DECOMPRESS_WITH_CRC  0x3
+#define EFT_FC_BLK_DATA_MOVE        0x4
+#endif /* NX_842 */
+
+#endif /* _NXU_H */
diff --git a/tools/testing/selftests/powerpc/nx-gzip/include/vas-api.h b/tools/testing/selftests/powerpc/nx-gzip/include/vas-api.h
new file mode 120000
index 000000000..77fb4c723
--- /dev/null
+++ b/tools/testing/selftests/powerpc/nx-gzip/include/vas-api.h
@@ -0,0 +1 @@
+../../../../../../arch/powerpc/include/uapi/asm/vas-api.h
\ No newline at end of file
diff --git a/tools/testing/selftests/powerpc/nx-gzip/nx-gzip-test.sh b/tools/testing/selftests/powerpc/nx-gzip/nx-gzip-test.sh
new file mode 100755
index 000000000..c7b46c5fd
--- /dev/null
+++ b/tools/testing/selftests/powerpc/nx-gzip/nx-gzip-test.sh
@@ -0,0 +1,46 @@
+#!/bin/bash
+# SPDX-License-Identifier: GPL-2.0-or-later
+
+if [[ ! -w /dev/crypto/nx-gzip ]]; then
+	echo "Can't access /dev/crypto/nx-gzip, skipping"
+	echo "skip: $0"
+	exit 4
+fi
+
+set -e
+
+function cleanup
+{
+	rm -f nx-tempfile*
+}
+
+trap cleanup EXIT
+
+function test_sizes
+{
+	local n=$1
+	local fname="nx-tempfile.$n"
+
+	for size in 4K 64K 1M 64M
+	do
+		echo "Testing $size ($n) ..."
+		dd if=/dev/urandom of=$fname bs=$size count=1
+		./gzfht_test $fname
+		./gunz_test ${fname}.nx.gz
+	done
+}
+
+echo "Doing basic test of different sizes ..."
+test_sizes 0
+
+echo "Running tests in parallel ..."
+for i in {1..16}
+do
+	test_sizes $i &
+done
+
+wait
+
+echo "OK"
+
+exit 0
diff --git a/tools/testing/selftests/powerpc/pmu/.gitignore b/tools/testing/selftests/powerpc/pmu/.gitignore
new file mode 100644
index 000000000..f69b1e264
--- /dev/null
+++ b/tools/testing/selftests/powerpc/pmu/.gitignore
@@ -0,0 +1,5 @@
+# SPDX-License-Identifier: GPL-2.0-only
+count_instructions
+l3_bank_test
+per_event_excludes
+count_stcx_fail
diff --git a/tools/testing/selftests/powerpc/pmu/Makefile b/tools/testing/selftests/powerpc/pmu/Makefile
new file mode 100644
index 000000000..904672fb7
--- /dev/null
+++ b/tools/testing/selftests/powerpc/pmu/Makefile
@@ -0,0 +1,50 @@
+# SPDX-License-Identifier: GPL-2.0
+noarg:
+	$(MAKE) -C ../
+
+TEST_GEN_PROGS := count_instructions count_stcx_fail l3_bank_test per_event_excludes
+EXTRA_SOURCES := ../harness.c event.c lib.c ../utils.c
+
+top_srcdir = ../../../../..
+include ../../lib.mk
+
+all: $(TEST_GEN_PROGS) ebb
+
+$(TEST_GEN_PROGS): $(EXTRA_SOURCES)
+
+# loop.S can only be built 64-bit
+$(OUTPUT)/count_instructions: CFLAGS += -m64
+$(OUTPUT)/count_instructions: loop.S count_instructions.c $(EXTRA_SOURCES)
+
+$(OUTPUT)/count_stcx_fail: CFLAGS += -m64
+$(OUTPUT)/count_stcx_fail: loop.S $(EXTRA_SOURCES)
+
+
+$(OUTPUT)/per_event_excludes: ../utils.c
+
+DEFAULT_RUN_TESTS := $(RUN_TESTS)
+override define RUN_TESTS
+	$(DEFAULT_RUN_TESTS)
+	TARGET=ebb; BUILD_TARGET=$$OUTPUT/$$TARGET; $(MAKE) OUTPUT=$$BUILD_TARGET -C $$TARGET run_tests
+endef
+
+DEFAULT_EMIT_TESTS := $(EMIT_TESTS)
+override define EMIT_TESTS
+	$(DEFAULT_EMIT_TESTS)
+	TARGET=ebb; BUILD_TARGET=$$OUTPUT/$$TARGET; $(MAKE) OUTPUT=$$BUILD_TARGET -s -C $$TARGET emit_tests
+endef
+
+DEFAULT_INSTALL_RULE := $(INSTALL_RULE)
+override define INSTALL_RULE
+	$(DEFAULT_INSTALL_RULE)
+	TARGET=ebb; BUILD_TARGET=$$OUTPUT/$$TARGET; $(MAKE) OUTPUT=$$BUILD_TARGET -C $$TARGET install
+endef
+
+clean:
+	$(RM) $(TEST_GEN_PROGS) $(OUTPUT)/loop.o
+	TARGET=ebb; BUILD_TARGET=$$OUTPUT/$$TARGET; $(MAKE) OUTPUT=$$BUILD_TARGET -C $$TARGET clean
+
+ebb:
+	TARGET=$@; BUILD_TARGET=$$OUTPUT/$$TARGET; mkdir -p $$BUILD_TARGET; $(MAKE) OUTPUT=$$BUILD_TARGET -k -C $$TARGET all
+
+.PHONY: all run_tests clean ebb
diff --git a/tools/testing/selftests/powerpc/pmu/count_instructions.c b/tools/testing/selftests/powerpc/pmu/count_instructions.c
new file mode 100644
index 000000000..a3984ef1e
--- /dev/null
+++ b/tools/testing/selftests/powerpc/pmu/count_instructions.c
@@ -0,0 +1,147 @@
+// SPDX-License-Identifier: GPL-2.0-only
+/*
+ * Copyright 2013, Michael Ellerman, IBM Corp.
+ */
+
+#define _GNU_SOURCE
+
+#include <stdio.h>
+#include <stdbool.h>
+#include <string.h>
+#include <sys/prctl.h>
+
+#include "event.h"
+#include "utils.h"
+#include "lib.h"
+
+extern void thirty_two_instruction_loop(u64 loops);
+
+static void setup_event(struct event *e, u64 config, char *name)
+{
+	event_init_opts(e, config, PERF_TYPE_HARDWARE, name);
+
+	e->attr.disabled = 1;
+	e->attr.exclude_kernel = 1;
+	e->attr.exclude_hv = 1;
+	e->attr.exclude_idle = 1;
+}
+
+static int do_count_loop(struct event *events, u64 instructions,
+			 u64 overhead, bool report)
+{
+	s64 difference, expected;
+	double percentage;
+
+	prctl(PR_TASK_PERF_EVENTS_ENABLE);
+
+	/* Run for 1M instructions */
+	thirty_two_instruction_loop(instructions >> 5);
+
+	prctl(PR_TASK_PERF_EVENTS_DISABLE);
+
+	event_read(&events[0]);
+	event_read(&events[1]);
+
+	expected = instructions + overhead;
+	difference = events[0].result.value - expected;
+	percentage = (double)difference / events[0].result.value * 100;
+
+	if (report) {
+		event_report(&events[0]);
+		event_report(&events[1]);
+
+		printf("Looped for %llu instructions, overhead %llu\n", instructions, overhead);
+		printf("Expected %llu\n", expected);
+		printf("Actual   %llu\n", events[0].result.value);
+		printf("Delta    %lld, %f%%\n", difference, percentage);
+	}
+
+	event_reset(&events[0]);
+	event_reset(&events[1]);
+
+	if (difference < 0)
+		difference = -difference;
+
+	/* Tolerate a difference below 0.0001 % */
+	difference *= 10000 * 100;
+	if (difference / events[0].result.value)
+		return -1;
+
+	return 0;
+}
+
+/* Count how many instructions it takes to do a null loop */
+static u64 determine_overhead(struct event *events)
+{
+	u64 current, overhead;
+	int i;
+
+	do_count_loop(events, 0, 0, false);
+	overhead = events[0].result.value;
+
+	for (i = 0; i < 100; i++) {
+		do_count_loop(events, 0, 0, false);
+		current = events[0].result.value;
+		if (current < overhead) {
+			printf("Replacing overhead %llu with %llu\n", overhead, current);
+			overhead = current;
+		}
+	}
+
+	return overhead;
+}
+
+static int test_body(void)
+{
+	struct event events[2];
+	u64 overhead;
+
+	setup_event(&events[0], PERF_COUNT_HW_INSTRUCTIONS, "instructions");
+	setup_event(&events[1], PERF_COUNT_HW_CPU_CYCLES, "cycles");
+
+	if (event_open(&events[0])) {
+		perror("perf_event_open");
+		return -1;
+	}
+
+	if (event_open_with_group(&events[1], events[0].fd)) {
+		perror("perf_event_open");
+		return -1;
+	}
+
+	overhead = determine_overhead(events);
+	printf("Overhead of null loop: %llu instructions\n", overhead);
+
+	/* Run for 1Mi instructions */
+	FAIL_IF(do_count_loop(events, 1000000, overhead, true));
+
+	/* Run for 10Mi instructions */
+	FAIL_IF(do_count_loop(events, 10000000, overhead, true));
+
+	/* Run for 100Mi instructions */
+	FAIL_IF(do_count_loop(events, 100000000, overhead, true));
+
+	/* Run for 1Bi instructions */
+	FAIL_IF(do_count_loop(events, 1000000000, overhead, true));
+
+	/* Run for 16Bi instructions */
+	FAIL_IF(do_count_loop(events, 16000000000, overhead, true));
+
+	/* Run for 64Bi instructions */
+	FAIL_IF(do_count_loop(events, 64000000000, overhead, true));
+
+	event_close(&events[0]);
+	event_close(&events[1]);
+
+	return 0;
+}
+
+static int count_instructions(void)
+{
+	return eat_cpu(test_body);
+}
+
+int main(void)
+{
+	return test_harness(count_instructions, "count_instructions");
+}
diff --git a/tools/testing/selftests/powerpc/pmu/count_stcx_fail.c b/tools/testing/selftests/powerpc/pmu/count_stcx_fail.c
new file mode 100644
index 000000000..2070a1e2b
--- /dev/null
+++ b/tools/testing/selftests/powerpc/pmu/count_stcx_fail.c
@@ -0,0 +1,164 @@
+/*
+ * Copyright 2013, Michael Ellerman, IBM Corp.
+ * Licensed under GPLv2.
+ */
+
+#define _GNU_SOURCE
+
+#include <stdio.h>
+#include <stdbool.h>
+#include <string.h>
+#include <sys/prctl.h>
+
+#include "event.h"
+#include "utils.h"
+#include "lib.h"
+
+extern void thirty_two_instruction_loop_with_ll_sc(u64 loops, u64 *ll_sc_target);
+
+static void setup_event(struct event *e, u64 config, int type, char *name)
+{
+	event_init_opts(e, config, type, name);
+
+	e->attr.disabled = 1;
+	e->attr.exclude_kernel = 1;
+	e->attr.exclude_hv = 1;
+	e->attr.exclude_idle = 1;
+}
+
+static int do_count_loop(struct event *events, u64 instructions,
+			 u64 overhead, bool report)
+{
+	s64 difference, expected;
+	double percentage;
+	u64 dummy;
+
+	prctl(PR_TASK_PERF_EVENTS_ENABLE);
+
+	/* Run for 1M instructions */
+	thirty_two_instruction_loop_with_ll_sc(instructions >> 5, &dummy);
+
+	prctl(PR_TASK_PERF_EVENTS_DISABLE);
+
+	event_read(&events[0]);
+	event_read(&events[1]);
+	event_read(&events[2]);
+
+	expected = instructions + overhead + (events[2].result.value * 10);
+	difference = events[0].result.value - expected;
+	percentage = (double)difference / events[0].result.value * 100;
+
+	if (report) {
+		printf("-----\n");
+		event_report(&events[0]);
+		event_report(&events[1]);
+		event_report(&events[2]);
+
+		printf("Looped for %llu instructions, overhead %llu\n", instructions, overhead);
+		printf("Expected %llu\n", expected);
+		printf("Actual   %llu\n", events[0].result.value);
+		printf("Delta    %lld, %f%%\n", difference, percentage);
+	}
+
+	event_reset(&events[0]);
+	event_reset(&events[1]);
+	event_reset(&events[2]);
+
+	if (difference < 0)
+		difference = -difference;
+
+	/* Tolerate a difference below 0.0001 % */
+	difference *= 10000 * 100;
+	if (difference / events[0].result.value)
+		return -1;
+
+	return 0;
+}
+
+/* Count how many instructions it takes to do a null loop */
+static u64 determine_overhead(struct event *events)
+{
+	u64 current, overhead;
+	int i;
+
+	do_count_loop(events, 0, 0, false);
+	overhead = events[0].result.value;
+
+	for (i = 0; i < 100; i++) {
+		do_count_loop(events, 0, 0, false);
+		current = events[0].result.value;
+		if (current < overhead) {
+			printf("Replacing overhead %llu with %llu\n", overhead, current);
+			overhead = current;
+		}
+	}
+
+	return overhead;
+}
+
+#define	PM_MRK_STCX_FAIL	0x03e158
+#define PM_STCX_FAIL	0x01e058
+
+static int test_body(void)
+{
+	struct event events[3];
+	u64 overhead;
+
+	// The STCX_FAIL event we use works on Power8 or later
+	SKIP_IF(!have_hwcap2(PPC_FEATURE2_ARCH_2_07));
+
+	setup_event(&events[0], PERF_COUNT_HW_INSTRUCTIONS, PERF_TYPE_HARDWARE, "instructions");
+	setup_event(&events[1], PERF_COUNT_HW_CPU_CYCLES, PERF_TYPE_HARDWARE, "cycles");
+	setup_event(&events[2], PM_STCX_FAIL, PERF_TYPE_RAW, "stcx_fail");
+
+	if (event_open(&events[0])) {
+		perror("perf_event_open");
+		return -1;
+	}
+
+	if (event_open_with_group(&events[1], events[0].fd)) {
+		perror("perf_event_open");
+		return -1;
+	}
+
+	if (event_open_with_group(&events[2], events[0].fd)) {
+		perror("perf_event_open");
+		return -1;
+	}
+
+	overhead = determine_overhead(events);
+	printf("Overhead of null loop: %llu instructions\n", overhead);
+
+	/* Run for 1Mi instructions */
+	FAIL_IF(do_count_loop(events, 1000000, overhead, true));
+
+	/* Run for 10Mi instructions */
+	FAIL_IF(do_count_loop(events, 10000000, overhead, true));
+
+	/* Run for 100Mi instructions */
+	FAIL_IF(do_count_loop(events, 100000000, overhead, true));
+
+	/* Run for 1Bi instructions */
+	FAIL_IF(do_count_loop(events, 1000000000, overhead, true));
+
+	/* Run for 16Bi instructions */
+	FAIL_IF(do_count_loop(events, 16000000000, overhead, true));
+
+	/* Run for 64Bi instructions */
+	FAIL_IF(do_count_loop(events, 64000000000, overhead, true));
+
+	event_close(&events[0]);
+	event_close(&events[1]);
+
+	return 0;
+}
+
+static int count_ll_sc(void)
+{
+	return eat_cpu(test_body);
+}
+
+int main(void)
+{
+	return test_harness(count_ll_sc, "count_ll_sc");
+}
diff --git a/tools/testing/selftests/powerpc/pmu/ebb/.gitignore b/tools/testing/selftests/powerpc/pmu/ebb/.gitignore
new file mode 100644
index 000000000..2920fb394
--- /dev/null
+++ b/tools/testing/selftests/powerpc/pmu/ebb/.gitignore
@@ -0,0 +1,23 @@
+# SPDX-License-Identifier: GPL-2.0-only
+reg_access_test
+event_attributes_test
+cycles_test
+cycles_with_freeze_test
+pmc56_overflow_test
+ebb_vs_cpu_event_test
+cpu_event_vs_ebb_test
+cpu_event_pinned_vs_ebb_test
+task_event_vs_ebb_test
+task_event_pinned_vs_ebb_test
+multi_ebb_procs_test
+multi_counter_test
+pmae_handling_test
+close_clears_pmcc_test
+instruction_count_test
+fork_cleanup_test
+ebb_on_child_test
+ebb_on_willing_child_test
+back_to_back_ebbs_test
+lost_exception_test
+no_handler_test
+cycles_with_mmcr2_test
diff --git a/tools/testing/selftests/powerpc/pmu/ebb/Makefile b/tools/testing/selftests/powerpc/pmu/ebb/Makefile
new file mode 100644
index 000000000..af3df79d8
--- /dev/null
+++ b/tools/testing/selftests/powerpc/pmu/ebb/Makefile
@@ -0,0 +1,37 @@
+# SPDX-License-Identifier: GPL-2.0
+include ../../../../../../scripts/Kbuild.include
+
+noarg:
+	$(MAKE) -C ../../
+
+# The EBB handler is 64-bit code and everything links against it
+CFLAGS += -m64
+
+TMPOUT = $(OUTPUT)/TMPDIR/
+# Toolchains may build PIE by default which breaks the assembly
+no-pie-option := $(call try-run, echo 'int main() { return 0; }' | \
+        $(CC) -Werror $(KBUILD_CPPFLAGS) $(CC_OPTION_CFLAGS) -no-pie -x c - -o "$$TMP", -no-pie)
+
+LDFLAGS += $(no-pie-option)
+
+TEST_GEN_PROGS := reg_access_test event_attributes_test cycles_test	\
+	 cycles_with_freeze_test pmc56_overflow_test		\
+	 ebb_vs_cpu_event_test cpu_event_vs_ebb_test		\
+	 cpu_event_pinned_vs_ebb_test task_event_vs_ebb_test	\
+	 task_event_pinned_vs_ebb_test multi_ebb_procs_test	\
+	 multi_counter_test pmae_handling_test			\
+	 close_clears_pmcc_test instruction_count_test		\
+	 fork_cleanup_test ebb_on_child_test			\
+	 ebb_on_willing_child_test back_to_back_ebbs_test	\
+	 lost_exception_test no_handler_test			\
+	 cycles_with_mmcr2_test
+
+top_srcdir = ../../../../../..
+include ../../../lib.mk
+
+$(TEST_GEN_PROGS): ../../harness.c ../../utils.c ../event.c ../lib.c \
+	       ebb.c ebb_handler.S trace.c busy_loop.S
+
+$(OUTPUT)/instruction_count_test: ../loop.S
+
+$(OUTPUT)/lost_exception_test: ../lib.c
diff --git a/tools/testing/selftests/powerpc/pmu/ebb/back_to_back_ebbs_test.c b/tools/testing/selftests/powerpc/pmu/ebb/back_to_back_ebbs_test.c
new file mode 100644
index 000000000..a26ac122c
--- /dev/null
+++ b/tools/testing/selftests/powerpc/pmu/ebb/back_to_back_ebbs_test.c
@@ -0,0 +1,106 @@
+// SPDX-License-Identifier: GPL-2.0-only
+/*
+ * Copyright 2014, Michael Ellerman, IBM Corp.
+ */
+
+#include <stdbool.h>
+#include <stdio.h>
+#include <stdlib.h>
+
+#include "ebb.h"
+
+
+#define NUMBER_OF_EBBS	50
+
+/*
+ * Test that if we overflow the counter while in the EBB handler, we take
+ * another EBB on exiting from the handler.
+ *
+ * We do this by counting with a stupidly low sample period, causing us to
+ * overflow the PMU while we're still in the EBB handler, leading to another
+ * EBB.
+ *
+ * We get out of what would otherwise be an infinite loop by leaving the
+ * counter frozen once we've taken enough EBBs.
+ */
+
+static void ebb_callee(void)
+{
+	uint64_t siar, val;
+
+	val = mfspr(SPRN_BESCR);
+	if (!(val & BESCR_PMEO)) {
+		ebb_state.stats.spurious++;
+		goto out;
+	}
+
+	ebb_state.stats.ebb_count++;
+	trace_log_counter(ebb_state.trace, ebb_state.stats.ebb_count);
+
+	/* Resets the PMC */
+	count_pmc(1, sample_period);
+
+out:
+	if (ebb_state.stats.ebb_count == NUMBER_OF_EBBS)
+		/* Reset but leave counters frozen */
+		reset_ebb_with_clear_mask(MMCR0_PMAO);
+	else
+		/* Unfreezes */
+		reset_ebb();
+
+	/* Do some stuff to chew some cycles and pop the counter */
+	siar = mfspr(SPRN_SIAR);
+	trace_log_reg(ebb_state.trace, SPRN_SIAR, siar);
+
+	val = mfspr(SPRN_PMC1);
+	trace_log_reg(ebb_state.trace, SPRN_PMC1, val);
+
+	val = mfspr(SPRN_MMCR0);
+	trace_log_reg(ebb_state.trace, SPRN_MMCR0, val);
+}
+
+int back_to_back_ebbs(void)
+{
+	struct event event;
+
+	SKIP_IF(!ebb_is_supported());
+
+	event_init_named(&event, 0x1001e, "cycles");
+	event_leader_ebb_init(&event);
+
+	event.attr.exclude_kernel = 1;
+	event.attr.exclude_hv = 1;
+	event.attr.exclude_idle = 1;
+
+	FAIL_IF(event_open(&event));
+
+	setup_ebb_handler(ebb_callee);
+
+	FAIL_IF(ebb_event_enable(&event));
+
+	sample_period = 5;
+
+	ebb_freeze_pmcs();
+	mtspr(SPRN_PMC1, pmc_sample_period(sample_period));
+	ebb_global_enable();
+	ebb_unfreeze_pmcs();
+
+	while (ebb_state.stats.ebb_count < NUMBER_OF_EBBS)
+		FAIL_IF(core_busy_loop());
+
+	ebb_global_disable();
+	ebb_freeze_pmcs();
+
+	dump_ebb_state();
+
+	event_close(&event);
+
+	FAIL_IF(ebb_state.stats.ebb_count != NUMBER_OF_EBBS);
+
+	return 0;
+}
+
+int main(void)
+{
+	return test_harness(back_to_back_ebbs, "back_to_back_ebbs");
+}
diff --git a/tools/testing/selftests/powerpc/pmu/ebb/busy_loop.S b/tools/testing/selftests/powerpc/pmu/ebb/busy_loop.S
new file mode 100644
index 000000000..4866a3a76
--- /dev/null
+++ b/tools/testing/selftests/powerpc/pmu/ebb/busy_loop.S
@@ -0,0 +1,271 @@
+/* SPDX-License-Identifier: GPL-2.0-only */
+/*
+ * Copyright 2014, Michael Ellerman, IBM Corp.
+ */
+
+#include <ppc-asm.h>
+
+	.text
+
+FUNC_START(core_busy_loop)
+	stdu	%r1, -168(%r1)
+	std	r14, 160(%r1)
+	std	r15, 152(%r1)
+	std	r16, 144(%r1)
+	std	r17, 136(%r1)
+	std	r18, 128(%r1)
+	std	r19, 120(%r1)
+	std	r20, 112(%r1)
+	std	r21, 104(%r1)
+	std	r22, 96(%r1)
+	std	r23, 88(%r1)
+	std	r24, 80(%r1)
+	std	r25, 72(%r1)
+	std	r26, 64(%r1)
+	std	r27, 56(%r1)
+	std	r28, 48(%r1)
+	std	r29, 40(%r1)
+	std	r30, 32(%r1)
+	std	r31, 24(%r1)
+
+	li	 r3, 0x3030
+	std	 r3, -96(%r1)
+	li	 r4, 0x4040
+	std	 r4, -104(%r1)
+	li	 r5, 0x5050
+	std	 r5, -112(%r1)
+	li	 r6, 0x6060
+	std	 r6, -120(%r1)
+	li	 r7, 0x7070
+	std	 r7, -128(%r1)
+	li	 r8, 0x0808
+	std	 r8, -136(%r1)
+	li	 r9, 0x0909
+	std	 r9, -144(%r1)
+	li	r10, 0x1010
+	std	r10, -152(%r1)
+	li	r11, 0x1111
+	std	r11, -160(%r1)
+	li	r14, 0x1414
+	std	r14, -168(%r1)
+	li	r15, 0x1515
+	std	r15, -176(%r1)
+	li	r16, 0x1616
+	std	r16, -184(%r1)
+	li	r17, 0x1717
+	std	r17, -192(%r1)
+	li	r18, 0x1818
+	std	r18, -200(%r1)
+	li	r19, 0x1919
+	std	r19, -208(%r1)
+	li	r20, 0x2020
+	std	r20, -216(%r1)
+	li	r21, 0x2121
+	std	r21, -224(%r1)
+	li	r22, 0x2222
+	std	r22, -232(%r1)
+	li	r23, 0x2323
+	std	r23, -240(%r1)
+	li	r24, 0x2424
+	std	r24, -248(%r1)
+	li	r25, 0x2525
+	std	r25, -256(%r1)
+	li	r26, 0x2626
+	std	r26, -264(%r1)
+	li	r27, 0x2727
+	std	r27, -272(%r1)
+	li	r28, 0x2828
+	std	r28, -280(%r1)
+	li	r29, 0x2929
+	std	r29, -288(%r1)
+	li	r30, 0x3030
+	li	r31, 0x3131
+
+	li	r3, 0
+0:	addi	r3, r3, 1
+	cmpwi	r3, 100
+	blt	0b
+
+	/* Return 1 (fail) unless we get through all the checks */
+	li	r3, 1
+
+	/* Check none of our registers have been corrupted */
+	cmpwi	r4,  0x4040
+	bne	1f
+	cmpwi	r5,  0x5050
+	bne	1f
+	cmpwi	r6,  0x6060
+	bne	1f
+	cmpwi	r7,  0x7070
+	bne	1f
+	cmpwi	r8,  0x0808
+	bne	1f
+	cmpwi	r9,  0x0909
+	bne	1f
+	cmpwi	r10, 0x1010
+	bne	1f
+	cmpwi	r11, 0x1111
+	bne	1f
+	cmpwi	r14, 0x1414
+	bne	1f
+	cmpwi	r15, 0x1515
+	bne	1f
+	cmpwi	r16, 0x1616
+	bne	1f
+	cmpwi	r17, 0x1717
+	bne	1f
+	cmpwi	r18, 0x1818
+	bne	1f
+	cmpwi	r19, 0x1919
+	bne	1f
+	cmpwi	r20, 0x2020
+	bne	1f
+	cmpwi	r21, 0x2121
+	bne	1f
+	cmpwi	r22, 0x2222
+	bne	1f
+	cmpwi	r23, 0x2323
+	bne	1f
+	cmpwi	r24, 0x2424
+	bne	1f
+	cmpwi	r25, 0x2525
+	bne	1f
+	cmpwi	r26, 0x2626
+	bne	1f
+	cmpwi	r27, 0x2727
+	bne	1f
+	cmpwi	r28, 0x2828
+	bne	1f
+	cmpwi	r29, 0x2929
+	bne	1f
+	cmpwi	r30, 0x3030
+	bne	1f
+	cmpwi	r31, 0x3131
+	bne	1f
+
+	/* Load junk into all our registers before we reload them from the stack. */
+	li	r3,  0xde
+	li	r4,  0xad
+	li	r5,  0xbe
+	li	r6,  0xef
+	li	r7,  0xde
+	li	r8,  0xad
+	li	r9,  0xbe
+	li	r10, 0xef
+	li	r11, 0xde
+	li	r14, 0xad
+	li	r15, 0xbe
+	li	r16, 0xef
+	li	r17, 0xde
+	li	r18, 0xad
+	li	r19, 0xbe
+	li	r20, 0xef
+	li	r21, 0xde
+	li	r22, 0xad
+	li	r23, 0xbe
+	li	r24, 0xef
+	li	r25, 0xde
+	li	r26, 0xad
+	li	r27, 0xbe
+	li	r28, 0xef
+	li	r29, 0xdd
+
+	ld	r3,	-96(%r1)
+	cmpwi	r3,  0x3030
+	bne	1f
+	ld	r4,	-104(%r1)
+	cmpwi	r4,  0x4040
+	bne	1f
+	ld	r5,	-112(%r1)
+	cmpwi	r5,  0x5050
+	bne	1f
+	ld	r6,	-120(%r1)
+	cmpwi	r6,  0x6060
+	bne	1f
+	ld	r7,	-128(%r1)
+	cmpwi	r7,  0x7070
+	bne	1f
+	ld	r8,	-136(%r1)
+	cmpwi	r8,  0x0808
+	bne	1f
+	ld	r9,	-144(%r1)
+	cmpwi	r9,  0x0909
+	bne	1f
+	ld	r10, -152(%r1)
+	cmpwi	r10, 0x1010
+	bne	1f
+	ld	r11, -160(%r1)
+	cmpwi	r11, 0x1111
+	bne	1f
+	ld	r14, -168(%r1)
+	cmpwi	r14, 0x1414
+	bne	1f
+	ld	r15, -176(%r1)
+	cmpwi	r15, 0x1515
+	bne	1f
+	ld	r16, -184(%r1)
+	cmpwi	r16, 0x1616
+	bne	1f
+	ld	r17, -192(%r1)
+	cmpwi	r17, 0x1717
+	bne	1f
+	ld	r18, -200(%r1)
+	cmpwi	r18, 0x1818
+	bne	1f
+	ld	r19, -208(%r1)
+	cmpwi	r19, 0x1919
+	bne	1f
+	ld	r20, -216(%r1)
+	cmpwi	r20, 0x2020
+	bne	1f
+	ld	r21, -224(%r1)
+	cmpwi	r21, 0x2121
+	bne	1f
+	ld	r22, -232(%r1)
+	cmpwi	r22, 0x2222
+	bne	1f
+	ld	r23, -240(%r1)
+	cmpwi	r23, 0x2323
+	bne	1f
+	ld	r24, -248(%r1)
+	cmpwi	r24, 0x2424
+	bne	1f
+	ld	r25, -256(%r1)
+	cmpwi	r25, 0x2525
+	bne	1f
+	ld	r26, -264(%r1)
+	cmpwi	r26, 0x2626
+	bne	1f
+	ld	r27, -272(%r1)
+	cmpwi	r27, 0x2727
+	bne	1f
+	ld	r28, -280(%r1)
+	cmpwi	r28, 0x2828
+	bne	1f
+	ld	r29, -288(%r1)
+	cmpwi	r29, 0x2929
+	bne	1f
+
+	/* Load 0 (success) to return */
+	li	r3, 0
+
+1:	ld	r14, 160(%r1)
+	ld	r15, 152(%r1)
+	ld	r16, 144(%r1)
+	ld	r17, 136(%r1)
+	ld	r18, 128(%r1)
+	ld	r19, 120(%r1)
+	ld	r20, 112(%r1)
+	ld	r21, 104(%r1)
+	ld	r22, 96(%r1)
+	ld	r23, 88(%r1)
+	ld	r24, 80(%r1)
+	ld	r25, 72(%r1)
+	ld	r26, 64(%r1)
+	ld	r27, 56(%r1)
+	ld	r28, 48(%r1)
+	ld	r29, 40(%r1)
+	ld	r30, 32(%r1)
+	ld	r31, 24(%r1)
+	addi	%r1, %r1, 168
+	blr
diff --git a/tools/testing/selftests/powerpc/pmu/ebb/close_clears_pmcc_test.c b/tools/testing/selftests/powerpc/pmu/ebb/close_clears_pmcc_test.c
new file mode 100644
index 000000000..ca9aeb0d8
--- /dev/null
+++ b/tools/testing/selftests/powerpc/pmu/ebb/close_clears_pmcc_test.c
@@ -0,0 +1,61 @@
+// SPDX-License-Identifier: GPL-2.0-only
+/*
+ * Copyright 2014, Michael Ellerman, IBM Corp.
+ */
+
+#include <stdio.h>
+#include <stdlib.h>
+#include <setjmp.h>
+#include <signal.h>
+
+#include "ebb.h"
+
+
+/*
+ * Test that closing the EBB event clears MMCR0_PMCC, preventing further access
+ * by userspace to the PMU hardware.
+ */
+
+int close_clears_pmcc(void)
+{
+	struct event event;
+
+	SKIP_IF(!ebb_is_supported());
+
+	event_init_named(&event, 0x1001e, "cycles");
+	event_leader_ebb_init(&event);
+
+	FAIL_IF(event_open(&event));
+
+	ebb_enable_pmc_counting(1);
+	setup_ebb_handler(standard_ebb_callee);
+	ebb_global_enable();
+	FAIL_IF(ebb_event_enable(&event));
+
+	mtspr(SPRN_PMC1, pmc_sample_period(sample_period));
+
+	while (ebb_state.stats.ebb_count < 1)
+		FAIL_IF(core_busy_loop());
+
+	ebb_global_disable();
+	event_close(&event);
+
+	FAIL_IF(ebb_state.stats.ebb_count == 0);
+
+	/* The real test is here, do we take a SIGILL when writing PMU regs now
+	 * that we have closed the event. We expect that we will. */
+
+	FAIL_IF(catch_sigill(write_pmc1));
+
+	/* We should still be able to read EBB regs though */
+	mfspr(SPRN_EBBHR);
+	mfspr(SPRN_EBBRR);
+	mfspr(SPRN_BESCR);
+
+	return 0;
+}
+
+int main(void)
+{
+	return test_harness(close_clears_pmcc, "close_clears_pmcc");
+}
diff --git a/tools/testing/selftests/powerpc/pmu/ebb/cpu_event_pinned_vs_ebb_test.c b/tools/testing/selftests/powerpc/pmu/ebb/cpu_event_pinned_vs_ebb_test.c
new file mode 100644
index 000000000..3cd33eb51
--- /dev/null
+++ b/tools/testing/selftests/powerpc/pmu/ebb/cpu_event_pinned_vs_ebb_test.c
@@ -0,0 +1,95 @@
+// SPDX-License-Identifier: GPL-2.0-only
+/*
+ * Copyright 2014, Michael Ellerman, IBM Corp.
+ */
+
+#include <signal.h>
+#include <stdio.h>
+#include <stdlib.h>
+#include <stdbool.h>
+#include <sys/types.h>
+#include <sys/wait.h>
+#include <unistd.h>
+
+#include "ebb.h"
+
+
+/*
+ * Tests a pinned cpu event vs an EBB - in that order. The pinned cpu event
+ * should remain and the EBB event should fail to enable.
+ */
+
+static int setup_cpu_event(struct event *event, int cpu)
+{
+	event_init_named(event, 0x400FA, "PM_RUN_INST_CMPL");
+
+	event->attr.pinned = 1;
+
+	event->attr.exclude_kernel = 1;
+	event->attr.exclude_hv = 1;
+	event->attr.exclude_idle = 1;
+
+	SKIP_IF(require_paranoia_below(1));
+	FAIL_IF(event_open_with_cpu(event, cpu));
+	FAIL_IF(event_enable(event));
+
+	return 0;
+}
+
+int cpu_event_pinned_vs_ebb(void)
+{
+	union pipe read_pipe, write_pipe;
+	struct event event;
+	int cpu, rc;
+	pid_t pid;
+
+	SKIP_IF(!ebb_is_supported());
+
+	cpu = pick_online_cpu();
+	FAIL_IF(cpu < 0);
+	FAIL_IF(bind_to_cpu(cpu));
+
+	FAIL_IF(pipe(read_pipe.fds) == -1);
+	FAIL_IF(pipe(write_pipe.fds) == -1);
+
+	pid = fork();
+	if (pid == 0) {
+		/* NB order of pipes looks reversed */
+		exit(ebb_child(write_pipe, read_pipe));
+	}
+
+	/* We setup the cpu event first */
+	rc = setup_cpu_event(&event, cpu);
+	if (rc) {
+		kill_child_and_wait(pid);
+		return rc;
+	}
+
+	/* Signal the child to install its EBB event and wait */
+	if (sync_with_child(read_pipe, write_pipe))
+		/* If it fails, wait for it to exit */
+		goto wait;
+
+	/* Signal the child to run */
+	FAIL_IF(sync_with_child(read_pipe, write_pipe));
+
+wait:
+	/* We expect it to fail to read the event */
+	FAIL_IF(wait_for_child(pid) != 2);
+
+	FAIL_IF(event_disable(&event));
+	FAIL_IF(event_read(&event));
+
+	event_report(&event);
+
+	/* The cpu event should have run */
+	FAIL_IF(event.result.value == 0);
+	FAIL_IF(event.result.enabled != event.result.running);
+
+	return 0;
+}
+
+int main(void)
+{
+	return test_harness(cpu_event_pinned_vs_ebb, "cpu_event_pinned_vs_ebb");
+}
diff --git a/tools/testing/selftests/powerpc/pmu/ebb/cpu_event_vs_ebb_test.c b/tools/testing/selftests/powerpc/pmu/ebb/cpu_event_vs_ebb_test.c
new file mode 100644
index 000000000..8466ef9d7
--- /dev/null
+++ b/tools/testing/selftests/powerpc/pmu/ebb/cpu_event_vs_ebb_test.c
@@ -0,0 +1,91 @@
+// SPDX-License-Identifier: GPL-2.0-only
+/*
+ * Copyright 2014, Michael Ellerman, IBM Corp.
+ */
+
+#include <signal.h>
+#include <stdio.h>
+#include <stdlib.h>
+#include <stdbool.h>
+#include <sys/types.h>
+#include <sys/wait.h>
+#include <unistd.h>
+
+#include "ebb.h"
+
+
+/*
+ * Tests a cpu event vs an EBB - in that order. The EBB should force the cpu
+ * event off the PMU.
+ */
+
+static int setup_cpu_event(struct event *event, int cpu)
+{
+	event_init_named(event, 0x400FA, "PM_RUN_INST_CMPL");
+
+	event->attr.exclude_kernel = 1;
+	event->attr.exclude_hv = 1;
+	event->attr.exclude_idle = 1;
+
+	SKIP_IF(require_paranoia_below(1));
+	FAIL_IF(event_open_with_cpu(event, cpu));
+	FAIL_IF(event_enable(event));
+
+	return 0;
+}
+
+int cpu_event_vs_ebb(void)
+{
+	union pipe read_pipe, write_pipe;
+	struct event event;
+	int cpu, rc;
+	pid_t pid;
+
+	SKIP_IF(!ebb_is_supported());
+
+	cpu = pick_online_cpu();
+	FAIL_IF(cpu < 0);
+	FAIL_IF(bind_to_cpu(cpu));
+
+	FAIL_IF(pipe(read_pipe.fds) == -1);
+	FAIL_IF(pipe(write_pipe.fds) == -1);
+
+	pid = fork();
+	if (pid == 0) {
+		/* NB order of pipes looks reversed */
+		exit(ebb_child(write_pipe, read_pipe));
+	}
+
+	/* We setup the cpu event first */
+	rc = setup_cpu_event(&event, cpu);
+	if (rc) {
+		kill_child_and_wait(pid);
+		return rc;
+	}
+
+	/* Signal the child to install its EBB event and wait */
+	if (sync_with_child(read_pipe, write_pipe))
+		/* If it fails, wait for it to exit */
+		goto wait;
+
+	/* Signal the child to run */
+	FAIL_IF(sync_with_child(read_pipe, write_pipe));
+
+wait:
+	/* We expect the child to succeed */
+	FAIL_IF(wait_for_child(pid));
+
+	FAIL_IF(event_disable(&event));
+	FAIL_IF(event_read(&event));
+
+	event_report(&event);
+
+	/* The cpu event may have run */
+
+	return 0;
+}
+
+int main(void)
+{
+	return test_harness(cpu_event_vs_ebb, "cpu_event_vs_ebb");
+}
diff --git a/tools/testing/selftests/powerpc/pmu/ebb/cycles_test.c b/tools/testing/selftests/powerpc/pmu/ebb/cycles_test.c
new file mode 100644
index 000000000..bb9f587fa
--- /dev/null
+++ b/tools/testing/selftests/powerpc/pmu/ebb/cycles_test.c
@@ -0,0 +1,58 @@
+// SPDX-License-Identifier: GPL-2.0-only
+/*
+ * Copyright 2014, Michael Ellerman, IBM Corp.
+ */
+
+#include <stdio.h>
+#include <stdlib.h>
+
+#include "ebb.h"
+
+
+/*
+ * Basic test that counts user cycles and takes EBBs.
+ */
+int cycles(void)
+{
+	struct event event;
+
+	SKIP_IF(!ebb_is_supported());
+
+	event_init_named(&event, 0x1001e, "cycles");
+	event_leader_ebb_init(&event);
+
+	event.attr.exclude_kernel = 1;
+	event.attr.exclude_hv = 1;
+	event.attr.exclude_idle = 1;
+
+	FAIL_IF(event_open(&event));
+
+	ebb_enable_pmc_counting(1);
+	setup_ebb_handler(standard_ebb_callee);
+	ebb_global_enable();
+	FAIL_IF(ebb_event_enable(&event));
+
+	mtspr(SPRN_PMC1, pmc_sample_period(sample_period));
+
+	while (ebb_state.stats.ebb_count < 10) {
+		FAIL_IF(core_busy_loop());
+		FAIL_IF(ebb_check_mmcr0());
+	}
+
+	ebb_global_disable();
+	ebb_freeze_pmcs();
+
+	dump_ebb_state();
+
+	event_close(&event);
+
+	FAIL_IF(ebb_state.stats.ebb_count == 0);
+	FAIL_IF(!ebb_check_count(1, sample_period, 100));
+
+	return 0;
+}
+
+int main(void)
+{
+	return test_harness(cycles, "cycles");
+}
diff --git a/tools/testing/selftests/powerpc/pmu/ebb/cycles_with_freeze_test.c b/tools/testing/selftests/powerpc/pmu/ebb/cycles_with_freeze_test.c
new file mode 100644
index 000000000..9ae795ce3
--- /dev/null
+++ b/tools/testing/selftests/powerpc/pmu/ebb/cycles_with_freeze_test.c
@@ -0,0 +1,117 @@
+// SPDX-License-Identifier: GPL-2.0-only
+/*
+ * Copyright 2014, Michael Ellerman, IBM Corp.
+ */
+
+#include <stdio.h>
+#include <stdlib.h>
+#include <stdbool.h>
+
+#include "ebb.h"
+
+
+/*
+ * Test of counting cycles while using MMCR0_FC (freeze counters) to only count
+ * parts of the code. This is complicated by the fact that FC is set by the
+ * hardware when the event overflows. We may take the EBB after we have set FC,
+ * so we have to be careful about whether we clear FC at the end of the EBB
+ * handler or not.
+ */
+
+static bool counters_frozen = false;
+static int ebbs_while_frozen = 0;
+
+static void ebb_callee(void)
+{
+	uint64_t mask, val;
+
+	mask = MMCR0_PMAO | MMCR0_FC;
+
+	val = mfspr(SPRN_BESCR);
+	if (!(val & BESCR_PMEO)) {
+		ebb_state.stats.spurious++;
+		goto out;
+	}
+
+	ebb_state.stats.ebb_count++;
+	trace_log_counter(ebb_state.trace, ebb_state.stats.ebb_count);
+
+	val = mfspr(SPRN_MMCR0);
+	trace_log_reg(ebb_state.trace, SPRN_MMCR0, val);
+
+	if (counters_frozen) {
+		trace_log_string(ebb_state.trace, "frozen");
+		ebbs_while_frozen++;
+		mask &= ~MMCR0_FC;
+	}
+
+	count_pmc(1, sample_period);
+out:
+	reset_ebb_with_clear_mask(mask);
+}
+
+int cycles_with_freeze(void)
+{
+	struct event event;
+	uint64_t val;
+	bool fc_cleared;
+
+	SKIP_IF(!ebb_is_supported());
+
+	event_init_named(&event, 0x1001e, "cycles");
+	event_leader_ebb_init(&event);
+
+	event.attr.exclude_kernel = 1;
+	event.attr.exclude_hv = 1;
+	event.attr.exclude_idle = 1;
+
+	FAIL_IF(event_open(&event));
+
+	setup_ebb_handler(ebb_callee);
+	ebb_global_enable();
+	FAIL_IF(ebb_event_enable(&event));
+
+	mtspr(SPRN_PMC1, pmc_sample_period(sample_period));
+
+	fc_cleared = false;
+
+	/* Make sure we loop until we take at least one EBB */
+	while ((ebb_state.stats.ebb_count < 20 && !fc_cleared) ||
+		ebb_state.stats.ebb_count < 1)
+	{
+		counters_frozen = false;
+		mb();
+		mtspr(SPRN_MMCR0, mfspr(SPRN_MMCR0) & ~MMCR0_FC);
+
+		FAIL_IF(core_busy_loop());
+
+		counters_frozen = true;
+		mb();
+		mtspr(SPRN_MMCR0, mfspr(SPRN_MMCR0) |  MMCR0_FC);
+
+		val = mfspr(SPRN_MMCR0);
+		if (! (val & MMCR0_FC)) {
+			printf("Outside of loop, FC NOT set MMCR0 0x%lx\n", val);
+			fc_cleared = true;
+		}
+	}
+
+	ebb_global_disable();
+	ebb_freeze_pmcs();
+
+	dump_ebb_state();
+
+	printf("EBBs while frozen %d\n", ebbs_while_frozen);
+
+	event_close(&event);
+
+	FAIL_IF(ebb_state.stats.ebb_count == 0);
+	FAIL_IF(fc_cleared);
+
+	return 0;
+}
+
+int main(void)
+{
+	return test_harness(cycles_with_freeze, "cycles_with_freeze");
+}
diff --git a/tools/testing/selftests/powerpc/pmu/ebb/cycles_with_mmcr2_test.c b/tools/testing/selftests/powerpc/pmu/ebb/cycles_with_mmcr2_test.c
new file mode 100644
index 000000000..4b45a2e70
--- /dev/null
+++ b/tools/testing/selftests/powerpc/pmu/ebb/cycles_with_mmcr2_test.c
@@ -0,0 +1,91 @@
+// SPDX-License-Identifier: GPL-2.0-only
+/*
+ * Copyright 2014, Michael Ellerman, IBM Corp.
+ */
+
+#include <stdio.h>
+#include <stdlib.h>
+#include <stdbool.h>
+
+#include "ebb.h"
+
+
+/*
+ * Test of counting cycles while manipulating the user accessible bits in MMCR2.
+ */
+
+/* We use two values because the first freezes PMC1 and so we would get no EBBs */
+#define MMCR2_EXPECTED_1 0x4020100804020000UL /* (FC1P|FC2P|FC3P|FC4P|FC5P|FC6P) */
+#define MMCR2_EXPECTED_2 0x0020100804020000UL /* (     FC2P|FC3P|FC4P|FC5P|FC6P) */
+
+
+int cycles_with_mmcr2(void)
+{
+	struct event event;
+	uint64_t val, expected[2], actual;
+	int i;
+	bool bad_mmcr2;
+
+	SKIP_IF(!ebb_is_supported());
+
+	event_init_named(&event, 0x1001e, "cycles");
+	event_leader_ebb_init(&event);
+
+	event.attr.exclude_kernel = 1;
+	event.attr.exclude_hv = 1;
+	event.attr.exclude_idle = 1;
+
+	FAIL_IF(event_open(&event));
+
+	ebb_enable_pmc_counting(1);
+	setup_ebb_handler(standard_ebb_callee);
+	ebb_global_enable();
+
+	FAIL_IF(ebb_event_enable(&event));
+
+	mtspr(SPRN_PMC1, pmc_sample_period(sample_period));
+
+	/* XXX Set of MMCR2 must be after enable */
+	expected[0] = MMCR2_EXPECTED_1;
+	expected[1] = MMCR2_EXPECTED_2;
+	i = 0;
+	bad_mmcr2 = false;
+
+	/* Make sure we loop until we take at least one EBB */
+	while ((ebb_state.stats.ebb_count < 20 && !bad_mmcr2) ||
+		ebb_state.stats.ebb_count < 1)
+	{
+		mtspr(SPRN_MMCR2, expected[i % 2]);
+
+		FAIL_IF(core_busy_loop());
+
+		val = mfspr(SPRN_MMCR2);
+		if (val != expected[i % 2]) {
+			bad_mmcr2 = true;
+			actual = val;
+		}
+
+		i++;
+	}
+
+	ebb_global_disable();
+	ebb_freeze_pmcs();
+
+	dump_ebb_state();
+
+	event_close(&event);
+
+	FAIL_IF(ebb_state.stats.ebb_count == 0);
+
+	if (bad_mmcr2)
+		printf("Bad MMCR2 value seen is 0x%lx\n", actual);
+
+	FAIL_IF(bad_mmcr2);
+
+	return 0;
+}
+
+int main(void)
+{
+	return test_harness(cycles_with_mmcr2, "cycles_with_mmcr2");
+}
diff --git a/tools/testing/selftests/powerpc/pmu/ebb/ebb.c b/tools/testing/selftests/powerpc/pmu/ebb/ebb.c
new file mode 100644
index 000000000..21537d6eb
--- /dev/null
+++ b/tools/testing/selftests/powerpc/pmu/ebb/ebb.c
@@ -0,0 +1,485 @@
+// SPDX-License-Identifier: GPL-2.0-only
+/*
+ * Copyright 2014, Michael Ellerman, IBM Corp.
+ */
+
+#define _GNU_SOURCE	/* For CPU_ZERO etc. */
+
+#include <sched.h>
+#include <sys/wait.h>
+#include <setjmp.h>
+#include <signal.h>
+#include <stdio.h>
+#include <stdlib.h>
+#include <string.h>
+#include <sys/ioctl.h>
+
+#include "trace.h"
+#include "ebb.h"
+
+
+void (*ebb_user_func)(void);
+
+void ebb_hook(void)
+{
+	if (ebb_user_func)
+		ebb_user_func();
+}
+
+struct ebb_state ebb_state;
+
+u64 sample_period = 0x40000000ull;
+
+void reset_ebb_with_clear_mask(unsigned long mmcr0_clear_mask)
+{
+	u64 val;
+
+	/* 2) clear MMCR0[PMAO] - docs say BESCR[PMEO] should do this */
+	/* 3) set MMCR0[PMAE]	- docs say BESCR[PME] should do this */
+	val = mfspr(SPRN_MMCR0);
+	mtspr(SPRN_MMCR0, (val & ~mmcr0_clear_mask) | MMCR0_PMAE);
+
+	/* 4) clear BESCR[PMEO] */
+	mtspr(SPRN_BESCRR, BESCR_PMEO);
+
+	/* 5) set BESCR[PME] */
+	mtspr(SPRN_BESCRS, BESCR_PME);
+
+	/* 6) rfebb 1 - done in our caller */
+}
+
+void reset_ebb(void)
+{
+	reset_ebb_with_clear_mask(MMCR0_PMAO | MMCR0_FC);
+}
+
+/* Called outside of the EBB handler to check MMCR0 is sane */
+int ebb_check_mmcr0(void)
+{
+	u64 val;
+
+	val = mfspr(SPRN_MMCR0);
+	if ((val & (MMCR0_FC | MMCR0_PMAO)) == MMCR0_FC) {
+		/* It's OK if we see FC & PMAO, but not FC by itself */
+		printf("Outside of loop, only FC set 0x%llx\n", val);
+		return 1;
+	}
+
+	return 0;
+}
+
+bool ebb_check_count(int pmc, u64 sample_period, int fudge)
+{
+	u64 count, upper, lower;
+
+	count = ebb_state.stats.pmc_count[PMC_INDEX(pmc)];
+
+	lower = ebb_state.stats.ebb_count * (sample_period - fudge);
+
+	if (count < lower) {
+		printf("PMC%d count (0x%llx) below lower limit 0x%llx (-0x%llx)\n",
+			pmc, count, lower, lower - count);
+		return false;
+	}
+
+	upper = ebb_state.stats.ebb_count * (sample_period + fudge);
+
+	if (count > upper) {
+		printf("PMC%d count (0x%llx) above upper limit 0x%llx (+0x%llx)\n",
+			pmc, count, upper, count - upper);
+		return false;
+	}
+
+	printf("PMC%d count (0x%llx) is between 0x%llx and 0x%llx delta +0x%llx/-0x%llx\n",
+		pmc, count, lower, upper, count - lower, upper - count);
+
+	return true;
+}
+
+void standard_ebb_callee(void)
+{
+	int found, i;
+	u64 val;
+
+	val = mfspr(SPRN_BESCR);
+	if (!(val & BESCR_PMEO)) {
+		ebb_state.stats.spurious++;
+		goto out;
+	}
+
+	ebb_state.stats.ebb_count++;
+	trace_log_counter(ebb_state.trace, ebb_state.stats.ebb_count);
+
+	val = mfspr(SPRN_MMCR0);
+	trace_log_reg(ebb_state.trace, SPRN_MMCR0, val);
+
+	found = 0;
+	for (i = 1; i <= 6; i++) {
+		if (ebb_state.pmc_enable[PMC_INDEX(i)])
+			found += count_pmc(i, sample_period);
+	}
+
+	if (!found)
+		ebb_state.stats.no_overflow++;
+
+out:
+	reset_ebb();
+}
+
+extern void ebb_handler(void);
+
+void setup_ebb_handler(void (*callee)(void))
+{
+	u64 entry;
+
+#if defined(_CALL_ELF) && _CALL_ELF == 2
+	entry = (u64)ebb_handler;
+#else
+	struct opd
+	{
+	    u64 entry;
+	    u64 toc;
+	} *opd;
+
+	opd = (struct opd *)ebb_handler;
+	entry = opd->entry;
+#endif
+	printf("EBB Handler is at %#llx\n", entry);
+
+	ebb_user_func = callee;
+
+	/* Ensure ebb_user_func is set before we set the handler */
+	mb();
+	mtspr(SPRN_EBBHR, entry);
+
+	/* Make sure the handler is set before we return */
+	mb();
+}
+
+void clear_ebb_stats(void)
+{
+	memset(&ebb_state.stats, 0, sizeof(ebb_state.stats));
+}
+
+void dump_summary_ebb_state(void)
+{
+	printf("ebb_state:\n"			\
+	       "  ebb_count    = %d\n"		\
+	       "  spurious     = %d\n"		\
+	       "  negative     = %d\n"		\
+	       "  no_overflow  = %d\n"		\
+	       "  pmc[1] count = 0x%llx\n"	\
+	       "  pmc[2] count = 0x%llx\n"	\
+	       "  pmc[3] count = 0x%llx\n"	\
+	       "  pmc[4] count = 0x%llx\n"	\
+	       "  pmc[5] count = 0x%llx\n"	\
+	       "  pmc[6] count = 0x%llx\n",
+		ebb_state.stats.ebb_count, ebb_state.stats.spurious,
+		ebb_state.stats.negative, ebb_state.stats.no_overflow,
+		ebb_state.stats.pmc_count[0], ebb_state.stats.pmc_count[1],
+		ebb_state.stats.pmc_count[2], ebb_state.stats.pmc_count[3],
+		ebb_state.stats.pmc_count[4], ebb_state.stats.pmc_count[5]);
+}
+
+static char *decode_mmcr0(u32 value)
+{
+	static char buf[16];
+
+	buf[0] = '\0';
+
+	if (value & (1 << 31))
+		strcat(buf, "FC ");
+	if (value & (1 << 26))
+		strcat(buf, "PMAE ");
+	if (value & (1 << 7))
+		strcat(buf, "PMAO ");
+
+	return buf;
+}
+
+static char *decode_bescr(u64 value)
+{
+	static char buf[16];
+
+	buf[0] = '\0';
+
+	if (value & (1ull << 63))
+		strcat(buf, "GE ");
+	if (value & (1ull << 32))
+		strcat(buf, "PMAE ");
+	if (value & 1)
+		strcat(buf, "PMAO ");
+
+	return buf;
+}
+
+void dump_ebb_hw_state(void)
+{
+	u64 bescr;
+	u32 mmcr0;
+
+	mmcr0 = mfspr(SPRN_MMCR0);
+	bescr = mfspr(SPRN_BESCR);
+
+	printf("HW state:\n"		\
+	       "MMCR0 0x%016x %s\n"	\
+	       "MMCR2 0x%016lx\n"	\
+	       "EBBHR 0x%016lx\n"	\
+	       "BESCR 0x%016llx %s\n"	\
+	       "PMC1  0x%016lx\n"	\
+	       "PMC2  0x%016lx\n"	\
+	       "PMC3  0x%016lx\n"	\
+	       "PMC4  0x%016lx\n"	\
+	       "PMC5  0x%016lx\n"	\
+	       "PMC6  0x%016lx\n"	\
+	       "SIAR  0x%016lx\n",
+	       mmcr0, decode_mmcr0(mmcr0), mfspr(SPRN_MMCR2),
+	       mfspr(SPRN_EBBHR), bescr, decode_bescr(bescr),
+	       mfspr(SPRN_PMC1), mfspr(SPRN_PMC2), mfspr(SPRN_PMC3),
+	       mfspr(SPRN_PMC4), mfspr(SPRN_PMC5), mfspr(SPRN_PMC6),
+	       mfspr(SPRN_SIAR));
+}
+
+void dump_ebb_state(void)
+{
+	dump_summary_ebb_state();
+
+	dump_ebb_hw_state();
+
+	trace_buffer_print(ebb_state.trace);
+}
+
+int count_pmc(int pmc, uint32_t sample_period)
+{
+	uint32_t start_value;
+	u64 val;
+
+	/* 0) Read PMC */
+	start_value = pmc_sample_period(sample_period);
+
+	val = read_pmc(pmc);
+	if (val < start_value)
+		ebb_state.stats.negative++;
+	else
+		ebb_state.stats.pmc_count[PMC_INDEX(pmc)] += val - start_value;
+
+	trace_log_reg(ebb_state.trace, SPRN_PMC1 + pmc - 1, val);
+
+	/* 1) Reset PMC */
+	write_pmc(pmc, start_value);
+
+	/* Report if we overflowed */
+	return val >= COUNTER_OVERFLOW;
+}
+
+int ebb_event_enable(struct event *e)
+{
+	int rc;
+
+	/* Ensure any SPR writes are ordered vs us */
+	mb();
+
+	rc = ioctl(e->fd, PERF_EVENT_IOC_ENABLE);
+	if (rc)
+		return rc;
+
+	rc = event_read(e);
+
+	/* Ditto */
+	mb();
+
+	return rc;
+}
+
+void ebb_freeze_pmcs(void)
+{
+	mtspr(SPRN_MMCR0, mfspr(SPRN_MMCR0) | MMCR0_FC);
+	mb();
+}
+
+void ebb_unfreeze_pmcs(void)
+{
+	/* Unfreeze counters */
+	mtspr(SPRN_MMCR0, mfspr(SPRN_MMCR0) & ~MMCR0_FC);
+	mb();
+}
+
+void ebb_global_enable(void)
+{
+	/* Enable EBBs globally and PMU EBBs */
+	mtspr(SPRN_BESCR, 0x8000000100000000ull);
+	mb();
+}
+
+void ebb_global_disable(void)
+{
+	/* Disable EBBs & freeze counters, events are still scheduled */
+	mtspr(SPRN_BESCRR, BESCR_PME);
+	mb();
+}
+
+bool ebb_is_supported(void)
+{
+#ifdef PPC_FEATURE2_EBB
+	/* EBB requires at least POWER8 */
+	return have_hwcap2(PPC_FEATURE2_EBB);
+#else
+	return false;
+#endif
+}
+
+void event_ebb_init(struct event *e)
+{
+	e->attr.config |= (1ull << 63);
+}
+
+void event_bhrb_init(struct event *e, unsigned ifm)
+{
+	e->attr.config |= (1ull << 62) | ((u64)ifm << 60);
+}
+
+void event_leader_ebb_init(struct event *e)
+{
+	event_ebb_init(e);
+
+	e->attr.exclusive = 1;
+	e->attr.pinned = 1;
+}
+
+int ebb_child(union pipe read_pipe, union pipe write_pipe)
+{
+	struct event event;
+	uint64_t val;
+
+	FAIL_IF(wait_for_parent(read_pipe));
+
+	event_init_named(&event, 0x1001e, "cycles");
+	event_leader_ebb_init(&event);
+
+	event.attr.exclude_kernel = 1;
+	event.attr.exclude_hv = 1;
+	event.attr.exclude_idle = 1;
+
+	FAIL_IF(event_open(&event));
+
+	ebb_enable_pmc_counting(1);
+	setup_ebb_handler(standard_ebb_callee);
+	ebb_global_enable();
+
+	FAIL_IF(event_enable(&event));
+
+	if (event_read(&event)) {
+		/*
+		 * Some tests expect to fail here, so don't report an error on
+		 * this line, and return a distinguisable error code. Tell the
+		 * parent an error happened.
+		 */
+		notify_parent_of_error(write_pipe);
+		return 2;
+	}
+
+	mtspr(SPRN_PMC1, pmc_sample_period(sample_period));
+
+	FAIL_IF(notify_parent(write_pipe));
+	FAIL_IF(wait_for_parent(read_pipe));
+	FAIL_IF(notify_parent(write_pipe));
+
+	while (ebb_state.stats.ebb_count < 20) {
+		FAIL_IF(core_busy_loop());
+
+		/* To try and hit SIGILL case */
+		val  = mfspr(SPRN_MMCRA);
+		val |= mfspr(SPRN_MMCR2);
+		val |= mfspr(SPRN_MMCR0);
+	}
+
+	ebb_global_disable();
+	ebb_freeze_pmcs();
+
+	dump_ebb_state();
+
+	event_close(&event);
+
+	FAIL_IF(ebb_state.stats.ebb_count == 0);
+
+	return 0;
+}
+
+static jmp_buf setjmp_env;
+
+static void sigill_handler(int signal)
+{
+	printf("Took sigill\n");
+	longjmp(setjmp_env, 1);
+}
+
+static struct sigaction sigill_action = {
+	.sa_handler = sigill_handler,
+};
+
+int catch_sigill(void (*func)(void))
+{
+	if (sigaction(SIGILL, &sigill_action, NULL)) {
+		perror("sigaction");
+		return 1;
+	}
+
+	if (setjmp(setjmp_env) == 0) {
+		func();
+		return 1;
+	}
+
+	return 0;
+}
+
+void write_pmc1(void)
+{
+	mtspr(SPRN_PMC1, 0);
+}
+
+void write_pmc(int pmc, u64 value)
+{
+	switch (pmc) {
+		case 1: mtspr(SPRN_PMC1, value); break;
+		case 2: mtspr(SPRN_PMC2, value); break;
+		case 3: mtspr(SPRN_PMC3, value); break;
+		case 4: mtspr(SPRN_PMC4, value); break;
+		case 5: mtspr(SPRN_PMC5, value); break;
+		case 6: mtspr(SPRN_PMC6, value); break;
+	}
+}
+
+u64 read_pmc(int pmc)
+{
+	switch (pmc) {
+		case 1: return mfspr(SPRN_PMC1);
+		case 2: return mfspr(SPRN_PMC2);
+		case 3: return mfspr(SPRN_PMC3);
+		case 4: return mfspr(SPRN_PMC4);
+		case 5: return mfspr(SPRN_PMC5);
+		case 6: return mfspr(SPRN_PMC6);
+	}
+
+	return 0;
+}
+
+static void term_handler(int signal)
+{
+	dump_summary_ebb_state();
+	dump_ebb_hw_state();
+	abort();
+}
+
+struct sigaction term_action = {
+	.sa_handler = term_handler,
+};
+
+static void __attribute__((constructor)) ebb_init(void)
+{
+	clear_ebb_stats();
+
+	if (sigaction(SIGTERM, &term_action, NULL))
+		perror("sigaction");
+
+	ebb_state.trace = trace_buffer_allocate(1 * 1024 * 1024);
+}
diff --git a/tools/testing/selftests/powerpc/pmu/ebb/ebb.h b/tools/testing/selftests/powerpc/pmu/ebb/ebb.h
new file mode 100644
index 000000000..b5bc2b616
--- /dev/null
+++ b/tools/testing/selftests/powerpc/pmu/ebb/ebb.h
@@ -0,0 +1,78 @@
+/* SPDX-License-Identifier: GPL-2.0-only */
+/*
+ * Copyright 2014, Michael Ellerman, IBM Corp.
+ */
+
+#ifndef _SELFTESTS_POWERPC_PMU_EBB_EBB_H
+#define _SELFTESTS_POWERPC_PMU_EBB_EBB_H
+
+#include "../event.h"
+#include "../lib.h"
+#include "trace.h"
+#include "reg.h"
+
+#define PMC_INDEX(pmc)	((pmc)-1)
+
+#define NUM_PMC_VALUES	128
+
+struct ebb_state
+{
+	struct {
+		u64 pmc_count[6];
+		volatile int ebb_count;
+		int spurious;
+		int negative;
+		int no_overflow;
+	} stats;
+
+	bool pmc_enable[6];
+	struct trace_buffer *trace;
+};
+
+extern struct ebb_state ebb_state;
+
+#define COUNTER_OVERFLOW 0x80000000ull
+
+static inline uint32_t pmc_sample_period(uint32_t value)
+{
+	return COUNTER_OVERFLOW - value;
+}
+
+static inline void ebb_enable_pmc_counting(int pmc)
+{
+	ebb_state.pmc_enable[PMC_INDEX(pmc)] = true;
+}
+
+bool ebb_check_count(int pmc, u64 sample_period, int fudge);
+void event_leader_ebb_init(struct event *e);
+void event_ebb_init(struct event *e);
+void event_bhrb_init(struct event *e, unsigned ifm);
+void setup_ebb_handler(void (*callee)(void));
+void standard_ebb_callee(void);
+int ebb_event_enable(struct event *e);
+void ebb_global_enable(void);
+void ebb_global_disable(void);
+bool ebb_is_supported(void);
+void ebb_freeze_pmcs(void);
+void ebb_unfreeze_pmcs(void);
+void event_ebb_init(struct event *e);
+void event_leader_ebb_init(struct event *e);
+int count_pmc(int pmc, uint32_t sample_period);
+void dump_ebb_state(void);
+void dump_summary_ebb_state(void);
+void dump_ebb_hw_state(void);
+void clear_ebb_stats(void);
+void write_pmc(int pmc, u64 value);
+u64 read_pmc(int pmc);
+void reset_ebb_with_clear_mask(unsigned long mmcr0_clear_mask);
+void reset_ebb(void);
+int ebb_check_mmcr0(void);
+
+extern u64 sample_period;
+
+int core_busy_loop(void);
+int ebb_child(union pipe read_pipe, union pipe write_pipe);
+int catch_sigill(void (*func)(void));
+void write_pmc1(void);
+
+#endif /* _SELFTESTS_POWERPC_PMU_EBB_EBB_H */
diff --git a/tools/testing/selftests/powerpc/pmu/ebb/ebb_handler.S b/tools/testing/selftests/powerpc/pmu/ebb/ebb_handler.S
new file mode 100644
index 000000000..c170398de
--- /dev/null
+++ b/tools/testing/selftests/powerpc/pmu/ebb/ebb_handler.S
@@ -0,0 +1,365 @@
+/* SPDX-License-Identifier: GPL-2.0-only */
+/*
+ * Copyright 2014, Michael Ellerman, IBM Corp.
+ */
+
+#include <ppc-asm.h>
+#include "reg.h"
+
+
+/* ppc-asm.h defines most of the reg aliases, but not r1/r2. */
+#define r1 1
+#define r2 2
+
+#define RFEBB   .long 0x4c000924
+
+/* Stack layout:
+ *
+ *                   ^
+ *  User stack       |
+ *  Back chain ------+	<- r1		<-------+
+ *  ...						|
+ *  Red zone / ABI Gap				|
+ *  ...						|
+ *  vr63	<+				|
+ *  vr0		 |				|
+ *  VSCR	 |				|
+ *  FSCR	 |				|
+ *  r31		 | Save area			|
+ *  r0		 |				|
+ *  XER		 |				|
+ *  CTR		 |				|
+ *  LR		 |				|
+ *  CCR		<+				|
+ *  ...		<+				|
+ *  LR		 | Caller frame			|
+ *  CCR		 |				|
+ *  Back chain	<+	<- updated r1	--------+
+ *
+ */
+
+#if defined(_CALL_ELF) && _CALL_ELF == 2
+#define ABIGAP		512
+#else
+#define ABIGAP		288
+#endif
+
+#define NR_GPR		32
+#define NR_SPR		6
+#define NR_VSR		64
+
+#define SAVE_AREA	((NR_GPR + NR_SPR) * 8 + (NR_VSR * 16))
+#define CALLER_FRAME	112
+
+#define STACK_FRAME	(ABIGAP + SAVE_AREA + CALLER_FRAME)
+
+#define CCR_SAVE	(CALLER_FRAME)
+#define LR_SAVE		(CCR_SAVE + 8)
+#define CTR_SAVE	(LR_SAVE  + 8)
+#define XER_SAVE	(CTR_SAVE + 8)
+#define GPR_SAVE(n)	(XER_SAVE + 8 + (8 * n))
+#define FSCR_SAVE	(GPR_SAVE(31) + 8)
+#define VSCR_SAVE	(FSCR_SAVE + 8)
+#define VSR_SAVE(n)	(VSCR_SAVE + 8 + (16 * n))
+
+#define SAVE_GPR(n)	std n,GPR_SAVE(n)(r1)
+#define REST_GPR(n)	ld  n,GPR_SAVE(n)(r1)
+#define TRASH_GPR(n)	lis n,0xaaaa
+
+#define SAVE_VSR(n, b)	li b, VSR_SAVE(n); stxvd2x n,b,r1
+#define LOAD_VSR(n, b)	li b, VSR_SAVE(n); lxvd2x  n,b,r1
+
+#define LOAD_REG_IMMEDIATE(reg,expr)	\
+	lis     reg,(expr)@highest;	\
+	ori     reg,reg,(expr)@higher;	\
+	rldicr  reg,reg,32,31;		\
+	oris    reg,reg,(expr)@h;	\
+	ori     reg,reg,(expr)@l;
+
+
+#if defined(_CALL_ELF) && _CALL_ELF == 2
+#define ENTRY_POINT(name) \
+	.type FUNC_NAME(name),@function; \
+	.globl FUNC_NAME(name); \
+	FUNC_NAME(name):
+
+#define RESTORE_TOC(name)	\
+	/* Restore our TOC pointer using our entry point */	\
+	LOAD_REG_IMMEDIATE(r12, name)				\
+0:	addis	r2,r12,(.TOC.-0b)@ha;				\
+	addi	r2,r2,(.TOC.-0b)@l;
+
+#else
+#define ENTRY_POINT(name) FUNC_START(name)
+#define RESTORE_TOC(name)	\
+	/* Restore our TOC pointer via our opd entry */	\
+	LOAD_REG_IMMEDIATE(r2, name)			\
+	ld      r2,8(r2);
+#endif
+
+    .text
+
+ENTRY_POINT(ebb_handler)
+    stdu    r1,-STACK_FRAME(r1)
+    SAVE_GPR(0)
+    mflr    r0
+    std     r0,LR_SAVE(r1)
+    mfcr    r0
+    std     r0,CCR_SAVE(r1)
+    mfctr   r0
+    std     r0,CTR_SAVE(r1)
+    mfxer   r0
+    std     r0,XER_SAVE(r1)
+    SAVE_GPR(2)
+    SAVE_GPR(3)
+    SAVE_GPR(4)
+    SAVE_GPR(5)
+    SAVE_GPR(6)
+    SAVE_GPR(7)
+    SAVE_GPR(8)
+    SAVE_GPR(9)
+    SAVE_GPR(10)
+    SAVE_GPR(11)
+    SAVE_GPR(12)
+    SAVE_GPR(13)
+    SAVE_GPR(14)
+    SAVE_GPR(15)
+    SAVE_GPR(16)
+    SAVE_GPR(17)
+    SAVE_GPR(18)
+    SAVE_GPR(19)
+    SAVE_GPR(20)
+    SAVE_GPR(21)
+    SAVE_GPR(22)
+    SAVE_GPR(23)
+    SAVE_GPR(24)
+    SAVE_GPR(25)
+    SAVE_GPR(26)
+    SAVE_GPR(27)
+    SAVE_GPR(28)
+    SAVE_GPR(29)
+    SAVE_GPR(30)
+    SAVE_GPR(31)
+    SAVE_VSR(0, r3)
+    mffs     f0
+    stfd     f0, FSCR_SAVE(r1)
+    mfvscr   f0
+    stfd     f0, VSCR_SAVE(r1)
+    SAVE_VSR(1,  r3)
+    SAVE_VSR(2,  r3)
+    SAVE_VSR(3,  r3)
+    SAVE_VSR(4,  r3)
+    SAVE_VSR(5,  r3)
+    SAVE_VSR(6,  r3)
+    SAVE_VSR(7,  r3)
+    SAVE_VSR(8,  r3)
+    SAVE_VSR(9,  r3)
+    SAVE_VSR(10, r3)
+    SAVE_VSR(11, r3)
+    SAVE_VSR(12, r3)
+    SAVE_VSR(13, r3)
+    SAVE_VSR(14, r3)
+    SAVE_VSR(15, r3)
+    SAVE_VSR(16, r3)
+    SAVE_VSR(17, r3)
+    SAVE_VSR(18, r3)
+    SAVE_VSR(19, r3)
+    SAVE_VSR(20, r3)
+    SAVE_VSR(21, r3)
+    SAVE_VSR(22, r3)
+    SAVE_VSR(23, r3)
+    SAVE_VSR(24, r3)
+    SAVE_VSR(25, r3)
+    SAVE_VSR(26, r3)
+    SAVE_VSR(27, r3)
+    SAVE_VSR(28, r3)
+    SAVE_VSR(29, r3)
+    SAVE_VSR(30, r3)
+    SAVE_VSR(31, r3)
+    SAVE_VSR(32, r3)
+    SAVE_VSR(33, r3)
+    SAVE_VSR(34, r3)
+    SAVE_VSR(35, r3)
+    SAVE_VSR(36, r3)
+    SAVE_VSR(37, r3)
+    SAVE_VSR(38, r3)
+    SAVE_VSR(39, r3)
+    SAVE_VSR(40, r3)
+    SAVE_VSR(41, r3)
+    SAVE_VSR(42, r3)
+    SAVE_VSR(43, r3)
+    SAVE_VSR(44, r3)
+    SAVE_VSR(45, r3)
+    SAVE_VSR(46, r3)
+    SAVE_VSR(47, r3)
+    SAVE_VSR(48, r3)
+    SAVE_VSR(49, r3)
+    SAVE_VSR(50, r3)
+    SAVE_VSR(51, r3)
+    SAVE_VSR(52, r3)
+    SAVE_VSR(53, r3)
+    SAVE_VSR(54, r3)
+    SAVE_VSR(55, r3)
+    SAVE_VSR(56, r3)
+    SAVE_VSR(57, r3)
+    SAVE_VSR(58, r3)
+    SAVE_VSR(59, r3)
+    SAVE_VSR(60, r3)
+    SAVE_VSR(61, r3)
+    SAVE_VSR(62, r3)
+    SAVE_VSR(63, r3)
+
+    TRASH_GPR(2)
+    TRASH_GPR(3)
+    TRASH_GPR(4)
+    TRASH_GPR(5)
+    TRASH_GPR(6)
+    TRASH_GPR(7)
+    TRASH_GPR(8)
+    TRASH_GPR(9)
+    TRASH_GPR(10)
+    TRASH_GPR(11)
+    TRASH_GPR(12)
+    TRASH_GPR(14)
+    TRASH_GPR(15)
+    TRASH_GPR(16)
+    TRASH_GPR(17)
+    TRASH_GPR(18)
+    TRASH_GPR(19)
+    TRASH_GPR(20)
+    TRASH_GPR(21)
+    TRASH_GPR(22)
+    TRASH_GPR(23)
+    TRASH_GPR(24)
+    TRASH_GPR(25)
+    TRASH_GPR(26)
+    TRASH_GPR(27)
+    TRASH_GPR(28)
+    TRASH_GPR(29)
+    TRASH_GPR(30)
+    TRASH_GPR(31)
+
+    RESTORE_TOC(ebb_handler)
+
+    /*
+     * r13 is our TLS pointer. We leave whatever value was in there when the
+     * EBB fired. That seems to be OK because once set the TLS pointer is not
+     * changed - but presumably that could change in future.
+     */
+
+    bl      ebb_hook
+    nop
+
+    /* r2 may be changed here but we don't care */
+
+    lfd      f0, FSCR_SAVE(r1)
+    mtfsf    0xff,f0
+    lfd      f0, VSCR_SAVE(r1)
+    mtvscr   f0
+    LOAD_VSR(0, r3)
+    LOAD_VSR(1,  r3)
+    LOAD_VSR(2,  r3)
+    LOAD_VSR(3,  r3)
+    LOAD_VSR(4,  r3)
+    LOAD_VSR(5,  r3)
+    LOAD_VSR(6,  r3)
+    LOAD_VSR(7,  r3)
+    LOAD_VSR(8,  r3)
+    LOAD_VSR(9,  r3)
+    LOAD_VSR(10, r3)
+    LOAD_VSR(11, r3)
+    LOAD_VSR(12, r3)
+    LOAD_VSR(13, r3)
+    LOAD_VSR(14, r3)
+    LOAD_VSR(15, r3)
+    LOAD_VSR(16, r3)
+    LOAD_VSR(17, r3)
+    LOAD_VSR(18, r3)
+    LOAD_VSR(19, r3)
+    LOAD_VSR(20, r3)
+    LOAD_VSR(21, r3)
+    LOAD_VSR(22, r3)
+    LOAD_VSR(23, r3)
+    LOAD_VSR(24, r3)
+    LOAD_VSR(25, r3)
+    LOAD_VSR(26, r3)
+    LOAD_VSR(27, r3)
+    LOAD_VSR(28, r3)
+    LOAD_VSR(29, r3)
+    LOAD_VSR(30, r3)
+    LOAD_VSR(31, r3)
+    LOAD_VSR(32, r3)
+    LOAD_VSR(33, r3)
+    LOAD_VSR(34, r3)
+    LOAD_VSR(35, r3)
+    LOAD_VSR(36, r3)
+    LOAD_VSR(37, r3)
+    LOAD_VSR(38, r3)
+    LOAD_VSR(39, r3)
+    LOAD_VSR(40, r3)
+    LOAD_VSR(41, r3)
+    LOAD_VSR(42, r3)
+    LOAD_VSR(43, r3)
+    LOAD_VSR(44, r3)
+    LOAD_VSR(45, r3)
+    LOAD_VSR(46, r3)
+    LOAD_VSR(47, r3)
+    LOAD_VSR(48, r3)
+    LOAD_VSR(49, r3)
+    LOAD_VSR(50, r3)
+    LOAD_VSR(51, r3)
+    LOAD_VSR(52, r3)
+    LOAD_VSR(53, r3)
+    LOAD_VSR(54, r3)
+    LOAD_VSR(55, r3)
+    LOAD_VSR(56, r3)
+    LOAD_VSR(57, r3)
+    LOAD_VSR(58, r3)
+    LOAD_VSR(59, r3)
+    LOAD_VSR(60, r3)
+    LOAD_VSR(61, r3)
+    LOAD_VSR(62, r3)
+    LOAD_VSR(63, r3)
+
+    ld      r0,XER_SAVE(r1)
+    mtxer   r0
+    ld      r0,CTR_SAVE(r1)
+    mtctr   r0
+    ld      r0,LR_SAVE(r1)
+    mtlr    r0
+    ld      r0,CCR_SAVE(r1)
+    mtcr    r0
+    REST_GPR(0)
+    REST_GPR(2)
+    REST_GPR(3)
+    REST_GPR(4)
+    REST_GPR(5)
+    REST_GPR(6)
+    REST_GPR(7)
+    REST_GPR(8)
+    REST_GPR(9)
+    REST_GPR(10)
+    REST_GPR(11)
+    REST_GPR(12)
+    REST_GPR(13)
+    REST_GPR(14)
+    REST_GPR(15)
+    REST_GPR(16)
+    REST_GPR(17)
+    REST_GPR(18)
+    REST_GPR(19)
+    REST_GPR(20)
+    REST_GPR(21)
+    REST_GPR(22)
+    REST_GPR(23)
+    REST_GPR(24)
+    REST_GPR(25)
+    REST_GPR(26)
+    REST_GPR(27)
+    REST_GPR(28)
+    REST_GPR(29)
+    REST_GPR(30)
+    REST_GPR(31)
+    addi    r1,r1,STACK_FRAME
+    RFEBB
+FUNC_END(ebb_handler)
diff --git a/tools/testing/selftests/powerpc/pmu/ebb/ebb_on_child_test.c b/tools/testing/selftests/powerpc/pmu/ebb/ebb_on_child_test.c
new file mode 100644
index 000000000..8980f054d
--- /dev/null
+++ b/tools/testing/selftests/powerpc/pmu/ebb/ebb_on_child_test.c
@@ -0,0 +1,88 @@
+// SPDX-License-Identifier: GPL-2.0-only
+/*
+ * Copyright 2014, Michael Ellerman, IBM Corp.
+ */
+
+#include <signal.h>
+#include <stdio.h>
+#include <stdlib.h>
+#include <stdbool.h>
+#include <sys/types.h>
+#include <sys/wait.h>
+#include <unistd.h>
+
+#include "ebb.h"
+
+
+/*
+ * Tests we can setup an EBB on our child. Nothing interesting happens, because
+ * even though the event is enabled and running the child hasn't enabled the
+ * actual delivery of the EBBs.
+ */
+
+static int victim_child(union pipe read_pipe, union pipe write_pipe)
+{
+	int i;
+
+	FAIL_IF(wait_for_parent(read_pipe));
+	FAIL_IF(notify_parent(write_pipe));
+
+	/* Parent creates EBB event */
+
+	FAIL_IF(wait_for_parent(read_pipe));
+	FAIL_IF(notify_parent(write_pipe));
+
+	/* Check the EBB is enabled by writing PMC1 */
+	write_pmc1();
+
+	/* EBB event is enabled here */
+	for (i = 0; i < 1000000; i++) ;
+
+	return 0;
+}
+
+int ebb_on_child(void)
+{
+	union pipe read_pipe, write_pipe;
+	struct event event;
+	pid_t pid;
+
+	SKIP_IF(!ebb_is_supported());
+
+	FAIL_IF(pipe(read_pipe.fds) == -1);
+	FAIL_IF(pipe(write_pipe.fds) == -1);
+
+	pid = fork();
+	if (pid == 0) {
+		/* NB order of pipes looks reversed */
+		exit(victim_child(write_pipe, read_pipe));
+	}
+
+	FAIL_IF(sync_with_child(read_pipe, write_pipe));
+
+	/* Child is running now */
+
+	event_init_named(&event, 0x1001e, "cycles");
+	event_leader_ebb_init(&event);
+
+	event.attr.exclude_kernel = 1;
+	event.attr.exclude_hv = 1;
+	event.attr.exclude_idle = 1;
+
+	FAIL_IF(event_open_with_pid(&event, pid));
+	FAIL_IF(ebb_event_enable(&event));
+
+	FAIL_IF(sync_with_child(read_pipe, write_pipe));
+
+	/* Child should just exit happily */
+	FAIL_IF(wait_for_child(pid));
+
+	event_close(&event);
+
+	return 0;
+}
+
+int main(void)
+{
+	return test_harness(ebb_on_child, "ebb_on_child");
+}
diff --git a/tools/testing/selftests/powerpc/pmu/ebb/ebb_on_willing_child_test.c b/tools/testing/selftests/powerpc/pmu/ebb/ebb_on_willing_child_test.c
new file mode 100644
index 000000000..b208bf6ad
--- /dev/null
+++ b/tools/testing/selftests/powerpc/pmu/ebb/ebb_on_willing_child_test.c
@@ -0,0 +1,92 @@
+// SPDX-License-Identifier: GPL-2.0-only
+/*
+ * Copyright 2014, Michael Ellerman, IBM Corp.
+ */
+
+#include <signal.h>
+#include <stdio.h>
+#include <stdlib.h>
+#include <stdbool.h>
+#include <sys/types.h>
+#include <sys/wait.h>
+#include <unistd.h>
+
+#include "ebb.h"
+
+
+/*
+ * Tests we can setup an EBB on our child. The child expects this and enables
+ * EBBs, which are then delivered to the child, even though the event is
+ * created by the parent.
+ */
+
+static int victim_child(union pipe read_pipe, union pipe write_pipe)
+{
+	FAIL_IF(wait_for_parent(read_pipe));
+
+	/* Setup our EBB handler, before the EBB event is created */
+	ebb_enable_pmc_counting(1);
+	setup_ebb_handler(standard_ebb_callee);
+	ebb_global_enable();
+
+	FAIL_IF(notify_parent(write_pipe));
+
+	while (ebb_state.stats.ebb_count < 20) {
+		FAIL_IF(core_busy_loop());
+	}
+
+	ebb_global_disable();
+	ebb_freeze_pmcs();
+
+	dump_ebb_state();
+
+	FAIL_IF(ebb_state.stats.ebb_count == 0);
+
+	return 0;
+}
+
+/* Tests we can setup an EBB on our child - if it's expecting it */
+int ebb_on_willing_child(void)
+{
+	union pipe read_pipe, write_pipe;
+	struct event event;
+	pid_t pid;
+
+	SKIP_IF(!ebb_is_supported());
+
+	FAIL_IF(pipe(read_pipe.fds) == -1);
+	FAIL_IF(pipe(write_pipe.fds) == -1);
+
+	pid = fork();
+	if (pid == 0) {
+		/* NB order of pipes looks reversed */
+		exit(victim_child(write_pipe, read_pipe));
+	}
+
+	/* Signal the child to setup its EBB handler */
+	FAIL_IF(sync_with_child(read_pipe, write_pipe));
+
+	/* Child is running now */
+
+	event_init_named(&event, 0x1001e, "cycles");
+	event_leader_ebb_init(&event);
+
+	event.attr.exclude_kernel = 1;
+	event.attr.exclude_hv = 1;
+	event.attr.exclude_idle = 1;
+
+	FAIL_IF(event_open_with_pid(&event, pid));
+	FAIL_IF(ebb_event_enable(&event));
+
+	/* Child show now take EBBs and then exit */
+	FAIL_IF(wait_for_child(pid));
+
+	event_close(&event);
+
+	return 0;
+}
+
+int main(void)
+{
+	return test_harness(ebb_on_willing_child, "ebb_on_willing_child");
+}
diff --git a/tools/testing/selftests/powerpc/pmu/ebb/ebb_vs_cpu_event_test.c b/tools/testing/selftests/powerpc/pmu/ebb/ebb_vs_cpu_event_test.c
new file mode 100644
index 000000000..4d822cb35
--- /dev/null
+++ b/tools/testing/selftests/powerpc/pmu/ebb/ebb_vs_cpu_event_test.c
@@ -0,0 +1,88 @@
+// SPDX-License-Identifier: GPL-2.0-only
+/*
+ * Copyright 2014, Michael Ellerman, IBM Corp.
+ */
+
+#include <signal.h>
+#include <stdio.h>
+#include <stdlib.h>
+#include <stdbool.h>
+#include <sys/types.h>
+#include <sys/wait.h>
+#include <unistd.h>
+
+#include "ebb.h"
+
+
+/*
+ * Tests an EBB vs a cpu event - in that order. The EBB should force the cpu
+ * event off the PMU.
+ */
+
+static int setup_cpu_event(struct event *event, int cpu)
+{
+	event_init_named(event, 0x400FA, "PM_RUN_INST_CMPL");
+
+	event->attr.exclude_kernel = 1;
+	event->attr.exclude_hv = 1;
+	event->attr.exclude_idle = 1;
+
+	SKIP_IF(require_paranoia_below(1));
+	FAIL_IF(event_open_with_cpu(event, cpu));
+	FAIL_IF(event_enable(event));
+
+	return 0;
+}
+
+int ebb_vs_cpu_event(void)
+{
+	union pipe read_pipe, write_pipe;
+	struct event event;
+	int cpu, rc;
+	pid_t pid;
+
+	SKIP_IF(!ebb_is_supported());
+
+	cpu = pick_online_cpu();
+	FAIL_IF(cpu < 0);
+	FAIL_IF(bind_to_cpu(cpu));
+
+	FAIL_IF(pipe(read_pipe.fds) == -1);
+	FAIL_IF(pipe(write_pipe.fds) == -1);
+
+	pid = fork();
+	if (pid == 0) {
+		/* NB order of pipes looks reversed */
+		exit(ebb_child(write_pipe, read_pipe));
+	}
+
+	/* Signal the child to install its EBB event and wait */
+	FAIL_IF(sync_with_child(read_pipe, write_pipe));
+
+	/* Now try to install our CPU event */
+	rc = setup_cpu_event(&event, cpu);
+	if (rc) {
+		kill_child_and_wait(pid);
+		return rc;
+	}
+
+	/* Signal the child to run */
+	FAIL_IF(sync_with_child(read_pipe, write_pipe));
+
+	/* .. and wait for it to complete */
+	FAIL_IF(wait_for_child(pid));
+	FAIL_IF(event_disable(&event));
+	FAIL_IF(event_read(&event));
+
+	event_report(&event);
+
+	/* The cpu event may have run, but we don't expect 100% */
+	FAIL_IF(event.result.enabled >= event.result.running);
+
+	return 0;
+}
+
+int main(void)
+{
+	return test_harness(ebb_vs_cpu_event, "ebb_vs_cpu_event");
+}
diff --git a/tools/testing/selftests/powerpc/pmu/ebb/event_attributes_test.c b/tools/testing/selftests/powerpc/pmu/ebb/event_attributes_test.c
new file mode 100644
index 000000000..6e6dd0bce
--- /dev/null
+++ b/tools/testing/selftests/powerpc/pmu/ebb/event_attributes_test.c
@@ -0,0 +1,133 @@
+// SPDX-License-Identifier: GPL-2.0-only
+/*
+ * Copyright 2014, Michael Ellerman, IBM Corp.
+ */
+
+#include <stdio.h>
+#include <stdlib.h>
+
+#include "ebb.h"
+
+
+/*
+ * Test various attributes of the EBB event are enforced.
+ */
+int event_attributes(void)
+{
+	struct event event, leader;
+
+	SKIP_IF(!ebb_is_supported());
+
+	event_init(&event, 0x1001e);
+	event_leader_ebb_init(&event);
+	/* Expected to succeed */
+	FAIL_IF(event_open(&event));
+	event_close(&event);
+
+
+	event_init(&event, 0x001e); /* CYCLES - no PMC specified */
+	event_leader_ebb_init(&event);
+	/* Expected to fail, no PMC specified */
+	FAIL_IF(event_open(&event) == 0);
+
+
+	event_init(&event, 0x2001e);
+	event_leader_ebb_init(&event);
+	event.attr.exclusive = 0;
+	/* Expected to fail, not exclusive */
+	FAIL_IF(event_open(&event) == 0);
+
+
+	event_init(&event, 0x3001e);
+	event_leader_ebb_init(&event);
+	event.attr.freq = 1;
+	/* Expected to fail, sets freq */
+	FAIL_IF(event_open(&event) == 0);
+
+
+	event_init(&event, 0x4001e);
+	event_leader_ebb_init(&event);
+	event.attr.sample_period = 1;
+	/* Expected to fail, sets sample_period */
+	FAIL_IF(event_open(&event) == 0);
+
+
+	event_init(&event, 0x1001e);
+	event_leader_ebb_init(&event);
+	event.attr.enable_on_exec = 1;
+	/* Expected to fail, sets enable_on_exec */
+	FAIL_IF(event_open(&event) == 0);
+
+
+	event_init(&event, 0x1001e);
+	event_leader_ebb_init(&event);
+	event.attr.inherit = 1;
+	/* Expected to fail, sets inherit */
+	FAIL_IF(event_open(&event) == 0);
+
+
+	event_init(&leader, 0x1001e);
+	event_leader_ebb_init(&leader);
+	FAIL_IF(event_open(&leader));
+
+	event_init(&event, 0x20002);
+	event_ebb_init(&event);
+
+	/* Expected to succeed */
+	FAIL_IF(event_open_with_group(&event, leader.fd));
+	event_close(&leader);
+	event_close(&event);
+
+
+	event_init(&leader, 0x1001e);
+	event_leader_ebb_init(&leader);
+	FAIL_IF(event_open(&leader));
+
+	event_init(&event, 0x20002);
+
+	/* Expected to fail, event doesn't request EBB, leader does */
+	FAIL_IF(event_open_with_group(&event, leader.fd) == 0);
+	event_close(&leader);
+
+
+	event_init(&leader, 0x1001e);
+	event_leader_ebb_init(&leader);
+	/* Clear the EBB flag */
+	leader.attr.config &= ~(1ull << 63);
+
+	FAIL_IF(event_open(&leader));
+
+	event_init(&event, 0x20002);
+	event_ebb_init(&event);
+
+	/* Expected to fail, leader doesn't request EBB */
+	FAIL_IF(event_open_with_group(&event, leader.fd) == 0);
+	event_close(&leader);
+
+
+	event_init(&leader, 0x1001e);
+	event_leader_ebb_init(&leader);
+	leader.attr.exclusive = 0;
+	/* Expected to fail, leader isn't exclusive */
+	FAIL_IF(event_open(&leader) == 0);
+
+
+	event_init(&leader, 0x1001e);
+	event_leader_ebb_init(&leader);
+	leader.attr.pinned = 0;
+	/* Expected to fail, leader isn't pinned */
+	FAIL_IF(event_open(&leader) == 0);
+
+	event_init(&event, 0x1001e);
+	event_leader_ebb_init(&event);
+	/* Expected to fail, not a task event */
+	SKIP_IF(require_paranoia_below(1));
+	FAIL_IF(event_open_with_cpu(&event, 0) == 0);
+
+	return 0;
+}
+
+int main(void)
+{
+	return test_harness(event_attributes, "event_attributes");
+}
diff --git a/tools/testing/selftests/powerpc/pmu/ebb/fixed_instruction_loop.S b/tools/testing/selftests/powerpc/pmu/ebb/fixed_instruction_loop.S
new file mode 100644
index 000000000..08a7b5f13
--- /dev/null
+++ b/tools/testing/selftests/powerpc/pmu/ebb/fixed_instruction_loop.S
@@ -0,0 +1,43 @@
+/* SPDX-License-Identifier: GPL-2.0-only */
+/*
+ * Copyright 2014, Michael Ellerman, IBM Corp.
+ */
+
+#include <ppc-asm.h>
+
+	.text
+
+FUNC_START(thirty_two_instruction_loop)
+	cmpwi	r3,0
+	beqlr
+	addi	r4,r3,1
+	addi	r4,r4,1
+	addi	r4,r4,1
+	addi	r4,r4,1
+	addi	r4,r4,1
+	addi	r4,r4,1
+	addi	r4,r4,1
+	addi	r4,r4,1
+	addi	r4,r4,1
+	addi	r4,r4,1
+	addi	r4,r4,1
+	addi	r4,r4,1
+	addi	r4,r4,1
+	addi	r4,r4,1
+	addi	r4,r4,1
+	addi	r4,r4,1
+	addi	r4,r4,1
+	addi	r4,r4,1
+	addi	r4,r4,1
+	addi	r4,r4,1
+	addi	r4,r4,1
+	addi	r4,r4,1
+	addi	r4,r4,1
+	addi	r4,r4,1
+	addi	r4,r4,1
+	addi	r4,r4,1
+	addi	r4,r4,1
+	addi	r4,r4,1	# 28 addi's
+	subi	r3,r3,1
+	b	FUNC_NAME(thirty_two_instruction_loop)
+FUNC_END(thirty_two_instruction_loop)
diff --git a/tools/testing/selftests/powerpc/pmu/ebb/fork_cleanup_test.c b/tools/testing/selftests/powerpc/pmu/ebb/fork_cleanup_test.c
new file mode 100644
index 000000000..2b25b5545
--- /dev/null
+++ b/tools/testing/selftests/powerpc/pmu/ebb/fork_cleanup_test.c
@@ -0,0 +1,80 @@
+// SPDX-License-Identifier: GPL-2.0-only
+/*
+ * Copyright 2014, Michael Ellerman, IBM Corp.
+ */
+
+#include <signal.h>
+#include <stdio.h>
+#include <stdlib.h>
+#include <stdbool.h>
+#include <sys/types.h>
+#include <sys/wait.h>
+#include <unistd.h>
+#include <setjmp.h>
+
+#include "ebb.h"
+
+
+/*
+ * Test that a fork clears the PMU state of the child. eg. BESCR/EBBHR/EBBRR
+ * are cleared, and MMCR0_PMCC is reset, preventing the child from accessing
+ * the PMU.
+ */
+
+static struct event event;
+
+static int child(void)
+{
+	/* Even though we have EBE=0 we can still see the EBB regs */
+	FAIL_IF(mfspr(SPRN_BESCR) != 0);
+	FAIL_IF(mfspr(SPRN_EBBHR) != 0);
+	FAIL_IF(mfspr(SPRN_EBBRR) != 0);
+
+	FAIL_IF(catch_sigill(write_pmc1));
+
+	/* We can still read from the event, though it is on our parent */
+	FAIL_IF(event_read(&event));
+
+	return 0;
+}
+
+/* Tests that fork clears EBB state */
+int fork_cleanup(void)
+{
+	pid_t pid;
+
+	SKIP_IF(!ebb_is_supported());
+
+	event_init_named(&event, 0x1001e, "cycles");
+	event_leader_ebb_init(&event);
+
+	FAIL_IF(event_open(&event));
+
+	ebb_enable_pmc_counting(1);
+	setup_ebb_handler(standard_ebb_callee);
+	ebb_global_enable();
+
+	FAIL_IF(ebb_event_enable(&event));
+
+	mtspr(SPRN_MMCR0, MMCR0_FC);
+	mtspr(SPRN_PMC1, pmc_sample_period(sample_period));
+
+	/* Don't need to actually take any EBBs */
+
+	pid = fork();
+	if (pid == 0)
+		exit(child());
+
+	/* Child does the actual testing */
+	FAIL_IF(wait_for_child(pid));
+
+	/* After fork */
+	event_close(&event);
+
+	return 0;
+}
+
+int main(void)
+{
+	return test_harness(fork_cleanup, "fork_cleanup");
+}
diff --git a/tools/testing/selftests/powerpc/pmu/ebb/instruction_count_test.c b/tools/testing/selftests/powerpc/pmu/ebb/instruction_count_test.c
new file mode 100644
index 000000000..eed338b18
--- /dev/null
+++ b/tools/testing/selftests/powerpc/pmu/ebb/instruction_count_test.c
@@ -0,0 +1,167 @@
+// SPDX-License-Identifier: GPL-2.0-only
+/*
+ * Copyright 2014, Michael Ellerman, IBM Corp.
+ */
+
+#define _GNU_SOURCE
+
+#include <stdio.h>
+#include <stdbool.h>
+#include <string.h>
+#include <sys/prctl.h>
+
+#include "ebb.h"
+
+
+/*
+ * Run a calibrated instruction loop and count instructions executed using
+ * EBBs. Make sure the counts look right.
+ */
+
+extern void thirty_two_instruction_loop(uint64_t loops);
+
+static bool counters_frozen = true;
+
+static int do_count_loop(struct event *event, uint64_t instructions,
+			 uint64_t overhead, bool report)
+{
+	int64_t difference, expected;
+	double percentage;
+
+	clear_ebb_stats();
+
+	counters_frozen = false;
+	mb();
+	mtspr(SPRN_MMCR0, mfspr(SPRN_MMCR0) & ~MMCR0_FC);
+
+	thirty_two_instruction_loop(instructions >> 5);
+
+	counters_frozen = true;
+	mb();
+	mtspr(SPRN_MMCR0, mfspr(SPRN_MMCR0) | MMCR0_FC);
+
+	count_pmc(4, sample_period);
+
+	event->result.value = ebb_state.stats.pmc_count[4-1];
+	expected = instructions + overhead;
+	difference = event->result.value - expected;
+	percentage = (double)difference / event->result.value * 100;
+
+	if (report) {
+		printf("Looped for %lu instructions, overhead %lu\n", instructions, overhead);
+		printf("Expected %lu\n", expected);
+		printf("Actual   %llu\n", event->result.value);
+		printf("Delta    %ld, %f%%\n", difference, percentage);
+		printf("Took %d EBBs\n", ebb_state.stats.ebb_count);
+	}
+
+	if (difference < 0)
+		difference = -difference;
+
+	/* Tolerate a difference of up to 0.0001 % */
+	difference *= 10000 * 100;
+	if (difference / event->result.value)
+		return -1;
+
+	return 0;
+}
+
+/* Count how many instructions it takes to do a null loop */
+static uint64_t determine_overhead(struct event *event)
+{
+	uint64_t current, overhead;
+	int i;
+
+	do_count_loop(event, 0, 0, false);
+	overhead = event->result.value;
+
+	for (i = 0; i < 100; i++) {
+		do_count_loop(event, 0, 0, false);
+		current = event->result.value;
+		if (current < overhead) {
+			printf("Replacing overhead %lu with %lu\n", overhead, current);
+			overhead = current;
+		}
+	}
+
+	return overhead;
+}
+
+static void pmc4_ebb_callee(void)
+{
+	uint64_t val;
+
+	val = mfspr(SPRN_BESCR);
+	if (!(val & BESCR_PMEO)) {
+		ebb_state.stats.spurious++;
+		goto out;
+	}
+
+	ebb_state.stats.ebb_count++;
+	count_pmc(4, sample_period);
+out:
+	if (counters_frozen)
+		reset_ebb_with_clear_mask(MMCR0_PMAO);
+	else
+		reset_ebb();
+}
+
+int instruction_count(void)
+{
+	struct event event;
+	uint64_t overhead;
+
+	SKIP_IF(!ebb_is_supported());
+
+	event_init_named(&event, 0x400FA, "PM_RUN_INST_CMPL");
+	event_leader_ebb_init(&event);
+	event.attr.exclude_kernel = 1;
+	event.attr.exclude_hv = 1;
+	event.attr.exclude_idle = 1;
+
+	FAIL_IF(event_open(&event));
+	FAIL_IF(ebb_event_enable(&event));
+
+	sample_period = COUNTER_OVERFLOW;
+
+	setup_ebb_handler(pmc4_ebb_callee);
+	mtspr(SPRN_MMCR0, mfspr(SPRN_MMCR0) & ~MMCR0_FC);
+	ebb_global_enable();
+
+	overhead = determine_overhead(&event);
+	printf("Overhead of null loop: %lu instructions\n", overhead);
+
+	/* Run for 1M instructions */
+	FAIL_IF(do_count_loop(&event, 0x100000, overhead, true));
+
+	/* Run for 10M instructions */
+	FAIL_IF(do_count_loop(&event, 0xa00000, overhead, true));
+
+	/* Run for 100M instructions */
+	FAIL_IF(do_count_loop(&event, 0x6400000, overhead, true));
+
+	/* Run for 1G instructions */
+	FAIL_IF(do_count_loop(&event, 0x40000000, overhead, true));
+
+	/* Run for 16G instructions */
+	FAIL_IF(do_count_loop(&event, 0x400000000, overhead, true));
+
+	/* Run for 64G instructions */
+	FAIL_IF(do_count_loop(&event, 0x1000000000, overhead, true));
+
+	/* Run for 128G instructions */
+	FAIL_IF(do_count_loop(&event, 0x2000000000, overhead, true));
+
+	ebb_global_disable();
+	event_close(&event);
+
+	printf("Finished OK\n");
+
+	return 0;
+}
+
+int main(void)
+{
+	test_harness_set_timeout(300);
+	return test_harness(instruction_count, "instruction_count");
+}
diff --git a/tools/testing/selftests/powerpc/pmu/ebb/lost_exception_test.c b/tools/testing/selftests/powerpc/pmu/ebb/lost_exception_test.c
new file mode 100644
index 000000000..ba2681a12
--- /dev/null
+++ b/tools/testing/selftests/powerpc/pmu/ebb/lost_exception_test.c
@@ -0,0 +1,102 @@
+// SPDX-License-Identifier: GPL-2.0-only
+/*
+ * Copyright 2014, Michael Ellerman, IBM Corp.
+ */
+
+#include <sched.h>
+#include <signal.h>
+#include <stdio.h>
+#include <stdlib.h>
+#include <sys/mman.h>
+
+#include "ebb.h"
+
+
+/*
+ * Test that tries to trigger CPU_FTR_PMAO_BUG. Which is a hardware defect
+ * where an exception triggers but we context switch before it is delivered and
+ * lose the exception.
+ */
+
+static int test_body(void)
+{
+	int i, orig_period, max_period;
+	struct event event;
+
+	SKIP_IF(!ebb_is_supported());
+
+	/* We use PMC4 to make sure the kernel switches all counters correctly */
+	event_init_named(&event, 0x40002, "instructions");
+	event_leader_ebb_init(&event);
+
+	event.attr.exclude_kernel = 1;
+	event.attr.exclude_hv = 1;
+	event.attr.exclude_idle = 1;
+
+	FAIL_IF(event_open(&event));
+
+	ebb_enable_pmc_counting(4);
+	setup_ebb_handler(standard_ebb_callee);
+	ebb_global_enable();
+	FAIL_IF(ebb_event_enable(&event));
+
+	/*
+	 * We want a low sample period, but we also want to get out of the EBB
+	 * handler without tripping up again.
+	 *
+	 * This value picked after much experimentation.
+	 */
+	orig_period = max_period = sample_period = 400;
+
+	mtspr(SPRN_PMC4, pmc_sample_period(sample_period));
+
+	while (ebb_state.stats.ebb_count < 1000000) {
+		/*
+		 * We are trying to get the EBB exception to race exactly with
+		 * us entering the kernel to do the syscall. We then need the
+		 * kernel to decide our timeslice is up and context switch to
+		 * the other thread. When we come back our EBB will have been
+		 * lost and we'll spin in this while loop forever.
+		 */
+
+		for (i = 0; i < 100000; i++)
+			sched_yield();
+
+		/* Change the sample period slightly to try and hit the race */
+		if (sample_period >= (orig_period + 200))
+			sample_period = orig_period;
+		else
+			sample_period++;
+
+		if (sample_period > max_period)
+			max_period = sample_period;
+	}
+
+	ebb_freeze_pmcs();
+	ebb_global_disable();
+
+	mtspr(SPRN_PMC4, 0xdead);
+
+	dump_summary_ebb_state();
+	dump_ebb_hw_state();
+
+	event_close(&event);
+
+	FAIL_IF(ebb_state.stats.ebb_count == 0);
+
+	/* We vary our sample period so we need extra fudge here */
+	FAIL_IF(!ebb_check_count(4, orig_period, 2 * (max_period - orig_period)));
+
+	return 0;
+}
+
+static int lost_exception(void)
+{
+	return eat_cpu(test_body);
+}
+
+int main(void)
+{
+	test_harness_set_timeout(300);
+	return test_harness(lost_exception, "lost_exception");
+}
diff --git a/tools/testing/selftests/powerpc/pmu/ebb/multi_counter_test.c b/tools/testing/selftests/powerpc/pmu/ebb/multi_counter_test.c
new file mode 100644
index 000000000..791d37ba3
--- /dev/null
+++ b/tools/testing/selftests/powerpc/pmu/ebb/multi_counter_test.c
@@ -0,0 +1,86 @@
+// SPDX-License-Identifier: GPL-2.0-only
+/*
+ * Copyright 2014, Michael Ellerman, IBM Corp.
+ */
+
+#include <stdio.h>
+#include <stdlib.h>
+#include <sys/ioctl.h>
+
+#include "ebb.h"
+
+
+/*
+ * Test counting multiple events using EBBs.
+ */
+int multi_counter(void)
+{
+	struct event events[6];
+	int i, group_fd;
+
+	SKIP_IF(!ebb_is_supported());
+
+	event_init_named(&events[0], 0x1001C, "PM_CMPLU_STALL_THRD");
+	event_init_named(&events[1], 0x2D016, "PM_CMPLU_STALL_FXU");
+	event_init_named(&events[2], 0x30006, "PM_CMPLU_STALL_OTHER_CMPL");
+	event_init_named(&events[3], 0x4000A, "PM_CMPLU_STALL");
+	event_init_named(&events[4], 0x600f4, "PM_RUN_CYC");
+	event_init_named(&events[5], 0x500fa, "PM_RUN_INST_CMPL");
+
+	event_leader_ebb_init(&events[0]);
+	for (i = 1; i < 6; i++)
+		event_ebb_init(&events[i]);
+
+	group_fd = -1;
+	for (i = 0; i < 6; i++) {
+		events[i].attr.exclude_kernel = 1;
+		events[i].attr.exclude_hv = 1;
+		events[i].attr.exclude_idle = 1;
+
+		FAIL_IF(event_open_with_group(&events[i], group_fd));
+		if (group_fd == -1)
+			group_fd = events[0].fd;
+	}
+
+	ebb_enable_pmc_counting(1);
+	ebb_enable_pmc_counting(2);
+	ebb_enable_pmc_counting(3);
+	ebb_enable_pmc_counting(4);
+	ebb_enable_pmc_counting(5);
+	ebb_enable_pmc_counting(6);
+	setup_ebb_handler(standard_ebb_callee);
+
+	FAIL_IF(ioctl(events[0].fd, PERF_EVENT_IOC_ENABLE, PERF_IOC_FLAG_GROUP));
+	FAIL_IF(event_read(&events[0]));
+
+	ebb_global_enable();
+
+	mtspr(SPRN_PMC1, pmc_sample_period(sample_period));
+	mtspr(SPRN_PMC2, pmc_sample_period(sample_period));
+	mtspr(SPRN_PMC3, pmc_sample_period(sample_period));
+	mtspr(SPRN_PMC4, pmc_sample_period(sample_period));
+	mtspr(SPRN_PMC5, pmc_sample_period(sample_period));
+	mtspr(SPRN_PMC6, pmc_sample_period(sample_period));
+
+	while (ebb_state.stats.ebb_count < 50) {
+		FAIL_IF(core_busy_loop());
+		FAIL_IF(ebb_check_mmcr0());
+	}
+
+	ebb_global_disable();
+	ebb_freeze_pmcs();
+
+	dump_ebb_state();
+
+	for (i = 0; i < 6; i++)
+		event_close(&events[i]);
+
+	FAIL_IF(ebb_state.stats.ebb_count == 0);
+
+	return 0;
+}
+
+int main(void)
+{
+	return test_harness(multi_counter, "multi_counter");
+}
diff --git a/tools/testing/selftests/powerpc/pmu/ebb/multi_ebb_procs_test.c b/tools/testing/selftests/powerpc/pmu/ebb/multi_ebb_procs_test.c
new file mode 100644
index 000000000..9b0f70d59
--- /dev/null
+++ b/tools/testing/selftests/powerpc/pmu/ebb/multi_ebb_procs_test.c
@@ -0,0 +1,109 @@
+// SPDX-License-Identifier: GPL-2.0-only
+/*
+ * Copyright 2014, Michael Ellerman, IBM Corp.
+ */
+
+#include <stdbool.h>
+#include <stdio.h>
+#include <stdlib.h>
+#include <signal.h>
+
+#include "ebb.h"
+
+
+/*
+ * Test running multiple EBB using processes at once on a single CPU. They
+ * should all run happily without interfering with each other.
+ */
+
+static bool child_should_exit;
+
+static void sigint_handler(int signal)
+{
+	child_should_exit = true;
+}
+
+struct sigaction sigint_action = {
+	.sa_handler = sigint_handler,
+};
+
+static int cycles_child(void)
+{
+	struct event event;
+
+	if (sigaction(SIGINT, &sigint_action, NULL)) {
+		perror("sigaction");
+		return 1;
+	}
+
+	event_init_named(&event, 0x1001e, "cycles");
+	event_leader_ebb_init(&event);
+
+	event.attr.exclude_kernel = 1;
+	event.attr.exclude_hv = 1;
+	event.attr.exclude_idle = 1;
+
+	FAIL_IF(event_open(&event));
+
+	ebb_enable_pmc_counting(1);
+	setup_ebb_handler(standard_ebb_callee);
+	ebb_global_enable();
+
+	FAIL_IF(ebb_event_enable(&event));
+
+	mtspr(SPRN_PMC1, pmc_sample_period(sample_period));
+
+	while (!child_should_exit) {
+		FAIL_IF(core_busy_loop());
+		FAIL_IF(ebb_check_mmcr0());
+	}
+
+	ebb_global_disable();
+	ebb_freeze_pmcs();
+
+	dump_summary_ebb_state();
+
+	event_close(&event);
+
+	FAIL_IF(ebb_state.stats.ebb_count == 0);
+
+	return 0;
+}
+
+#define NR_CHILDREN	4
+
+int multi_ebb_procs(void)
+{
+	pid_t pids[NR_CHILDREN];
+	int cpu, rc, i;
+
+	SKIP_IF(!ebb_is_supported());
+
+	cpu = pick_online_cpu();
+	FAIL_IF(cpu < 0);
+	FAIL_IF(bind_to_cpu(cpu));
+
+	for (i = 0; i < NR_CHILDREN; i++) {
+		pids[i] = fork();
+		if (pids[i] == 0)
+			exit(cycles_child());
+	}
+
+	/* Have them all run for "a while" */
+	sleep(10);
+
+	rc = 0;
+	for (i = 0; i < NR_CHILDREN; i++) {
+		/* Tell them to stop */
+		kill(pids[i], SIGINT);
+		/* And wait */
+		rc |= wait_for_child(pids[i]);
+	}
+
+	return rc;
+}
+
+int main(void)
+{
+	return test_harness(multi_ebb_procs, "multi_ebb_procs");
+}
diff --git a/tools/testing/selftests/powerpc/pmu/ebb/no_handler_test.c b/tools/testing/selftests/powerpc/pmu/ebb/no_handler_test.c
new file mode 100644
index 000000000..01e827c31
--- /dev/null
+++ b/tools/testing/selftests/powerpc/pmu/ebb/no_handler_test.c
@@ -0,0 +1,61 @@
+// SPDX-License-Identifier: GPL-2.0-only
+/*
+ * Copyright 2014, Michael Ellerman, IBM Corp.
+ */
+
+#include <stdio.h>
+#include <stdlib.h>
+#include <setjmp.h>
+#include <signal.h>
+
+#include "ebb.h"
+
+
+/* Test that things work sanely if we have no handler */
+
+static int no_handler_test(void)
+{
+	struct event event;
+	u64 val;
+	int i;
+
+	SKIP_IF(!ebb_is_supported());
+
+	event_init_named(&event, 0x1001e, "cycles");
+	event_leader_ebb_init(&event);
+
+	event.attr.exclude_kernel = 1;
+	event.attr.exclude_hv = 1;
+	event.attr.exclude_idle = 1;
+
+	FAIL_IF(event_open(&event));
+	FAIL_IF(ebb_event_enable(&event));
+
+	val = mfspr(SPRN_EBBHR);
+	FAIL_IF(val != 0);
+
+	/* Make sure it overflows quickly */
+	sample_period = 1000;
+	mtspr(SPRN_PMC1, pmc_sample_period(sample_period));
+
+	/* Spin to make sure the event has time to overflow */
+	for (i = 0; i < 1000; i++)
+		mb();
+
+	dump_ebb_state();
+
+	/* We expect to see the PMU frozen & PMAO set */
+	val = mfspr(SPRN_MMCR0);
+	FAIL_IF(val != 0x0000000080000080);
+
+	event_close(&event);
+
+	/* The real test is that we never took an EBB at 0x0 */
+
+	return 0;
+}
+
+int main(void)
+{
+	return test_harness(no_handler_test,"no_handler_test");
+}
diff --git a/tools/testing/selftests/powerpc/pmu/ebb/pmae_handling_test.c b/tools/testing/selftests/powerpc/pmu/ebb/pmae_handling_test.c
new file mode 100644
index 000000000..2904c741e
--- /dev/null
+++ b/tools/testing/selftests/powerpc/pmu/ebb/pmae_handling_test.c
@@ -0,0 +1,106 @@
+// SPDX-License-Identifier: GPL-2.0-only
+/*
+ * Copyright 2014, Michael Ellerman, IBM Corp.
+ */
+
+#include <sched.h>
+#include <signal.h>
+#include <stdbool.h>
+#include <stdio.h>
+#include <stdlib.h>
+
+#include "ebb.h"
+
+
+/*
+ * Test that the kernel properly handles PMAE across context switches.
+ *
+ * We test this by calling into the kernel inside our EBB handler, where PMAE
+ * is clear. A cpu eater companion thread is running on the same CPU as us to
+ * encourage the scheduler to switch us.
+ *
+ * The kernel must make sure that when it context switches us back in, it
+ * honours the fact that we had PMAE clear.
+ *
+ * Observed to hit the failing case on the first EBB with a broken kernel.
+ */
+
+static bool mmcr0_mismatch;
+static uint64_t before, after;
+
+static void syscall_ebb_callee(void)
+{
+	uint64_t val;
+
+	val = mfspr(SPRN_BESCR);
+	if (!(val & BESCR_PMEO)) {
+		ebb_state.stats.spurious++;
+		goto out;
+	}
+
+	ebb_state.stats.ebb_count++;
+	count_pmc(1, sample_period);
+
+	before = mfspr(SPRN_MMCR0);
+
+	/* Try and get ourselves scheduled, to force a PMU context switch */
+	sched_yield();
+
+	after = mfspr(SPRN_MMCR0);
+	if (before != after)
+		mmcr0_mismatch = true;
+
+out:
+	reset_ebb();
+}
+
+static int test_body(void)
+{
+	struct event event;
+
+	SKIP_IF(!ebb_is_supported());
+
+	event_init_named(&event, 0x1001e, "cycles");
+	event_leader_ebb_init(&event);
+
+	event.attr.exclude_kernel = 1;
+	event.attr.exclude_hv = 1;
+	event.attr.exclude_idle = 1;
+
+	FAIL_IF(event_open(&event));
+
+	setup_ebb_handler(syscall_ebb_callee);
+	ebb_global_enable();
+
+	FAIL_IF(ebb_event_enable(&event));
+
+	mtspr(SPRN_PMC1, pmc_sample_period(sample_period));
+
+	while (ebb_state.stats.ebb_count < 20 && !mmcr0_mismatch)
+		FAIL_IF(core_busy_loop());
+
+	ebb_global_disable();
+	ebb_freeze_pmcs();
+
+	dump_ebb_state();
+
+	if (mmcr0_mismatch)
+		printf("Saw MMCR0 before 0x%lx after 0x%lx\n", before, after);
+
+	event_close(&event);
+
+	FAIL_IF(ebb_state.stats.ebb_count == 0);
+	FAIL_IF(mmcr0_mismatch);
+
+	return 0;
+}
+
+int pmae_handling(void)
+{
+	return eat_cpu(test_body);
+}
+
+int main(void)
+{
+	return test_harness(pmae_handling, "pmae_handling");
+}
diff --git a/tools/testing/selftests/powerpc/pmu/ebb/pmc56_overflow_test.c b/tools/testing/selftests/powerpc/pmu/ebb/pmc56_overflow_test.c
new file mode 100644
index 000000000..b29f8ba22
--- /dev/null
+++ b/tools/testing/selftests/powerpc/pmu/ebb/pmc56_overflow_test.c
@@ -0,0 +1,93 @@
+// SPDX-License-Identifier: GPL-2.0-only
+/*
+ * Copyright 2014, Michael Ellerman, IBM Corp.
+ */
+
+#include <stdio.h>
+#include <stdlib.h>
+
+#include "ebb.h"
+
+
+/*
+ * Test that PMC5 & 6 are frozen (ie. don't overflow) when they are not being
+ * used. Tests the MMCR0_FC56 logic in the kernel.
+ */
+
+static int pmc56_overflowed;
+
+static void ebb_callee(void)
+{
+	uint64_t val;
+
+	val = mfspr(SPRN_BESCR);
+	if (!(val & BESCR_PMEO)) {
+		ebb_state.stats.spurious++;
+		goto out;
+	}
+
+	ebb_state.stats.ebb_count++;
+	count_pmc(2, sample_period);
+
+	val = mfspr(SPRN_PMC5);
+	if (val >= COUNTER_OVERFLOW)
+		pmc56_overflowed++;
+
+	count_pmc(5, COUNTER_OVERFLOW);
+
+	val = mfspr(SPRN_PMC6);
+	if (val >= COUNTER_OVERFLOW)
+		pmc56_overflowed++;
+
+	count_pmc(6, COUNTER_OVERFLOW);
+
+out:
+	reset_ebb();
+}
+
+int pmc56_overflow(void)
+{
+	struct event event;
+
+	SKIP_IF(!ebb_is_supported());
+
+	/* Use PMC2 so we set PMCjCE, which enables PMC5/6 */
+	event_init(&event, 0x2001e);
+	event_leader_ebb_init(&event);
+
+	event.attr.exclude_kernel = 1;
+	event.attr.exclude_hv = 1;
+	event.attr.exclude_idle = 1;
+
+	FAIL_IF(event_open(&event));
+
+	setup_ebb_handler(ebb_callee);
+	ebb_global_enable();
+
+	FAIL_IF(ebb_event_enable(&event));
+
+	mtspr(SPRN_PMC2, pmc_sample_period(sample_period));
+	mtspr(SPRN_PMC5, 0);
+	mtspr(SPRN_PMC6, 0);
+
+	while (ebb_state.stats.ebb_count < 10)
+		FAIL_IF(core_busy_loop());
+
+	ebb_global_disable();
+	ebb_freeze_pmcs();
+
+	dump_ebb_state();
+
+	printf("PMC5/6 overflow %d\n", pmc56_overflowed);
+
+	event_close(&event);
+
+	FAIL_IF(ebb_state.stats.ebb_count == 0 || pmc56_overflowed != 0);
+
+	return 0;
+}
+
+int main(void)
+{
+	return test_harness(pmc56_overflow, "pmc56_overflow");
+}
diff --git a/tools/testing/selftests/powerpc/pmu/ebb/reg_access_test.c b/tools/testing/selftests/powerpc/pmu/ebb/reg_access_test.c
new file mode 100644
index 000000000..bd1ace9a0
--- /dev/null
+++ b/tools/testing/selftests/powerpc/pmu/ebb/reg_access_test.c
@@ -0,0 +1,40 @@
+// SPDX-License-Identifier: GPL-2.0-only
+/*
+ * Copyright 2014, Michael Ellerman, IBM Corp.
+ */
+
+#include <stdio.h>
+#include <stdlib.h>
+
+#include "ebb.h"
+
+
+/*
+ * Test basic access to the EBB regs, they should be user accessible with no
+ * kernel interaction required.
+ */
+int reg_access(void)
+{
+	uint64_t val, expected;
+
+	SKIP_IF(!ebb_is_supported());
+
+	expected = 0x8000000100000000ull;
+	mtspr(SPRN_BESCR, expected);
+	val = mfspr(SPRN_BESCR);
+
+	FAIL_IF(val != expected);
+
+	expected = 0x0000000001000000ull;
+	mtspr(SPRN_EBBHR, expected);
+	val = mfspr(SPRN_EBBHR);
+
+	FAIL_IF(val != expected);
+
+	return 0;
+}
+
+int main(void)
+{
+	return test_harness(reg_access, "reg_access");
+}
diff --git a/tools/testing/selftests/powerpc/pmu/ebb/task_event_pinned_vs_ebb_test.c b/tools/testing/selftests/powerpc/pmu/ebb/task_event_pinned_vs_ebb_test.c
new file mode 100644
index 000000000..0aa2aefd3
--- /dev/null
+++ b/tools/testing/selftests/powerpc/pmu/ebb/task_event_pinned_vs_ebb_test.c
@@ -0,0 +1,93 @@
+// SPDX-License-Identifier: GPL-2.0-only
+/*
+ * Copyright 2014, Michael Ellerman, IBM Corp.
+ */
+
+#include <signal.h>
+#include <stdio.h>
+#include <stdlib.h>
+#include <stdbool.h>
+#include <sys/types.h>
+#include <sys/wait.h>
+#include <unistd.h>
+
+#include "ebb.h"
+
+
+/*
+ * Tests a pinned per-task event vs an EBB - in that order. The pinned per-task
+ * event should prevent the EBB event from being enabled.
+ */
+
+static int setup_child_event(struct event *event, pid_t child_pid)
+{
+	event_init_named(event, 0x400FA, "PM_RUN_INST_CMPL");
+
+	event->attr.pinned = 1;
+
+	event->attr.exclude_kernel = 1;
+	event->attr.exclude_hv = 1;
+	event->attr.exclude_idle = 1;
+
+	FAIL_IF(event_open_with_pid(event, child_pid));
+	FAIL_IF(event_enable(event));
+
+	return 0;
+}
+
+int task_event_pinned_vs_ebb(void)
+{
+	union pipe read_pipe, write_pipe;
+	struct event event;
+	pid_t pid;
+	int rc;
+
+	SKIP_IF(!ebb_is_supported());
+
+	FAIL_IF(pipe(read_pipe.fds) == -1);
+	FAIL_IF(pipe(write_pipe.fds) == -1);
+
+	pid = fork();
+	if (pid == 0) {
+		/* NB order of pipes looks reversed */
+		exit(ebb_child(write_pipe, read_pipe));
+	}
+
+	/* We setup the task event first */
+	rc = setup_child_event(&event, pid);
+	if (rc) {
+		kill_child_and_wait(pid);
+		return rc;
+	}
+
+	/* Signal the child to install its EBB event and wait */
+	if (sync_with_child(read_pipe, write_pipe))
+		/* If it fails, wait for it to exit */
+		goto wait;
+
+	/* Signal the child to run */
+	FAIL_IF(sync_with_child(read_pipe, write_pipe));
+
+wait:
+	/* We expect it to fail to read the event */
+	FAIL_IF(wait_for_child(pid) != 2);
+	FAIL_IF(event_disable(&event));
+	FAIL_IF(event_read(&event));
+
+	event_report(&event);
+
+	FAIL_IF(event.result.value == 0);
+	/*
+	 * For reasons I don't understand enabled is usually just slightly
+	 * lower than running. Would be good to confirm why.
+	 */
+	FAIL_IF(event.result.enabled == 0);
+	FAIL_IF(event.result.running == 0);
+
+	return 0;
+}
+
+int main(void)
+{
+	return test_harness(task_event_pinned_vs_ebb, "task_event_pinned_vs_ebb");
+}
diff --git a/tools/testing/selftests/powerpc/pmu/ebb/task_event_vs_ebb_test.c b/tools/testing/selftests/powerpc/pmu/ebb/task_event_vs_ebb_test.c
new file mode 100644
index 000000000..3e9d95ad9
--- /dev/null
+++ b/tools/testing/selftests/powerpc/pmu/ebb/task_event_vs_ebb_test.c
@@ -0,0 +1,85 @@
+// SPDX-License-Identifier: GPL-2.0-only
+/*
+ * Copyright 2014, Michael Ellerman, IBM Corp.
+ */
+
+#include <signal.h>
+#include <stdio.h>
+#include <stdlib.h>
+#include <stdbool.h>
+#include <sys/types.h>
+#include <sys/wait.h>
+#include <unistd.h>
+
+#include "ebb.h"
+
+
+/*
+ * Tests a per-task event vs an EBB - in that order. The EBB should push the
+ * per-task event off the PMU.
+ */
+
+static int setup_child_event(struct event *event, pid_t child_pid)
+{
+	event_init_named(event, 0x400FA, "PM_RUN_INST_CMPL");
+
+	event->attr.exclude_kernel = 1;
+	event->attr.exclude_hv = 1;
+	event->attr.exclude_idle = 1;
+
+	FAIL_IF(event_open_with_pid(event, child_pid));
+	FAIL_IF(event_enable(event));
+
+	return 0;
+}
+
+int task_event_vs_ebb(void)
+{
+	union pipe read_pipe, write_pipe;
+	struct event event;
+	pid_t pid;
+	int rc;
+
+	SKIP_IF(!ebb_is_supported());
+
+	FAIL_IF(pipe(read_pipe.fds) == -1);
+	FAIL_IF(pipe(write_pipe.fds) == -1);
+
+	pid = fork();
+	if (pid == 0) {
+		/* NB order of pipes looks reversed */
+		exit(ebb_child(write_pipe, read_pipe));
+	}
+
+	/* We setup the task event first */
+	rc = setup_child_event(&event, pid);
+	if (rc) {
+		kill_child_and_wait(pid);
+		return rc;
+	}
+
+	/* Signal the child to install its EBB event and wait */
+	if (sync_with_child(read_pipe, write_pipe))
+		/* If it fails, wait for it to exit */
+		goto wait;
+
+	/* Signal the child to run */
+	FAIL_IF(sync_with_child(read_pipe, write_pipe));
+
+wait:
+	/* The EBB event should push the task event off so the child should succeed */
+	FAIL_IF(wait_for_child(pid));
+	FAIL_IF(event_disable(&event));
+	FAIL_IF(event_read(&event));
+
+	event_report(&event);
+
+	/* The task event may have run, or not so we can't assert anything about it */
+
+	return 0;
+}
+
+int main(void)
+{
+	return test_harness(task_event_vs_ebb, "task_event_vs_ebb");
+}
diff --git a/tools/testing/selftests/powerpc/pmu/ebb/trace.c b/tools/testing/selftests/powerpc/pmu/ebb/trace.c
new file mode 100644
index 000000000..0c59f66a6
--- /dev/null
+++ b/tools/testing/selftests/powerpc/pmu/ebb/trace.c
@@ -0,0 +1,300 @@
+// SPDX-License-Identifier: GPL-2.0-only
+/*
+ * Copyright 2014, Michael Ellerman, IBM Corp.
+ */
+
+#include <errno.h>
+#include <stdio.h>
+#include <stdlib.h>
+#include <string.h>
+#include <sys/mman.h>
+
+#include "trace.h"
+
+
+struct trace_buffer *trace_buffer_allocate(u64 size)
+{
+	struct trace_buffer *tb;
+
+	if (size < sizeof(*tb)) {
+		fprintf(stderr, "Error: trace buffer too small\n");
+		return NULL;
+	}
+
+	tb = mmap(NULL, size, PROT_READ | PROT_WRITE,
+		  MAP_ANONYMOUS | MAP_PRIVATE, -1, 0);
+	if (tb == MAP_FAILED) {
+		perror("mmap");
+		return NULL;
+	}
+
+	tb->size = size;
+	tb->tail = tb->data;
+	tb->overflow = false;
+
+	return tb;
+}
+
+static bool trace_check_bounds(struct trace_buffer *tb, void *p)
+{
+	return p < ((void *)tb + tb->size);
+}
+
+static bool trace_check_alloc(struct trace_buffer *tb, void *p)
+{
+	/*
+	 * If we ever overflowed don't allow any more input. This prevents us
+	 * from dropping a large item and then later logging a small one. The
+	 * buffer should just stop when overflow happened, not be patchy. If
+	 * you're overflowing, make your buffer bigger.
+	 */
+	if (tb->overflow)
+		return false;
+
+	if (!trace_check_bounds(tb, p)) {
+		tb->overflow = true;
+		return false;
+	}
+
+	return true;
+}
+
+static void *trace_alloc(struct trace_buffer *tb, int bytes)
+{
+	void *p, *newtail;
+
+	p = tb->tail;
+	newtail = tb->tail + bytes;
+	if (!trace_check_alloc(tb, newtail))
+		return NULL;
+
+	tb->tail = newtail;
+
+	return p;
+}
+
+static struct trace_entry *trace_alloc_entry(struct trace_buffer *tb, int payload_size)
+{
+	struct trace_entry *e;
+
+	e = trace_alloc(tb, sizeof(*e) + payload_size);
+	if (e)
+		e->length = payload_size;
+
+	return e;
+}
+
+int trace_log_reg(struct trace_buffer *tb, u64 reg, u64 value)
+{
+	struct trace_entry *e;
+	u64 *p;
+
+	e = trace_alloc_entry(tb, sizeof(reg) + sizeof(value));
+	if (!e)
+		return -ENOSPC;
+
+	e->type = TRACE_TYPE_REG;
+	p = (u64 *)e->data;
+	*p++ = reg;
+	*p++ = value;
+
+	return 0;
+}
+
+int trace_log_counter(struct trace_buffer *tb, u64 value)
+{
+	struct trace_entry *e;
+	u64 *p;
+
+	e = trace_alloc_entry(tb, sizeof(value));
+	if (!e)
+		return -ENOSPC;
+
+	e->type = TRACE_TYPE_COUNTER;
+	p = (u64 *)e->data;
+	*p++ = value;
+
+	return 0;
+}
+
+int trace_log_string(struct trace_buffer *tb, char *str)
+{
+	struct trace_entry *e;
+	char *p;
+	int len;
+
+	len = strlen(str);
+
+	/* We NULL terminate to make printing easier */
+	e = trace_alloc_entry(tb, len + 1);
+	if (!e)
+		return -ENOSPC;
+
+	e->type = TRACE_TYPE_STRING;
+	p = (char *)e->data;
+	memcpy(p, str, len);
+	p += len;
+	*p = '\0';
+
+	return 0;
+}
+
+int trace_log_indent(struct trace_buffer *tb)
+{
+	struct trace_entry *e;
+
+	e = trace_alloc_entry(tb, 0);
+	if (!e)
+		return -ENOSPC;
+
+	e->type = TRACE_TYPE_INDENT;
+
+	return 0;
+}
+
+int trace_log_outdent(struct trace_buffer *tb)
+{
+	struct trace_entry *e;
+
+	e = trace_alloc_entry(tb, 0);
+	if (!e)
+		return -ENOSPC;
+
+	e->type = TRACE_TYPE_OUTDENT;
+
+	return 0;
+}
+
+static void trace_print_header(int seq, int prefix)
+{
+	printf("%*s[%d]: ", prefix, "", seq);
+}
+
+static char *trace_decode_reg(int reg)
+{
+	switch (reg) {
+		case 769: return "SPRN_MMCR2"; break;
+		case 770: return "SPRN_MMCRA"; break;
+		case 779: return "SPRN_MMCR0"; break;
+		case 804: return "SPRN_EBBHR"; break;
+		case 805: return "SPRN_EBBRR"; break;
+		case 806: return "SPRN_BESCR"; break;
+		case 800: return "SPRN_BESCRS"; break;
+		case 801: return "SPRN_BESCRSU"; break;
+		case 802: return "SPRN_BESCRR"; break;
+		case 803: return "SPRN_BESCRRU"; break;
+		case 771: return "SPRN_PMC1"; break;
+		case 772: return "SPRN_PMC2"; break;
+		case 773: return "SPRN_PMC3"; break;
+		case 774: return "SPRN_PMC4"; break;
+		case 775: return "SPRN_PMC5"; break;
+		case 776: return "SPRN_PMC6"; break;
+		case 780: return "SPRN_SIAR"; break;
+		case 781: return "SPRN_SDAR"; break;
+		case 768: return "SPRN_SIER"; break;
+	}
+
+	return NULL;
+}
+
+static void trace_print_reg(struct trace_entry *e)
+{
+	u64 *p, *reg, *value;
+	char *name;
+
+	p = (u64 *)e->data;
+	reg = p++;
+	value = p;
+
+	name = trace_decode_reg(*reg);
+	if (name)
+		printf("register %-10s = 0x%016llx\n", name, *value);
+	else
+		printf("register %lld = 0x%016llx\n", *reg, *value);
+}
+
+static void trace_print_counter(struct trace_entry *e)
+{
+	u64 *value;
+
+	value = (u64 *)e->data;
+	printf("counter = %lld\n", *value);
+}
+
+static void trace_print_string(struct trace_entry *e)
+{
+	char *str;
+
+	str = (char *)e->data;
+	puts(str);
+}
+
+#define BASE_PREFIX	2
+#define PREFIX_DELTA	8
+
+static void trace_print_entry(struct trace_entry *e, int seq, int *prefix)
+{
+	switch (e->type) {
+	case TRACE_TYPE_REG:
+		trace_print_header(seq, *prefix);
+		trace_print_reg(e);
+		break;
+	case TRACE_TYPE_COUNTER:
+		trace_print_header(seq, *prefix);
+		trace_print_counter(e);
+		break;
+	case TRACE_TYPE_STRING:
+		trace_print_header(seq, *prefix);
+		trace_print_string(e);
+		break;
+	case TRACE_TYPE_INDENT:
+		trace_print_header(seq, *prefix);
+		puts("{");
+		*prefix += PREFIX_DELTA;
+		break;
+	case TRACE_TYPE_OUTDENT:
+		*prefix -= PREFIX_DELTA;
+		if (*prefix < BASE_PREFIX)
+			*prefix = BASE_PREFIX;
+		trace_print_header(seq, *prefix);
+		puts("}");
+		break;
+	default:
+		trace_print_header(seq, *prefix);
+		printf("entry @ %p type %d\n", e, e->type);
+		break;
+	}
+}
+
+void trace_buffer_print(struct trace_buffer *tb)
+{
+	struct trace_entry *e;
+	int i, prefix;
+	void *p;
+
+	printf("Trace buffer dump:\n");
+	printf("  address  %p \n", tb);
+	printf("  tail     %p\n", tb->tail);
+	printf("  size     %llu\n", tb->size);
+	printf("  overflow %s\n", tb->overflow ? "TRUE" : "false");
+	printf("  Content:\n");
+
+	p = tb->data;
+
+	i = 0;
+	prefix = BASE_PREFIX;
+
+	while (trace_check_bounds(tb, p) && p < tb->tail) {
+		e = p;
+
+		trace_print_entry(e, i, &prefix);
+
+		i++;
+		p = (void *)e + sizeof(*e) + e->length;
+	}
+}
+
+void trace_print_location(struct trace_buffer *tb)
+{
+	printf("Trace buffer 0x%llx bytes @ %p\n", tb->size, tb);
+}
diff --git a/tools/testing/selftests/powerpc/pmu/ebb/trace.h b/tools/testing/selftests/powerpc/pmu/ebb/trace.h
new file mode 100644
index 000000000..da2a3be54
--- /dev/null
+++ b/tools/testing/selftests/powerpc/pmu/ebb/trace.h
@@ -0,0 +1,41 @@
+/* SPDX-License-Identifier: GPL-2.0-only */
+/*
+ * Copyright 2014, Michael Ellerman, IBM Corp.
+ */
+
+#ifndef _SELFTESTS_POWERPC_PMU_EBB_TRACE_H
+#define _SELFTESTS_POWERPC_PMU_EBB_TRACE_H
+
+#include "utils.h"
+
+#define TRACE_TYPE_REG		1
+#define TRACE_TYPE_COUNTER	2
+#define TRACE_TYPE_STRING	3
+#define TRACE_TYPE_INDENT	4
+#define TRACE_TYPE_OUTDENT	5
+
+struct trace_entry
+{
+	u8 type;
+	u8 length;
+	u8 data[];
+};
+
+struct trace_buffer
+{
+	u64  size;
+	bool overflow;
+	void *tail;
+	u8   data[];
+};
+
+struct trace_buffer *trace_buffer_allocate(u64 size);
+int trace_log_reg(struct trace_buffer *tb, u64 reg, u64 value);
+int trace_log_counter(struct trace_buffer *tb, u64 value);
+int trace_log_string(struct trace_buffer *tb, char *str);
+int trace_log_indent(struct trace_buffer *tb);
+int trace_log_outdent(struct trace_buffer *tb);
+void trace_buffer_print(struct trace_buffer *tb);
+void trace_print_location(struct trace_buffer *tb);
+
+#endif /* _SELFTESTS_POWERPC_PMU_EBB_TRACE_H */
diff --git a/tools/testing/selftests/powerpc/pmu/event.c b/tools/testing/selftests/powerpc/pmu/event.c
new file mode 100644
index 000000000..48e3a413b
--- /dev/null
+++ b/tools/testing/selftests/powerpc/pmu/event.c
@@ -0,0 +1,131 @@
+// SPDX-License-Identifier: GPL-2.0-only
+/*
+ * Copyright 2013, Michael Ellerman, IBM Corp.
+ */
+
+#define _GNU_SOURCE
+#include <unistd.h>
+#include <sys/syscall.h>
+#include <string.h>
+#include <stdio.h>
+#include <sys/ioctl.h>
+
+#include "event.h"
+
+
+int perf_event_open(struct perf_event_attr *attr, pid_t pid, int cpu,
+		int group_fd, unsigned long flags)
+{
+	return syscall(__NR_perf_event_open, attr, pid, cpu,
+			   group_fd, flags);
+}
+
+void event_init_opts(struct event *e, u64 config, int type, char *name)
+{
+	memset(e, 0, sizeof(*e));
+
+	e->name = name;
+
+	e->attr.type = type;
+	e->attr.config = config;
+	e->attr.size = sizeof(e->attr);
+	/* This has to match the structure layout in the header */
+	e->attr.read_format = PERF_FORMAT_TOTAL_TIME_ENABLED | \
+				  PERF_FORMAT_TOTAL_TIME_RUNNING;
+}
+
+void event_init_named(struct event *e, u64 config, char *name)
+{
+	event_init_opts(e, config, PERF_TYPE_RAW, name);
+}
+
+void event_init(struct event *e, u64 config)
+{
+	event_init_opts(e, config, PERF_TYPE_RAW, "event");
+}
+
+#define PERF_CURRENT_PID	0
+#define PERF_NO_PID		-1
+#define PERF_NO_CPU		-1
+#define PERF_NO_GROUP		-1
+
+int event_open_with_options(struct event *e, pid_t pid, int cpu, int group_fd)
+{
+	e->fd = perf_event_open(&e->attr, pid, cpu, group_fd, 0);
+	if (e->fd == -1) {
+		perror("perf_event_open");
+		return -1;
+	}
+
+	return 0;
+}
+
+int event_open_with_group(struct event *e, int group_fd)
+{
+	return event_open_with_options(e, PERF_CURRENT_PID, PERF_NO_CPU, group_fd);
+}
+
+int event_open_with_pid(struct event *e, pid_t pid)
+{
+	return event_open_with_options(e, pid, PERF_NO_CPU, PERF_NO_GROUP);
+}
+
+int event_open_with_cpu(struct event *e, int cpu)
+{
+	return event_open_with_options(e, PERF_NO_PID, cpu, PERF_NO_GROUP);
+}
+
+int event_open(struct event *e)
+{
+	return event_open_with_options(e, PERF_CURRENT_PID, PERF_NO_CPU, PERF_NO_GROUP);
+}
+
+void event_close(struct event *e)
+{
+	close(e->fd);
+}
+
+int event_enable(struct event *e)
+{
+	return ioctl(e->fd, PERF_EVENT_IOC_ENABLE);
+}
+
+int event_disable(struct event *e)
+{
+	return ioctl(e->fd, PERF_EVENT_IOC_DISABLE);
+}
+
+int event_reset(struct event *e)
+{
+	return ioctl(e->fd, PERF_EVENT_IOC_RESET);
+}
+
+int event_read(struct event *e)
+{
+	int rc;
+
+	rc = read(e->fd, &e->result, sizeof(e->result));
+	if (rc != sizeof(e->result)) {
+		fprintf(stderr, "read error on event %p!\n", e);
+		return -1;
+	}
+
+	return 0;
+}
+
+void event_report_justified(struct event *e, int name_width, int result_width)
+{
+	printf("%*s: result %*llu ", name_width, e->name, result_width,
+	       e->result.value);
+
+	if (e->result.running == e->result.enabled)
+		printf("running/enabled %llu\n", e->result.running);
+	else
+		printf("running %llu enabled %llu\n", e->result.running,
+			e->result.enabled);
+}
+
+void event_report(struct event *e)
+{
+	event_report_justified(e, 0, 0);
+}
diff --git a/tools/testing/selftests/powerpc/pmu/event.h b/tools/testing/selftests/powerpc/pmu/event.h
new file mode 100644
index 000000000..302eaab51
--- /dev/null
+++ b/tools/testing/selftests/powerpc/pmu/event.h
@@ -0,0 +1,43 @@
+/* SPDX-License-Identifier: GPL-2.0-only */
+/*
+ * Copyright 2013, Michael Ellerman, IBM Corp.
+ */
+
+#ifndef _SELFTESTS_POWERPC_PMU_EVENT_H
+#define _SELFTESTS_POWERPC_PMU_EVENT_H
+
+#include <unistd.h>
+#include <linux/perf_event.h>
+
+#include "utils.h"
+
+
+struct event {
+	struct perf_event_attr attr;
+	char *name;
+	int fd;
+	/* This must match the read_format we use */
+	struct {
+		u64 value;
+		u64 running;
+		u64 enabled;
+	} result;
+};
+
+void event_init(struct event *e, u64 config);
+void event_init_named(struct event *e, u64 config, char *name);
+void event_init_opts(struct event *e, u64 config, int type, char *name);
+int event_open_with_options(struct event *e, pid_t pid, int cpu, int group_fd);
+int event_open_with_group(struct event *e, int group_fd);
+int event_open_with_pid(struct event *e, pid_t pid);
+int event_open_with_cpu(struct event *e, int cpu);
+int event_open(struct event *e);
+void event_close(struct event *e);
+int event_enable(struct event *e);
+int event_disable(struct event *e);
+int event_reset(struct event *e);
+int event_read(struct event *e);
+void event_report_justified(struct event *e, int name_width, int result_width);
+void event_report(struct event *e);
+
+#endif /* _SELFTESTS_POWERPC_PMU_EVENT_H */
diff --git a/tools/testing/selftests/powerpc/pmu/l3_bank_test.c b/tools/testing/selftests/powerpc/pmu/l3_bank_test.c
new file mode 100644
index 000000000..a5dfa9bf3
--- /dev/null
+++ b/tools/testing/selftests/powerpc/pmu/l3_bank_test.c
@@ -0,0 +1,51 @@
+// SPDX-License-Identifier: GPL-2.0-only
+/*
+ * Copyright 2014, Michael Ellerman, IBM Corp.
+ */
+
+#include <stdio.h>
+#include <stdlib.h>
+
+#include "event.h"
+#include "utils.h"
+
+#define MALLOC_SIZE     (0x10000 * 10)  /* Ought to be enough .. */
+
+/*
+ * Tests that the L3 bank handling is correct. We fixed it in commit e9aaac1.
+ */
+static int l3_bank_test(void)
+{
+	struct event event;
+	char *p;
+	int i;
+
+	// The L3 bank logic is only used on Power8 or later
+	SKIP_IF(!have_hwcap2(PPC_FEATURE2_ARCH_2_07));
+
+	p = malloc(MALLOC_SIZE);
+	FAIL_IF(!p);
+
+	event_init(&event, 0x84918F);
+
+	FAIL_IF(event_open(&event));
+
+	for (i = 0; i < MALLOC_SIZE; i += 0x10000)
+		p[i] = i;
+
+	event_read(&event);
+	event_report(&event);
+
+	FAIL_IF(event.result.running == 0);
+	FAIL_IF(event.result.enabled == 0);
+
+	event_close(&event);
+	free(p);
+
+	return 0;
+}
+
+int main(void)
+{
+	return test_harness(l3_bank_test, "l3_bank_test");
+}
diff --git a/tools/testing/selftests/powerpc/pmu/lib.c b/tools/testing/selftests/powerpc/pmu/lib.c
new file mode 100644
index 000000000..88690b97b
--- /dev/null
+++ b/tools/testing/selftests/powerpc/pmu/lib.c
@@ -0,0 +1,227 @@
+// SPDX-License-Identifier: GPL-2.0-only
+/*
+ * Copyright 2014, Michael Ellerman, IBM Corp.
+ */
+
+#define _GNU_SOURCE	/* For CPU_ZERO etc. */
+
+#include <errno.h>
+#include <sched.h>
+#include <setjmp.h>
+#include <stdlib.h>
+#include <sys/wait.h>
+
+#include "utils.h"
+#include "lib.h"
+
+
+int bind_to_cpu(int cpu)
+{
+	cpu_set_t mask;
+
+	printf("Binding to cpu %d\n", cpu);
+
+	CPU_ZERO(&mask);
+	CPU_SET(cpu, &mask);
+
+	return sched_setaffinity(0, sizeof(mask), &mask);
+}
+
+#define PARENT_TOKEN	0xAA
+#define CHILD_TOKEN	0x55
+
+int sync_with_child(union pipe read_pipe, union pipe write_pipe)
+{
+	char c = PARENT_TOKEN;
+
+	FAIL_IF(write(write_pipe.write_fd, &c, 1) != 1);
+	FAIL_IF(read(read_pipe.read_fd, &c, 1) != 1);
+	if (c != CHILD_TOKEN) /* sometimes expected */
+		return 1;
+
+	return 0;
+}
+
+int wait_for_parent(union pipe read_pipe)
+{
+	char c;
+
+	FAIL_IF(read(read_pipe.read_fd, &c, 1) != 1);
+	FAIL_IF(c != PARENT_TOKEN);
+
+	return 0;
+}
+
+int notify_parent(union pipe write_pipe)
+{
+	char c = CHILD_TOKEN;
+
+	FAIL_IF(write(write_pipe.write_fd, &c, 1) != 1);
+
+	return 0;
+}
+
+int notify_parent_of_error(union pipe write_pipe)
+{
+	char c = ~CHILD_TOKEN;
+
+	FAIL_IF(write(write_pipe.write_fd, &c, 1) != 1);
+
+	return 0;
+}
+
+int wait_for_child(pid_t child_pid)
+{
+	int rc;
+
+	if (waitpid(child_pid, &rc, 0) == -1) {
+		perror("waitpid");
+		return 1;
+	}
+
+	if (WIFEXITED(rc))
+		rc = WEXITSTATUS(rc);
+	else
+		rc = 1; /* Signal or other */
+
+	return rc;
+}
+
+int kill_child_and_wait(pid_t child_pid)
+{
+	kill(child_pid, SIGTERM);
+
+	return wait_for_child(child_pid);
+}
+
+static int eat_cpu_child(union pipe read_pipe, union pipe write_pipe)
+{
+	volatile int i = 0;
+
+	/*
+	 * We are just here to eat cpu and die. So make sure we can be killed,
+	 * and also don't do any custom SIGTERM handling.
+	 */
+	signal(SIGTERM, SIG_DFL);
+
+	notify_parent(write_pipe);
+	wait_for_parent(read_pipe);
+
+	/* Soak up cpu forever */
+	while (1) i++;
+
+	return 0;
+}
+
+pid_t eat_cpu(int (test_function)(void))
+{
+	union pipe read_pipe, write_pipe;
+	int cpu, rc;
+	pid_t pid;
+
+	cpu = pick_online_cpu();
+	FAIL_IF(cpu < 0);
+	FAIL_IF(bind_to_cpu(cpu));
+
+	if (pipe(read_pipe.fds) == -1)
+		return -1;
+
+	if (pipe(write_pipe.fds) == -1)
+		return -1;
+
+	pid = fork();
+	if (pid == 0)
+		exit(eat_cpu_child(write_pipe, read_pipe));
+
+	if (sync_with_child(read_pipe, write_pipe)) {
+		rc = -1;
+		goto out;
+	}
+
+	printf("main test running as pid %d\n", getpid());
+
+	rc = test_function();
+out:
+	kill(pid, SIGKILL);
+
+	return rc;
+}
+
+struct addr_range libc, vdso;
+
+int parse_proc_maps(void)
+{
+	unsigned long start, end;
+	char execute, name[128];
+	FILE *f;
+	int rc;
+
+	f = fopen("/proc/self/maps", "r");
+	if (!f) {
+		perror("fopen");
+		return -1;
+	}
+
+	do {
+		/* This skips line with no executable which is what we want */
+		rc = fscanf(f, "%lx-%lx %*c%*c%c%*c %*x %*d:%*d %*d %127s\n",
+			    &start, &end, &execute, name);
+		if (rc <= 0)
+			break;
+
+		if (execute != 'x')
+			continue;
+
+		if (strstr(name, "libc")) {
+			libc.first = start;
+			libc.last = end - 1;
+		} else if (strstr(name, "[vdso]")) {
+			vdso.first = start;
+			vdso.last = end - 1;
+		}
+	} while(1);
+
+	fclose(f);
+
+	return 0;
+}
+
+#define PARANOID_PATH	"/proc/sys/kernel/perf_event_paranoid"
+
+bool require_paranoia_below(int level)
+{
+	long current;
+	char *end, buf[16];
+	FILE *f;
+	bool rc;
+
+	rc = false;
+
+	f = fopen(PARANOID_PATH, "r");
+	if (!f) {
+		perror("fopen");
+		goto out;
+	}
+
+	if (!fgets(buf, sizeof(buf), f)) {
+		printf("Couldn't read " PARANOID_PATH "?\n");
+		goto out_close;
+	}
+
+	current = strtol(buf, &end, 10);
+
+	if (end == buf) {
+		printf("Couldn't parse " PARANOID_PATH "?\n");
+		goto out_close;
+	}
+
+	if (current >= level)
+		goto out_close;
+
+	rc = true;
+out_close:
+	fclose(f);
+out:
+	return rc;
+}
+
diff --git a/tools/testing/selftests/powerpc/pmu/lib.h b/tools/testing/selftests/powerpc/pmu/lib.h
new file mode 100644
index 000000000..bf1bec013
--- /dev/null
+++ b/tools/testing/selftests/powerpc/pmu/lib.h
@@ -0,0 +1,41 @@
+/* SPDX-License-Identifier: GPL-2.0-only */
+/*
+ * Copyright 2014, Michael Ellerman, IBM Corp.
+ */
+
+#ifndef __SELFTESTS_POWERPC_PMU_LIB_H
+#define __SELFTESTS_POWERPC_PMU_LIB_H
+
+#include <stdbool.h>
+#include <stdio.h>
+#include <stdint.h>
+#include <string.h>
+#include <unistd.h>
+
+union pipe {
+	struct {
+		int read_fd;
+		int write_fd;
+	};
+	int fds[2];
+};
+
+extern int bind_to_cpu(int cpu);
+extern int kill_child_and_wait(pid_t child_pid);
+extern int wait_for_child(pid_t child_pid);
+extern int sync_with_child(union pipe read_pipe, union pipe write_pipe);
+extern int wait_for_parent(union pipe read_pipe);
+extern int notify_parent(union pipe write_pipe);
+extern int notify_parent_of_error(union pipe write_pipe);
+extern pid_t eat_cpu(int (test_function)(void));
+extern bool require_paranoia_below(int level);
+
+struct addr_range {
+	uint64_t first, last;
+};
+
+extern struct addr_range libc, vdso;
+
+int parse_proc_maps(void);
+
+#endif /* __SELFTESTS_POWERPC_PMU_LIB_H */
diff --git a/tools/testing/selftests/powerpc/pmu/loop.S b/tools/testing/selftests/powerpc/pmu/loop.S
new file mode 100644
index 000000000..c52ba09b6
--- /dev/null
+++ b/tools/testing/selftests/powerpc/pmu/loop.S
@@ -0,0 +1,78 @@
+/* SPDX-License-Identifier: GPL-2.0-only */
+/*
+ * Copyright 2013, Michael Ellerman, IBM Corp.
+ */
+
+#include <ppc-asm.h>
+
+	.text
+
+FUNC_START(thirty_two_instruction_loop)
+	cmpdi	r3,0
+	beqlr
+	addi	r4,r3,1
+	addi	r4,r4,1
+	addi	r4,r4,1
+	addi	r4,r4,1
+	addi	r4,r4,1
+	addi	r4,r4,1
+	addi	r4,r4,1
+	addi	r4,r4,1
+	addi	r4,r4,1
+	addi	r4,r4,1
+	addi	r4,r4,1
+	addi	r4,r4,1
+	addi	r4,r4,1
+	addi	r4,r4,1
+	addi	r4,r4,1
+	addi	r4,r4,1
+	addi	r4,r4,1
+	addi	r4,r4,1
+	addi	r4,r4,1
+	addi	r4,r4,1
+	addi	r4,r4,1
+	addi	r4,r4,1
+	addi	r4,r4,1
+	addi	r4,r4,1
+	addi	r4,r4,1
+	addi	r4,r4,1
+	addi	r4,r4,1
+	addi	r4,r4,1	# 28 addi's
+	subi	r3,r3,1
+	b	FUNC_NAME(thirty_two_instruction_loop)
+FUNC_END(thirty_two_instruction_loop)
+
+FUNC_START(thirty_two_instruction_loop_with_ll_sc)
+	cmpdi	r3,0
+	beqlr
+	addi	r5,r5,1
+	addi	r5,r5,1
+	addi	r5,r5,1		# 5
+	addi	r5,r5,1
+	addi	r5,r5,1
+	addi	r5,r5,1
+	addi	r5,r5,1
+1:	ldarx	r6,0,r4		# 10
+	addi	r5,r5,1
+	addi	r5,r5,1
+	addi	r5,r5,1
+	addi	r5,r5,1
+	addi	r5,r5,1		# 15
+	addi	r5,r5,1
+	addi	r5,r5,1
+	stdcx.	r6,0,r4
+	bne-	1b
+	addi	r5,r5,1		# 20
+	addi	r5,r5,1
+	addi	r5,r5,1
+	addi	r5,r5,1
+	addi	r5,r5,1
+	addi	r5,r5,1		# 25
+	addi	r5,r5,1
+	addi	r5,r5,1
+	addi	r5,r5,1
+	addi	r5,r5,1
+	addi	r5,r5,1		# 30
+	subi	r3,r3,1
+	b	FUNC_NAME(thirty_two_instruction_loop_with_ll_sc)
+FUNC_END(thirty_two_instruction_loop_with_ll_sc)
diff --git a/tools/testing/selftests/powerpc/pmu/per_event_excludes.c b/tools/testing/selftests/powerpc/pmu/per_event_excludes.c
new file mode 100644
index 000000000..ad32a09a6
--- /dev/null
+++ b/tools/testing/selftests/powerpc/pmu/per_event_excludes.c
@@ -0,0 +1,111 @@
+// SPDX-License-Identifier: GPL-2.0-only
+/*
+ * Copyright 2014, Michael Ellerman, IBM Corp.
+ */
+
+#define _GNU_SOURCE
+
+#include <elf.h>
+#include <limits.h>
+#include <stdio.h>
+#include <stdbool.h>
+#include <string.h>
+#include <sys/prctl.h>
+
+#include "event.h"
+#include "lib.h"
+#include "utils.h"
+
+/*
+ * Test that per-event excludes work.
+ */
+
+static int per_event_excludes(void)
+{
+	struct event *e, events[4];
+	int i;
+
+	SKIP_IF(!have_hwcap2(PPC_FEATURE2_ARCH_2_07));
+
+	/*
+	 * We need to create the events disabled, otherwise the running/enabled
+	 * counts don't match up.
+	 */
+	e = &events[0];
+	event_init_opts(e, PERF_COUNT_HW_INSTRUCTIONS,
+			PERF_TYPE_HARDWARE, "instructions");
+	e->attr.disabled = 1;
+
+	e = &events[1];
+	event_init_opts(e, PERF_COUNT_HW_INSTRUCTIONS,
+			PERF_TYPE_HARDWARE, "instructions(k)");
+	e->attr.disabled = 1;
+	e->attr.exclude_user = 1;
+	e->attr.exclude_hv = 1;
+
+	e = &events[2];
+	event_init_opts(e, PERF_COUNT_HW_INSTRUCTIONS,
+			PERF_TYPE_HARDWARE, "instructions(h)");
+	e->attr.disabled = 1;
+	e->attr.exclude_user = 1;
+	e->attr.exclude_kernel = 1;
+
+	e = &events[3];
+	event_init_opts(e, PERF_COUNT_HW_INSTRUCTIONS,
+			PERF_TYPE_HARDWARE, "instructions(u)");
+	e->attr.disabled = 1;
+	e->attr.exclude_hv = 1;
+	e->attr.exclude_kernel = 1;
+
+	FAIL_IF(event_open(&events[0]));
+
+	/*
+	 * The open here will fail if we don't have per event exclude support,
+	 * because the second event has an incompatible set of exclude settings
+	 * and we're asking for the events to be in a group.
+	 */
+	for (i = 1; i < 4; i++)
+		FAIL_IF(event_open_with_group(&events[i], events[0].fd));
+
+	/*
+	 * Even though the above will fail without per-event excludes we keep
+	 * testing in order to be thorough.
+	 */
+	prctl(PR_TASK_PERF_EVENTS_ENABLE);
+
+	/* Spin for a while */
+	for (i = 0; i < INT_MAX; i++)
+		asm volatile("" : : : "memory");
+
+	prctl(PR_TASK_PERF_EVENTS_DISABLE);
+
+	for (i = 0; i < 4; i++) {
+		FAIL_IF(event_read(&events[i]));
+		event_report(&events[i]);
+	}
+
+	/*
+	 * We should see that all events have enabled == running. That
+	 * shows that they were all on the PMU at once.
+	 */
+	for (i = 0; i < 4; i++)
+		FAIL_IF(events[i].result.running != events[i].result.enabled);
+
+	/*
+	 * We can also check that the result for instructions is >= all the
+	 * other counts. That's because it is counting all instructions while
+	 * the others are counting a subset.
+	 */
+	for (i = 1; i < 4; i++)
+		FAIL_IF(events[0].result.value < events[i].result.value);
+
+	for (i = 0; i < 4; i++)
+		event_close(&events[i]);
+
+	return 0;
+}
+
+int main(void)
+{
+	return test_harness(per_event_excludes, "per_event_excludes");
+}
diff --git a/tools/testing/selftests/powerpc/primitives/.gitignore b/tools/testing/selftests/powerpc/primitives/.gitignore
new file mode 100644
index 000000000..1e5c04e24
--- /dev/null
+++ b/tools/testing/selftests/powerpc/primitives/.gitignore
@@ -0,0 +1,2 @@
+# SPDX-License-Identifier: GPL-2.0-only
+load_unaligned_zeropad
diff --git a/tools/testing/selftests/powerpc/primitives/Makefile b/tools/testing/selftests/powerpc/primitives/Makefile
new file mode 100644
index 000000000..9b9491a63
--- /dev/null
+++ b/tools/testing/selftests/powerpc/primitives/Makefile
@@ -0,0 +1,9 @@
+# SPDX-License-Identifier: GPL-2.0-only
+CFLAGS += -I$(CURDIR)
+
+TEST_GEN_PROGS := load_unaligned_zeropad
+
+top_srcdir = ../../../../..
+include ../../lib.mk
+
+$(TEST_GEN_PROGS): ../harness.c
diff --git a/tools/testing/selftests/powerpc/primitives/asm/asm-compat.h b/tools/testing/selftests/powerpc/primitives/asm/asm-compat.h
new file mode 120000
index 000000000..b14255e15
--- /dev/null
+++ b/tools/testing/selftests/powerpc/primitives/asm/asm-compat.h
@@ -0,0 +1 @@
+../.././../../../../arch/powerpc/include/asm/asm-compat.h
\ No newline at end of file
diff --git a/tools/testing/selftests/powerpc/primitives/asm/asm-const.h b/tools/testing/selftests/powerpc/primitives/asm/asm-const.h
new file mode 120000
index 000000000..18d8be13e
--- /dev/null
+++ b/tools/testing/selftests/powerpc/primitives/asm/asm-const.h
@@ -0,0 +1 @@
+../../../../../../arch/powerpc/include/asm/asm-const.h
\ No newline at end of file
diff --git a/tools/testing/selftests/powerpc/primitives/asm/feature-fixups.h b/tools/testing/selftests/powerpc/primitives/asm/feature-fixups.h
new file mode 120000
index 000000000..8dc6d4d46
--- /dev/null
+++ b/tools/testing/selftests/powerpc/primitives/asm/feature-fixups.h
@@ -0,0 +1 @@
+../../../../../../arch/powerpc/include/asm/feature-fixups.h
\ No newline at end of file
diff --git a/tools/testing/selftests/powerpc/primitives/asm/firmware.h b/tools/testing/selftests/powerpc/primitives/asm/firmware.h
new file mode 100644
index 000000000..e69de29bb
--- /dev/null
+++ b/tools/testing/selftests/powerpc/primitives/asm/firmware.h
diff --git a/tools/testing/selftests/powerpc/primitives/asm/ppc-opcode.h b/tools/testing/selftests/powerpc/primitives/asm/ppc-opcode.h
new file mode 100644
index 000000000..e69de29bb
--- /dev/null
+++ b/tools/testing/selftests/powerpc/primitives/asm/ppc-opcode.h
diff --git a/tools/testing/selftests/powerpc/primitives/asm/ppc_asm.h b/tools/testing/selftests/powerpc/primitives/asm/ppc_asm.h
new file mode 120000
index 000000000..66c819322
--- /dev/null
+++ b/tools/testing/selftests/powerpc/primitives/asm/ppc_asm.h
@@ -0,0 +1 @@
+../../../../../../arch/powerpc/include/asm/ppc_asm.h
\ No newline at end of file
diff --git a/tools/testing/selftests/powerpc/primitives/asm/processor.h b/tools/testing/selftests/powerpc/primitives/asm/processor.h
new file mode 100644
index 000000000..e69de29bb
--- /dev/null
+++ b/tools/testing/selftests/powerpc/primitives/asm/processor.h
diff --git a/tools/testing/selftests/powerpc/primitives/linux/stringify.h b/tools/testing/selftests/powerpc/primitives/linux/stringify.h
new file mode 100644
index 000000000..e69de29bb
--- /dev/null
+++ b/tools/testing/selftests/powerpc/primitives/linux/stringify.h
diff --git a/tools/testing/selftests/powerpc/primitives/load_unaligned_zeropad.c b/tools/testing/selftests/powerpc/primitives/load_unaligned_zeropad.c
new file mode 100644
index 000000000..1439c8c7f
--- /dev/null
+++ b/tools/testing/selftests/powerpc/primitives/load_unaligned_zeropad.c
@@ -0,0 +1,147 @@
+// SPDX-License-Identifier: GPL-2.0-or-later
+/*
+ * Userspace test harness for load_unaligned_zeropad. Creates two
+ * pages and uses mprotect to prevent access to the second page and
+ * a SEGV handler that walks the exception tables and runs the fixup
+ * routine.
+ *
+ * The results are compared against a normal load that is that is
+ * performed while access to the second page is enabled via mprotect.
+ *
+ * Copyright (C) 2014 Anton Blanchard <anton@au.ibm.com>, IBM
+ */
+
+#include <stdlib.h>
+#include <string.h>
+#include <stdio.h>
+#include <stdbool.h>
+#include <signal.h>
+#include <unistd.h>
+#include <sys/mman.h>
+
+#define FIXUP_SECTION ".ex_fixup"
+
+static inline unsigned long __fls(unsigned long x);
+
+#include "word-at-a-time.h"
+
+#include "utils.h"
+
+static inline unsigned long __fls(unsigned long x)
+{
+	int lz;
+
+	asm (PPC_CNTLZL "%0,%1" : "=r" (lz) : "r" (x));
+	return sizeof(unsigned long) - 1 - lz;
+}
+
+static int page_size;
+static char *mem_region;
+
+static int protect_region(void)
+{
+	if (mprotect(mem_region + page_size, page_size, PROT_NONE)) {
+		perror("mprotect");
+		return 1;
+	}
+
+	return 0;
+}
+
+static int unprotect_region(void)
+{
+	if (mprotect(mem_region + page_size, page_size, PROT_READ|PROT_WRITE)) {
+		perror("mprotect");
+		return 1;
+	}
+
+	return 0;
+}
+
+extern char __start___ex_table[];
+extern char __stop___ex_table[];
+
+struct extbl_entry {
+	int insn;
+	int fixup;
+};
+
+static void segv_handler(int signr, siginfo_t *info, void *ptr)
+{
+	ucontext_t *uc = (ucontext_t *)ptr;
+	unsigned long addr = (unsigned long)info->si_addr;
+	unsigned long *ip = &UCONTEXT_NIA(uc);
+	struct extbl_entry *entry = (struct extbl_entry *)__start___ex_table;
+
+	while (entry < (struct extbl_entry *)__stop___ex_table) {
+		unsigned long insn, fixup;
+
+		insn  = (unsigned long)&entry->insn + entry->insn;
+		fixup = (unsigned long)&entry->fixup + entry->fixup;
+
+		if (insn == *ip) {
+			*ip = fixup;
+			return;
+		}
+	}
+
+	printf("No exception table match for NIA %lx ADDR %lx\n", *ip, addr);
+	abort();
+}
+
+static void setup_segv_handler(void)
+{
+	struct sigaction action;
+
+	memset(&action, 0, sizeof(action));
+	action.sa_sigaction = segv_handler;
+	action.sa_flags = SA_SIGINFO;
+	sigaction(SIGSEGV, &action, NULL);
+}
+
+static int do_one_test(char *p, int page_offset)
+{
+	unsigned long should;
+	unsigned long got;
+
+	FAIL_IF(unprotect_region());
+	should = *(unsigned long *)p;
+	FAIL_IF(protect_region());
+
+	got = load_unaligned_zeropad(p);
+
+	if (should != got) {
+		printf("offset %u load_unaligned_zeropad returned 0x%lx, should be 0x%lx\n", page_offset, got, should);
+		return 1;
+	}
+
+	return 0;
+}
+
+static int test_body(void)
+{
+	unsigned long i;
+
+	page_size = getpagesize();
+	mem_region = mmap(NULL, page_size * 2, PROT_READ|PROT_WRITE,
+		MAP_PRIVATE|MAP_ANONYMOUS, -1, 0);
+
+	FAIL_IF(mem_region == MAP_FAILED);
+
+	for (i = 0; i < page_size; i++)
+		mem_region[i] = i;
+
+	memset(mem_region+page_size, 0, page_size);
+
+	setup_segv_handler();
+
+	for (i = 0; i < page_size; i++)
+		FAIL_IF(do_one_test(mem_region+i, i));
+
+	return 0;
+}
+
+int main(void)
+{
+	return test_harness(test_body, "load_unaligned_zeropad");
+}
diff --git a/tools/testing/selftests/powerpc/primitives/word-at-a-time.h b/tools/testing/selftests/powerpc/primitives/word-at-a-time.h
new file mode 120000
index 000000000..eb74401b5
--- /dev/null
+++ b/tools/testing/selftests/powerpc/primitives/word-at-a-time.h
@@ -0,0 +1 @@
+../../../../../arch/powerpc/include/asm/word-at-a-time.h
\ No newline at end of file
diff --git a/tools/testing/selftests/powerpc/ptrace/.gitignore b/tools/testing/selftests/powerpc/ptrace/.gitignore
new file mode 100644
index 000000000..0e96150b7
--- /dev/null
+++ b/tools/testing/selftests/powerpc/ptrace/.gitignore
@@ -0,0 +1,16 @@
+# SPDX-License-Identifier: GPL-2.0-only
+ptrace-gpr
+ptrace-tm-gpr
+ptrace-tm-spd-gpr
+ptrace-tar
+ptrace-tm-tar
+ptrace-tm-spd-tar
+ptrace-vsx
+ptrace-tm-vsx
+ptrace-tm-spd-vsx
+ptrace-tm-spr
+ptrace-hwbreak
+perf-hwbreak
+core-pkey
+ptrace-pkey
+ptrace-syscall
diff --git a/tools/testing/selftests/powerpc/ptrace/Makefile b/tools/testing/selftests/powerpc/ptrace/Makefile
new file mode 100644
index 000000000..8d3f006c9
--- /dev/null
+++ b/tools/testing/selftests/powerpc/ptrace/Makefile
@@ -0,0 +1,15 @@
+# SPDX-License-Identifier: GPL-2.0
+TEST_GEN_PROGS := ptrace-gpr ptrace-tm-gpr ptrace-tm-spd-gpr \
+              ptrace-tar ptrace-tm-tar ptrace-tm-spd-tar ptrace-vsx ptrace-tm-vsx \
+              ptrace-tm-spd-vsx ptrace-tm-spr ptrace-hwbreak ptrace-pkey core-pkey \
+              perf-hwbreak ptrace-syscall
+
+top_srcdir = ../../../../..
+include ../../lib.mk
+
+CFLAGS += -m64 -I../../../../../usr/include -I../tm -mhtm -fno-pie
+
+$(OUTPUT)/ptrace-pkey $(OUTPUT)/core-pkey: child.h
+$(OUTPUT)/ptrace-pkey $(OUTPUT)/core-pkey: LDLIBS += -pthread
+
+$(TEST_GEN_PROGS): ../harness.c ../utils.c ../lib/reg.S ptrace.h
diff --git a/tools/testing/selftests/powerpc/ptrace/child.h b/tools/testing/selftests/powerpc/ptrace/child.h
new file mode 100644
index 000000000..d7275b7b3
--- /dev/null
+++ b/tools/testing/selftests/powerpc/ptrace/child.h
@@ -0,0 +1,139 @@
+// SPDX-License-Identifier: GPL-2.0+
+/*
+ * Helper functions to sync execution between parent and child processes.
+ *
+ * Copyright 2018, Thiago Jung Bauermann, IBM Corporation.
+ */
+#include <stdio.h>
+#include <stdbool.h>
+#include <semaphore.h>
+
+/*
+ * Information in a shared memory location for synchronization between child and
+ * parent.
+ */
+struct child_sync {
+	/* The parent waits on this semaphore. */
+	sem_t sem_parent;
+
+	/* If true, the child should give up as well. */
+	bool parent_gave_up;
+
+	/* The child waits on this semaphore. */
+	sem_t sem_child;
+
+	/* If true, the parent should give up as well. */
+	bool child_gave_up;
+};
+
+#define CHILD_FAIL_IF(x, sync)						\
+	do {								\
+		if (x) {						\
+			fprintf(stderr,					\
+				"[FAIL] Test FAILED on line %d\n", __LINE__); \
+			(sync)->child_gave_up = true;			\
+			prod_parent(sync);				\
+			return 1;					\
+		}							\
+	} while (0)
+
+#define PARENT_FAIL_IF(x, sync)						\
+	do {								\
+		if (x) {						\
+			fprintf(stderr,					\
+				"[FAIL] Test FAILED on line %d\n", __LINE__); \
+			(sync)->parent_gave_up = true;			\
+			prod_child(sync);				\
+			return 1;					\
+		}							\
+	} while (0)
+
+#define PARENT_SKIP_IF_UNSUPPORTED(x, sync)				\
+	do {								\
+		if ((x) == -1 && (errno == ENODEV || errno == EINVAL)) { \
+			(sync)->parent_gave_up = true;			\
+			prod_child(sync);				\
+			SKIP_IF(1);					\
+		}							\
+	} while (0)
+
+int init_child_sync(struct child_sync *sync)
+{
+	int ret;
+
+	ret = sem_init(&sync->sem_parent, 1, 0);
+	if (ret) {
+		perror("Semaphore initialization failed");
+		return 1;
+	}
+
+	ret = sem_init(&sync->sem_child, 1, 0);
+	if (ret) {
+		perror("Semaphore initialization failed");
+		return 1;
+	}
+
+	return 0;
+}
+
+void destroy_child_sync(struct child_sync *sync)
+{
+	sem_destroy(&sync->sem_parent);
+	sem_destroy(&sync->sem_child);
+}
+
+int wait_child(struct child_sync *sync)
+{
+	int ret;
+
+	/* Wait until the child prods us. */
+	ret = sem_wait(&sync->sem_parent);
+	if (ret) {
+		perror("Error waiting for child");
+		return 1;
+	}
+
+	return sync->child_gave_up;
+}
+
+int prod_child(struct child_sync *sync)
+{
+	int ret;
+
+	/* Unblock the child now. */
+	ret = sem_post(&sync->sem_child);
+	if (ret) {
+		perror("Error prodding child");
+		return 1;
+	}
+
+	return 0;
+}
+
+int wait_parent(struct child_sync *sync)
+{
+	int ret;
+
+	/* Wait until the parent prods us. */
+	ret = sem_wait(&sync->sem_child);
+	if (ret) {
+		perror("Error waiting for parent");
+		return 1;
+	}
+
+	return sync->parent_gave_up;
+}
+
+int prod_parent(struct child_sync *sync)
+{
+	int ret;
+
+	/* Unblock the parent now. */
+	ret = sem_post(&sync->sem_parent);
+	if (ret) {
+		perror("Error prodding parent");
+		return 1;
+	}
+
+	return 0;
+}
diff --git a/tools/testing/selftests/powerpc/ptrace/core-pkey.c b/tools/testing/selftests/powerpc/ptrace/core-pkey.c
new file mode 100644
index 000000000..bbc05ffc5
--- /dev/null
+++ b/tools/testing/selftests/powerpc/ptrace/core-pkey.c
@@ -0,0 +1,462 @@
+// SPDX-License-Identifier: GPL-2.0+
+/*
+ * Ptrace test for Memory Protection Key registers
+ *
+ * Copyright (C) 2015 Anshuman Khandual, IBM Corporation.
+ * Copyright (C) 2018 IBM Corporation.
+ */
+#include <limits.h>
+#include <linux/kernel.h>
+#include <sys/mman.h>
+#include <sys/types.h>
+#include <sys/stat.h>
+#include <sys/time.h>
+#include <sys/resource.h>
+#include <fcntl.h>
+#include <unistd.h>
+#include "ptrace.h"
+#include "child.h"
+
+#ifndef __NR_pkey_alloc
+#define __NR_pkey_alloc		384
+#endif
+
+#ifndef __NR_pkey_free
+#define __NR_pkey_free		385
+#endif
+
+#ifndef NT_PPC_PKEY
+#define NT_PPC_PKEY		0x110
+#endif
+
+#ifndef PKEY_DISABLE_EXECUTE
+#define PKEY_DISABLE_EXECUTE	0x4
+#endif
+
+#define AMR_BITS_PER_PKEY 2
+#define PKEY_REG_BITS (sizeof(u64) * 8)
+#define pkeyshift(pkey) (PKEY_REG_BITS - ((pkey + 1) * AMR_BITS_PER_PKEY))
+
+#define CORE_FILE_LIMIT	(5 * 1024 * 1024)	/* 5 MB should be enough */
+
+static const char core_pattern_file[] = "/proc/sys/kernel/core_pattern";
+
+static const char user_write[] = "[User Write (Running)]";
+static const char core_read_running[] = "[Core Read (Running)]";
+
+/* Information shared between the parent and the child. */
+struct shared_info {
+	struct child_sync child_sync;
+
+	/* AMR value the parent expects to read in the core file. */
+	unsigned long amr;
+
+	/* IAMR value the parent expects to read in the core file. */
+	unsigned long iamr;
+
+	/* UAMOR value the parent expects to read in the core file. */
+	unsigned long uamor;
+
+	/* When the child crashed. */
+	time_t core_time;
+};
+
+static int sys_pkey_alloc(unsigned long flags, unsigned long init_access_rights)
+{
+	return syscall(__NR_pkey_alloc, flags, init_access_rights);
+}
+
+static int sys_pkey_free(int pkey)
+{
+	return syscall(__NR_pkey_free, pkey);
+}
+
+static int increase_core_file_limit(void)
+{
+	struct rlimit rlim;
+	int ret;
+
+	ret = getrlimit(RLIMIT_CORE, &rlim);
+	FAIL_IF(ret);
+
+	if (rlim.rlim_cur != RLIM_INFINITY && rlim.rlim_cur < CORE_FILE_LIMIT) {
+		rlim.rlim_cur = CORE_FILE_LIMIT;
+
+		if (rlim.rlim_max != RLIM_INFINITY &&
+		    rlim.rlim_max < CORE_FILE_LIMIT)
+			rlim.rlim_max = CORE_FILE_LIMIT;
+
+		ret = setrlimit(RLIMIT_CORE, &rlim);
+		FAIL_IF(ret);
+	}
+
+	ret = getrlimit(RLIMIT_FSIZE, &rlim);
+	FAIL_IF(ret);
+
+	if (rlim.rlim_cur != RLIM_INFINITY && rlim.rlim_cur < CORE_FILE_LIMIT) {
+		rlim.rlim_cur = CORE_FILE_LIMIT;
+
+		if (rlim.rlim_max != RLIM_INFINITY &&
+		    rlim.rlim_max < CORE_FILE_LIMIT)
+			rlim.rlim_max = CORE_FILE_LIMIT;
+
+		ret = setrlimit(RLIMIT_FSIZE, &rlim);
+		FAIL_IF(ret);
+	}
+
+	return TEST_PASS;
+}
+
+static int child(struct shared_info *info)
+{
+	bool disable_execute = true;
+	int pkey1, pkey2, pkey3;
+	int *ptr, ret;
+
+	/* Wait until parent fills out the initial register values. */
+	ret = wait_parent(&info->child_sync);
+	if (ret)
+		return ret;
+
+	ret = increase_core_file_limit();
+	FAIL_IF(ret);
+
+	/* Get some pkeys so that we can change their bits in the AMR. */
+	pkey1 = sys_pkey_alloc(0, PKEY_DISABLE_EXECUTE);
+	if (pkey1 < 0) {
+		pkey1 = sys_pkey_alloc(0, 0);
+		FAIL_IF(pkey1 < 0);
+
+		disable_execute = false;
+	}
+
+	pkey2 = sys_pkey_alloc(0, 0);
+	FAIL_IF(pkey2 < 0);
+
+	pkey3 = sys_pkey_alloc(0, 0);
+	FAIL_IF(pkey3 < 0);
+
+	info->amr |= 3ul << pkeyshift(pkey1) | 2ul << pkeyshift(pkey2);
+
+	if (disable_execute)
+		info->iamr |= 1ul << pkeyshift(pkey1);
+	else
+		info->iamr &= ~(1ul << pkeyshift(pkey1));
+
+	info->iamr &= ~(1ul << pkeyshift(pkey2) | 1ul << pkeyshift(pkey3));
+
+	info->uamor |= 3ul << pkeyshift(pkey1) | 3ul << pkeyshift(pkey2);
+
+	printf("%-30s AMR: %016lx pkey1: %d pkey2: %d pkey3: %d\n",
+	       user_write, info->amr, pkey1, pkey2, pkey3);
+
+	set_amr(info->amr);
+
+	/*
+	 * We won't use pkey3. This tests whether the kernel restores the UAMOR
+	 * permissions after a key is freed.
+	 */
+	sys_pkey_free(pkey3);
+
+	info->core_time = time(NULL);
+
+	/* Crash. */
+	ptr = 0;
+	*ptr = 1;
+
+	/* Shouldn't get here. */
+	FAIL_IF(true);
+
+	return TEST_FAIL;
+}
+
+/* Return file size if filename exists and pass sanity check, or zero if not. */
+static off_t try_core_file(const char *filename, struct shared_info *info,
+			   pid_t pid)
+{
+	struct stat buf;
+	int ret;
+
+	ret = stat(filename, &buf);
+	if (ret == -1)
+		return TEST_FAIL;
+
+	/* Make sure we're not using a stale core file. */
+	return buf.st_mtime >= info->core_time ? buf.st_size : TEST_FAIL;
+}
+
+static Elf64_Nhdr *next_note(Elf64_Nhdr *nhdr)
+{
+	return (void *) nhdr + sizeof(*nhdr) +
+		__ALIGN_KERNEL(nhdr->n_namesz, 4) +
+		__ALIGN_KERNEL(nhdr->n_descsz, 4);
+}
+
+static int check_core_file(struct shared_info *info, Elf64_Ehdr *ehdr,
+			   off_t core_size)
+{
+	unsigned long *regs;
+	Elf64_Phdr *phdr;
+	Elf64_Nhdr *nhdr;
+	size_t phdr_size;
+	void *p = ehdr, *note;
+	int ret;
+
+	ret = memcmp(ehdr->e_ident, ELFMAG, SELFMAG);
+	FAIL_IF(ret);
+
+	FAIL_IF(ehdr->e_type != ET_CORE);
+	FAIL_IF(ehdr->e_machine != EM_PPC64);
+	FAIL_IF(ehdr->e_phoff == 0 || ehdr->e_phnum == 0);
+
+	/*
+	 * e_phnum is at most 65535 so calculating the size of the
+	 * program header cannot overflow.
+	 */
+	phdr_size = sizeof(*phdr) * ehdr->e_phnum;
+
+	/* Sanity check the program header table location. */
+	FAIL_IF(ehdr->e_phoff + phdr_size < ehdr->e_phoff);
+	FAIL_IF(ehdr->e_phoff + phdr_size > core_size);
+
+	/* Find the PT_NOTE segment. */
+	for (phdr = p + ehdr->e_phoff;
+	     (void *) phdr < p + ehdr->e_phoff + phdr_size;
+	     phdr += ehdr->e_phentsize)
+		if (phdr->p_type == PT_NOTE)
+			break;
+
+	FAIL_IF((void *) phdr >= p + ehdr->e_phoff + phdr_size);
+
+	/* Find the NT_PPC_PKEY note. */
+	for (nhdr = p + phdr->p_offset;
+	     (void *) nhdr < p + phdr->p_offset + phdr->p_filesz;
+	     nhdr = next_note(nhdr))
+		if (nhdr->n_type == NT_PPC_PKEY)
+			break;
+
+	FAIL_IF((void *) nhdr >= p + phdr->p_offset + phdr->p_filesz);
+	FAIL_IF(nhdr->n_descsz == 0);
+
+	p = nhdr;
+	note = p + sizeof(*nhdr) + __ALIGN_KERNEL(nhdr->n_namesz, 4);
+
+	regs = (unsigned long *) note;
+
+	printf("%-30s AMR: %016lx IAMR: %016lx UAMOR: %016lx\n",
+	       core_read_running, regs[0], regs[1], regs[2]);
+
+	FAIL_IF(regs[0] != info->amr);
+	FAIL_IF(regs[1] != info->iamr);
+	FAIL_IF(regs[2] != info->uamor);
+
+	return TEST_PASS;
+}
+
+static int parent(struct shared_info *info, pid_t pid)
+{
+	char *filenames, *filename[3];
+	int fd, i, ret, status;
+	unsigned long regs[3];
+	off_t core_size;
+	void *core;
+
+	/*
+	 * Get the initial values for AMR, IAMR and UAMOR and communicate them
+	 * to the child.
+	 */
+	ret = ptrace_read_regs(pid, NT_PPC_PKEY, regs, 3);
+	PARENT_SKIP_IF_UNSUPPORTED(ret, &info->child_sync);
+	PARENT_FAIL_IF(ret, &info->child_sync);
+
+	info->amr = regs[0];
+	info->iamr = regs[1];
+	info->uamor = regs[2];
+
+	/* Wake up child so that it can set itself up. */
+	ret = prod_child(&info->child_sync);
+	PARENT_FAIL_IF(ret, &info->child_sync);
+
+	ret = wait(&status);
+	if (ret != pid) {
+		printf("Child's exit status not captured\n");
+		return TEST_FAIL;
+	} else if (!WIFSIGNALED(status) || !WCOREDUMP(status)) {
+		printf("Child didn't dump core\n");
+		return TEST_FAIL;
+	}
+
+	/* Construct array of core file names to try. */
+
+	filename[0] = filenames = malloc(PATH_MAX);
+	if (!filenames) {
+		perror("Error allocating memory");
+		return TEST_FAIL;
+	}
+
+	ret = snprintf(filename[0], PATH_MAX, "core-pkey.%d", pid);
+	if (ret < 0 || ret >= PATH_MAX) {
+		ret = TEST_FAIL;
+		goto out;
+	}
+
+	filename[1] = filename[0] + ret + 1;
+	ret = snprintf(filename[1], PATH_MAX - ret - 1, "core.%d", pid);
+	if (ret < 0 || ret >= PATH_MAX - ret - 1) {
+		ret = TEST_FAIL;
+		goto out;
+	}
+	filename[2] = "core";
+
+	for (i = 0; i < 3; i++) {
+		core_size = try_core_file(filename[i], info, pid);
+		if (core_size != TEST_FAIL)
+			break;
+	}
+
+	if (i == 3) {
+		printf("Couldn't find core file\n");
+		ret = TEST_FAIL;
+		goto out;
+	}
+
+	fd = open(filename[i], O_RDONLY);
+	if (fd == -1) {
+		perror("Error opening core file");
+		ret = TEST_FAIL;
+		goto out;
+	}
+
+	core = mmap(NULL, core_size, PROT_READ, MAP_PRIVATE, fd, 0);
+	if (core == (void *) -1) {
+		perror("Error mmaping core file");
+		ret = TEST_FAIL;
+		goto out;
+	}
+
+	ret = check_core_file(info, core, core_size);
+
+	munmap(core, core_size);
+	close(fd);
+	unlink(filename[i]);
+
+ out:
+	free(filenames);
+
+	return ret;
+}
+
+static int write_core_pattern(const char *core_pattern)
+{
+	size_t len = strlen(core_pattern), ret;
+	FILE *f;
+
+	f = fopen(core_pattern_file, "w");
+	SKIP_IF_MSG(!f, "Try with root privileges");
+
+	ret = fwrite(core_pattern, 1, len, f);
+	fclose(f);
+	if (ret != len) {
+		perror("Error writing to core_pattern file");
+		return TEST_FAIL;
+	}
+
+	return TEST_PASS;
+}
+
+static int setup_core_pattern(char **core_pattern_, bool *changed_)
+{
+	FILE *f;
+	char *core_pattern;
+	int ret;
+
+	core_pattern = malloc(PATH_MAX);
+	if (!core_pattern) {
+		perror("Error allocating memory");
+		return TEST_FAIL;
+	}
+
+	f = fopen(core_pattern_file, "r");
+	if (!f) {
+		perror("Error opening core_pattern file");
+		ret = TEST_FAIL;
+		goto out;
+	}
+
+	ret = fread(core_pattern, 1, PATH_MAX, f);
+	fclose(f);
+	if (!ret) {
+		perror("Error reading core_pattern file");
+		ret = TEST_FAIL;
+		goto out;
+	}
+
+	/* Check whether we can predict the name of the core file. */
+	if (!strcmp(core_pattern, "core") || !strcmp(core_pattern, "core.%p"))
+		*changed_ = false;
+	else {
+		ret = write_core_pattern("core-pkey.%p");
+		if (ret)
+			goto out;
+
+		*changed_ = true;
+	}
+
+	*core_pattern_ = core_pattern;
+	ret = TEST_PASS;
+
+ out:
+	if (ret)
+		free(core_pattern);
+
+	return ret;
+}
+
+static int core_pkey(void)
+{
+	char *core_pattern;
+	bool changed_core_pattern;
+	struct shared_info *info;
+	int shm_id;
+	int ret;
+	pid_t pid;
+
+	ret = setup_core_pattern(&core_pattern, &changed_core_pattern);
+	if (ret)
+		return ret;
+
+	shm_id = shmget(IPC_PRIVATE, sizeof(*info), 0777 | IPC_CREAT);
+	info = shmat(shm_id, NULL, 0);
+
+	ret = init_child_sync(&info->child_sync);
+	if (ret)
+		return ret;
+
+	pid = fork();
+	if (pid < 0) {
+		perror("fork() failed");
+		ret = TEST_FAIL;
+	} else if (pid == 0)
+		ret = child(info);
+	else
+		ret = parent(info, pid);
+
+	shmdt(info);
+
+	if (pid) {
+		destroy_child_sync(&info->child_sync);
+		shmctl(shm_id, IPC_RMID, NULL);
+
+		if (changed_core_pattern)
+			write_core_pattern(core_pattern);
+	}
+
+	free(core_pattern);
+
+	return ret;
+}
+
+int main(int argc, char *argv[])
+{
+	return test_harness(core_pkey, "core_pkey");
+}
diff --git a/tools/testing/selftests/powerpc/ptrace/perf-hwbreak.c b/tools/testing/selftests/powerpc/ptrace/perf-hwbreak.c
new file mode 100644
index 000000000..c1f324afd
--- /dev/null
+++ b/tools/testing/selftests/powerpc/ptrace/perf-hwbreak.c
@@ -0,0 +1,308 @@
+// SPDX-License-Identifier: GPL-2.0-or-later
+/*
+ * perf events self profiling example test case for hw breakpoints.
+ *
+ * This tests perf PERF_TYPE_BREAKPOINT parameters
+ * 1) tests all variants of the break on read/write flags
+ * 2) tests exclude_user == 0 and 1
+ * 3) test array matches (if DAWR is supported))
+ * 4) test different numbers of breakpoints matches
+ *
+ * Configure this breakpoint, then read and write the data a number of
+ * times. Then check the output count from perf is as expected.
+ *
+ * Based on:
+ *   http://ozlabs.org/~anton/junkcode/perf_events_example1.c
+ *
+ * Copyright (C) 2018 Michael Neuling, IBM Corporation.
+ */
+
+#include <unistd.h>
+#include <assert.h>
+#include <stdio.h>
+#include <stdlib.h>
+#include <string.h>
+#include <sys/ioctl.h>
+#include <elf.h>
+#include <pthread.h>
+#include <sys/syscall.h>
+#include <linux/perf_event.h>
+#include <linux/hw_breakpoint.h>
+#include "utils.h"
+
+#define MAX_LOOPS 10000
+
+#define DAWR_LENGTH_MAX ((0x3f + 1) * 8)
+
+static inline int sys_perf_event_open(struct perf_event_attr *attr, pid_t pid,
+				      int cpu, int group_fd,
+				      unsigned long flags)
+{
+	attr->size = sizeof(*attr);
+	return syscall(__NR_perf_event_open, attr, pid, cpu, group_fd, flags);
+}
+
+static inline bool breakpoint_test(int len)
+{
+	struct perf_event_attr attr;
+	int fd;
+
+	/* setup counters */
+	memset(&attr, 0, sizeof(attr));
+	attr.disabled = 1;
+	attr.type = PERF_TYPE_BREAKPOINT;
+	attr.bp_type = HW_BREAKPOINT_R;
+	/* bp_addr can point anywhere but needs to be aligned */
+	attr.bp_addr = (__u64)(&attr) & 0xfffffffffffff800;
+	attr.bp_len = len;
+	fd = sys_perf_event_open(&attr, 0, -1, -1, 0);
+	if (fd < 0)
+		return false;
+	close(fd);
+	return true;
+}
+
+static inline bool perf_breakpoint_supported(void)
+{
+	return breakpoint_test(4);
+}
+
+static inline bool dawr_supported(void)
+{
+	return breakpoint_test(DAWR_LENGTH_MAX);
+}
+
+static int runtestsingle(int readwriteflag, int exclude_user, int arraytest)
+{
+	int i,j;
+	struct perf_event_attr attr;
+	size_t res;
+	unsigned long long breaks, needed;
+	int readint;
+	int readintarraybig[2*DAWR_LENGTH_MAX/sizeof(int)];
+	int *readintalign;
+	volatile int *ptr;
+	int break_fd;
+	int loop_num = MAX_LOOPS - (rand() % 100); /* provide some variability */
+	volatile int *k;
+
+	/* align to 0x400 boundary as required by DAWR */
+	readintalign = (int *)(((unsigned long)readintarraybig + 0x7ff) &
+			       0xfffffffffffff800);
+
+	ptr = &readint;
+	if (arraytest)
+		ptr = &readintalign[0];
+
+	/* setup counters */
+	memset(&attr, 0, sizeof(attr));
+	attr.disabled = 1;
+	attr.type = PERF_TYPE_BREAKPOINT;
+	attr.bp_type = readwriteflag;
+	attr.bp_addr = (__u64)ptr;
+	attr.bp_len = sizeof(int);
+	if (arraytest)
+		attr.bp_len = DAWR_LENGTH_MAX;
+	attr.exclude_user = exclude_user;
+	break_fd = sys_perf_event_open(&attr, 0, -1, -1, 0);
+	if (break_fd < 0) {
+		perror("sys_perf_event_open");
+		exit(1);
+	}
+
+	/* start counters */
+	ioctl(break_fd, PERF_EVENT_IOC_ENABLE);
+
+	/* Test a bunch of reads and writes */
+	k = &readint;
+	for (i = 0; i < loop_num; i++) {
+		if (arraytest)
+			k = &(readintalign[i % (DAWR_LENGTH_MAX/sizeof(int))]);
+
+		j = *k;
+		*k = j;
+	}
+
+	/* stop counters */
+	ioctl(break_fd, PERF_EVENT_IOC_DISABLE);
+
+	/* read and check counters */
+	res = read(break_fd, &breaks, sizeof(unsigned long long));
+	assert(res == sizeof(unsigned long long));
+	/* we read and write each loop, so subtract the ones we are counting */
+	needed = 0;
+	if (readwriteflag & HW_BREAKPOINT_R)
+		needed += loop_num;
+	if (readwriteflag & HW_BREAKPOINT_W)
+		needed += loop_num;
+	needed = needed * (1 - exclude_user);
+	printf("TESTED: addr:0x%lx brks:% 8lld loops:% 8i rw:%i !user:%i array:%i\n",
+	       (unsigned long int)ptr, breaks, loop_num, readwriteflag, exclude_user, arraytest);
+	if (breaks != needed) {
+		printf("FAILED: 0x%lx brks:%lld needed:%lli %i %i %i\n\n",
+		       (unsigned long int)ptr, breaks, needed, loop_num, readwriteflag, exclude_user);
+		return 1;
+	}
+	close(break_fd);
+
+	return 0;
+}
+
+static int runtest_dar_outside(void)
+{
+	void *target;
+	volatile __u16 temp16;
+	volatile __u64 temp64;
+	struct perf_event_attr attr;
+	int break_fd;
+	unsigned long long breaks;
+	int fail = 0;
+	size_t res;
+
+	target = malloc(8);
+	if (!target) {
+		perror("malloc failed");
+		exit(EXIT_FAILURE);
+	}
+
+	/* setup counters */
+	memset(&attr, 0, sizeof(attr));
+	attr.disabled = 1;
+	attr.type = PERF_TYPE_BREAKPOINT;
+	attr.exclude_kernel = 1;
+	attr.exclude_hv = 1;
+	attr.exclude_guest = 1;
+	attr.bp_type = HW_BREAKPOINT_RW;
+	/* watch middle half of target array */
+	attr.bp_addr = (__u64)(target + 2);
+	attr.bp_len = 4;
+	break_fd = sys_perf_event_open(&attr, 0, -1, -1, 0);
+	if (break_fd < 0) {
+		free(target);
+		perror("sys_perf_event_open");
+		exit(EXIT_FAILURE);
+	}
+
+	/* Shouldn't hit. */
+	ioctl(break_fd, PERF_EVENT_IOC_RESET);
+	ioctl(break_fd, PERF_EVENT_IOC_ENABLE);
+	temp16 = *((__u16 *)target);
+	*((__u16 *)target) = temp16;
+	ioctl(break_fd, PERF_EVENT_IOC_DISABLE);
+	res = read(break_fd, &breaks, sizeof(unsigned long long));
+	assert(res == sizeof(unsigned long long));
+	if (breaks == 0) {
+		printf("TESTED: No overlap\n");
+	} else {
+		printf("FAILED: No overlap: %lld != 0\n", breaks);
+		fail = 1;
+	}
+
+	/* Hit */
+	ioctl(break_fd, PERF_EVENT_IOC_RESET);
+	ioctl(break_fd, PERF_EVENT_IOC_ENABLE);
+	temp16 = *((__u16 *)(target + 1));
+	*((__u16 *)(target + 1)) = temp16;
+	ioctl(break_fd, PERF_EVENT_IOC_DISABLE);
+	res = read(break_fd, &breaks, sizeof(unsigned long long));
+	assert(res == sizeof(unsigned long long));
+	if (breaks == 2) {
+		printf("TESTED: Partial overlap\n");
+	} else {
+		printf("FAILED: Partial overlap: %lld != 2\n", breaks);
+		fail = 1;
+	}
+
+	/* Hit */
+	ioctl(break_fd, PERF_EVENT_IOC_RESET);
+	ioctl(break_fd, PERF_EVENT_IOC_ENABLE);
+	temp16 = *((__u16 *)(target + 5));
+	*((__u16 *)(target + 5)) = temp16;
+	ioctl(break_fd, PERF_EVENT_IOC_DISABLE);
+	res = read(break_fd, &breaks, sizeof(unsigned long long));
+	assert(res == sizeof(unsigned long long));
+	if (breaks == 2) {
+		printf("TESTED: Partial overlap\n");
+	} else {
+		printf("FAILED: Partial overlap: %lld != 2\n", breaks);
+		fail = 1;
+	}
+
+	/* Shouldn't Hit */
+	ioctl(break_fd, PERF_EVENT_IOC_RESET);
+	ioctl(break_fd, PERF_EVENT_IOC_ENABLE);
+	temp16 = *((__u16 *)(target + 6));
+	*((__u16 *)(target + 6)) = temp16;
+	ioctl(break_fd, PERF_EVENT_IOC_DISABLE);
+	res = read(break_fd, &breaks, sizeof(unsigned long long));
+	assert(res == sizeof(unsigned long long));
+	if (breaks == 0) {
+		printf("TESTED: No overlap\n");
+	} else {
+		printf("FAILED: No overlap: %lld != 0\n", breaks);
+		fail = 1;
+	}
+
+	/* Hit */
+	ioctl(break_fd, PERF_EVENT_IOC_RESET);
+	ioctl(break_fd, PERF_EVENT_IOC_ENABLE);
+	temp64 = *((__u64 *)target);
+	*((__u64 *)target) = temp64;
+	ioctl(break_fd, PERF_EVENT_IOC_DISABLE);
+	res = read(break_fd, &breaks, sizeof(unsigned long long));
+	assert(res == sizeof(unsigned long long));
+	if (breaks == 2) {
+		printf("TESTED: Full overlap\n");
+	} else {
+		printf("FAILED: Full overlap: %lld != 2\n", breaks);
+		fail = 1;
+	}
+
+	free(target);
+	close(break_fd);
+	return fail;
+}
+
+static int runtest(void)
+{
+	int rwflag;
+	int exclude_user;
+	int ret;
+
+	/*
+	 * perf defines rwflag as two bits read and write and at least
+	 * one must be set.  So range 1-3.
+	 */
+	for (rwflag = 1 ; rwflag < 4; rwflag++) {
+		for (exclude_user = 0 ; exclude_user < 2; exclude_user++) {
+			ret = runtestsingle(rwflag, exclude_user, 0);
+			if (ret)
+				return ret;
+
+			/* if we have the dawr, we can do an array test */
+			if (!dawr_supported())
+				continue;
+			ret = runtestsingle(rwflag, exclude_user, 1);
+			if (ret)
+				return ret;
+		}
+	}
+
+	ret = runtest_dar_outside();
+	return ret;
+}
+
+
+static int perf_hwbreak(void)
+{
+	srand ( time(NULL) );
+
+	SKIP_IF(!perf_breakpoint_supported());
+
+	return runtest();
+}
+
+int main(int argc, char *argv[], char **envp)
+{
+	return test_harness(perf_hwbreak, "perf_hwbreak");
+}
diff --git a/tools/testing/selftests/powerpc/ptrace/ptrace-gpr.c b/tools/testing/selftests/powerpc/ptrace/ptrace-gpr.c
new file mode 100644
index 000000000..17cd480c8
--- /dev/null
+++ b/tools/testing/selftests/powerpc/ptrace/ptrace-gpr.c
@@ -0,0 +1,119 @@
+// SPDX-License-Identifier: GPL-2.0-or-later
+/*
+ * Ptrace test for GPR/FPR registers
+ *
+ * Copyright (C) 2015 Anshuman Khandual, IBM Corporation.
+ */
+#include "ptrace.h"
+#include "ptrace-gpr.h"
+#include "reg.h"
+
+/* Tracer and Tracee Shared Data */
+int shm_id;
+int *cptr, *pptr;
+
+float a = FPR_1;
+float b = FPR_2;
+float c = FPR_3;
+
+void gpr(void)
+{
+	unsigned long gpr_buf[18];
+	float fpr_buf[32];
+
+	cptr = (int *)shmat(shm_id, NULL, 0);
+
+	asm __volatile__(
+		ASM_LOAD_GPR_IMMED(gpr_1)
+		ASM_LOAD_FPR_SINGLE_PRECISION(flt_1)
+		:
+		: [gpr_1]"i"(GPR_1), [flt_1] "b" (&a)
+		: "memory", "r6", "r7", "r8", "r9", "r10",
+		"r11", "r12", "r13", "r14", "r15", "r16", "r17",
+		"r18", "r19", "r20", "r21", "r22", "r23", "r24",
+		"r25", "r26", "r27", "r28", "r29", "r30", "r31"
+		);
+
+	cptr[1] = 1;
+
+	while (!cptr[0])
+		asm volatile("" : : : "memory");
+
+	shmdt((void *)cptr);
+	store_gpr(gpr_buf);
+	store_fpr_single_precision(fpr_buf);
+
+	if (validate_gpr(gpr_buf, GPR_3))
+		exit(1);
+
+	if (validate_fpr_float(fpr_buf, c))
+		exit(1);
+
+	exit(0);
+}
+
+int trace_gpr(pid_t child)
+{
+	unsigned long gpr[18];
+	unsigned long fpr[32];
+
+	FAIL_IF(start_trace(child));
+	FAIL_IF(show_gpr(child, gpr));
+	FAIL_IF(validate_gpr(gpr, GPR_1));
+	FAIL_IF(show_fpr(child, fpr));
+	FAIL_IF(validate_fpr(fpr, FPR_1_REP));
+	FAIL_IF(write_gpr(child, GPR_3));
+	FAIL_IF(write_fpr(child, FPR_3_REP));
+	FAIL_IF(stop_trace(child));
+
+	return TEST_PASS;
+}
+
+int ptrace_gpr(void)
+{
+	pid_t pid;
+	int ret, status;
+
+	shm_id = shmget(IPC_PRIVATE, sizeof(int) * 2, 0777|IPC_CREAT);
+	pid = fork();
+	if (pid < 0) {
+		perror("fork() failed");
+		return TEST_FAIL;
+	}
+	if (pid == 0)
+		gpr();
+
+	if (pid) {
+		pptr = (int *)shmat(shm_id, NULL, 0);
+		while (!pptr[1])
+			asm volatile("" : : : "memory");
+
+		ret = trace_gpr(pid);
+		if (ret) {
+			kill(pid, SIGTERM);
+			shmdt((void *)pptr);
+			shmctl(shm_id, IPC_RMID, NULL);
+			return TEST_FAIL;
+		}
+
+		pptr[0] = 1;
+		shmdt((void *)pptr);
+
+		ret = wait(&status);
+		shmctl(shm_id, IPC_RMID, NULL);
+		if (ret != pid) {
+			printf("Child's exit status not captured\n");
+			return TEST_FAIL;
+		}
+
+		return (WIFEXITED(status) && WEXITSTATUS(status)) ? TEST_FAIL :
+			TEST_PASS;
+	}
+
+	return TEST_PASS;
+}
+
+int main(int argc, char *argv[])
+{
+	return test_harness(ptrace_gpr, "ptrace_gpr");
+}
diff --git a/tools/testing/selftests/powerpc/ptrace/ptrace-gpr.h b/tools/testing/selftests/powerpc/ptrace/ptrace-gpr.h
new file mode 100644
index 000000000..c5cd53181
--- /dev/null
+++ b/tools/testing/selftests/powerpc/ptrace/ptrace-gpr.h
@@ -0,0 +1,70 @@
+/* SPDX-License-Identifier: GPL-2.0-or-later */
+/*
+ * Copyright (C) 2015 Anshuman Khandual, IBM Corporation.
+ */
+#define GPR_1	1
+#define GPR_2	2
+#define GPR_3	3
+#define GPR_4	4
+
+#define FPR_1	0.001
+#define FPR_2	0.002
+#define FPR_3	0.003
+#define FPR_4	0.004
+
+#define FPR_1_REP 0x3f50624de0000000
+#define FPR_2_REP 0x3f60624de0000000
+#define FPR_3_REP 0x3f689374c0000000
+#define FPR_4_REP 0x3f70624de0000000
+
+/* Buffer must have 18 elements */
+int validate_gpr(unsigned long *gpr, unsigned long val)
+{
+	int i, found = 1;
+
+	for (i = 0; i < 18; i++) {
+		if (gpr[i] != val) {
+			printf("GPR[%d]: %lx Expected: %lx\n",
+				i+14, gpr[i], val);
+			found = 0;
+		}
+	}
+
+	if (!found)
+		return TEST_FAIL;
+	return TEST_PASS;
+}
+
+/* Buffer must have 32 elements */
+int validate_fpr(unsigned long *fpr, unsigned long val)
+{
+	int i, found = 1;
+
+	for (i = 0; i < 32; i++) {
+		if (fpr[i] != val) {
+			printf("FPR[%d]: %lx Expected: %lx\n", i, fpr[i], val);
+			found = 0;
+		}
+	}
+
+	if (!found)
+		return TEST_FAIL;
+	return TEST_PASS;
+}
+
+/* Buffer must have 32 elements */
+int validate_fpr_float(float *fpr, float val)
+{
+	int i, found = 1;
+
+	for (i = 0; i < 32; i++) {
+		if (fpr[i] != val) {
+			printf("FPR[%d]: %f Expected: %f\n", i, fpr[i], val);
+			found = 0;
+		}
+	}
+
+	if (!found)
+		return TEST_FAIL;
+	return TEST_PASS;
+}
diff --git a/tools/testing/selftests/powerpc/ptrace/ptrace-hwbreak.c b/tools/testing/selftests/powerpc/ptrace/ptrace-hwbreak.c
new file mode 100644
index 000000000..2e0d86e06
--- /dev/null
+++ b/tools/testing/selftests/powerpc/ptrace/ptrace-hwbreak.c
@@ -0,0 +1,550 @@
+// SPDX-License-Identifier: GPL-2.0+
+
+/*
+ * Ptrace test for hw breakpoints
+ *
+ * Based on tools/testing/selftests/breakpoints/breakpoint_test.c
+ *
+ * This test forks and the parent then traces the child doing various
+ * types of ptrace enabled breakpoints
+ *
+ * Copyright (C) 2018 Michael Neuling, IBM Corporation.
+ */
+
+#include <sys/ptrace.h>
+#include <unistd.h>
+#include <stddef.h>
+#include <sys/user.h>
+#include <stdio.h>
+#include <stdlib.h>
+#include <signal.h>
+#include <sys/types.h>
+#include <sys/wait.h>
+#include <sys/syscall.h>
+#include <linux/limits.h>
+#include "ptrace.h"
+
+#define SPRN_PVR	0x11F
+#define PVR_8xx		0x00500000
+
+bool is_8xx;
+
+/*
+ * Use volatile on all global var so that compiler doesn't
+ * optimise their load/stores. Otherwise selftest can fail.
+ */
+static volatile __u64 glvar;
+
+#define DAWR_MAX_LEN 512
+static volatile __u8 big_var[DAWR_MAX_LEN] __attribute__((aligned(512)));
+
+#define A_LEN 6
+#define B_LEN 6
+struct gstruct {
+	__u8 a[A_LEN]; /* double word aligned */
+	__u8 b[B_LEN]; /* double word unaligned */
+};
+static volatile struct gstruct gstruct __attribute__((aligned(512)));
+
+static volatile char cwd[PATH_MAX] __attribute__((aligned(8)));
+
+static void get_dbginfo(pid_t child_pid, struct ppc_debug_info *dbginfo)
+{
+	if (ptrace(PPC_PTRACE_GETHWDBGINFO, child_pid, NULL, dbginfo)) {
+		perror("Can't get breakpoint info");
+		exit(-1);
+	}
+}
+
+static bool dawr_present(struct ppc_debug_info *dbginfo)
+{
+	return !!(dbginfo->features & PPC_DEBUG_FEATURE_DATA_BP_DAWR);
+}
+
+static void write_var(int len)
+{
+	__u8 *pcvar;
+	__u16 *psvar;
+	__u32 *pivar;
+	__u64 *plvar;
+
+	switch (len) {
+	case 1:
+		pcvar = (__u8 *)&glvar;
+		*pcvar = 0xff;
+		break;
+	case 2:
+		psvar = (__u16 *)&glvar;
+		*psvar = 0xffff;
+		break;
+	case 4:
+		pivar = (__u32 *)&glvar;
+		*pivar = 0xffffffff;
+		break;
+	case 8:
+		plvar = (__u64 *)&glvar;
+		*plvar = 0xffffffffffffffffLL;
+		break;
+	}
+}
+
+static void read_var(int len)
+{
+	__u8 cvar __attribute__((unused));
+	__u16 svar __attribute__((unused));
+	__u32 ivar __attribute__((unused));
+	__u64 lvar __attribute__((unused));
+
+	switch (len) {
+	case 1:
+		cvar = (__u8)glvar;
+		break;
+	case 2:
+		svar = (__u16)glvar;
+		break;
+	case 4:
+		ivar = (__u32)glvar;
+		break;
+	case 8:
+		lvar = (__u64)glvar;
+		break;
+	}
+}
+
+static void test_workload(void)
+{
+	__u8 cvar __attribute__((unused));
+	__u32 ivar __attribute__((unused));
+	int len = 0;
+
+	if (ptrace(PTRACE_TRACEME, 0, NULL, 0)) {
+		perror("Child can't be traced?");
+		exit(-1);
+	}
+
+	/* Wake up father so that it sets up the first test */
+	kill(getpid(), SIGUSR1);
+
+	/* PTRACE_SET_DEBUGREG, WO test */
+	for (len = 1; len <= sizeof(glvar); len <<= 1)
+		write_var(len);
+
+	/* PTRACE_SET_DEBUGREG, RO test */
+	for (len = 1; len <= sizeof(glvar); len <<= 1)
+		read_var(len);
+
+	/* PTRACE_SET_DEBUGREG, RW test */
+	for (len = 1; len <= sizeof(glvar); len <<= 1) {
+		if (rand() % 2)
+			read_var(len);
+		else
+			write_var(len);
+	}
+
+	/* PTRACE_SET_DEBUGREG, Kernel Access Userspace test */
+	syscall(__NR_getcwd, &cwd, PATH_MAX);
+
+	/* PPC_PTRACE_SETHWDEBUG, MODE_EXACT, WO test */
+	write_var(1);
+
+	/* PPC_PTRACE_SETHWDEBUG, MODE_EXACT, RO test */
+	read_var(1);
+
+	/* PPC_PTRACE_SETHWDEBUG, MODE_EXACT, RW test */
+	if (rand() % 2)
+		write_var(1);
+	else
+		read_var(1);
+
+	/* PPC_PTRACE_SETHWDEBUG, MODE_EXACT, Kernel Access Userspace test */
+	syscall(__NR_getcwd, &cwd, PATH_MAX);
+
+	/* PPC_PTRACE_SETHWDEBUG, MODE_RANGE, DW ALIGNED, WO test */
+	gstruct.a[rand() % A_LEN] = 'a';
+
+	/* PPC_PTRACE_SETHWDEBUG, MODE_RANGE, DW ALIGNED, RO test */
+	cvar = gstruct.a[rand() % A_LEN];
+
+	/* PPC_PTRACE_SETHWDEBUG, MODE_RANGE, DW ALIGNED, RW test */
+	if (rand() % 2)
+		gstruct.a[rand() % A_LEN] = 'a';
+	else
+		cvar = gstruct.a[rand() % A_LEN];
+
+	/* PPC_PTRACE_SETHWDEBUG, MODE_RANGE, DW UNALIGNED, WO test */
+	gstruct.b[rand() % B_LEN] = 'b';
+
+	/* PPC_PTRACE_SETHWDEBUG, MODE_RANGE, DW UNALIGNED, RO test */
+	cvar = gstruct.b[rand() % B_LEN];
+
+	/* PPC_PTRACE_SETHWDEBUG, MODE_RANGE, DW UNALIGNED, RW test */
+	if (rand() % 2)
+		gstruct.b[rand() % B_LEN] = 'b';
+	else
+		cvar = gstruct.b[rand() % B_LEN];
+
+	/* PPC_PTRACE_SETHWDEBUG, MODE_RANGE, DW UNALIGNED, DAR OUTSIDE, RW test */
+	if (rand() % 2)
+		*((int *)(gstruct.a + 4)) = 10;
+	else
+		ivar = *((int *)(gstruct.a + 4));
+
+	/* PPC_PTRACE_SETHWDEBUG. DAWR_MAX_LEN. RW test */
+	if (rand() % 2)
+		big_var[rand() % DAWR_MAX_LEN] = 'a';
+	else
+		cvar = big_var[rand() % DAWR_MAX_LEN];
+}
+
+static void check_success(pid_t child_pid, const char *name, const char *type,
+			  unsigned long saddr, int len)
+{
+	int status;
+	siginfo_t siginfo;
+	unsigned long eaddr = (saddr + len - 1) | 0x7;
+
+	saddr &= ~0x7;
+
+	/* Wait for the child to SIGTRAP */
+	wait(&status);
+
+	ptrace(PTRACE_GETSIGINFO, child_pid, NULL, &siginfo);
+
+	if (!WIFSTOPPED(status) || WSTOPSIG(status) != SIGTRAP ||
+	    (unsigned long)siginfo.si_addr < saddr ||
+	    (unsigned long)siginfo.si_addr > eaddr) {
+		printf("%s, %s, len: %d: Fail\n", name, type, len);
+		exit(-1);
+	}
+
+	printf("%s, %s, len: %d: Ok\n", name, type, len);
+
+	if (!is_8xx) {
+		/*
+		 * For ptrace registered watchpoint, signal is generated
+		 * before executing load/store. Singlestep the instruction
+		 * and then continue the test.
+		 */
+		ptrace(PTRACE_SINGLESTEP, child_pid, NULL, 0);
+		wait(NULL);
+	}
+}
+
+static void ptrace_set_debugreg(pid_t child_pid, unsigned long wp_addr)
+{
+	if (ptrace(PTRACE_SET_DEBUGREG, child_pid, 0, wp_addr)) {
+		perror("PTRACE_SET_DEBUGREG failed");
+		exit(-1);
+	}
+}
+
+static int ptrace_sethwdebug(pid_t child_pid, struct ppc_hw_breakpoint *info)
+{
+	int wh = ptrace(PPC_PTRACE_SETHWDEBUG, child_pid, 0, info);
+
+	if (wh <= 0) {
+		perror("PPC_PTRACE_SETHWDEBUG failed");
+		exit(-1);
+	}
+	return wh;
+}
+
+static void ptrace_delhwdebug(pid_t child_pid, int wh)
+{
+	if (ptrace(PPC_PTRACE_DELHWDEBUG, child_pid, 0, wh) < 0) {
+		perror("PPC_PTRACE_DELHWDEBUG failed");
+		exit(-1);
+	}
+}
+
+#define DABR_READ_SHIFT		0
+#define DABR_WRITE_SHIFT	1
+#define DABR_TRANSLATION_SHIFT	2
+
+static int test_set_debugreg(pid_t child_pid)
+{
+	unsigned long wp_addr = (unsigned long)&glvar;
+	char *name = "PTRACE_SET_DEBUGREG";
+	int len;
+
+	/* PTRACE_SET_DEBUGREG, WO test*/
+	wp_addr &= ~0x7UL;
+	wp_addr |= (1UL << DABR_WRITE_SHIFT);
+	wp_addr |= (1UL << DABR_TRANSLATION_SHIFT);
+	for (len = 1; len <= sizeof(glvar); len <<= 1) {
+		ptrace_set_debugreg(child_pid, wp_addr);
+		ptrace(PTRACE_CONT, child_pid, NULL, 0);
+		check_success(child_pid, name, "WO", wp_addr, len);
+	}
+
+	/* PTRACE_SET_DEBUGREG, RO test */
+	wp_addr &= ~0x7UL;
+	wp_addr |= (1UL << DABR_READ_SHIFT);
+	wp_addr |= (1UL << DABR_TRANSLATION_SHIFT);
+	for (len = 1; len <= sizeof(glvar); len <<= 1) {
+		ptrace_set_debugreg(child_pid, wp_addr);
+		ptrace(PTRACE_CONT, child_pid, NULL, 0);
+		check_success(child_pid, name, "RO", wp_addr, len);
+	}
+
+	/* PTRACE_SET_DEBUGREG, RW test */
+	wp_addr &= ~0x7UL;
+	wp_addr |= (1Ul << DABR_READ_SHIFT);
+	wp_addr |= (1UL << DABR_WRITE_SHIFT);
+	wp_addr |= (1UL << DABR_TRANSLATION_SHIFT);
+	for (len = 1; len <= sizeof(glvar); len <<= 1) {
+		ptrace_set_debugreg(child_pid, wp_addr);
+		ptrace(PTRACE_CONT, child_pid, NULL, 0);
+		check_success(child_pid, name, "RW", wp_addr, len);
+	}
+
+	ptrace_set_debugreg(child_pid, 0);
+	return 0;
+}
+
+static int test_set_debugreg_kernel_userspace(pid_t child_pid)
+{
+	unsigned long wp_addr = (unsigned long)cwd;
+	char *name = "PTRACE_SET_DEBUGREG";
+
+	/* PTRACE_SET_DEBUGREG, Kernel Access Userspace test */
+	wp_addr &= ~0x7UL;
+	wp_addr |= (1Ul << DABR_READ_SHIFT);
+	wp_addr |= (1UL << DABR_WRITE_SHIFT);
+	wp_addr |= (1UL << DABR_TRANSLATION_SHIFT);
+	ptrace_set_debugreg(child_pid, wp_addr);
+	ptrace(PTRACE_CONT, child_pid, NULL, 0);
+	check_success(child_pid, name, "Kernel Access Userspace", wp_addr, 8);
+
+	ptrace_set_debugreg(child_pid, 0);
+	return 0;
+}
+
+static void get_ppc_hw_breakpoint(struct ppc_hw_breakpoint *info, int type,
+				  unsigned long addr, int len)
+{
+	info->version = 1;
+	info->trigger_type = type;
+	info->condition_mode = PPC_BREAKPOINT_CONDITION_NONE;
+	info->addr = (__u64)addr;
+	info->addr2 = (__u64)addr + len;
+	info->condition_value = 0;
+	if (!len)
+		info->addr_mode = PPC_BREAKPOINT_MODE_EXACT;
+	else
+		info->addr_mode = PPC_BREAKPOINT_MODE_RANGE_INCLUSIVE;
+}
+
+static void test_sethwdebug_exact(pid_t child_pid)
+{
+	struct ppc_hw_breakpoint info;
+	unsigned long wp_addr = (unsigned long)&glvar;
+	char *name = "PPC_PTRACE_SETHWDEBUG, MODE_EXACT";
+	int len = 1; /* hardcoded in kernel */
+	int wh;
+
+	/* PPC_PTRACE_SETHWDEBUG, MODE_EXACT, WO test */
+	get_ppc_hw_breakpoint(&info, PPC_BREAKPOINT_TRIGGER_WRITE, wp_addr, 0);
+	wh = ptrace_sethwdebug(child_pid, &info);
+	ptrace(PTRACE_CONT, child_pid, NULL, 0);
+	check_success(child_pid, name, "WO", wp_addr, len);
+	ptrace_delhwdebug(child_pid, wh);
+
+	/* PPC_PTRACE_SETHWDEBUG, MODE_EXACT, RO test */
+	get_ppc_hw_breakpoint(&info, PPC_BREAKPOINT_TRIGGER_READ, wp_addr, 0);
+	wh = ptrace_sethwdebug(child_pid, &info);
+	ptrace(PTRACE_CONT, child_pid, NULL, 0);
+	check_success(child_pid, name, "RO", wp_addr, len);
+	ptrace_delhwdebug(child_pid, wh);
+
+	/* PPC_PTRACE_SETHWDEBUG, MODE_EXACT, RW test */
+	get_ppc_hw_breakpoint(&info, PPC_BREAKPOINT_TRIGGER_RW, wp_addr, 0);
+	wh = ptrace_sethwdebug(child_pid, &info);
+	ptrace(PTRACE_CONT, child_pid, NULL, 0);
+	check_success(child_pid, name, "RW", wp_addr, len);
+	ptrace_delhwdebug(child_pid, wh);
+}
+
+static void test_sethwdebug_exact_kernel_userspace(pid_t child_pid)
+{
+	struct ppc_hw_breakpoint info;
+	unsigned long wp_addr = (unsigned long)&cwd;
+	char *name = "PPC_PTRACE_SETHWDEBUG, MODE_EXACT";
+	int len = 1; /* hardcoded in kernel */
+	int wh;
+
+	/* PPC_PTRACE_SETHWDEBUG, MODE_EXACT, Kernel Access Userspace test */
+	get_ppc_hw_breakpoint(&info, PPC_BREAKPOINT_TRIGGER_WRITE, wp_addr, 0);
+	wh = ptrace_sethwdebug(child_pid, &info);
+	ptrace(PTRACE_CONT, child_pid, NULL, 0);
+	check_success(child_pid, name, "Kernel Access Userspace", wp_addr, len);
+	ptrace_delhwdebug(child_pid, wh);
+}
+
+static void test_sethwdebug_range_aligned(pid_t child_pid)
+{
+	struct ppc_hw_breakpoint info;
+	unsigned long wp_addr;
+	char *name = "PPC_PTRACE_SETHWDEBUG, MODE_RANGE, DW ALIGNED";
+	int len;
+	int wh;
+
+	/* PPC_PTRACE_SETHWDEBUG, MODE_RANGE, DW ALIGNED, WO test */
+	wp_addr = (unsigned long)&gstruct.a;
+	len = A_LEN;
+	get_ppc_hw_breakpoint(&info, PPC_BREAKPOINT_TRIGGER_WRITE, wp_addr, len);
+	wh = ptrace_sethwdebug(child_pid, &info);
+	ptrace(PTRACE_CONT, child_pid, NULL, 0);
+	check_success(child_pid, name, "WO", wp_addr, len);
+	ptrace_delhwdebug(child_pid, wh);
+
+	/* PPC_PTRACE_SETHWDEBUG, MODE_RANGE, DW ALIGNED, RO test */
+	wp_addr = (unsigned long)&gstruct.a;
+	len = A_LEN;
+	get_ppc_hw_breakpoint(&info, PPC_BREAKPOINT_TRIGGER_READ, wp_addr, len);
+	wh = ptrace_sethwdebug(child_pid, &info);
+	ptrace(PTRACE_CONT, child_pid, NULL, 0);
+	check_success(child_pid, name, "RO", wp_addr, len);
+	ptrace_delhwdebug(child_pid, wh);
+
+	/* PPC_PTRACE_SETHWDEBUG, MODE_RANGE, DW ALIGNED, RW test */
+	wp_addr = (unsigned long)&gstruct.a;
+	len = A_LEN;
+	get_ppc_hw_breakpoint(&info, PPC_BREAKPOINT_TRIGGER_RW, wp_addr, len);
+	wh = ptrace_sethwdebug(child_pid, &info);
+	ptrace(PTRACE_CONT, child_pid, NULL, 0);
+	check_success(child_pid, name, "RW", wp_addr, len);
+	ptrace_delhwdebug(child_pid, wh);
+}
+
+static void test_sethwdebug_range_unaligned(pid_t child_pid)
+{
+	struct ppc_hw_breakpoint info;
+	unsigned long wp_addr;
+	char *name = "PPC_PTRACE_SETHWDEBUG, MODE_RANGE, DW UNALIGNED";
+	int len;
+	int wh;
+
+	/* PPC_PTRACE_SETHWDEBUG, MODE_RANGE, DW UNALIGNED, WO test */
+	wp_addr = (unsigned long)&gstruct.b;
+	len = B_LEN;
+	get_ppc_hw_breakpoint(&info, PPC_BREAKPOINT_TRIGGER_WRITE, wp_addr, len);
+	wh = ptrace_sethwdebug(child_pid, &info);
+	ptrace(PTRACE_CONT, child_pid, NULL, 0);
+	check_success(child_pid, name, "WO", wp_addr, len);
+	ptrace_delhwdebug(child_pid, wh);
+
+	/* PPC_PTRACE_SETHWDEBUG, MODE_RANGE, DW UNALIGNED, RO test */
+	wp_addr = (unsigned long)&gstruct.b;
+	len = B_LEN;
+	get_ppc_hw_breakpoint(&info, PPC_BREAKPOINT_TRIGGER_READ, wp_addr, len);
+	wh = ptrace_sethwdebug(child_pid, &info);
+	ptrace(PTRACE_CONT, child_pid, NULL, 0);
+	check_success(child_pid, name, "RO", wp_addr, len);
+	ptrace_delhwdebug(child_pid, wh);
+
+	/* PPC_PTRACE_SETHWDEBUG, MODE_RANGE, DW UNALIGNED, RW test */
+	wp_addr = (unsigned long)&gstruct.b;
+	len = B_LEN;
+	get_ppc_hw_breakpoint(&info, PPC_BREAKPOINT_TRIGGER_RW, wp_addr, len);
+	wh = ptrace_sethwdebug(child_pid, &info);
+	ptrace(PTRACE_CONT, child_pid, NULL, 0);
+	check_success(child_pid, name, "RW", wp_addr, len);
+	ptrace_delhwdebug(child_pid, wh);
+
+}
+
+static void test_sethwdebug_range_unaligned_dar(pid_t child_pid)
+{
+	struct ppc_hw_breakpoint info;
+	unsigned long wp_addr;
+	char *name = "PPC_PTRACE_SETHWDEBUG, MODE_RANGE, DW UNALIGNED, DAR OUTSIDE";
+	int len;
+	int wh;
+
+	/* PPC_PTRACE_SETHWDEBUG, MODE_RANGE, DW UNALIGNED, DAR OUTSIDE, RW test */
+	wp_addr = (unsigned long)&gstruct.b;
+	len = B_LEN;
+	get_ppc_hw_breakpoint(&info, PPC_BREAKPOINT_TRIGGER_WRITE, wp_addr, len);
+	wh = ptrace_sethwdebug(child_pid, &info);
+	ptrace(PTRACE_CONT, child_pid, NULL, 0);
+	check_success(child_pid, name, "RW", wp_addr, len);
+	ptrace_delhwdebug(child_pid, wh);
+}
+
+static void test_sethwdebug_dawr_max_range(pid_t child_pid)
+{
+	struct ppc_hw_breakpoint info;
+	unsigned long wp_addr;
+	char *name = "PPC_PTRACE_SETHWDEBUG, DAWR_MAX_LEN";
+	int len;
+	int wh;
+
+	/* PPC_PTRACE_SETHWDEBUG, DAWR_MAX_LEN, RW test */
+	wp_addr = (unsigned long)big_var;
+	len = DAWR_MAX_LEN;
+	get_ppc_hw_breakpoint(&info, PPC_BREAKPOINT_TRIGGER_RW, wp_addr, len);
+	wh = ptrace_sethwdebug(child_pid, &info);
+	ptrace(PTRACE_CONT, child_pid, NULL, 0);
+	check_success(child_pid, name, "RW", wp_addr, len);
+	ptrace_delhwdebug(child_pid, wh);
+}
+
+/* Set the breakpoints and check the child successfully trigger them */
+static void
+run_tests(pid_t child_pid, struct ppc_debug_info *dbginfo, bool dawr)
+{
+	test_set_debugreg(child_pid);
+	test_set_debugreg_kernel_userspace(child_pid);
+	test_sethwdebug_exact(child_pid);
+	test_sethwdebug_exact_kernel_userspace(child_pid);
+	if (dbginfo->features & PPC_DEBUG_FEATURE_DATA_BP_RANGE) {
+		test_sethwdebug_range_aligned(child_pid);
+		if (dawr || is_8xx) {
+			test_sethwdebug_range_unaligned(child_pid);
+			test_sethwdebug_range_unaligned_dar(child_pid);
+			test_sethwdebug_dawr_max_range(child_pid);
+		}
+	}
+}
+
+static int ptrace_hwbreak(void)
+{
+	pid_t child_pid;
+	struct ppc_debug_info dbginfo;
+	bool dawr;
+
+	child_pid = fork();
+	if (!child_pid) {
+		test_workload();
+		return 0;
+	}
+
+	wait(NULL);
+
+	get_dbginfo(child_pid, &dbginfo);
+	SKIP_IF(dbginfo.num_data_bps == 0);
+
+	dawr = dawr_present(&dbginfo);
+	run_tests(child_pid, &dbginfo, dawr);
+
+	/* Let the child exit first. */
+	ptrace(PTRACE_CONT, child_pid, NULL, 0);
+	wait(NULL);
+
+	/*
+	 * Testcases exits immediately with -1 on any failure. If
+	 * it has reached here, it means all tests were successful.
+	 */
+	return TEST_PASS;
+}
+
+int main(int argc, char **argv, char **envp)
+{
+	int pvr = 0;
+	asm __volatile__ ("mfspr %0,%1" : "=r"(pvr) : "i"(SPRN_PVR));
+	if (pvr == PVR_8xx)
+		is_8xx = true;
+
+	return test_harness(ptrace_hwbreak, "ptrace-hwbreak");
+}
diff --git a/tools/testing/selftests/powerpc/ptrace/ptrace-pkey.c b/tools/testing/selftests/powerpc/ptrace/ptrace-pkey.c
new file mode 100644
index 000000000..bc454f899
--- /dev/null
+++ b/tools/testing/selftests/powerpc/ptrace/ptrace-pkey.c
@@ -0,0 +1,330 @@
+// SPDX-License-Identifier: GPL-2.0+
+/*
+ * Ptrace test for Memory Protection Key registers
+ *
+ * Copyright (C) 2015 Anshuman Khandual, IBM Corporation.
+ * Copyright (C) 2018 IBM Corporation.
+ */
+#include "ptrace.h"
+#include "child.h"
+
+#ifndef __NR_pkey_alloc
+#define __NR_pkey_alloc		384
+#endif
+
+#ifndef __NR_pkey_free
+#define __NR_pkey_free		385
+#endif
+
+#ifndef NT_PPC_PKEY
+#define NT_PPC_PKEY		0x110
+#endif
+
+#ifndef PKEY_DISABLE_EXECUTE
+#define PKEY_DISABLE_EXECUTE	0x4
+#endif
+
+#define AMR_BITS_PER_PKEY 2
+#define PKEY_REG_BITS (sizeof(u64) * 8)
+#define pkeyshift(pkey) (PKEY_REG_BITS - ((pkey + 1) * AMR_BITS_PER_PKEY))
+
+static const char user_read[] = "[User Read (Running)]";
+static const char user_write[] = "[User Write (Running)]";
+static const char ptrace_read_running[] = "[Ptrace Read (Running)]";
+static const char ptrace_write_running[] = "[Ptrace Write (Running)]";
+
+/* Information shared between the parent and the child. */
+struct shared_info {
+	struct child_sync child_sync;
+
+	/* AMR value the parent expects to read from the child. */
+	unsigned long amr1;
+
+	/* AMR value the parent is expected to write to the child. */
+	unsigned long amr2;
+
+	/* AMR value that ptrace should refuse to write to the child. */
+	unsigned long invalid_amr;
+
+	/* IAMR value the parent expects to read from the child. */
+	unsigned long expected_iamr;
+
+	/* UAMOR value the parent expects to read from the child. */
+	unsigned long expected_uamor;
+
+	/*
+	 * IAMR and UAMOR values that ptrace should refuse to write to the child
+	 * (even though they're valid ones) because userspace doesn't have
+	 * access to those registers.
+	 */
+	unsigned long invalid_iamr;
+	unsigned long invalid_uamor;
+};
+
+static int sys_pkey_alloc(unsigned long flags, unsigned long init_access_rights)
+{
+	return syscall(__NR_pkey_alloc, flags, init_access_rights);
+}
+
+static int child(struct shared_info *info)
+{
+	unsigned long reg;
+	bool disable_execute = true;
+	int pkey1, pkey2, pkey3;
+	int ret;
+
+	/* Wait until parent fills out the initial register values. */
+	ret = wait_parent(&info->child_sync);
+	if (ret)
+		return ret;
+
+	/* Get some pkeys so that we can change their bits in the AMR. */
+	pkey1 = sys_pkey_alloc(0, PKEY_DISABLE_EXECUTE);
+	if (pkey1 < 0) {
+		pkey1 = sys_pkey_alloc(0, 0);
+		CHILD_FAIL_IF(pkey1 < 0, &info->child_sync);
+
+		disable_execute = false;
+	}
+
+	pkey2 = sys_pkey_alloc(0, 0);
+	CHILD_FAIL_IF(pkey2 < 0, &info->child_sync);
+
+	pkey3 = sys_pkey_alloc(0, 0);
+	CHILD_FAIL_IF(pkey3 < 0, &info->child_sync);
+
+	info->amr1 |= 3ul << pkeyshift(pkey1);
+	info->amr2 |= 3ul << pkeyshift(pkey2);
+	/*
+	 * invalid amr value where we try to force write
+	 * things which are deined by a uamor setting.
+	 */
+	info->invalid_amr = info->amr2 | (~0x0UL & ~info->expected_uamor);
+
+	/*
+	 * if PKEY_DISABLE_EXECUTE succeeded we should update the expected_iamr
+	 */
+	if (disable_execute)
+		info->expected_iamr |= 1ul << pkeyshift(pkey1);
+	else
+		info->expected_iamr &= ~(1ul << pkeyshift(pkey1));
+
+	/*
+	 * We allocated pkey2 and pkey 3 above. Clear the IAMR bits.
+	 */
+	info->expected_iamr &= ~(1ul << pkeyshift(pkey2));
+	info->expected_iamr &= ~(1ul << pkeyshift(pkey3));
+
+	/*
+	 * Create an IAMR value different from expected value.
+	 * Kernel will reject an IAMR and UAMOR change.
+	 */
+	info->invalid_iamr = info->expected_iamr | (1ul << pkeyshift(pkey1) | 1ul << pkeyshift(pkey2));
+	info->invalid_uamor = info->expected_uamor & ~(0x3ul << pkeyshift(pkey1));
+
+	printf("%-30s AMR: %016lx pkey1: %d pkey2: %d pkey3: %d\n",
+	       user_write, info->amr1, pkey1, pkey2, pkey3);
+
+	set_amr(info->amr1);
+
+	/* Wait for parent to read our AMR value and write a new one. */
+	ret = prod_parent(&info->child_sync);
+	CHILD_FAIL_IF(ret, &info->child_sync);
+
+	ret = wait_parent(&info->child_sync);
+	if (ret)
+		return ret;
+
+	reg = mfspr(SPRN_AMR);
+
+	printf("%-30s AMR: %016lx\n", user_read, reg);
+
+	CHILD_FAIL_IF(reg != info->amr2, &info->child_sync);
+
+	/*
+	 * Wait for parent to try to write an invalid AMR value.
+	 */
+	ret = prod_parent(&info->child_sync);
+	CHILD_FAIL_IF(ret, &info->child_sync);
+
+	ret = wait_parent(&info->child_sync);
+	if (ret)
+		return ret;
+
+	reg = mfspr(SPRN_AMR);
+
+	printf("%-30s AMR: %016lx\n", user_read, reg);
+
+	CHILD_FAIL_IF(reg != info->amr2, &info->child_sync);
+
+	/*
+	 * Wait for parent to try to write an IAMR and a UAMOR value. We can't
+	 * verify them, but we can verify that the AMR didn't change.
+	 */
+	ret = prod_parent(&info->child_sync);
+	CHILD_FAIL_IF(ret, &info->child_sync);
+
+	ret = wait_parent(&info->child_sync);
+	if (ret)
+		return ret;
+
+	reg = mfspr(SPRN_AMR);
+
+	printf("%-30s AMR: %016lx\n", user_read, reg);
+
+	CHILD_FAIL_IF(reg != info->amr2, &info->child_sync);
+
+	/* Now let parent now that we are finished. */
+
+	ret = prod_parent(&info->child_sync);
+	CHILD_FAIL_IF(ret, &info->child_sync);
+
+	return TEST_PASS;
+}
+
+static int parent(struct shared_info *info, pid_t pid)
+{
+	unsigned long regs[3];
+	int ret, status;
+
+	/*
+	 * Get the initial values for AMR, IAMR and UAMOR and communicate them
+	 * to the child.
+	 */
+	ret = ptrace_read_regs(pid, NT_PPC_PKEY, regs, 3);
+	PARENT_SKIP_IF_UNSUPPORTED(ret, &info->child_sync);
+	PARENT_FAIL_IF(ret, &info->child_sync);
+
+	info->amr1 = info->amr2 = regs[0];
+	info->expected_iamr = regs[1];
+	info->expected_uamor = regs[2];
+
+	/* Wake up child so that it can set itself up. */
+	ret = prod_child(&info->child_sync);
+	PARENT_FAIL_IF(ret, &info->child_sync);
+
+	ret = wait_child(&info->child_sync);
+	if (ret)
+		return ret;
+
+	/* Verify that we can read the pkey registers from the child. */
+	ret = ptrace_read_regs(pid, NT_PPC_PKEY, regs, 3);
+	PARENT_FAIL_IF(ret, &info->child_sync);
+
+	printf("%-30s AMR: %016lx IAMR: %016lx UAMOR: %016lx\n",
+	       ptrace_read_running, regs[0], regs[1], regs[2]);
+
+	PARENT_FAIL_IF(regs[0] != info->amr1, &info->child_sync);
+	PARENT_FAIL_IF(regs[1] != info->expected_iamr, &info->child_sync);
+	PARENT_FAIL_IF(regs[2] != info->expected_uamor, &info->child_sync);
+
+	/* Write valid AMR value in child. */
+	ret = ptrace_write_regs(pid, NT_PPC_PKEY, &info->amr2, 1);
+	PARENT_FAIL_IF(ret, &info->child_sync);
+
+	printf("%-30s AMR: %016lx\n", ptrace_write_running, info->amr2);
+
+	/* Wake up child so that it can verify it changed. */
+	ret = prod_child(&info->child_sync);
+	PARENT_FAIL_IF(ret, &info->child_sync);
+
+	ret = wait_child(&info->child_sync);
+	if (ret)
+		return ret;
+
+	/* Write invalid AMR value in child. */
+	ret = ptrace_write_regs(pid, NT_PPC_PKEY, &info->invalid_amr, 1);
+	PARENT_FAIL_IF(ret, &info->child_sync);
+
+	printf("%-30s AMR: %016lx\n", ptrace_write_running, info->invalid_amr);
+
+	/* Wake up child so that it can verify it didn't change. */
+	ret = prod_child(&info->child_sync);
+	PARENT_FAIL_IF(ret, &info->child_sync);
+
+	ret = wait_child(&info->child_sync);
+	if (ret)
+		return ret;
+
+	/* Try to write to IAMR. */
+	regs[0] = info->amr1;
+	regs[1] = info->invalid_iamr;
+	ret = ptrace_write_regs(pid, NT_PPC_PKEY, regs, 2);
+	PARENT_FAIL_IF(!ret, &info->child_sync);
+
+	printf("%-30s AMR: %016lx IAMR: %016lx\n",
+	       ptrace_write_running, regs[0], regs[1]);
+
+	/* Try to write to IAMR and UAMOR. */
+	regs[2] = info->invalid_uamor;
+	ret = ptrace_write_regs(pid, NT_PPC_PKEY, regs, 3);
+	PARENT_FAIL_IF(!ret, &info->child_sync);
+
+	printf("%-30s AMR: %016lx IAMR: %016lx UAMOR: %016lx\n",
+	       ptrace_write_running, regs[0], regs[1], regs[2]);
+
+	/* Verify that all registers still have their expected values. */
+	ret = ptrace_read_regs(pid, NT_PPC_PKEY, regs, 3);
+	PARENT_FAIL_IF(ret, &info->child_sync);
+
+	printf("%-30s AMR: %016lx IAMR: %016lx UAMOR: %016lx\n",
+	       ptrace_read_running, regs[0], regs[1], regs[2]);
+
+	PARENT_FAIL_IF(regs[0] != info->amr2, &info->child_sync);
+	PARENT_FAIL_IF(regs[1] != info->expected_iamr, &info->child_sync);
+	PARENT_FAIL_IF(regs[2] != info->expected_uamor, &info->child_sync);
+
+	/* Wake up child so that it can verify AMR didn't change and wrap up. */
+	ret = prod_child(&info->child_sync);
+	PARENT_FAIL_IF(ret, &info->child_sync);
+
+	ret = wait(&status);
+	if (ret != pid) {
+		printf("Child's exit status not captured\n");
+		ret = TEST_PASS;
+	} else if (!WIFEXITED(status)) {
+		printf("Child exited abnormally\n");
+		ret = TEST_FAIL;
+	} else
+		ret = WEXITSTATUS(status) ? TEST_FAIL : TEST_PASS;
+
+	return ret;
+}
+
+static int ptrace_pkey(void)
+{
+	struct shared_info *info;
+	int shm_id;
+	int ret;
+	pid_t pid;
+
+	shm_id = shmget(IPC_PRIVATE, sizeof(*info), 0777 | IPC_CREAT);
+	info = shmat(shm_id, NULL, 0);
+
+	ret = init_child_sync(&info->child_sync);
+	if (ret)
+		return ret;
+
+	pid = fork();
+	if (pid < 0) {
+		perror("fork() failed");
+		ret = TEST_FAIL;
+	} else if (pid == 0)
+		ret = child(info);
+	else
+		ret = parent(info, pid);
+
+	shmdt(info);
+
+	if (pid) {
+		destroy_child_sync(&info->child_sync);
+		shmctl(shm_id, IPC_RMID, NULL);
+	}
+
+	return ret;
+}
+
+int main(int argc, char *argv[])
+{
+	return test_harness(ptrace_pkey, "ptrace_pkey");
+}
diff --git a/tools/testing/selftests/powerpc/ptrace/ptrace-syscall.c b/tools/testing/selftests/powerpc/ptrace/ptrace-syscall.c
new file mode 100644
index 000000000..3353210dc
--- /dev/null
+++ b/tools/testing/selftests/powerpc/ptrace/ptrace-syscall.c
@@ -0,0 +1,228 @@
+// SPDX-License-Identifier: GPL-2.0
+/*
+ * A ptrace test for testing PTRACE_SYSEMU, PTRACE_SETREGS and
+ * PTRACE_GETREG.  This test basically create a child process that executes
+ * syscalls and the parent process check if it is being traced appropriated.
+ *
+ * This test is heavily based on tools/testing/selftests/x86/ptrace_syscall.c
+ * test, and it was adapted to run on Powerpc by
+ * Breno Leitao <leitao@debian.org>
+ */
+#define _GNU_SOURCE
+
+#include <sys/ptrace.h>
+#include <sys/types.h>
+#include <sys/wait.h>
+#include <sys/syscall.h>
+#include <sys/user.h>
+#include <unistd.h>
+#include <errno.h>
+#include <stddef.h>
+#include <stdio.h>
+#include <err.h>
+#include <string.h>
+#include <sys/auxv.h>
+#include "utils.h"
+
+/* Bitness-agnostic defines for user_regs_struct fields. */
+#define user_syscall_nr	gpr[0]
+#define user_arg0		gpr[3]
+#define user_arg1		gpr[4]
+#define user_arg2		gpr[5]
+#define user_arg3		gpr[6]
+#define user_arg4		gpr[7]
+#define user_arg5		gpr[8]
+#define user_ip		nip
+
+#define PTRACE_SYSEMU		0x1d
+
+static int nerrs;
+
+static void wait_trap(pid_t chld)
+{
+	siginfo_t si;
+
+	if (waitid(P_PID, chld, &si, WEXITED|WSTOPPED) != 0)
+		err(1, "waitid");
+	if (si.si_pid != chld)
+		errx(1, "got unexpected pid in event\n");
+	if (si.si_code != CLD_TRAPPED)
+		errx(1, "got unexpected event type %d\n", si.si_code);
+}
+
+static void test_ptrace_syscall_restart(void)
+{
+	int status;
+	struct pt_regs regs;
+	pid_t chld;
+
+	printf("[RUN]\tptrace-induced syscall restart\n");
+
+	chld = fork();
+	if (chld < 0)
+		err(1, "fork");
+
+	/*
+	 * Child process is running 4 syscalls after ptrace.
+	 *
+	 * 1) getpid()
+	 * 2) gettid()
+	 * 3) tgkill() -> Send SIGSTOP
+	 * 4) gettid() -> Where the tests will happen essentially
+	 */
+	if (chld == 0) {
+		if (ptrace(PTRACE_TRACEME, 0, 0, 0) != 0)
+			err(1, "PTRACE_TRACEME");
+
+		pid_t pid = getpid(), tid = syscall(SYS_gettid);
+
+		printf("\tChild will make one syscall\n");
+		syscall(SYS_tgkill, pid, tid, SIGSTOP);
+
+		syscall(SYS_gettid, 10, 11, 12, 13, 14, 15);
+		_exit(0);
+	}
+	/* Parent process below */
+
+	/* Wait for SIGSTOP sent by tgkill above. */
+	if (waitpid(chld, &status, 0) != chld || !WIFSTOPPED(status))
+		err(1, "waitpid");
+
+	printf("[RUN]\tSYSEMU\n");
+	if (ptrace(PTRACE_SYSEMU, chld, 0, 0) != 0)
+		err(1, "PTRACE_SYSEMU");
+	wait_trap(chld);
+
+	if (ptrace(PTRACE_GETREGS, chld, 0, &regs) != 0)
+		err(1, "PTRACE_GETREGS");
+
+	/*
+	 * Ptrace trapped prior to executing the syscall, thus r3 still has
+	 * the syscall number instead of the sys_gettid() result
+	 */
+	if (regs.user_syscall_nr != SYS_gettid ||
+	    regs.user_arg0 != 10 || regs.user_arg1 != 11 ||
+	    regs.user_arg2 != 12 || regs.user_arg3 != 13 ||
+	    regs.user_arg4 != 14 || regs.user_arg5 != 15) {
+		printf("[FAIL]\tInitial args are wrong (nr=%lu, args=%lu %lu %lu %lu %lu %lu)\n",
+			(unsigned long)regs.user_syscall_nr,
+			(unsigned long)regs.user_arg0,
+			(unsigned long)regs.user_arg1,
+			(unsigned long)regs.user_arg2,
+			(unsigned long)regs.user_arg3,
+			(unsigned long)regs.user_arg4,
+			(unsigned long)regs.user_arg5);
+		 nerrs++;
+	} else {
+		printf("[OK]\tInitial nr and args are correct\n"); }
+
+	printf("[RUN]\tRestart the syscall (ip = 0x%lx)\n",
+	       (unsigned long)regs.user_ip);
+
+	/*
+	 * Rewind to retry the same syscall again. This will basically test
+	 * the rewind process together with PTRACE_SETREGS and PTRACE_GETREGS.
+	 */
+	regs.user_ip -= 4;
+	if (ptrace(PTRACE_SETREGS, chld, 0, &regs) != 0)
+		err(1, "PTRACE_SETREGS");
+
+	if (ptrace(PTRACE_SYSEMU, chld, 0, 0) != 0)
+		err(1, "PTRACE_SYSEMU");
+	wait_trap(chld);
+
+	if (ptrace(PTRACE_GETREGS, chld, 0, &regs) != 0)
+		err(1, "PTRACE_GETREGS");
+
+	if (regs.user_syscall_nr != SYS_gettid ||
+	    regs.user_arg0 != 10 || regs.user_arg1 != 11 ||
+	    regs.user_arg2 != 12 || regs.user_arg3 != 13 ||
+	    regs.user_arg4 != 14 || regs.user_arg5 != 15) {
+		printf("[FAIL]\tRestart nr or args are wrong (nr=%lu, args=%lu %lu %lu %lu %lu %lu)\n",
+			(unsigned long)regs.user_syscall_nr,
+			(unsigned long)regs.user_arg0,
+			(unsigned long)regs.user_arg1,
+			(unsigned long)regs.user_arg2,
+			(unsigned long)regs.user_arg3,
+			(unsigned long)regs.user_arg4,
+			(unsigned long)regs.user_arg5);
+		nerrs++;
+	} else {
+		printf("[OK]\tRestarted nr and args are correct\n");
+	}
+
+	printf("[RUN]\tChange nr and args and restart the syscall (ip = 0x%lx)\n",
+	       (unsigned long)regs.user_ip);
+
+	/*
+	 * Inject a new syscall (getpid) in the same place the previous
+	 * syscall (gettid), rewind and re-execute.
+	 */
+	regs.user_syscall_nr = SYS_getpid;
+	regs.user_arg0 = 20;
+	regs.user_arg1 = 21;
+	regs.user_arg2 = 22;
+	regs.user_arg3 = 23;
+	regs.user_arg4 = 24;
+	regs.user_arg5 = 25;
+	regs.user_ip -= 4;
+
+	if (ptrace(PTRACE_SETREGS, chld, 0, &regs) != 0)
+		err(1, "PTRACE_SETREGS");
+
+	if (ptrace(PTRACE_SYSEMU, chld, 0, 0) != 0)
+		err(1, "PTRACE_SYSEMU");
+	wait_trap(chld);
+
+	if (ptrace(PTRACE_GETREGS, chld, 0, &regs) != 0)
+		err(1, "PTRACE_GETREGS");
+
+	/* Check that ptrace stopped at the new syscall that was
+	 * injected, and guarantee that it haven't executed, i.e, user_args
+	 * contain the arguments and not the syscall return value, for
+	 * instance.
+	 */
+	if (regs.user_syscall_nr != SYS_getpid
+		|| regs.user_arg0 != 20 || regs.user_arg1 != 21
+		|| regs.user_arg2 != 22 || regs.user_arg3 != 23
+		|| regs.user_arg4 != 24 || regs.user_arg5 != 25) {
+
+		printf("[FAIL]\tRestart nr or args are wrong (nr=%lu, args=%lu %lu %lu %lu %lu %lu)\n",
+			(unsigned long)regs.user_syscall_nr,
+			(unsigned long)regs.user_arg0,
+			(unsigned long)regs.user_arg1,
+			(unsigned long)regs.user_arg2,
+			(unsigned long)regs.user_arg3,
+			(unsigned long)regs.user_arg4,
+			(unsigned long)regs.user_arg5);
+		nerrs++;
+	} else {
+		printf("[OK]\tReplacement nr and args are correct\n");
+	}
+
+	if (ptrace(PTRACE_CONT, chld, 0, 0) != 0)
+		err(1, "PTRACE_CONT");
+
+	if (waitpid(chld, &status, 0) != chld)
+		err(1, "waitpid");
+
+	/* Guarantee that the process executed properly, returning 0 */
+	if (!WIFEXITED(status) || WEXITSTATUS(status) != 0) {
+		printf("[FAIL]\tChild failed\n");
+		nerrs++;
+	} else {
+		printf("[OK]\tChild exited cleanly\n");
+	}
+}
+
+int ptrace_syscall(void)
+{
+	test_ptrace_syscall_restart();
+
+	return nerrs;
+}
+
+int main(void)
+{
+	return test_harness(ptrace_syscall, "ptrace_syscall");
+}
diff --git a/tools/testing/selftests/powerpc/ptrace/ptrace-tar.c b/tools/testing/selftests/powerpc/ptrace/ptrace-tar.c
new file mode 100644
index 000000000..4436ca9d3
--- /dev/null
+++ b/tools/testing/selftests/powerpc/ptrace/ptrace-tar.c
@@ -0,0 +1,134 @@
+// SPDX-License-Identifier: GPL-2.0-or-later
+/*
+ * Ptrace test for TAR, PPR, DSCR registers
+ *
+ * Copyright (C) 2015 Anshuman Khandual, IBM Corporation.
+ */
+#include "ptrace.h"
+#include "ptrace-tar.h"
+
+/* Tracer and Tracee Shared Data */
+int shm_id;
+int *cptr;
+int *pptr;
+
+void tar(void)
+{
+	unsigned long reg[3];
+	int ret;
+
+	cptr = (int *)shmat(shm_id, NULL, 0);
+	printf("%-30s TAR: %u PPR: %lx DSCR: %u\n",
+			user_write, TAR_1, PPR_1, DSCR_1);
+
+	mtspr(SPRN_TAR, TAR_1);
+	mtspr(SPRN_PPR, PPR_1);
+	mtspr(SPRN_DSCR, DSCR_1);
+
+	cptr[2] = 1;
+
+	/* Wait on parent */
+	while (!cptr[0])
+		asm volatile("" : : : "memory");
+
+	reg[0] = mfspr(SPRN_TAR);
+	reg[1] = mfspr(SPRN_PPR);
+	reg[2] = mfspr(SPRN_DSCR);
+
+	printf("%-30s TAR: %lu PPR: %lx DSCR: %lu\n",
+			user_read, reg[0], reg[1], reg[2]);
+
+	/* Unblock the parent now */
+	cptr[1] = 1;
+	shmdt((int *)cptr);
+
+	ret = validate_tar_registers(reg, TAR_2, PPR_2, DSCR_2);
+	if (ret)
+		exit(1);
+	exit(0);
+}
+
+int trace_tar(pid_t child)
+{
+	unsigned long reg[3];
+
+	FAIL_IF(start_trace(child));
+	FAIL_IF(show_tar_registers(child, reg));
+	printf("%-30s TAR: %lu PPR: %lx DSCR: %lu\n",
+			ptrace_read_running, reg[0], reg[1], reg[2]);
+
+	FAIL_IF(validate_tar_registers(reg, TAR_1, PPR_1, DSCR_1));
+	FAIL_IF(stop_trace(child));
+	return TEST_PASS;
+}
+
+int trace_tar_write(pid_t child)
+{
+	FAIL_IF(start_trace(child));
+	FAIL_IF(write_tar_registers(child, TAR_2, PPR_2, DSCR_2));
+	printf("%-30s TAR: %u PPR: %lx DSCR: %u\n",
+			ptrace_write_running, TAR_2, PPR_2, DSCR_2);
+
+	FAIL_IF(stop_trace(child));
+	return TEST_PASS;
+}
+
+int ptrace_tar(void)
+{
+	pid_t pid;
+	int ret, status;
+
+	// TAR was added in v2.07
+	SKIP_IF(!have_hwcap2(PPC_FEATURE2_ARCH_2_07));
+
+	shm_id = shmget(IPC_PRIVATE, sizeof(int) * 3, 0777|IPC_CREAT);
+	pid = fork();
+	if (pid < 0) {
+		perror("fork() failed");
+		return TEST_FAIL;
+	}
+
+	if (pid == 0)
+		tar();
+
+	if (pid) {
+		pptr = (int *)shmat(shm_id, NULL, 0);
+		pptr[0] = 0;
+		pptr[1] = 0;
+
+		while (!pptr[2])
+			asm volatile("" : : : "memory");
+		ret = trace_tar(pid);
+		if (ret)
+			return ret;
+
+		ret = trace_tar_write(pid);
+		if (ret)
+			return ret;
+
+		/* Unblock the child now */
+		pptr[0] = 1;
+
+		/* Wait on child */
+		while (!pptr[1])
+			asm volatile("" : : : "memory");
+
+		shmdt((int *)pptr);
+
+		ret = wait(&status);
+		shmctl(shm_id, IPC_RMID, NULL);
+		if (ret != pid) {
+			printf("Child's exit status not captured\n");
+			return TEST_PASS;
+		}
+
+		return (WIFEXITED(status) && WEXITSTATUS(status)) ? TEST_FAIL :
+			TEST_PASS;
+	}
+	return TEST_PASS;
+}
+
+int main(int argc, char *argv[])
+{
+	return test_harness(ptrace_tar, "ptrace_tar");
+}
diff --git a/tools/testing/selftests/powerpc/ptrace/ptrace-tar.h b/tools/testing/selftests/powerpc/ptrace/ptrace-tar.h
new file mode 100644
index 000000000..d6a4c0aab
--- /dev/null
+++ b/tools/testing/selftests/powerpc/ptrace/ptrace-tar.h
@@ -0,0 +1,46 @@
+/* SPDX-License-Identifier: GPL-2.0-or-later */
+/*
+ * Copyright (C) 2015 Anshuman Khandual, IBM Corporation.
+ */
+#define TAR_1   10
+#define TAR_2   20
+#define TAR_3   30
+#define TAR_4   40
+#define TAR_5   50
+
+#define DSCR_1  100
+#define DSCR_2  200
+#define DSCR_3  300
+#define DSCR_4  400
+#define DSCR_5  500
+
+#define PPR_1   0x4000000000000         /* or 31,31,31*/
+#define PPR_2   0x8000000000000         /* or 1,1,1 */
+#define PPR_3   0xc000000000000         /* or 6,6,6 */
+#define PPR_4   0x10000000000000        /* or 2,2,2 */
+
+char *user_read = "[User Read (Running)]";
+char *user_write = "[User Write (Running)]";
+char *ptrace_read_running = "[Ptrace Read (Running)]";
+char *ptrace_write_running = "[Ptrace Write (Running)]";
+char *ptrace_read_ckpt = "[Ptrace Read (Checkpointed)]";
+char *ptrace_write_ckpt = "[Ptrace Write (Checkpointed)]";
+
+int validate_tar_registers(unsigned long *reg, unsigned long tar,
+				unsigned long ppr, unsigned long dscr)
+{
+	int match = 1;
+
+	if (reg[0] != tar)
+		match = 0;
+
+	if (reg[1] != ppr)
+		match = 0;
+
+	if (reg[2] != dscr)
+		match = 0;
+
+	if (!match)
+		return TEST_FAIL;
+	return TEST_PASS;
+}
diff --git a/tools/testing/selftests/powerpc/ptrace/ptrace-tm-gpr.c b/tools/testing/selftests/powerpc/ptrace/ptrace-tm-gpr.c
new file mode 100644
index 000000000..82f7bdc2e
--- /dev/null
+++ b/tools/testing/selftests/powerpc/ptrace/ptrace-tm-gpr.c
@@ -0,0 +1,154 @@
+// SPDX-License-Identifier: GPL-2.0-or-later
+/*
+ * Ptrace test for GPR/FPR registers in TM context
+ *
+ * Copyright (C) 2015 Anshuman Khandual, IBM Corporation.
+ */
+#include "ptrace.h"
+#include "ptrace-gpr.h"
+#include "tm.h"
+
+/* Tracer and Tracee Shared Data */
+int shm_id;
+unsigned long *cptr, *pptr;
+
+float a = FPR_1;
+float b = FPR_2;
+float c = FPR_3;
+
+void tm_gpr(void)
+{
+	unsigned long gpr_buf[18];
+	unsigned long result, texasr;
+	float fpr_buf[32];
+
+	printf("Starting the child\n");
+	cptr = (unsigned long *)shmat(shm_id, NULL, 0);
+
+trans:
+	cptr[1] = 0;
+	asm __volatile__(
+		ASM_LOAD_GPR_IMMED(gpr_1)
+		ASM_LOAD_FPR_SINGLE_PRECISION(flt_1)
+		"1: ;"
+		"tbegin.;"
+		"beq 2f;"
+		ASM_LOAD_GPR_IMMED(gpr_2)
+		ASM_LOAD_FPR_SINGLE_PRECISION(flt_2)
+		"tsuspend.;"
+		"li 7, 1;"
+		"stw 7, 0(%[cptr1]);"
+		"tresume.;"
+		"b .;"
+
+		"tend.;"
+		"li 0, 0;"
+		"ori %[res], 0, 0;"
+		"b 3f;"
+
+		/* Transaction abort handler */
+		"2: ;"
+		"li 0, 1;"
+		"ori %[res], 0, 0;"
+		"mfspr %[texasr], %[sprn_texasr];"
+
+		"3: ;"
+		: [res] "=r" (result), [texasr] "=r" (texasr)
+		: [gpr_1]"i"(GPR_1), [gpr_2]"i"(GPR_2),
+		[sprn_texasr] "i" (SPRN_TEXASR), [flt_1] "b" (&a),
+		[flt_2] "b" (&b), [cptr1] "b" (&cptr[1])
+		: "memory", "r7", "r8", "r9", "r10",
+		"r11", "r12", "r13", "r14", "r15", "r16",
+		"r17", "r18", "r19", "r20", "r21", "r22",
+		"r23", "r24", "r25", "r26", "r27", "r28",
+		"r29", "r30", "r31"
+		);
+
+	if (result) {
+		if (!cptr[0])
+			goto trans;
+
+		shmdt((void *)cptr);
+		store_gpr(gpr_buf);
+		store_fpr_single_precision(fpr_buf);
+
+		if (validate_gpr(gpr_buf, GPR_3))
+			exit(1);
+
+		if (validate_fpr_float(fpr_buf, c))
+			exit(1);
+
+		exit(0);
+	}
+	shmdt((void *)cptr);
+	exit(1);
+}
+
+int trace_tm_gpr(pid_t child)
+{
+	unsigned long gpr[18];
+	unsigned long fpr[32];
+
+	FAIL_IF(start_trace(child));
+	FAIL_IF(show_gpr(child, gpr));
+	FAIL_IF(validate_gpr(gpr, GPR_2));
+	FAIL_IF(show_fpr(child, fpr));
+	FAIL_IF(validate_fpr(fpr, FPR_2_REP));
+	FAIL_IF(show_ckpt_fpr(child, fpr));
+	FAIL_IF(validate_fpr(fpr, FPR_1_REP));
+	FAIL_IF(show_ckpt_gpr(child, gpr));
+	FAIL_IF(validate_gpr(gpr, GPR_1));
+	FAIL_IF(write_ckpt_gpr(child, GPR_3));
+	FAIL_IF(write_ckpt_fpr(child, FPR_3_REP));
+
+	pptr[0] = 1;
+	FAIL_IF(stop_trace(child));
+
+	return TEST_PASS;
+}
+
+int ptrace_tm_gpr(void)
+{
+	pid_t pid;
+	int ret, status;
+
+	SKIP_IF(!have_htm());
+	shm_id = shmget(IPC_PRIVATE, sizeof(int) * 2, 0777|IPC_CREAT);
+	pid = fork();
+	if (pid < 0) {
+		perror("fork() failed");
+		return TEST_FAIL;
+	}
+	if (pid == 0)
+		tm_gpr();
+
+	if (pid) {
+		pptr = (unsigned long *)shmat(shm_id, NULL, 0);
+
+		while (!pptr[1])
+			asm volatile("" : : : "memory");
+		ret = trace_tm_gpr(pid);
+		if (ret) {
+			kill(pid, SIGTERM);
+			return TEST_FAIL;
+		}
+
+		shmdt((void *)pptr);
+
+		ret = wait(&status);
+		shmctl(shm_id, IPC_RMID, NULL);
+		if (ret != pid) {
+			printf("Child's exit status not captured\n");
+			return TEST_FAIL;
+		}
+
+		return (WIFEXITED(status) && WEXITSTATUS(status)) ? TEST_FAIL :
+			TEST_PASS;
+	}
+	return TEST_PASS;
+}
+
+int main(int argc, char *argv[])
+{
+	return test_harness(ptrace_tm_gpr, "ptrace_tm_gpr");
+}
diff --git a/tools/testing/selftests/powerpc/ptrace/ptrace-tm-spd-gpr.c b/tools/testing/selftests/powerpc/ptrace/ptrace-tm-spd-gpr.c
new file mode 100644
index 000000000..ad65be6e8
--- /dev/null
+++ b/tools/testing/selftests/powerpc/ptrace/ptrace-tm-spd-gpr.c
@@ -0,0 +1,165 @@
+// SPDX-License-Identifier: GPL-2.0-or-later
+/*
+ * Ptrace test for GPR/FPR registers in TM Suspend context
+ *
+ * Copyright (C) 2015 Anshuman Khandual, IBM Corporation.
+ */
+#include "ptrace.h"
+#include "ptrace-gpr.h"
+#include "tm.h"
+
+/* Tracer and Tracee Shared Data */
+int shm_id;
+int *cptr, *pptr;
+
+float a = FPR_1;
+float b = FPR_2;
+float c = FPR_3;
+float d = FPR_4;
+
+__attribute__((used)) void wait_parent(void)
+{
+	cptr[2] = 1;
+	while (!cptr[1])
+		asm volatile("" : : : "memory");
+}
+
+void tm_spd_gpr(void)
+{
+	unsigned long gpr_buf[18];
+	unsigned long result, texasr;
+	float fpr_buf[32];
+
+	cptr = (int *)shmat(shm_id, NULL, 0);
+
+trans:
+	cptr[2] = 0;
+	asm __volatile__(
+		ASM_LOAD_GPR_IMMED(gpr_1)
+		ASM_LOAD_FPR_SINGLE_PRECISION(flt_1)
+
+		"1: ;"
+		"tbegin.;"
+		"beq 2f;"
+
+		ASM_LOAD_GPR_IMMED(gpr_2)
+		"tsuspend.;"
+		ASM_LOAD_GPR_IMMED(gpr_4)
+		ASM_LOAD_FPR_SINGLE_PRECISION(flt_4)
+
+		"bl wait_parent;"
+		"tresume.;"
+		"tend.;"
+		"li 0, 0;"
+		"ori %[res], 0, 0;"
+		"b 3f;"
+
+		/* Transaction abort handler */
+		"2: ;"
+		"li 0, 1;"
+		"ori %[res], 0, 0;"
+		"mfspr %[texasr], %[sprn_texasr];"
+
+		"3: ;"
+		: [res] "=r" (result), [texasr] "=r" (texasr)
+		: [gpr_1]"i"(GPR_1), [gpr_2]"i"(GPR_2), [gpr_4]"i"(GPR_4),
+		[sprn_texasr] "i" (SPRN_TEXASR), [flt_1] "b" (&a),
+		[flt_4] "b" (&d)
+		: "memory", "r5", "r6", "r7",
+		"r8", "r9", "r10", "r11", "r12", "r13", "r14", "r15",
+		"r16", "r17", "r18", "r19", "r20", "r21", "r22", "r23",
+		"r24", "r25", "r26", "r27", "r28", "r29", "r30", "r31"
+		);
+
+	if (result) {
+		if (!cptr[0])
+			goto trans;
+
+		shmdt((void *)cptr);
+		store_gpr(gpr_buf);
+		store_fpr_single_precision(fpr_buf);
+
+		if (validate_gpr(gpr_buf, GPR_3))
+			exit(1);
+
+		if (validate_fpr_float(fpr_buf, c))
+			exit(1);
+		exit(0);
+	}
+	shmdt((void *)cptr);
+	exit(1);
+}
+
+int trace_tm_spd_gpr(pid_t child)
+{
+	unsigned long gpr[18];
+	unsigned long fpr[32];
+
+	FAIL_IF(start_trace(child));
+	FAIL_IF(show_gpr(child, gpr));
+	FAIL_IF(validate_gpr(gpr, GPR_4));
+	FAIL_IF(show_fpr(child, fpr));
+	FAIL_IF(validate_fpr(fpr, FPR_4_REP));
+	FAIL_IF(show_ckpt_fpr(child, fpr));
+	FAIL_IF(validate_fpr(fpr, FPR_1_REP));
+	FAIL_IF(show_ckpt_gpr(child, gpr));
+	FAIL_IF(validate_gpr(gpr, GPR_1));
+	FAIL_IF(write_ckpt_gpr(child, GPR_3));
+	FAIL_IF(write_ckpt_fpr(child, FPR_3_REP));
+
+	pptr[0] = 1;
+	pptr[1] = 1;
+	FAIL_IF(stop_trace(child));
+	return TEST_PASS;
+}
+
+int ptrace_tm_spd_gpr(void)
+{
+	pid_t pid;
+	int ret, status;
+
+	SKIP_IF(!have_htm());
+	shm_id = shmget(IPC_PRIVATE, sizeof(int) * 3, 0777|IPC_CREAT);
+	pid = fork();
+	if (pid < 0) {
+		perror("fork() failed");
+		return TEST_FAIL;
+	}
+
+	if (pid == 0)
+		tm_spd_gpr();
+
+	if (pid) {
+		pptr = (int *)shmat(shm_id, NULL, 0);
+		pptr[0] = 0;
+		pptr[1] = 0;
+
+		while (!pptr[2])
+			asm volatile("" : : : "memory");
+		ret = trace_tm_spd_gpr(pid);
+		if (ret) {
+			kill(pid, SIGTERM);
+			shmdt((void *)pptr);
+			shmctl(shm_id, IPC_RMID, NULL);
+			return TEST_FAIL;
+		}
+
+		shmdt((void *)pptr);
+
+		ret = wait(&status);
+		shmctl(shm_id, IPC_RMID, NULL);
+		if (ret != pid) {
+			printf("Child's exit status not captured\n");
+			return TEST_FAIL;
+		}
+
+		return (WIFEXITED(status) && WEXITSTATUS(status)) ? TEST_FAIL :
+			TEST_PASS;
+	}
+	return TEST_PASS;
+}
+
+int main(int argc, char *argv[])
+{
+	return test_harness(ptrace_tm_spd_gpr, "ptrace_tm_spd_gpr");
+}
diff --git a/tools/testing/selftests/powerpc/ptrace/ptrace-tm-spd-tar.c b/tools/testing/selftests/powerpc/ptrace/ptrace-tm-spd-tar.c
new file mode 100644
index 000000000..2ecfa1158
--- /dev/null
+++ b/tools/testing/selftests/powerpc/ptrace/ptrace-tm-spd-tar.c
@@ -0,0 +1,170 @@
+// SPDX-License-Identifier: GPL-2.0-or-later
+/*
+ * Ptrace test for TAR, PPR, DSCR registers in the TM Suspend context
+ *
+ * Copyright (C) 2015 Anshuman Khandual, IBM Corporation.
+ */
+#include "ptrace.h"
+#include "tm.h"
+#include "ptrace-tar.h"
+
+int shm_id;
+int *cptr, *pptr;
+
+__attribute__((used)) void wait_parent(void)
+{
+	cptr[2] = 1;
+	while (!cptr[1])
+		asm volatile("" : : : "memory");
+}
+
+void tm_spd_tar(void)
+{
+	unsigned long result, texasr;
+	unsigned long regs[3];
+	int ret;
+
+	cptr = (int *)shmat(shm_id, NULL, 0);
+
+trans:
+	cptr[2] = 0;
+	asm __volatile__(
+		"li	4, %[tar_1];"
+		"mtspr %[sprn_tar],  4;"	/* TAR_1 */
+		"li	4, %[dscr_1];"
+		"mtspr %[sprn_dscr], 4;"	/* DSCR_1 */
+		"or     31,31,31;"		/* PPR_1*/
+
+		"1: ;"
+		"tbegin.;"
+		"beq 2f;"
+
+		"li	4, %[tar_2];"
+		"mtspr %[sprn_tar],  4;"	/* TAR_2 */
+		"li	4, %[dscr_2];"
+		"mtspr %[sprn_dscr], 4;"	/* DSCR_2 */
+		"or     1,1,1;"			/* PPR_2 */
+
+		"tsuspend.;"
+		"li	4, %[tar_3];"
+		"mtspr %[sprn_tar],  4;"	/* TAR_3 */
+		"li	4, %[dscr_3];"
+		"mtspr %[sprn_dscr], 4;"	/* DSCR_3 */
+		"or     6,6,6;"			/* PPR_3 */
+		"bl wait_parent;"
+		"tresume.;"
+
+		"tend.;"
+		"li 0, 0;"
+		"ori %[res], 0, 0;"
+		"b 3f;"
+
+		/* Transaction abort handler */
+		"2: ;"
+		"li 0, 1;"
+		"ori %[res], 0, 0;"
+		"mfspr %[texasr], %[sprn_texasr];"
+
+		"3: ;"
+
+		: [res] "=r" (result), [texasr] "=r" (texasr)
+		: [sprn_dscr]"i"(SPRN_DSCR),
+		[sprn_tar]"i"(SPRN_TAR), [sprn_ppr]"i"(SPRN_PPR),
+		[sprn_texasr]"i"(SPRN_TEXASR), [tar_1]"i"(TAR_1),
+		[dscr_1]"i"(DSCR_1), [tar_2]"i"(TAR_2), [dscr_2]"i"(DSCR_2),
+		[tar_3]"i"(TAR_3), [dscr_3]"i"(DSCR_3)
+		: "memory", "r0", "r3", "r4", "r5", "r6", "lr"
+		);
+
+	/* TM failed, analyse */
+	if (result) {
+		if (!cptr[0])
+			goto trans;
+
+		regs[0] = mfspr(SPRN_TAR);
+		regs[1] = mfspr(SPRN_PPR);
+		regs[2] = mfspr(SPRN_DSCR);
+
+		shmdt(&cptr);
+		printf("%-30s TAR: %lu PPR: %lx DSCR: %lu\n",
+				user_read, regs[0], regs[1], regs[2]);
+
+		ret = validate_tar_registers(regs, TAR_4, PPR_4, DSCR_4);
+		if (ret)
+			exit(1);
+		exit(0);
+	}
+	shmdt(&cptr);
+	exit(1);
+}
+
+int trace_tm_spd_tar(pid_t child)
+{
+	unsigned long regs[3];
+
+	FAIL_IF(start_trace(child));
+	FAIL_IF(show_tar_registers(child, regs));
+	printf("%-30s TAR: %lu PPR: %lx DSCR: %lu\n",
+			ptrace_read_running, regs[0], regs[1], regs[2]);
+
+	FAIL_IF(validate_tar_registers(regs, TAR_3, PPR_3, DSCR_3));
+	FAIL_IF(show_tm_checkpointed_state(child, regs));
+	printf("%-30s TAR: %lu PPR: %lx DSCR: %lu\n",
+			ptrace_read_ckpt, regs[0], regs[1], regs[2]);
+
+	FAIL_IF(validate_tar_registers(regs, TAR_1, PPR_1, DSCR_1));
+	FAIL_IF(write_ckpt_tar_registers(child, TAR_4, PPR_4, DSCR_4));
+	printf("%-30s TAR: %u PPR: %lx DSCR: %u\n",
+			ptrace_write_ckpt, TAR_4, PPR_4, DSCR_4);
+
+	pptr[0] = 1;
+	pptr[1] = 1;
+	FAIL_IF(stop_trace(child));
+	return TEST_PASS;
+}
+
+int ptrace_tm_spd_tar(void)
+{
+	pid_t pid;
+	int ret, status;
+
+	SKIP_IF(!have_htm());
+	shm_id = shmget(IPC_PRIVATE, sizeof(int) * 3, 0777|IPC_CREAT);
+	pid = fork();
+	if (pid == 0)
+		tm_spd_tar();
+
+	pptr = (int *)shmat(shm_id, NULL, 0);
+	pptr[0] = 0;
+	pptr[1] = 0;
+
+	if (pid) {
+		while (!pptr[2])
+			asm volatile("" : : : "memory");
+		ret = trace_tm_spd_tar(pid);
+		if (ret) {
+			kill(pid, SIGTERM);
+			shmdt(&pptr);
+			shmctl(shm_id, IPC_RMID, NULL);
+			return TEST_FAIL;
+		}
+
+		shmdt(&pptr);
+
+		ret = wait(&status);
+		shmctl(shm_id, IPC_RMID, NULL);
+		if (ret != pid) {
+			printf("Child's exit status not captured\n");
+			return TEST_FAIL;
+		}
+
+		return (WIFEXITED(status) && WEXITSTATUS(status)) ? TEST_FAIL :
+			TEST_PASS;
+	}
+	return TEST_PASS;
+}
+
+int main(int argc, char *argv[])
+{
+	return test_harness(ptrace_tm_spd_tar, "ptrace_tm_spd_tar");
+}
diff --git a/tools/testing/selftests/powerpc/ptrace/ptrace-tm-spd-vsx.c b/tools/testing/selftests/powerpc/ptrace/ptrace-tm-spd-vsx.c
new file mode 100644
index 000000000..6f7fb51f0
--- /dev/null
+++ b/tools/testing/selftests/powerpc/ptrace/ptrace-tm-spd-vsx.c
@@ -0,0 +1,180 @@
+// SPDX-License-Identifier: GPL-2.0-or-later
+/*
+ * Ptrace test for VMX/VSX registers in the TM Suspend context
+ *
+ * Copyright (C) 2015 Anshuman Khandual, IBM Corporation.
+ */
+#include "ptrace.h"
+#include "tm.h"
+#include "ptrace-vsx.h"
+
+int shm_id;
+int *cptr, *pptr;
+
+unsigned long fp_load[VEC_MAX];
+unsigned long fp_load_new[VEC_MAX];
+unsigned long fp_store[VEC_MAX];
+unsigned long fp_load_ckpt[VEC_MAX];
+unsigned long fp_load_ckpt_new[VEC_MAX];
+
+__attribute__((used)) void load_vsx(void)
+{
+	loadvsx(fp_load, 0);
+}
+
+__attribute__((used)) void load_vsx_new(void)
+{
+	loadvsx(fp_load_new, 0);
+}
+
+__attribute__((used)) void load_vsx_ckpt(void)
+{
+	loadvsx(fp_load_ckpt, 0);
+}
+
+__attribute__((used)) void wait_parent(void)
+{
+	cptr[2] = 1;
+	while (!cptr[1])
+		asm volatile("" : : : "memory");
+}
+
+void tm_spd_vsx(void)
+{
+	unsigned long result, texasr;
+	int ret;
+
+	cptr = (int *)shmat(shm_id, NULL, 0);
+
+trans:
+	cptr[2] = 0;
+	asm __volatile__(
+		"bl load_vsx_ckpt;"
+
+		"1: ;"
+		"tbegin.;"
+		"beq 2f;"
+
+		"bl load_vsx_new;"
+		"tsuspend.;"
+		"bl load_vsx;"
+		"bl wait_parent;"
+		"tresume.;"
+
+		"tend.;"
+		"li 0, 0;"
+		"ori %[res], 0, 0;"
+		"b 3f;"
+
+		"2: ;"
+		"li 0, 1;"
+		"ori %[res], 0, 0;"
+		"mfspr %[texasr], %[sprn_texasr];"
+
+		"3: ;"
+		: [res] "=r" (result), [texasr] "=r" (texasr)
+		: [sprn_texasr] "i"  (SPRN_TEXASR)
+		: "memory", "r0", "r3", "r4",
+		  "r7", "r8", "r9", "r10", "r11", "lr"
+		);
+
+	if (result) {
+		if (!cptr[0])
+			goto trans;
+		shmdt((void *)cptr);
+
+		storevsx(fp_store, 0);
+		ret = compare_vsx_vmx(fp_store, fp_load_ckpt_new);
+		if (ret)
+			exit(1);
+		exit(0);
+	}
+	shmdt((void *)cptr);
+	exit(1);
+}
+
+int trace_tm_spd_vsx(pid_t child)
+{
+	unsigned long vsx[VSX_MAX];
+	unsigned long vmx[VMX_MAX + 2][2];
+
+	FAIL_IF(start_trace(child));
+	FAIL_IF(show_vsx(child, vsx));
+	FAIL_IF(validate_vsx(vsx, fp_load));
+	FAIL_IF(show_vmx(child, vmx));
+	FAIL_IF(validate_vmx(vmx, fp_load));
+	FAIL_IF(show_vsx_ckpt(child, vsx));
+	FAIL_IF(validate_vsx(vsx, fp_load_ckpt));
+	FAIL_IF(show_vmx_ckpt(child, vmx));
+	FAIL_IF(validate_vmx(vmx, fp_load_ckpt));
+
+	memset(vsx, 0, sizeof(vsx));
+	memset(vmx, 0, sizeof(vmx));
+
+	load_vsx_vmx(fp_load_ckpt_new, vsx, vmx);
+
+	FAIL_IF(write_vsx_ckpt(child, vsx));
+	FAIL_IF(write_vmx_ckpt(child, vmx));
+
+	pptr[0] = 1;
+	pptr[1] = 1;
+	FAIL_IF(stop_trace(child));
+
+	return TEST_PASS;
+}
+
+int ptrace_tm_spd_vsx(void)
+{
+	pid_t pid;
+	int ret, status, i;
+
+	SKIP_IF(!have_htm());
+	shm_id = shmget(IPC_PRIVATE, sizeof(int) * 3, 0777|IPC_CREAT);
+
+	for (i = 0; i < 128; i++) {
+		fp_load[i] = 1 + rand();
+		fp_load_new[i] = 1 + 2 * rand();
+		fp_load_ckpt[i] = 1 + 3 * rand();
+		fp_load_ckpt_new[i] = 1 + 4 * rand();
+	}
+
+	pid = fork();
+	if (pid < 0) {
+		perror("fork() failed");
+		return TEST_FAIL;
+	}
+
+	if (pid == 0)
+		tm_spd_vsx();
+
+	if (pid) {
+		pptr = (int *)shmat(shm_id, NULL, 0);
+		while (!pptr[2])
+			asm volatile("" : : : "memory");
+
+		ret = trace_tm_spd_vsx(pid);
+		if (ret) {
+			kill(pid, SIGKILL);
+			shmdt((void *)pptr);
+			shmctl(shm_id, IPC_RMID, NULL);
+			return TEST_FAIL;
+		}
+
+		shmdt((void *)pptr);
+		ret = wait(&status);
+		shmctl(shm_id, IPC_RMID, NULL);
+		if (ret != pid) {
+			printf("Child's exit status not captured\n");
+			return TEST_FAIL;
+		}
+
+		return (WIFEXITED(status) && WEXITSTATUS(status)) ? TEST_FAIL :
+			TEST_PASS;
+	}
+	return TEST_PASS;
+}
+
+int main(int argc, char *argv[])
+{
+	return test_harness(ptrace_tm_spd_vsx, "ptrace_tm_spd_vsx");
+}
diff --git a/tools/testing/selftests/powerpc/ptrace/ptrace-tm-spr.c b/tools/testing/selftests/powerpc/ptrace/ptrace-tm-spr.c
new file mode 100644
index 000000000..068bfed2e
--- /dev/null
+++ b/tools/testing/selftests/powerpc/ptrace/ptrace-tm-spr.c
@@ -0,0 +1,163 @@
+// SPDX-License-Identifier: GPL-2.0-or-later
+/*
+ * Ptrace test TM SPR registers
+ *
+ * Copyright (C) 2015 Anshuman Khandual, IBM Corporation.
+ */
+#include "ptrace.h"
+#include "tm.h"
+
+/* Tracee and tracer shared data */
+struct shared {
+	int flag;
+	struct tm_spr_regs regs;
+};
+unsigned long tfhar;
+
+int shm_id;
+struct shared *cptr, *pptr;
+
+int shm_id1;
+int *cptr1, *pptr1;
+
+#define TM_KVM_SCHED   0xe0000001ac000001
+int validate_tm_spr(struct tm_spr_regs *regs)
+{
+	FAIL_IF(regs->tm_tfhar != tfhar);
+	FAIL_IF((regs->tm_texasr == TM_KVM_SCHED) && (regs->tm_tfiar != 0));
+
+	return TEST_PASS;
+}
+
+void tm_spr(void)
+{
+	unsigned long result, texasr;
+	int ret;
+
+	cptr = (struct shared *)shmat(shm_id, NULL, 0);
+	cptr1 = (int *)shmat(shm_id1, NULL, 0);
+
+trans:
+	cptr1[0] = 0;
+	asm __volatile__(
+		"1: ;"
+		/* TM failover handler should follow "tbegin.;" */
+		"mflr 31;"
+		"bl 4f;"	/* $ = TFHAR - 12 */
+		"4: ;"
+		"mflr %[tfhar];"
+		"mtlr 31;"
+
+		"tbegin.;"
+		"beq 2f;"
+
+		"tsuspend.;"
+		"li 8, 1;"
+		"sth 8, 0(%[cptr1]);"
+		"tresume.;"
+		"b .;"
+
+		"tend.;"
+		"li 0, 0;"
+		"ori %[res], 0, 0;"
+		"b 3f;"
+
+		"2: ;"
+
+		"li 0, 1;"
+		"ori %[res], 0, 0;"
+		"mfspr %[texasr], %[sprn_texasr];"
+
+		"3: ;"
+		: [tfhar] "=r" (tfhar), [res] "=r" (result),
+		[texasr] "=r" (texasr), [cptr1] "=b" (cptr1)
+		: [sprn_texasr] "i"  (SPRN_TEXASR)
+		: "memory", "r0", "r8", "r31"
+		);
+
+	/* There are 2 32bit instructions before tbegin. */
+	tfhar += 12;
+
+	if (result) {
+		if (!cptr->flag)
+			goto trans;
+
+		ret = validate_tm_spr((struct tm_spr_regs *)&cptr->regs);
+		shmdt((void *)cptr);
+		shmdt((void *)cptr1);
+		if (ret)
+			exit(1);
+		exit(0);
+	}
+	shmdt((void *)cptr);
+	shmdt((void *)cptr1);
+	exit(1);
+}
+
+int trace_tm_spr(pid_t child)
+{
+	FAIL_IF(start_trace(child));
+	FAIL_IF(show_tm_spr(child, (struct tm_spr_regs *)&pptr->regs));
+
+	printf("TFHAR: %lx TEXASR: %lx TFIAR: %lx\n", pptr->regs.tm_tfhar,
+				pptr->regs.tm_texasr, pptr->regs.tm_tfiar);
+
+	pptr->flag = 1;
+	FAIL_IF(stop_trace(child));
+
+	return TEST_PASS;
+}
+
+int ptrace_tm_spr(void)
+{
+	pid_t pid;
+	int ret, status;
+
+	SKIP_IF(!have_htm());
+	shm_id = shmget(IPC_PRIVATE, sizeof(struct shared), 0777|IPC_CREAT);
+	shm_id1 = shmget(IPC_PRIVATE, sizeof(int), 0777|IPC_CREAT);
+	pid = fork();
+	if (pid < 0) {
+		perror("fork() failed");
+		return TEST_FAIL;
+	}
+
+	if (pid == 0)
+		tm_spr();
+
+	if (pid) {
+		pptr = (struct shared *)shmat(shm_id, NULL, 0);
+		pptr1 = (int *)shmat(shm_id1, NULL, 0);
+
+		while (!pptr1[0])
+			asm volatile("" : : : "memory");
+		ret = trace_tm_spr(pid);
+		if (ret) {
+			kill(pid, SIGKILL);
+			shmdt((void *)pptr);
+			shmdt((void *)pptr1);
+			shmctl(shm_id, IPC_RMID, NULL);
+			shmctl(shm_id1, IPC_RMID, NULL);
+			return TEST_FAIL;
+		}
+
+		shmdt((void *)pptr);
+		shmdt((void *)pptr1);
+		ret = wait(&status);
+		shmctl(shm_id, IPC_RMID, NULL);
+		shmctl(shm_id1, IPC_RMID, NULL);
+		if (ret != pid) {
+			printf("Child's exit status not captured\n");
+			return TEST_FAIL;
+		}
+
+		return (WIFEXITED(status) && WEXITSTATUS(status)) ? TEST_FAIL :
+			TEST_PASS;
+	}
+	return TEST_PASS;
+}
+
+int main(int argc, char *argv[])
+{
+	return test_harness(ptrace_tm_spr, "ptrace_tm_spr");
+}
diff --git a/tools/testing/selftests/powerpc/ptrace/ptrace-tm-tar.c b/tools/testing/selftests/powerpc/ptrace/ptrace-tm-tar.c
new file mode 100644
index 000000000..46ef378a1
--- /dev/null
+++ b/tools/testing/selftests/powerpc/ptrace/ptrace-tm-tar.c
@@ -0,0 +1,156 @@
+// SPDX-License-Identifier: GPL-2.0-or-later
+/*
+ * Ptrace test for TAR, PPR, DSCR registers in the TM context
+ *
+ * Copyright (C) 2015 Anshuman Khandual, IBM Corporation.
+ */
+#include "ptrace.h"
+#include "tm.h"
+#include "ptrace-tar.h"
+
+int shm_id;
+unsigned long *cptr, *pptr;
+
+
+void tm_tar(void)
+{
+	unsigned long result, texasr;
+	unsigned long regs[3];
+	int ret;
+
+	cptr = (unsigned long *)shmat(shm_id, NULL, 0);
+
+trans:
+	cptr[1] = 0;
+	asm __volatile__(
+		"li	4, %[tar_1];"
+		"mtspr %[sprn_tar],  4;"	/* TAR_1 */
+		"li	4, %[dscr_1];"
+		"mtspr %[sprn_dscr], 4;"	/* DSCR_1 */
+		"or     31,31,31;"		/* PPR_1*/
+
+		"1: ;"
+		"tbegin.;"
+		"beq 2f;"
+
+		"li	4, %[tar_2];"
+		"mtspr %[sprn_tar],  4;"	/* TAR_2 */
+		"li	4, %[dscr_2];"
+		"mtspr %[sprn_dscr], 4;"	/* DSCR_2 */
+		"or     1,1,1;"			/* PPR_2 */
+		"tsuspend.;"
+		"li 0, 1;"
+		"stw 0, 0(%[cptr1]);"
+		"tresume.;"
+		"b .;"
+
+		"tend.;"
+		"li 0, 0;"
+		"ori %[res], 0, 0;"
+		"b 3f;"
+
+		/* Transaction abort handler */
+		"2: ;"
+		"li 0, 1;"
+		"ori %[res], 0, 0;"
+		"mfspr %[texasr], %[sprn_texasr];"
+
+		"3: ;"
+
+		: [res] "=r" (result), [texasr] "=r" (texasr)
+		: [sprn_dscr]"i"(SPRN_DSCR), [sprn_tar]"i"(SPRN_TAR),
+		[sprn_ppr]"i"(SPRN_PPR), [sprn_texasr]"i"(SPRN_TEXASR),
+		[tar_1]"i"(TAR_1), [dscr_1]"i"(DSCR_1), [tar_2]"i"(TAR_2),
+		[dscr_2]"i"(DSCR_2), [cptr1] "b" (&cptr[1])
+		: "memory", "r0", "r3", "r4", "r5", "r6"
+		);
+
+	/* TM failed, analyse */
+	if (result) {
+		if (!cptr[0])
+			goto trans;
+
+		regs[0] = mfspr(SPRN_TAR);
+		regs[1] = mfspr(SPRN_PPR);
+		regs[2] = mfspr(SPRN_DSCR);
+
+		shmdt(&cptr);
+		printf("%-30s TAR: %lu PPR: %lx DSCR: %lu\n",
+				user_read, regs[0], regs[1], regs[2]);
+
+		ret = validate_tar_registers(regs, TAR_4, PPR_4, DSCR_4);
+		if (ret)
+			exit(1);
+		exit(0);
+	}
+	shmdt(&cptr);
+	exit(1);
+}
+
+int trace_tm_tar(pid_t child)
+{
+	unsigned long regs[3];
+
+	FAIL_IF(start_trace(child));
+	FAIL_IF(show_tar_registers(child, regs));
+	printf("%-30s TAR: %lu PPR: %lx DSCR: %lu\n",
+			ptrace_read_running, regs[0], regs[1], regs[2]);
+
+	FAIL_IF(validate_tar_registers(regs, TAR_2, PPR_2, DSCR_2));
+	FAIL_IF(show_tm_checkpointed_state(child, regs));
+	printf("%-30s TAR: %lu PPR: %lx DSCR: %lu\n",
+			ptrace_read_ckpt, regs[0], regs[1], regs[2]);
+
+	FAIL_IF(validate_tar_registers(regs, TAR_1, PPR_1, DSCR_1));
+	FAIL_IF(write_ckpt_tar_registers(child, TAR_4, PPR_4, DSCR_4));
+	printf("%-30s TAR: %u PPR: %lx DSCR: %u\n",
+			ptrace_write_ckpt, TAR_4, PPR_4, DSCR_4);
+
+	pptr[0] = 1;
+	FAIL_IF(stop_trace(child));
+	return TEST_PASS;
+}
+
+int ptrace_tm_tar(void)
+{
+	pid_t pid;
+	int ret, status;
+
+	SKIP_IF(!have_htm());
+	shm_id = shmget(IPC_PRIVATE, sizeof(int) * 2, 0777|IPC_CREAT);
+	pid = fork();
+	if (pid == 0)
+		tm_tar();
+
+	pptr = (unsigned long *)shmat(shm_id, NULL, 0);
+	pptr[0] = 0;
+
+	if (pid) {
+		while (!pptr[1])
+			asm volatile("" : : : "memory");
+		ret = trace_tm_tar(pid);
+		if (ret) {
+			kill(pid, SIGTERM);
+			shmdt(&pptr);
+			shmctl(shm_id, IPC_RMID, NULL);
+			return TEST_FAIL;
+		}
+		shmdt(&pptr);
+
+		ret = wait(&status);
+		shmctl(shm_id, IPC_RMID, NULL);
+		if (ret != pid) {
+			printf("Child's exit status not captured\n");
+			return TEST_FAIL;
+		}
+
+		return (WIFEXITED(status) && WEXITSTATUS(status)) ? TEST_FAIL :
+			TEST_PASS;
+	}
+	return TEST_PASS;
+}
+
+int main(int argc, char *argv[])
+{
+	return test_harness(ptrace_tm_tar, "ptrace_tm_tar");
+}
diff --git a/tools/testing/selftests/powerpc/ptrace/ptrace-tm-vsx.c b/tools/testing/selftests/powerpc/ptrace/ptrace-tm-vsx.c
new file mode 100644
index 000000000..70ca01234
--- /dev/null
+++ b/tools/testing/selftests/powerpc/ptrace/ptrace-tm-vsx.c
@@ -0,0 +1,163 @@
+// SPDX-License-Identifier: GPL-2.0-or-later
+/*
+ * Ptrace test for VMX/VSX registers in the TM context
+ *
+ * Copyright (C) 2015 Anshuman Khandual, IBM Corporation.
+ */
+#include "ptrace.h"
+#include "tm.h"
+#include "ptrace-vsx.h"
+
+int shm_id;
+unsigned long *cptr, *pptr;
+
+unsigned long fp_load[VEC_MAX];
+unsigned long fp_store[VEC_MAX];
+unsigned long fp_load_ckpt[VEC_MAX];
+unsigned long fp_load_ckpt_new[VEC_MAX];
+
+__attribute__((used)) void load_vsx(void)
+{
+	loadvsx(fp_load, 0);
+}
+
+__attribute__((used)) void load_vsx_ckpt(void)
+{
+	loadvsx(fp_load_ckpt, 0);
+}
+
+void tm_vsx(void)
+{
+	unsigned long result, texasr;
+	int ret;
+
+	cptr = (unsigned long *)shmat(shm_id, NULL, 0);
+
+trans:
+	cptr[1] = 0;
+	asm __volatile__(
+		"bl load_vsx_ckpt;"
+
+		"1: ;"
+		"tbegin.;"
+		"beq 2f;"
+
+		"bl load_vsx;"
+		"tsuspend.;"
+		"li 7, 1;"
+		"stw 7, 0(%[cptr1]);"
+		"tresume.;"
+		"b .;"
+
+		"tend.;"
+		"li 0, 0;"
+		"ori %[res], 0, 0;"
+		"b 3f;"
+
+		"2: ;"
+		"li 0, 1;"
+		"ori %[res], 0, 0;"
+		"mfspr %[texasr], %[sprn_texasr];"
+
+		"3: ;"
+		: [res] "=r" (result), [texasr] "=r" (texasr)
+		: [sprn_texasr] "i"  (SPRN_TEXASR), [cptr1] "b" (&cptr[1])
+		: "memory", "r0", "r3", "r4",
+		  "r7", "r8", "r9", "r10", "r11", "lr"
+		);
+
+	if (result) {
+		if (!cptr[0])
+			goto trans;
+
+		shmdt((void *)cptr);
+		storevsx(fp_store, 0);
+		ret = compare_vsx_vmx(fp_store, fp_load_ckpt_new);
+		if (ret)
+			exit(1);
+		exit(0);
+	}
+	shmdt((void *)cptr);
+	exit(1);
+}
+
+int trace_tm_vsx(pid_t child)
+{
+	unsigned long vsx[VSX_MAX];
+	unsigned long vmx[VMX_MAX + 2][2];
+
+	FAIL_IF(start_trace(child));
+	FAIL_IF(show_vsx(child, vsx));
+	FAIL_IF(validate_vsx(vsx, fp_load));
+	FAIL_IF(show_vmx(child, vmx));
+	FAIL_IF(validate_vmx(vmx, fp_load));
+	FAIL_IF(show_vsx_ckpt(child, vsx));
+	FAIL_IF(validate_vsx(vsx, fp_load_ckpt));
+	FAIL_IF(show_vmx_ckpt(child, vmx));
+	FAIL_IF(validate_vmx(vmx, fp_load_ckpt));
+	memset(vsx, 0, sizeof(vsx));
+	memset(vmx, 0, sizeof(vmx));
+
+	load_vsx_vmx(fp_load_ckpt_new, vsx, vmx);
+
+	FAIL_IF(write_vsx_ckpt(child, vsx));
+	FAIL_IF(write_vmx_ckpt(child, vmx));
+	pptr[0] = 1;
+	FAIL_IF(stop_trace(child));
+	return TEST_PASS;
+}
+
+int ptrace_tm_vsx(void)
+{
+	pid_t pid;
+	int ret, status, i;
+
+	SKIP_IF(!have_htm());
+	shm_id = shmget(IPC_PRIVATE, sizeof(int) * 2, 0777|IPC_CREAT);
+
+	for (i = 0; i < 128; i++) {
+		fp_load[i] = 1 + rand();
+		fp_load_ckpt[i] = 1 + 2 * rand();
+		fp_load_ckpt_new[i] = 1 + 3 * rand();
+	}
+
+	pid = fork();
+	if (pid < 0) {
+		perror("fork() failed");
+		return TEST_FAIL;
+	}
+
+	if (pid == 0)
+		tm_vsx();
+
+	if (pid) {
+		pptr = (unsigned long *)shmat(shm_id, NULL, 0);
+		while (!pptr[1])
+			asm volatile("" : : : "memory");
+
+		ret = trace_tm_vsx(pid);
+		if (ret) {
+			kill(pid, SIGKILL);
+			shmdt((void *)pptr);
+			shmctl(shm_id, IPC_RMID, NULL);
+			return TEST_FAIL;
+		}
+
+		shmdt((void *)pptr);
+		ret = wait(&status);
+		shmctl(shm_id, IPC_RMID, NULL);
+		if (ret != pid) {
+			printf("Child's exit status not captured\n");
+			return TEST_FAIL;
+		}
+
+		return (WIFEXITED(status) && WEXITSTATUS(status)) ? TEST_FAIL :
+			TEST_PASS;
+	}
+	return TEST_PASS;
+}
+
+int main(int argc, char *argv[])
+{
+	return test_harness(ptrace_tm_vsx, "ptrace_tm_vsx");
+}
diff --git a/tools/testing/selftests/powerpc/ptrace/ptrace-vsx.c b/tools/testing/selftests/powerpc/ptrace/ptrace-vsx.c
new file mode 100644
index 000000000..cb9875f76
--- /dev/null
+++ b/tools/testing/selftests/powerpc/ptrace/ptrace-vsx.c
@@ -0,0 +1,115 @@
+// SPDX-License-Identifier: GPL-2.0-or-later
+/*
+ * Ptrace test for VMX/VSX registers
+ *
+ * Copyright (C) 2015 Anshuman Khandual, IBM Corporation.
+ */
+#include "ptrace.h"
+#include "ptrace-vsx.h"
+
+/* Tracer and Tracee Shared Data */
+int shm_id;
+int *cptr, *pptr;
+
+unsigned long fp_load[VEC_MAX];
+unsigned long fp_load_new[VEC_MAX];
+unsigned long fp_store[VEC_MAX];
+
+void vsx(void)
+{
+	int ret;
+
+	cptr = (int *)shmat(shm_id, NULL, 0);
+	loadvsx(fp_load, 0);
+	cptr[1] = 1;
+
+	while (!cptr[0])
+		asm volatile("" : : : "memory");
+	shmdt((void *) cptr);
+
+	storevsx(fp_store, 0);
+	ret = compare_vsx_vmx(fp_store, fp_load_new);
+	if (ret)
+		exit(1);
+	exit(0);
+}
+
+int trace_vsx(pid_t child)
+{
+	unsigned long vsx[VSX_MAX];
+	unsigned long vmx[VMX_MAX + 2][2];
+
+	FAIL_IF(start_trace(child));
+	FAIL_IF(show_vsx(child, vsx));
+	FAIL_IF(validate_vsx(vsx, fp_load));
+	FAIL_IF(show_vmx(child, vmx));
+	FAIL_IF(validate_vmx(vmx, fp_load));
+
+	memset(vsx, 0, sizeof(vsx));
+	memset(vmx, 0, sizeof(vmx));
+	load_vsx_vmx(fp_load_new, vsx, vmx);
+
+	FAIL_IF(write_vsx(child, vsx));
+	FAIL_IF(write_vmx(child, vmx));
+	FAIL_IF(stop_trace(child));
+
+	return TEST_PASS;
+}
+
+int ptrace_vsx(void)
+{
+	pid_t pid;
+	int ret, status, i;
+
+	SKIP_IF(!have_hwcap(PPC_FEATURE_HAS_VSX));
+
+	shm_id = shmget(IPC_PRIVATE, sizeof(int) * 2, 0777|IPC_CREAT);
+
+	for (i = 0; i < VEC_MAX; i++)
+		fp_load[i] = i + rand();
+
+	for (i = 0; i < VEC_MAX; i++)
+		fp_load_new[i] = i + 2 * rand();
+
+	pid = fork();
+	if (pid < 0) {
+		perror("fork() failed");
+		return TEST_FAIL;
+	}
+
+	if (pid == 0)
+		vsx();
+
+	if (pid) {
+		pptr = (int *)shmat(shm_id, NULL, 0);
+		while (!pptr[1])
+			asm volatile("" : : : "memory");
+
+		ret = trace_vsx(pid);
+		if (ret) {
+			kill(pid, SIGTERM);
+			shmdt((void *)pptr);
+			shmctl(shm_id, IPC_RMID, NULL);
+			return TEST_FAIL;
+		}
+
+		pptr[0] = 1;
+		shmdt((void *)pptr);
+
+		ret = wait(&status);
+		shmctl(shm_id, IPC_RMID, NULL);
+		if (ret != pid) {
+			printf("Child's exit status not captured\n");
+			return TEST_FAIL;
+		}
+
+		return (WIFEXITED(status) && WEXITSTATUS(status)) ? TEST_FAIL :
+			TEST_PASS;
+	}
+	return TEST_PASS;
+}
+
+int main(int argc, char *argv[])
+{
+	return test_harness(ptrace_vsx, "ptrace_vsx");
+}
diff --git a/tools/testing/selftests/powerpc/ptrace/ptrace-vsx.h b/tools/testing/selftests/powerpc/ptrace/ptrace-vsx.h
new file mode 100644
index 000000000..663348521
--- /dev/null
+++ b/tools/testing/selftests/powerpc/ptrace/ptrace-vsx.h
@@ -0,0 +1,123 @@
+/* SPDX-License-Identifier: GPL-2.0-or-later */
+/*
+ * Copyright (C) 2015 Anshuman Khandual, IBM Corporation.
+ */
+#define VEC_MAX 128
+#define VSX_MAX 32
+#define VMX_MAX 32
+
+/*
+ * unsigned long vsx[32]
+ * unsigned long load[128]
+ */
+int validate_vsx(unsigned long *vsx, unsigned long *load)
+{
+	int i;
+
+	for (i = 0; i < VSX_MAX; i++) {
+		if (vsx[i] != load[2 * i + 1]) {
+			printf("vsx[%d]: %lx load[%d] %lx\n",
+					i, vsx[i], 2 * i + 1, load[2 * i + 1]);
+			return TEST_FAIL;
+		}
+	}
+	return TEST_PASS;
+}
+
+/*
+ * unsigned long vmx[32][2]
+ * unsigned long load[128]
+ */
+int validate_vmx(unsigned long vmx[][2], unsigned long *load)
+{
+	int i;
+
+	for (i = 0; i < VMX_MAX; i++) {
+		#if __BYTE_ORDER__ == __ORDER_BIG_ENDIAN__
+		if ((vmx[i][0] != load[64 + 2 * i]) ||
+				(vmx[i][1] != load[65 + 2 * i])) {
+			printf("vmx[%d][0]: %lx load[%d] %lx\n",
+					i, vmx[i][0], 64 + 2 * i,
+					load[64 + 2 * i]);
+			printf("vmx[%d][1]: %lx load[%d] %lx\n",
+					i, vmx[i][1], 65 + 2 * i,
+					load[65 + 2 * i]);
+			return TEST_FAIL;
+		}
+		#else  /*
+			* In LE each value pair is stored in an
+			* alternate manner.
+			*/
+		if ((vmx[i][0] != load[65 + 2 * i]) ||
+				(vmx[i][1] != load[64 + 2 * i])) {
+			printf("vmx[%d][0]: %lx load[%d] %lx\n",
+					i, vmx[i][0], 65 + 2 * i,
+					load[65 + 2 * i]);
+			printf("vmx[%d][1]: %lx load[%d] %lx\n",
+					i, vmx[i][1], 64 + 2 * i,
+					load[64 + 2 * i]);
+			return TEST_FAIL;
+		}
+		#endif
+	}
+	return TEST_PASS;
+}
+
+/*
+ * unsigned long store[128]
+ * unsigned long load[128]
+ */
+int compare_vsx_vmx(unsigned long *store, unsigned long *load)
+{
+	int i;
+
+	for (i = 0; i < VSX_MAX; i++) {
+		if (store[1 + 2 * i] != load[1 + 2 * i]) {
+			printf("store[%d]: %lx load[%d] %lx\n",
+					1 + 2 * i, store[i],
+					1 + 2 * i, load[i]);
+			return TEST_FAIL;
+		}
+	}
+
+	#if __BYTE_ORDER__ == __ORDER_BIG_ENDIAN__
+	for (i = 64; i < VEC_MAX; i++) {
+		if (store[i] != load[i]) {
+			printf("store[%d]: %lx load[%d] %lx\n",
+					i, store[i], i, load[i]);
+			return TEST_FAIL;
+		}
+	}
+	#else	/* In LE each value pair is stored in an alternate manner */
+	for (i = 64; i < VEC_MAX; i++) {
+		if (!(i % 2) && (store[i] != load[i+1])) {
+			printf("store[%d]: %lx load[%d] %lx\n",
+					i, store[i], i+1, load[i+1]);
+			return TEST_FAIL;
+		}
+		if ((i % 2) && (store[i] != load[i-1])) {
+			printf("here store[%d]: %lx load[%d] %lx\n",
+					i, store[i], i-1, load[i-1]);
+			return TEST_FAIL;
+		}
+	}
+	#endif
+	return TEST_PASS;
+}
+
+void load_vsx_vmx(unsigned long *load, unsigned long *vsx,
+		unsigned long vmx[][2])
+{
+	int i;
+
+	for (i = 0; i < VSX_MAX; i++)
+		vsx[i] = load[1 + 2 * i];
+
+	for (i = 0; i < VMX_MAX; i++) {
+		vmx[i][0] = load[64 + 2 * i];
+		vmx[i][1] = load[65 + 2 * i];
+	}
+}
+
+void loadvsx(void *p, int tmp);
+void storevsx(void *p, int tmp);
diff --git a/tools/testing/selftests/powerpc/ptrace/ptrace.h b/tools/testing/selftests/powerpc/ptrace/ptrace.h
new file mode 100644
index 000000000..5181ad9b4
--- /dev/null
+++ b/tools/testing/selftests/powerpc/ptrace/ptrace.h
@@ -0,0 +1,745 @@
+/* SPDX-License-Identifier: GPL-2.0-or-later */
+/*
+ * Ptrace interface test helper functions
+ *
+ * Copyright (C) 2015 Anshuman Khandual, IBM Corporation.
+ */
+#include <inttypes.h>
+#include <unistd.h>
+#include <stdlib.h>
+#include <string.h>
+#include <malloc.h>
+#include <errno.h>
+#include <time.h>
+#include <sys/ptrace.h>
+#include <sys/ioctl.h>
+#include <sys/uio.h>
+#include <sys/types.h>
+#include <sys/wait.h>
+#include <sys/signal.h>
+#include <sys/ipc.h>
+#include <sys/shm.h>
+#include <sys/user.h>
+#include <linux/elf.h>
+#include <linux/types.h>
+#include <linux/auxvec.h>
+#include "reg.h"
+#include "utils.h"
+
+#define TEST_PASS 0
+#define TEST_FAIL 1
+
+struct fpr_regs {
+	unsigned long fpr[32];
+	unsigned long fpscr;
+};
+
+struct tm_spr_regs {
+	unsigned long tm_tfhar;
+	unsigned long tm_texasr;
+	unsigned long tm_tfiar;
+};
+
+#ifndef NT_PPC_TAR
+#define NT_PPC_TAR	0x103
+#define NT_PPC_PPR	0x104
+#define NT_PPC_DSCR	0x105
+#define NT_PPC_EBB	0x106
+#define NT_PPC_PMU	0x107
+#define NT_PPC_TM_CGPR	0x108
+#define NT_PPC_TM_CFPR	0x109
+#define NT_PPC_TM_CVMX	0x10a
+#define NT_PPC_TM_CVSX	0x10b
+#define NT_PPC_TM_SPR	0x10c
+#define NT_PPC_TM_CTAR	0x10d
+#define NT_PPC_TM_CPPR	0x10e
+#define NT_PPC_TM_CDSCR	0x10f
+#endif
+
+/* Basic ptrace operations */
+int start_trace(pid_t child)
+{
+	int ret;
+
+	ret = ptrace(PTRACE_ATTACH, child, NULL, NULL);
+	if (ret) {
+		perror("ptrace(PTRACE_ATTACH) failed");
+		return TEST_FAIL;
+	}
+	ret = waitpid(child, NULL, 0);
+	if (ret != child) {
+		perror("waitpid() failed");
+		return TEST_FAIL;
+	}
+	return TEST_PASS;
+}
+
+int stop_trace(pid_t child)
+{
+	int ret;
+
+	ret = ptrace(PTRACE_DETACH, child, NULL, NULL);
+	if (ret) {
+		perror("ptrace(PTRACE_DETACH) failed");
+		return TEST_FAIL;
+	}
+	return TEST_PASS;
+}
+
+int cont_trace(pid_t child)
+{
+	int ret;
+
+	ret = ptrace(PTRACE_CONT, child, NULL, NULL);
+	if (ret) {
+		perror("ptrace(PTRACE_CONT) failed");
+		return TEST_FAIL;
+	}
+	return TEST_PASS;
+}
+
+int ptrace_read_regs(pid_t child, unsigned long type, unsigned long regs[],
+		     int n)
+{
+	struct iovec iov;
+	long ret;
+
+	FAIL_IF(start_trace(child));
+
+	iov.iov_base = regs;
+	iov.iov_len = n * sizeof(unsigned long);
+
+	ret = ptrace(PTRACE_GETREGSET, child, type, &iov);
+	if (ret)
+		return ret;
+
+	FAIL_IF(stop_trace(child));
+
+	return TEST_PASS;
+}
+
+long ptrace_write_regs(pid_t child, unsigned long type, unsigned long regs[],
+		       int n)
+{
+	struct iovec iov;
+	long ret;
+
+	FAIL_IF(start_trace(child));
+
+	iov.iov_base = regs;
+	iov.iov_len = n * sizeof(unsigned long);
+
+	ret = ptrace(PTRACE_SETREGSET, child, type, &iov);
+
+	FAIL_IF(stop_trace(child));
+
+	return ret;
+}
+
+/* TAR, PPR, DSCR */
+int show_tar_registers(pid_t child, unsigned long *out)
+{
+	struct iovec iov;
+	unsigned long *reg;
+	int ret;
+
+	reg = malloc(sizeof(unsigned long));
+	if (!reg) {
+		perror("malloc() failed");
+		return TEST_FAIL;
+	}
+	iov.iov_base = (u64 *) reg;
+	iov.iov_len = sizeof(unsigned long);
+
+	ret = ptrace(PTRACE_GETREGSET, child, NT_PPC_TAR, &iov);
+	if (ret) {
+		perror("ptrace(PTRACE_GETREGSET) failed");
+		goto fail;
+	}
+	if (out)
+		out[0] = *reg;
+
+	ret = ptrace(PTRACE_GETREGSET, child, NT_PPC_PPR, &iov);
+	if (ret) {
+		perror("ptrace(PTRACE_GETREGSET) failed");
+		goto fail;
+	}
+	if (out)
+		out[1] = *reg;
+
+	ret = ptrace(PTRACE_GETREGSET, child, NT_PPC_DSCR, &iov);
+	if (ret) {
+		perror("ptrace(PTRACE_GETREGSET) failed");
+		goto fail;
+	}
+	if (out)
+		out[2] = *reg;
+
+	free(reg);
+	return TEST_PASS;
+fail:
+	free(reg);
+	return TEST_FAIL;
+}
+
+int write_tar_registers(pid_t child, unsigned long tar,
+		unsigned long ppr, unsigned long dscr)
+{
+	struct iovec iov;
+	unsigned long *reg;
+	int ret;
+
+	reg = malloc(sizeof(unsigned long));
+	if (!reg) {
+		perror("malloc() failed");
+		return TEST_FAIL;
+	}
+
+	iov.iov_base = (u64 *) reg;
+	iov.iov_len = sizeof(unsigned long);
+
+	*reg = tar;
+	ret = ptrace(PTRACE_SETREGSET, child, NT_PPC_TAR, &iov);
+	if (ret) {
+		perror("ptrace(PTRACE_SETREGSET) failed");
+		goto fail;
+	}
+
+	*reg = ppr;
+	ret = ptrace(PTRACE_SETREGSET, child, NT_PPC_PPR, &iov);
+	if (ret) {
+		perror("ptrace(PTRACE_SETREGSET) failed");
+		goto fail;
+	}
+
+	*reg = dscr;
+	ret = ptrace(PTRACE_SETREGSET, child, NT_PPC_DSCR, &iov);
+	if (ret) {
+		perror("ptrace(PTRACE_SETREGSET) failed");
+		goto fail;
+	}
+
+	free(reg);
+	return TEST_PASS;
+fail:
+	free(reg);
+	return TEST_FAIL;
+}
+
+int show_tm_checkpointed_state(pid_t child, unsigned long *out)
+{
+	struct iovec iov;
+	unsigned long *reg;
+	int ret;
+
+	reg = malloc(sizeof(unsigned long));
+	if (!reg) {
+		perror("malloc() failed");
+		return TEST_FAIL;
+	}
+
+	iov.iov_base = (u64 *) reg;
+	iov.iov_len = sizeof(unsigned long);
+
+	ret = ptrace(PTRACE_GETREGSET, child, NT_PPC_TM_CTAR, &iov);
+	if (ret) {
+		perror("ptrace(PTRACE_GETREGSET) failed");
+		goto fail;
+	}
+	if (out)
+		out[0] = *reg;
+
+	ret = ptrace(PTRACE_GETREGSET, child, NT_PPC_TM_CPPR, &iov);
+	if (ret) {
+		perror("ptrace(PTRACE_GETREGSET) failed");
+		goto fail;
+	}
+	if (out)
+		out[1] = *reg;
+
+	ret = ptrace(PTRACE_GETREGSET, child, NT_PPC_TM_CDSCR, &iov);
+	if (ret) {
+		perror("ptrace(PTRACE_GETREGSET) failed");
+		goto fail;
+	}
+	if (out)
+		out[2] = *reg;
+
+	free(reg);
+	return TEST_PASS;
+
+fail:
+	free(reg);
+	return TEST_FAIL;
+}
+
+int write_ckpt_tar_registers(pid_t child, unsigned long tar,
+		unsigned long ppr, unsigned long dscr)
+{
+	struct iovec iov;
+	unsigned long *reg;
+	int ret;
+
+	reg = malloc(sizeof(unsigned long));
+	if (!reg) {
+		perror("malloc() failed");
+		return TEST_FAIL;
+	}
+
+	iov.iov_base = (u64 *) reg;
+	iov.iov_len = sizeof(unsigned long);
+
+	*reg = tar;
+	ret = ptrace(PTRACE_SETREGSET, child, NT_PPC_TM_CTAR, &iov);
+	if (ret) {
+		perror("ptrace(PTRACE_GETREGSET) failed");
+		goto fail;
+	}
+
+	*reg = ppr;
+	ret = ptrace(PTRACE_SETREGSET, child, NT_PPC_TM_CPPR, &iov);
+	if (ret) {
+		perror("ptrace(PTRACE_GETREGSET) failed");
+		goto fail;
+	}
+
+	*reg = dscr;
+	ret = ptrace(PTRACE_SETREGSET, child, NT_PPC_TM_CDSCR, &iov);
+	if (ret) {
+		perror("ptrace(PTRACE_GETREGSET) failed");
+		goto fail;
+	}
+
+	free(reg);
+	return TEST_PASS;
+fail:
+	free(reg);
+	return TEST_FAIL;
+}
+
+/* FPR */
+int show_fpr(pid_t child, unsigned long *fpr)
+{
+	struct fpr_regs *regs;
+	int ret, i;
+
+	regs = (struct fpr_regs *) malloc(sizeof(struct fpr_regs));
+	ret = ptrace(PTRACE_GETFPREGS, child, NULL, regs);
+	if (ret) {
+		perror("ptrace(PTRACE_GETREGSET) failed");
+		return TEST_FAIL;
+	}
+
+	if (fpr) {
+		for (i = 0; i < 32; i++)
+			fpr[i] = regs->fpr[i];
+	}
+	return TEST_PASS;
+}
+
+int write_fpr(pid_t child, unsigned long val)
+{
+	struct fpr_regs *regs;
+	int ret, i;
+
+	regs = (struct fpr_regs *) malloc(sizeof(struct fpr_regs));
+	ret = ptrace(PTRACE_GETFPREGS, child, NULL, regs);
+	if (ret) {
+		perror("ptrace(PTRACE_GETREGSET) failed");
+		return TEST_FAIL;
+	}
+
+	for (i = 0; i < 32; i++)
+		regs->fpr[i] = val;
+
+	ret = ptrace(PTRACE_SETFPREGS, child, NULL, regs);
+	if (ret) {
+		perror("ptrace(PTRACE_GETREGSET) failed");
+		return TEST_FAIL;
+	}
+	return TEST_PASS;
+}
+
+int show_ckpt_fpr(pid_t child, unsigned long *fpr)
+{
+	struct fpr_regs *regs;
+	struct iovec iov;
+	int ret, i;
+
+	regs = (struct fpr_regs *) malloc(sizeof(struct fpr_regs));
+	iov.iov_base = regs;
+	iov.iov_len = sizeof(struct fpr_regs);
+
+	ret = ptrace(PTRACE_GETREGSET, child, NT_PPC_TM_CFPR, &iov);
+	if (ret) {
+		perror("ptrace(PTRACE_GETREGSET) failed");
+		return TEST_FAIL;
+	}
+
+	if (fpr) {
+		for (i = 0; i < 32; i++)
+			fpr[i] = regs->fpr[i];
+	}
+
+	return TEST_PASS;
+}
+
+int write_ckpt_fpr(pid_t child, unsigned long val)
+{
+	struct fpr_regs *regs;
+	struct iovec iov;
+	int ret, i;
+
+	regs = (struct fpr_regs *) malloc(sizeof(struct fpr_regs));
+	iov.iov_base = regs;
+	iov.iov_len = sizeof(struct fpr_regs);
+
+	ret = ptrace(PTRACE_GETREGSET, child, NT_PPC_TM_CFPR, &iov);
+	if (ret) {
+		perror("ptrace(PTRACE_GETREGSET) failed");
+		return TEST_FAIL;
+	}
+
+	for (i = 0; i < 32; i++)
+		regs->fpr[i] = val;
+
+	ret = ptrace(PTRACE_SETREGSET, child, NT_PPC_TM_CFPR, &iov);
+	if (ret) {
+		perror("ptrace(PTRACE_GETREGSET) failed");
+		return TEST_FAIL;
+	}
+	return TEST_PASS;
+}
+
+/* GPR */
+int show_gpr(pid_t child, unsigned long *gpr)
+{
+	struct pt_regs *regs;
+	int ret, i;
+
+	regs = (struct pt_regs *) malloc(sizeof(struct pt_regs));
+	if (!regs) {
+		perror("malloc() failed");
+		return TEST_FAIL;
+	}
+
+	ret = ptrace(PTRACE_GETREGS, child, NULL, regs);
+	if (ret) {
+		perror("ptrace(PTRACE_GETREGSET) failed");
+		return TEST_FAIL;
+	}
+
+	if (gpr) {
+		for (i = 14; i < 32; i++)
+			gpr[i-14] = regs->gpr[i];
+	}
+
+	return TEST_PASS;
+}
+
+int write_gpr(pid_t child, unsigned long val)
+{
+	struct pt_regs *regs;
+	int i, ret;
+
+	regs = (struct pt_regs *) malloc(sizeof(struct pt_regs));
+	if (!regs) {
+		perror("malloc() failed");
+		return TEST_FAIL;
+	}
+
+	ret = ptrace(PTRACE_GETREGS, child, NULL, regs);
+	if (ret) {
+		perror("ptrace(PTRACE_GETREGSET) failed");
+		return TEST_FAIL;
+	}
+
+	for (i = 14; i < 32; i++)
+		regs->gpr[i] = val;
+
+	ret = ptrace(PTRACE_SETREGS, child, NULL, regs);
+	if (ret) {
+		perror("ptrace(PTRACE_GETREGSET) failed");
+		return TEST_FAIL;
+	}
+	return TEST_PASS;
+}
+
+int show_ckpt_gpr(pid_t child, unsigned long *gpr)
+{
+	struct pt_regs *regs;
+	struct iovec iov;
+	int ret, i;
+
+	regs = (struct pt_regs *) malloc(sizeof(struct pt_regs));
+	if (!regs) {
+		perror("malloc() failed");
+		return TEST_FAIL;
+	}
+
+	iov.iov_base = (u64 *) regs;
+	iov.iov_len = sizeof(struct pt_regs);
+
+	ret = ptrace(PTRACE_GETREGSET, child, NT_PPC_TM_CGPR, &iov);
+	if (ret) {
+		perror("ptrace(PTRACE_GETREGSET) failed");
+		return TEST_FAIL;
+	}
+
+	if (gpr) {
+		for (i = 14; i < 32; i++)
+			gpr[i-14] = regs->gpr[i];
+	}
+
+	return TEST_PASS;
+}
+
+int write_ckpt_gpr(pid_t child, unsigned long val)
+{
+	struct pt_regs *regs;
+	struct iovec iov;
+	int ret, i;
+
+	regs = (struct pt_regs *) malloc(sizeof(struct pt_regs));
+	if (!regs) {
+		perror("malloc() failed\n");
+		return TEST_FAIL;
+	}
+	iov.iov_base = (u64 *) regs;
+	iov.iov_len = sizeof(struct pt_regs);
+
+	ret = ptrace(PTRACE_GETREGSET, child, NT_PPC_TM_CGPR, &iov);
+	if (ret) {
+		perror("ptrace(PTRACE_GETREGSET) failed");
+		return TEST_FAIL;
+	}
+
+	for (i = 14; i < 32; i++)
+		regs->gpr[i] = val;
+
+	ret = ptrace(PTRACE_SETREGSET, child, NT_PPC_TM_CGPR, &iov);
+	if (ret) {
+		perror("ptrace(PTRACE_GETREGSET) failed");
+		return TEST_FAIL;
+	}
+	return TEST_PASS;
+}
+
+/* VMX */
+int show_vmx(pid_t child, unsigned long vmx[][2])
+{
+	int ret;
+
+	ret = ptrace(PTRACE_GETVRREGS, child, 0, vmx);
+	if (ret) {
+		perror("ptrace(PTRACE_GETVRREGS) failed");
+		return TEST_FAIL;
+	}
+	return TEST_PASS;
+}
+
+int show_vmx_ckpt(pid_t child, unsigned long vmx[][2])
+{
+	unsigned long regs[34][2];
+	struct iovec iov;
+	int ret;
+
+	iov.iov_base = (u64 *) regs;
+	iov.iov_len = sizeof(regs);
+	ret = ptrace(PTRACE_GETREGSET, child, NT_PPC_TM_CVMX, &iov);
+	if (ret) {
+		perror("ptrace(PTRACE_GETREGSET, NT_PPC_TM_CVMX) failed");
+		return TEST_FAIL;
+	}
+	memcpy(vmx, regs, sizeof(regs));
+	return TEST_PASS;
+}
+
+
+int write_vmx(pid_t child, unsigned long vmx[][2])
+{
+	int ret;
+
+	ret = ptrace(PTRACE_SETVRREGS, child, 0, vmx);
+	if (ret) {
+		perror("ptrace(PTRACE_SETVRREGS) failed");
+		return TEST_FAIL;
+	}
+	return TEST_PASS;
+}
+
+int write_vmx_ckpt(pid_t child, unsigned long vmx[][2])
+{
+	unsigned long regs[34][2];
+	struct iovec iov;
+	int ret;
+
+	memcpy(regs, vmx, sizeof(regs));
+	iov.iov_base = (u64 *) regs;
+	iov.iov_len = sizeof(regs);
+	ret = ptrace(PTRACE_SETREGSET, child, NT_PPC_TM_CVMX, &iov);
+	if (ret) {
+		perror("ptrace(PTRACE_SETREGSET, NT_PPC_TM_CVMX) failed");
+		return TEST_FAIL;
+	}
+	return TEST_PASS;
+}
+
+/* VSX */
+int show_vsx(pid_t child, unsigned long *vsx)
+{
+	int ret;
+
+	ret = ptrace(PTRACE_GETVSRREGS, child, 0, vsx);
+	if (ret) {
+		perror("ptrace(PTRACE_GETVSRREGS) failed");
+		return TEST_FAIL;
+	}
+	return TEST_PASS;
+}
+
+int show_vsx_ckpt(pid_t child, unsigned long *vsx)
+{
+	unsigned long regs[32];
+	struct iovec iov;
+	int ret;
+
+	iov.iov_base = (u64 *) regs;
+	iov.iov_len = sizeof(regs);
+	ret = ptrace(PTRACE_GETREGSET, child, NT_PPC_TM_CVSX, &iov);
+	if (ret) {
+		perror("ptrace(PTRACE_GETREGSET, NT_PPC_TM_CVSX) failed");
+		return TEST_FAIL;
+	}
+	memcpy(vsx, regs, sizeof(regs));
+	return TEST_PASS;
+}
+
+int write_vsx(pid_t child, unsigned long *vsx)
+{
+	int ret;
+
+	ret = ptrace(PTRACE_SETVSRREGS, child, 0, vsx);
+	if (ret) {
+		perror("ptrace(PTRACE_SETVSRREGS) failed");
+		return TEST_FAIL;
+	}
+	return TEST_PASS;
+}
+
+int write_vsx_ckpt(pid_t child, unsigned long *vsx)
+{
+	unsigned long regs[32];
+	struct iovec iov;
+	int ret;
+
+	memcpy(regs, vsx, sizeof(regs));
+	iov.iov_base = (u64 *) regs;
+	iov.iov_len = sizeof(regs);
+	ret = ptrace(PTRACE_SETREGSET, child, NT_PPC_TM_CVSX, &iov);
+	if (ret) {
+		perror("ptrace(PTRACE_SETREGSET, NT_PPC_TM_CVSX) failed");
+		return TEST_FAIL;
+	}
+	return TEST_PASS;
+}
+
+/* TM SPR */
+int show_tm_spr(pid_t child, struct tm_spr_regs *out)
+{
+	struct tm_spr_regs *regs;
+	struct iovec iov;
+	int ret;
+
+	regs = (struct tm_spr_regs *) malloc(sizeof(struct tm_spr_regs));
+	if (!regs) {
+		perror("malloc() failed");
+		return TEST_FAIL;
+	}
+
+	iov.iov_base = (u64 *) regs;
+	iov.iov_len = sizeof(struct tm_spr_regs);
+
+	ret = ptrace(PTRACE_GETREGSET, child, NT_PPC_TM_SPR, &iov);
+	if (ret) {
+		perror("ptrace(PTRACE_GETREGSET) failed");
+		return TEST_FAIL;
+	}
+
+	if (out)
+		memcpy(out, regs, sizeof(struct tm_spr_regs));
+
+	return TEST_PASS;
+}
+
+
+
+/* Analyse TEXASR after TM failure */
+inline unsigned long get_tfiar(void)
+{
+	unsigned long ret;
+
+	asm volatile("mfspr %0,%1" : "=r" (ret) : "i" (SPRN_TFIAR));
+	return ret;
+}
+
+void analyse_texasr(unsigned long texasr)
+{
+	printf("TEXASR: %16lx\t", texasr);
+
+	if (texasr & TEXASR_FP)
+		printf("TEXASR_FP  ");
+
+	if (texasr & TEXASR_DA)
+		printf("TEXASR_DA  ");
+
+	if (texasr & TEXASR_NO)
+		printf("TEXASR_NO  ");
+
+	if (texasr & TEXASR_FO)
+		printf("TEXASR_FO  ");
+
+	if (texasr & TEXASR_SIC)
+		printf("TEXASR_SIC  ");
+
+	if (texasr & TEXASR_NTC)
+		printf("TEXASR_NTC  ");
+
+	if (texasr & TEXASR_TC)
+		printf("TEXASR_TC  ");
+
+	if (texasr & TEXASR_TIC)
+		printf("TEXASR_TIC  ");
+
+	if (texasr & TEXASR_IC)
+		printf("TEXASR_IC  ");
+
+	if (texasr & TEXASR_IFC)
+		printf("TEXASR_IFC  ");
+
+	if (texasr & TEXASR_ABT)
+		printf("TEXASR_ABT  ");
+
+	if (texasr & TEXASR_SPD)
+		printf("TEXASR_SPD  ");
+
+	if (texasr & TEXASR_HV)
+		printf("TEXASR_HV  ");
+
+	if (texasr & TEXASR_PR)
+		printf("TEXASR_PR  ");
+
+	if (texasr & TEXASR_FS)
+		printf("TEXASR_FS  ");
+
+	if (texasr & TEXASR_TE)
+		printf("TEXASR_TE  ");
+
+	if (texasr & TEXASR_ROT)
+		printf("TEXASR_ROT  ");
+
+	printf("TFIAR :%lx\n", get_tfiar());
+}
+
+void store_gpr(unsigned long *addr);
+void store_fpr(float *addr);
diff --git a/tools/testing/selftests/powerpc/scripts/hmi.sh b/tools/testing/selftests/powerpc/scripts/hmi.sh
new file mode 100755
index 000000000..dcdb392e8
--- /dev/null
+++ b/tools/testing/selftests/powerpc/scripts/hmi.sh
@@ -0,0 +1,82 @@
+#!/bin/sh
+# SPDX-License-Identifier: GPL-2.0-only
+#
+# Copyright 2015, Daniel Axtens, IBM Corporation
+#
+
+
+# do we have ./getscom, ./putscom?
+if [ -x ./getscom ] && [ -x ./putscom ]; then
+	GETSCOM=./getscom
+	PUTSCOM=./putscom
+elif which getscom > /dev/null; then
+	GETSCOM=$(which getscom)
+	PUTSCOM=$(which putscom)
+else
+	cat <<EOF
+Can't find getscom/putscom in . or \$PATH.
+See https://github.com/open-power/skiboot.
+The tool is in external/xscom-utils
+EOF
+	exit 1
+fi
+
+# We will get 8 HMI events per injection
+# todo: deal with things being offline
+expected_hmis=8
+COUNT_HMIS() {
+    dmesg | grep -c 'Harmless Hypervisor Maintenance interrupt'
+}
+
+# massively expand snooze delay, allowing injection on all cores
+ppc64_cpu --smt-snooze-delay=1000000000
+
+# when we exit, restore it
+trap "ppc64_cpu --smt-snooze-delay=100" 0 1
+
+# for each chip+core combination
+# todo - less fragile parsing
+egrep -o 'OCC: Chip [0-9a-f]+ Core [0-9a-f]' < /sys/firmware/opal/msglog |
+while read chipcore; do
+	chip=$(echo "$chipcore"|awk '{print $3}')
+	core=$(echo "$chipcore"|awk '{print $5}')
+	fir="0x1${core}013100"
+
+	# verify that Core FIR is zero as expected
+	if [ "$($GETSCOM -c 0x${chip} $fir)" != 0 ]; then
+		echo "FIR was not zero before injection for chip $chip, core $core. Aborting!"
+		echo "Result of $GETSCOM -c 0x${chip} $fir:"
+		$GETSCOM -c 0x${chip} $fir
+		echo "If you get a -5 error, the core may be in idle state. Try stress-ng."
+		echo "Otherwise, try $PUTSCOM -c 0x${chip} $fir 0"
+		exit 1
+	fi
+
+	# keep track of the number of HMIs handled
+	old_hmis=$(COUNT_HMIS)
+
+	# do injection, adding a marker to dmesg for clarity
+	echo "Injecting HMI on core $core, chip $chip" | tee /dev/kmsg
+	# inject a RegFile recoverable error
+	if ! $PUTSCOM -c 0x${chip} $fir 2000000000000000 > /dev/null; then
+		echo "Error injecting. Aborting!"
+		exit 1
+	fi
+
+	# now we want to wait for all the HMIs to be processed
+	# we expect one per thread on the core
+	i=0;
+	new_hmis=$(COUNT_HMIS)
+	while [ $new_hmis -lt $((old_hmis + expected_hmis)) ] && [ $i -lt 12 ]; do
+	    echo "Seen $((new_hmis - old_hmis)) HMI(s) out of $expected_hmis expected, sleeping"
+	    sleep 5;
+	    i=$((i + 1))
+	    new_hmis=$(COUNT_HMIS)
+	done
+	if [ $i = 12 ]; then
+	    echo "Haven't seen expected $expected_hmis recoveries after 1 min. Aborting."
+	    exit 1
+	fi
+	echo "Processed $expected_hmis events; presumed success. Check dmesg."
+	echo ""
+done
diff --git a/tools/testing/selftests/powerpc/security/.gitignore b/tools/testing/selftests/powerpc/security/.gitignore
new file mode 100644
index 000000000..4257a1f15
--- /dev/null
+++ b/tools/testing/selftests/powerpc/security/.gitignore
@@ -0,0 +1,3 @@
+# SPDX-License-Identifier: GPL-2.0-only
+rfi_flush
+entry_flush
diff --git a/tools/testing/selftests/powerpc/security/Makefile b/tools/testing/selftests/powerpc/security/Makefile
new file mode 100644
index 000000000..f25e854fe
--- /dev/null
+++ b/tools/testing/selftests/powerpc/security/Makefile
@@ -0,0 +1,15 @@
+# SPDX-License-Identifier: GPL-2.0+
+
+TEST_GEN_PROGS := rfi_flush entry_flush spectre_v2
+top_srcdir = ../../../../..
+
+CFLAGS += -I../../../../../usr/include
+
+include ../../lib.mk
+
+$(TEST_GEN_PROGS): ../harness.c ../utils.c
+
+$(OUTPUT)/spectre_v2: CFLAGS += -m64
+$(OUTPUT)/spectre_v2: ../pmu/event.c branch_loops.S
+$(OUTPUT)/rfi_flush: flush_utils.c
+$(OUTPUT)/entry_flush: flush_utils.c
diff --git a/tools/testing/selftests/powerpc/security/branch_loops.S b/tools/testing/selftests/powerpc/security/branch_loops.S
new file mode 100644
index 000000000..22e9204e3
--- /dev/null
+++ b/tools/testing/selftests/powerpc/security/branch_loops.S
@@ -0,0 +1,82 @@
+// SPDX-License-Identifier: GPL-2.0+
+
+/*
+ * Copyright 2019, Michael Ellerman, IBM Corp.
+ */
+
+#include <ppc-asm.h>
+
+	.data
+
+jump_table:
+	.long	0x0
+	.long	(.Lstate_1 - .Lstate_0)
+	.long	(.Lstate_2 - .Lstate_0)
+	.long	(.Lstate_3 - .Lstate_0)
+	.long	(.Lstate_4 - .Lstate_0)
+	.long	(.Lstate_5 - .Lstate_0)
+	.long	(.Lstate_6 - .Lstate_0)
+	.long	(.Lstate_7 - .Lstate_0)
+
+	.text
+
+#define ITER_SHIFT	31
+
+.macro state number
+	.balign	32
+.Lstate_\number:
+	.if	\number==7
+	li	r3, 0
+	.else
+	li	r3, \number+1
+	.endif
+	b	.Lloop
+.endm
+
+FUNC_START(pattern_cache_loop)
+	li	r3, 0
+	li	r4, 1
+	sldi	r4, r4, ITER_SHIFT
+
+.Lloop:	cmpdi	r4, 0
+	beqlr
+
+	addi	r4, r4, -1
+
+	ld	r6, jump_table@got(%r2)
+	sldi	r5, r3, 2
+	lwax	r6, r5, r6
+	ld	r7, .Lstate_0@got(%r2)
+	add	r6, r6, r7
+	mtctr	r6
+	bctr
+
+	state	0
+	state	1
+	state	2
+	state	3
+	state	4
+	state	5
+	state	6
+	state	7
+
+FUNC_END(pattern_cache_loop)
+
+
+FUNC_START(indirect_branch_loop)
+	li	r3, 1
+	sldi	r3, r3, ITER_SHIFT
+
+1:	cmpdi	r3, 0
+	beqlr
+
+	addi	r3, r3, -1
+
+	ld	r4, 2f@got(%r2)
+	mtctr	r4
+	bctr
+
+	.balign 32
+2:	b	1b
+
+FUNC_END(indirect_branch_loop)
diff --git a/tools/testing/selftests/powerpc/security/entry_flush.c b/tools/testing/selftests/powerpc/security/entry_flush.c
new file mode 100644
index 000000000..68ce377b2
--- /dev/null
+++ b/tools/testing/selftests/powerpc/security/entry_flush.c
@@ -0,0 +1,139 @@
+// SPDX-License-Identifier: GPL-2.0+
+
+/*
+ * Copyright 2018 IBM Corporation.
+ */
+
+#define __SANE_USERSPACE_TYPES__
+
+#include <sys/types.h>
+#include <stdint.h>
+#include <malloc.h>
+#include <unistd.h>
+#include <signal.h>
+#include <stdlib.h>
+#include <string.h>
+#include <stdio.h>
+#include "utils.h"
+#include "flush_utils.h"
+
+int entry_flush_test(void)
+{
+	char *p;
+	int repetitions = 10;
+	int fd, passes = 0, iter, rc = 0;
+	struct perf_event_read v;
+	__u64 l1d_misses_total = 0;
+	unsigned long iterations = 100000, zero_size = 24 * 1024;
+	unsigned long l1d_misses_expected;
+	int rfi_flush_orig;
+	int entry_flush, entry_flush_orig;
+
+	SKIP_IF(geteuid() != 0);
+
+	// The PMU event we use only works on Power7 or later
+	SKIP_IF(!have_hwcap(PPC_FEATURE_ARCH_2_06));
+
+	if (read_debugfs_file("powerpc/rfi_flush", &rfi_flush_orig) < 0) {
+		perror("Unable to read powerpc/rfi_flush debugfs file");
+		SKIP_IF(1);
+	}
+
+	if (read_debugfs_file("powerpc/entry_flush", &entry_flush_orig) < 0) {
+		perror("Unable to read powerpc/entry_flush debugfs file");
+		SKIP_IF(1);
+	}
+
+	if (rfi_flush_orig != 0) {
+		if (write_debugfs_file("powerpc/rfi_flush", 0) < 0) {
+			perror("error writing to powerpc/rfi_flush debugfs file");
+			FAIL_IF(1);
+		}
+	}
+
+	entry_flush = entry_flush_orig;
+
+	fd = perf_event_open_counter(PERF_TYPE_HW_CACHE, PERF_L1D_READ_MISS_CONFIG, -1);
+	FAIL_IF(fd < 0);
+
+	p = (char *)memalign(zero_size, CACHELINE_SIZE);
+
+	FAIL_IF(perf_event_enable(fd));
+
+	// disable L1 prefetching
+	set_dscr(1);
+
+	iter = repetitions;
+
+	/*
+	 * We expect to see l1d miss for each cacheline access when entry_flush
+	 * is set. Allow a small variation on this.
+	 */
+	l1d_misses_expected = iterations * (zero_size / CACHELINE_SIZE - 2);
+
+again:
+	FAIL_IF(perf_event_reset(fd));
+
+	syscall_loop(p, iterations, zero_size);
+
+	FAIL_IF(read(fd, &v, sizeof(v)) != sizeof(v));
+
+	if (entry_flush && v.l1d_misses >= l1d_misses_expected)
+		passes++;
+	else if (!entry_flush && v.l1d_misses < (l1d_misses_expected / 2))
+		passes++;
+
+	l1d_misses_total += v.l1d_misses;
+
+	while (--iter)
+		goto again;
+
+	if (passes < repetitions) {
+		printf("FAIL (L1D misses with entry_flush=%d: %llu %c %lu) [%d/%d failures]\n",
+		       entry_flush, l1d_misses_total, entry_flush ? '<' : '>',
+		       entry_flush ? repetitions * l1d_misses_expected :
+		       repetitions * l1d_misses_expected / 2,
+		       repetitions - passes, repetitions);
+		rc = 1;
+	} else {
+		printf("PASS (L1D misses with entry_flush=%d: %llu %c %lu) [%d/%d pass]\n",
+		       entry_flush, l1d_misses_total, entry_flush ? '>' : '<',
+		       entry_flush ? repetitions * l1d_misses_expected :
+		       repetitions * l1d_misses_expected / 2,
+		       passes, repetitions);
+	}
+
+	if (entry_flush == entry_flush_orig) {
+		entry_flush = !entry_flush_orig;
+		if (write_debugfs_file("powerpc/entry_flush", entry_flush) < 0) {
+			perror("error writing to powerpc/entry_flush debugfs file");
+			return 1;
+		}
+		iter = repetitions;
+		l1d_misses_total = 0;
+		passes = 0;
+		goto again;
+	}
+
+	perf_event_disable(fd);
+	close(fd);
+
+	set_dscr(0);
+
+	if (write_debugfs_file("powerpc/rfi_flush", rfi_flush_orig) < 0) {
+		perror("unable to restore original value of powerpc/rfi_flush debugfs file");
+		return 1;
+	}
+
+	if (write_debugfs_file("powerpc/entry_flush", entry_flush_orig) < 0) {
+		perror("unable to restore original value of powerpc/entry_flush debugfs file");
+		return 1;
+	}
+
+	return rc;
+}
+
+int main(int argc, char *argv[])
+{
+	return test_harness(entry_flush_test, "entry_flush_test");
+}
diff --git a/tools/testing/selftests/powerpc/security/flush_utils.c b/tools/testing/selftests/powerpc/security/flush_utils.c
new file mode 100644
index 000000000..0c3c4c40c
--- /dev/null
+++ b/tools/testing/selftests/powerpc/security/flush_utils.c
@@ -0,0 +1,70 @@
+// SPDX-License-Identifier: GPL-2.0+
+
+/*
+ * Copyright 2018 IBM Corporation.
+ */
+
+#define __SANE_USERSPACE_TYPES__
+
+#include <sys/types.h>
+#include <stdint.h>
+#include <unistd.h>
+#include <signal.h>
+#include <stdlib.h>
+#include <string.h>
+#include <stdio.h>
+#include "utils.h"
+#include "flush_utils.h"
+
+static inline __u64 load(void *addr)
+{
+	__u64 tmp;
+
+	asm volatile("ld %0,0(%1)" : "=r"(tmp) : "b"(addr));
+
+	return tmp;
+}
+
+void syscall_loop(char *p, unsigned long iterations,
+		  unsigned long zero_size)
+{
+	for (unsigned long i = 0; i < iterations; i++) {
+		for (unsigned long j = 0; j < zero_size; j += CACHELINE_SIZE)
+			load(p + j);
+		getppid();
+	}
+}
+
+static void sigill_handler(int signr, siginfo_t *info, void *unused)
+{
+	static int warned;
+	ucontext_t *ctx = (ucontext_t *)unused;
+	unsigned long *pc = &UCONTEXT_NIA(ctx);
+
+	/* mtspr 3,RS to check for move to DSCR below */
+	if ((*((unsigned int *)*pc) & 0xfc1fffff) == 0x7c0303a6) {
+		if (!warned++)
+			printf("WARNING: Skipping over dscr setup. Consider running 'ppc64_cpu --dscr=1' manually.\n");
+		*pc += 4;
+	} else {
+		printf("SIGILL at %p\n", pc);
+		abort();
+	}
+}
+
+void set_dscr(unsigned long val)
+{
+	static int init;
+	struct sigaction sa;
+
+	if (!init) {
+		memset(&sa, 0, sizeof(sa));
+		sa.sa_sigaction = sigill_handler;
+		sa.sa_flags = SA_SIGINFO;
+		if (sigaction(SIGILL, &sa, NULL))
+			perror("sigill_handler");
+		init = 1;
+	}
+
+	asm volatile("mtspr %1,%0" : : "r" (val), "i" (SPRN_DSCR));
+}
diff --git a/tools/testing/selftests/powerpc/security/flush_utils.h b/tools/testing/selftests/powerpc/security/flush_utils.h
new file mode 100644
index 000000000..7a3d60292
--- /dev/null
+++ b/tools/testing/selftests/powerpc/security/flush_utils.h
@@ -0,0 +1,21 @@
+/* SPDX-License-Identifier: GPL-2.0+ */
+
+/*
+ * Copyright 2018 IBM Corporation.
+ */
+
+#ifndef _SELFTESTS_POWERPC_SECURITY_FLUSH_UTILS_H
+#define _SELFTESTS_POWERPC_SECURITY_FLUSH_UTILS_H
+
+#define CACHELINE_SIZE 128
+
+#define PERF_L1D_READ_MISS_CONFIG	((PERF_COUNT_HW_CACHE_L1D) | 		\
+					(PERF_COUNT_HW_CACHE_OP_READ << 8) |	\
+					(PERF_COUNT_HW_CACHE_RESULT_MISS << 16))
+
+void syscall_loop(char *p, unsigned long iterations,
+		  unsigned long zero_size);
+
+void set_dscr(unsigned long val);
+
+#endif /* _SELFTESTS_POWERPC_SECURITY_FLUSH_UTILS_H */
diff --git a/tools/testing/selftests/powerpc/security/rfi_flush.c b/tools/testing/selftests/powerpc/security/rfi_flush.c
new file mode 100644
index 000000000..f73484a64
--- /dev/null
+++ b/tools/testing/selftests/powerpc/security/rfi_flush.c
@@ -0,0 +1,142 @@
+// SPDX-License-Identifier: GPL-2.0+
+
+/*
+ * Copyright 2018 IBM Corporation.
+ */
+
+#define __SANE_USERSPACE_TYPES__
+
+#include <sys/types.h>
+#include <stdint.h>
+#include <malloc.h>
+#include <unistd.h>
+#include <stdlib.h>
+#include <string.h>
+#include <stdio.h>
+#include "utils.h"
+#include "flush_utils.h"
+
+
+int rfi_flush_test(void)
+{
+	char *p;
+	int repetitions = 10;
+	int fd, passes = 0, iter, rc = 0;
+	struct perf_event_read v;
+	__u64 l1d_misses_total = 0;
+	unsigned long iterations = 100000, zero_size = 24 * 1024;
+	unsigned long l1d_misses_expected;
+	int rfi_flush_orig, rfi_flush;
+	int have_entry_flush, entry_flush_orig;
+
+	SKIP_IF(geteuid() != 0);
+
+	// The PMU event we use only works on Power7 or later
+	SKIP_IF(!have_hwcap(PPC_FEATURE_ARCH_2_06));
+
+	if (read_debugfs_file("powerpc/rfi_flush", &rfi_flush_orig) < 0) {
+		perror("Unable to read powerpc/rfi_flush debugfs file");
+		SKIP_IF(1);
+	}
+
+	if (read_debugfs_file("powerpc/entry_flush", &entry_flush_orig) < 0) {
+		have_entry_flush = 0;
+	} else {
+		have_entry_flush = 1;
+
+		if (entry_flush_orig != 0) {
+			if (write_debugfs_file("powerpc/entry_flush", 0) < 0) {
+				perror("error writing to powerpc/entry_flush debugfs file");
+				return 1;
+			}
+		}
+	}
+
+	rfi_flush = rfi_flush_orig;
+
+	fd = perf_event_open_counter(PERF_TYPE_HW_CACHE, PERF_L1D_READ_MISS_CONFIG, -1);
+	FAIL_IF(fd < 0);
+
+	p = (char *)memalign(zero_size, CACHELINE_SIZE);
+
+	FAIL_IF(perf_event_enable(fd));
+
+	// disable L1 prefetching
+	set_dscr(1);
+
+	iter = repetitions;
+
+	/*
+	 * We expect to see l1d miss for each cacheline access when rfi_flush
+	 * is set. Allow a small variation on this.
+	 */
+	l1d_misses_expected = iterations * (zero_size / CACHELINE_SIZE - 2);
+
+again:
+	FAIL_IF(perf_event_reset(fd));
+
+	syscall_loop(p, iterations, zero_size);
+
+	FAIL_IF(read(fd, &v, sizeof(v)) != sizeof(v));
+
+	if (rfi_flush && v.l1d_misses >= l1d_misses_expected)
+		passes++;
+	else if (!rfi_flush && v.l1d_misses < (l1d_misses_expected / 2))
+		passes++;
+
+	l1d_misses_total += v.l1d_misses;
+
+	while (--iter)
+		goto again;
+
+	if (passes < repetitions) {
+		printf("FAIL (L1D misses with rfi_flush=%d: %llu %c %lu) [%d/%d failures]\n",
+		       rfi_flush, l1d_misses_total, rfi_flush ? '<' : '>',
+		       rfi_flush ? repetitions * l1d_misses_expected :
+		       repetitions * l1d_misses_expected / 2,
+		       repetitions - passes, repetitions);
+		rc = 1;
+	} else
+		printf("PASS (L1D misses with rfi_flush=%d: %llu %c %lu) [%d/%d pass]\n",
+		       rfi_flush, l1d_misses_total, rfi_flush ? '>' : '<',
+		       rfi_flush ? repetitions * l1d_misses_expected :
+		       repetitions * l1d_misses_expected / 2,
+		       passes, repetitions);
+
+	if (rfi_flush == rfi_flush_orig) {
+		rfi_flush = !rfi_flush_orig;
+		if (write_debugfs_file("powerpc/rfi_flush", rfi_flush) < 0) {
+			perror("error writing to powerpc/rfi_flush debugfs file");
+			return 1;
+		}
+		iter = repetitions;
+		l1d_misses_total = 0;
+		passes = 0;
+		goto again;
+	}
+
+	perf_event_disable(fd);
+	close(fd);
+
+	set_dscr(0);
+
+	if (write_debugfs_file("powerpc/rfi_flush", rfi_flush_orig) < 0) {
+		perror("unable to restore original value of powerpc/rfi_flush debugfs file");
+		return 1;
+	}
+
+	if (have_entry_flush) {
+		if (write_debugfs_file("powerpc/entry_flush", entry_flush_orig) < 0) {
+			perror("unable to restore original value of powerpc/entry_flush "
+			       "debugfs file");
+			return 1;
+		}
+	}
+
+	return rc;
+}
+
+int main(int argc, char *argv[])
+{
+	return test_harness(rfi_flush_test, "rfi_flush_test");
+}
diff --git a/tools/testing/selftests/powerpc/security/spectre_v2.c b/tools/testing/selftests/powerpc/security/spectre_v2.c
new file mode 100644
index 000000000..83647b827
--- /dev/null
+++ b/tools/testing/selftests/powerpc/security/spectre_v2.c
@@ -0,0 +1,231 @@
+// SPDX-License-Identifier: GPL-2.0+
+
+/*
+ * Copyright 2018-2019 IBM Corporation.
+ */
+
+#define __SANE_USERSPACE_TYPES__
+
+#include <sys/types.h>
+#include <stdint.h>
+#include <malloc.h>
+#include <unistd.h>
+#include <stdlib.h>
+#include <string.h>
+#include <stdio.h>
+#include <sys/prctl.h>
+#include "utils.h"
+
+#include "../pmu/event.h"
+
+
+extern void pattern_cache_loop(void);
+extern void indirect_branch_loop(void);
+
+static int do_count_loop(struct event *events, bool is_p9, s64 *miss_percent)
+{
+	u64 pred, mpred;
+
+	prctl(PR_TASK_PERF_EVENTS_ENABLE);
+
+	if (is_p9)
+		pattern_cache_loop();
+	else
+		indirect_branch_loop();
+
+	prctl(PR_TASK_PERF_EVENTS_DISABLE);
+
+	event_read(&events[0]);
+	event_read(&events[1]);
+
+	// We could scale all the events by running/enabled but we're lazy
+	// As long as the PMU is uncontended they should all run
+	FAIL_IF(events[0].result.running != events[0].result.enabled);
+	FAIL_IF(events[1].result.running != events[1].result.enabled);
+
+	pred =  events[0].result.value;
+	mpred = events[1].result.value;
+
+	if (is_p9) {
+		event_read(&events[2]);
+		event_read(&events[3]);
+		FAIL_IF(events[2].result.running != events[2].result.enabled);
+		FAIL_IF(events[3].result.running != events[3].result.enabled);
+
+		pred  += events[2].result.value;
+		mpred += events[3].result.value;
+	}
+
+	*miss_percent = 100 * mpred / pred;
+
+	return 0;
+}
+
+static void setup_event(struct event *e, u64 config, char *name)
+{
+	event_init_named(e, config, name);
+
+	e->attr.disabled = 1;
+	e->attr.exclude_kernel = 1;
+	e->attr.exclude_hv = 1;
+	e->attr.exclude_idle = 1;
+}
+
+enum spectre_v2_state {
+	VULNERABLE = 0,
+	UNKNOWN = 1,		// Works with FAIL_IF()
+	NOT_AFFECTED,
+	BRANCH_SERIALISATION,
+	COUNT_CACHE_DISABLED,
+	COUNT_CACHE_FLUSH_SW,
+	COUNT_CACHE_FLUSH_HW,
+	BTB_FLUSH,
+};
+
+static enum spectre_v2_state get_sysfs_state(void)
+{
+	enum spectre_v2_state state = UNKNOWN;
+	char buf[256];
+	int len;
+
+	memset(buf, 0, sizeof(buf));
+	FAIL_IF(read_sysfs_file("devices/system/cpu/vulnerabilities/spectre_v2", buf, sizeof(buf)));
+
+	// Make sure it's NULL terminated
+	buf[sizeof(buf) - 1] = '\0';
+
+	// Trim the trailing newline
+	len = strlen(buf);
+	FAIL_IF(len < 1);
+	buf[len - 1] = '\0';
+
+	printf("sysfs reports: '%s'\n", buf);
+
+	// Order matters
+	if (strstr(buf, "Vulnerable"))
+		state = VULNERABLE;
+	else if (strstr(buf, "Not affected"))
+		state = NOT_AFFECTED;
+	else if (strstr(buf, "Indirect branch serialisation (kernel only)"))
+		state = BRANCH_SERIALISATION;
+	else if (strstr(buf, "Indirect branch cache disabled"))
+		state = COUNT_CACHE_DISABLED;
+	else if (strstr(buf, "Software count cache flush (hardware accelerated)"))
+		state = COUNT_CACHE_FLUSH_HW;
+	else if (strstr(buf, "Software count cache flush"))
+		state = COUNT_CACHE_FLUSH_SW;
+	else if (strstr(buf, "Branch predictor state flush"))
+		state = BTB_FLUSH;
+
+	return state;
+}
+
+#define PM_BR_PRED_CCACHE	0x040a4	// P8 + P9
+#define PM_BR_MPRED_CCACHE	0x040ac	// P8 + P9
+#define PM_BR_PRED_PCACHE	0x048a0	// P9 only
+#define PM_BR_MPRED_PCACHE	0x048b0	// P9 only
+
+#define SPRN_PVR 287
+
+int spectre_v2_test(void)
+{
+	enum spectre_v2_state state;
+	struct event events[4];
+	s64 miss_percent;
+	bool is_p9;
+
+	// The PMU events we use only work on Power8 or later
+	SKIP_IF(!have_hwcap2(PPC_FEATURE2_ARCH_2_07));
+
+	state = get_sysfs_state();
+	if (state == UNKNOWN) {
+		printf("Error: couldn't determine spectre_v2 mitigation state?\n");
+		return -1;
+	}
+
+	memset(events, 0, sizeof(events));
+
+	setup_event(&events[0], PM_BR_PRED_CCACHE,  "PM_BR_PRED_CCACHE");
+	setup_event(&events[1], PM_BR_MPRED_CCACHE, "PM_BR_MPRED_CCACHE");
+	FAIL_IF(event_open(&events[0]));
+	FAIL_IF(event_open_with_group(&events[1], events[0].fd) == -1);
+
+	is_p9 = ((mfspr(SPRN_PVR) >>  16) & 0xFFFF) == 0x4e;
+
+	if (is_p9) {
+		// Count pattern cache too
+		setup_event(&events[2], PM_BR_PRED_PCACHE,  "PM_BR_PRED_PCACHE");
+		setup_event(&events[3], PM_BR_MPRED_PCACHE, "PM_BR_MPRED_PCACHE");
+
+		FAIL_IF(event_open_with_group(&events[2], events[0].fd) == -1);
+		FAIL_IF(event_open_with_group(&events[3], events[0].fd) == -1);
+	}
+
+	FAIL_IF(do_count_loop(events, is_p9, &miss_percent));
+
+	event_report_justified(&events[0], 18, 10);
+	event_report_justified(&events[1], 18, 10);
+	event_close(&events[0]);
+	event_close(&events[1]);
+
+	if (is_p9) {
+		event_report_justified(&events[2], 18, 10);
+		event_report_justified(&events[3], 18, 10);
+		event_close(&events[2]);
+		event_close(&events[3]);
+	}
+
+	printf("Miss percent %lld %%\n", miss_percent);
+
+	switch (state) {
+	case VULNERABLE:
+	case NOT_AFFECTED:
+	case COUNT_CACHE_FLUSH_SW:
+	case COUNT_CACHE_FLUSH_HW:
+		// These should all not affect userspace branch prediction
+		if (miss_percent > 15) {
+			printf("Branch misses > 15%% unexpected in this configuration!\n");
+			printf("Possible mis-match between reported & actual mitigation\n");
+			/*
+			 * Such a mismatch may be caused by a guest system
+			 * reporting as vulnerable when the host is mitigated.
+			 * Return skip code to avoid detecting this as an error.
+			 * We are not vulnerable and reporting otherwise, so
+			 * missing such a mismatch is safe.
+			 */
+			if (miss_percent > 95)
+				return 4;
+
+			return 1;
+		}
+		break;
+	case BRANCH_SERIALISATION:
+		// This seems to affect userspace branch prediction a bit?
+		if (miss_percent > 25) {
+			printf("Branch misses > 25%% unexpected in this configuration!\n");
+			printf("Possible mis-match between reported & actual mitigation\n");
+			return 1;
+		}
+		break;
+	case COUNT_CACHE_DISABLED:
+		if (miss_percent < 95) {
+			printf("Branch misses < 20%% unexpected in this configuration!\n");
+			printf("Possible mis-match between reported & actual mitigation\n");
+			return 1;
+		}
+		break;
+	case UNKNOWN:
+	case BTB_FLUSH:
+		printf("Not sure!\n");
+		return 1;
+	}
+
+	printf("OK - Measured branch prediction rates match reported spectre v2 mitigation.\n");
+
+	return 0;
+}
+
+int main(int argc, char *argv[])
+{
+	return test_harness(spectre_v2_test, "spectre_v2");
+}
diff --git a/tools/testing/selftests/powerpc/signal/.gitignore b/tools/testing/selftests/powerpc/signal/.gitignore
new file mode 100644
index 000000000..405b53640
--- /dev/null
+++ b/tools/testing/selftests/powerpc/signal/.gitignore
@@ -0,0 +1,5 @@
+# SPDX-License-Identifier: GPL-2.0-only
+signal
+signal_tm
+sigfuz
+sigreturn_vdso
diff --git a/tools/testing/selftests/powerpc/signal/Makefile b/tools/testing/selftests/powerpc/signal/Makefile
new file mode 100644
index 000000000..d6ae54663
--- /dev/null
+++ b/tools/testing/selftests/powerpc/signal/Makefile
@@ -0,0 +1,13 @@
+# SPDX-License-Identifier: GPL-2.0
+TEST_GEN_PROGS := signal signal_tm sigfuz sigreturn_vdso sig_sc_double_restart
+
+CFLAGS += -maltivec
+$(OUTPUT)/signal_tm: CFLAGS += -mhtm
+$(OUTPUT)/sigfuz: CFLAGS += -pthread -m64
+
+TEST_FILES := settings
+
+top_srcdir = ../../../../..
+include ../../lib.mk
+
+$(TEST_GEN_PROGS): ../harness.c ../utils.c signal.S
diff --git a/tools/testing/selftests/powerpc/signal/settings b/tools/testing/selftests/powerpc/signal/settings
new file mode 100644
index 000000000..e7b941753
--- /dev/null
+++ b/tools/testing/selftests/powerpc/signal/settings
@@ -0,0 +1 @@
+timeout=0
diff --git a/tools/testing/selftests/powerpc/signal/sig_sc_double_restart.c b/tools/testing/selftests/powerpc/signal/sig_sc_double_restart.c
new file mode 100644
index 000000000..e39722646
--- /dev/null
+++ b/tools/testing/selftests/powerpc/signal/sig_sc_double_restart.c
@@ -0,0 +1,174 @@
+// SPDX-License-Identifier: GPL-2.0-or-later
+/*
+ * Test that a syscall does not get restarted twice, handled by trap_norestart()
+ *
+ * Based on Al's description, and a test for the bug fixed in this commit:
+ *
+ * commit 9a81c16b527528ad307843be5571111aa8d35a80
+ * Author: Al Viro <viro@zeniv.linux.org.uk>
+ * Date:   Mon Sep 20 21:48:57 2010 +0100
+ *
+ *  powerpc: fix double syscall restarts
+ *
+ *  Make sigreturn zero regs->trap, make do_signal() do the same on all
+ *  paths.  As it is, signal interrupting e.g. read() from fd 512 (==
+ *  ERESTARTSYS) with another signal getting unblocked when the first
+ *  handler finishes will lead to restart one insn earlier than it ought
+ *  to.  Same for multiple signals with in-kernel handlers interrupting
+ *  that sucker at the same time.  Same for multiple signals of any kind
+ *  interrupting that sucker on 64bit...
+ */
+#define _GNU_SOURCE
+#include <sys/types.h>
+#include <sys/wait.h>
+#include <sys/syscall.h>
+#include <unistd.h>
+#include <signal.h>
+#include <errno.h>
+#include <stdlib.h>
+#include <stdio.h>
+#include <string.h>
+
+#include "utils.h"
+
+static void SIGUSR1_handler(int sig)
+{
+	kill(getpid(), SIGUSR2);
+	/*
+	 * SIGUSR2 is blocked until the handler exits, at which point it will
+	 * be raised again and think there is a restart to be done because the
+	 * pending restarted syscall has 512 (ERESTARTSYS) in r3. The second
+	 * restart will retreat NIP another 4 bytes to fail case branch.
+	 */
+}
+
+static void SIGUSR2_handler(int sig)
+{
+}
+
+static ssize_t raw_read(int fd, void *buf, size_t count)
+{
+	register long nr asm("r0") = __NR_read;
+	register long _fd asm("r3") = fd;
+	register void *_buf asm("r4") = buf;
+	register size_t _count asm("r5") = count;
+
+	asm volatile(
+"		b	0f		\n"
+"		b	1f		\n"
+"	0:	sc	0		\n"
+"		bns	2f		\n"
+"		neg	%0,%0		\n"
+"		b	2f		\n"
+"	1:				\n"
+"		li	%0,%4		\n"
+"	2:				\n"
+		: "+r"(_fd), "+r"(nr), "+r"(_buf), "+r"(_count)
+		: "i"(-ENOANO)
+		: "memory", "r6", "r7", "r8", "r9", "r10", "r11", "r12", "ctr", "cr0");
+
+	if (_fd < 0) {
+		errno = -_fd;
+		_fd = -1;
+	}
+
+	return _fd;
+}
+
+#define DATA "test 123"
+#define DLEN (strlen(DATA)+1)
+
+int test_restart(void)
+{
+	int pipefd[2];
+	pid_t pid;
+	char buf[512];
+
+	if (pipe(pipefd) == -1) {
+		perror("pipe");
+		exit(EXIT_FAILURE);
+	}
+
+	pid = fork();
+	if (pid == -1) {
+		perror("fork");
+		exit(EXIT_FAILURE);
+	}
+
+	if (pid == 0) { /* Child reads from pipe */
+		struct sigaction act;
+		int fd;
+
+		memset(&act, 0, sizeof(act));
+		sigaddset(&act.sa_mask, SIGUSR2);
+		act.sa_handler = SIGUSR1_handler;
+		act.sa_flags = SA_RESTART;
+		if (sigaction(SIGUSR1, &act, NULL) == -1) {
+			perror("sigaction");
+			exit(EXIT_FAILURE);
+		}
+
+		memset(&act, 0, sizeof(act));
+		act.sa_handler = SIGUSR2_handler;
+		act.sa_flags = SA_RESTART;
+		if (sigaction(SIGUSR2, &act, NULL) == -1) {
+			perror("sigaction");
+			exit(EXIT_FAILURE);
+		}
+
+		/* Let's get ERESTARTSYS into r3 */
+		while ((fd = dup(pipefd[0])) != 512) {
+			if (fd == -1) {
+				perror("dup");
+				exit(EXIT_FAILURE);
+			}
+		}
+
+		if (raw_read(fd, buf, 512) == -1) {
+			if (errno == ENOANO) {
+				fprintf(stderr, "Double restart moved restart before sc instruction.\n");
+				_exit(EXIT_FAILURE);
+			}
+			perror("read");
+			exit(EXIT_FAILURE);
+		}
+
+		if (strncmp(buf, DATA, DLEN)) {
+			fprintf(stderr, "bad test string %s\n", buf);
+			exit(EXIT_FAILURE);
+		}
+
+		return 0;
+
+	} else {
+		int wstatus;
+
+		usleep(100000);		/* Hack to get reader waiting */
+		kill(pid, SIGUSR1);
+		usleep(100000);
+		if (write(pipefd[1], DATA, DLEN) != DLEN) {
+			perror("write");
+			exit(EXIT_FAILURE);
+		}
+		close(pipefd[0]);
+		close(pipefd[1]);
+		if (wait(&wstatus) == -1) {
+			perror("wait");
+			exit(EXIT_FAILURE);
+		}
+		if (!WIFEXITED(wstatus)) {
+			fprintf(stderr, "child exited abnormally\n");
+			exit(EXIT_FAILURE);
+		}
+
+		FAIL_IF(WEXITSTATUS(wstatus) != EXIT_SUCCESS);
+
+		return 0;
+	}
+}
+
+int main(void)
+{
+	test_harness_set_timeout(10);
+	return test_harness(test_restart, "sig sys restart");
+}
diff --git a/tools/testing/selftests/powerpc/signal/sigfuz.c b/tools/testing/selftests/powerpc/signal/sigfuz.c
new file mode 100644
index 000000000..08f9afe3b
--- /dev/null
+++ b/tools/testing/selftests/powerpc/signal/sigfuz.c
@@ -0,0 +1,325 @@
+// SPDX-License-Identifier: GPL-2.0
+/*
+ * Copyright 2018, Breno Leitao, IBM Corp.
+ * Licensed under GPLv2.
+ *
+ * Sigfuz(tm): A PowerPC TM-aware signal fuzzer.
+ *
+ * This is a new selftest that raises SIGUSR1 signals and handles it in a set
+ * of different ways, trying to create different scenario for testing
+ * purpose.
+ *
+ * This test works raising a signal and calling sigreturn interleaved with
+ * TM operations, as starting, suspending and terminating a transaction. The
+ * test depends on random numbers, and, based on them, it sets different TM
+ * states.
+ *
+ * Other than that, the test fills out the user context struct that is passed
+ * to the sigreturn system call with random data, in order to make sure that
+ * the signal handler syscall can handle different and invalid states
+ * properly.
+ *
+ * This selftest has command line parameters to control what kind of tests the
+ * user wants to run, as for example, if a transaction should be started prior
+ * to signal being raised, or, after the signal being raised and before the
+ * sigreturn. If no parameter is given, the default is enabling all options.
+ *
+ * This test does not check if the user context is being read and set
+ * properly by the kernel. Its purpose, at this time, is basically
+ * guaranteeing that the kernel does not crash on invalid scenarios.
+ */
+
+#include <stdio.h>
+#include <limits.h>
+#include <sys/wait.h>
+#include <unistd.h>
+#include <stdlib.h>
+#include <signal.h>
+#include <string.h>
+#include <ucontext.h>
+#include <sys/mman.h>
+#include <pthread.h>
+#include "utils.h"
+
+/* Selftest defaults */
+#define COUNT_MAX	600		/* Number of interactions */
+#define THREADS		16		/* Number of threads */
+
+/* Arguments options */
+#define ARG_MESS_WITH_TM_AT	0x1
+#define ARG_MESS_WITH_TM_BEFORE	0x2
+#define ARG_MESS_WITH_MSR_AT	0x4
+#define ARG_FOREVER		0x10
+#define ARG_COMPLETE		(ARG_MESS_WITH_TM_AT |		\
+				ARG_MESS_WITH_TM_BEFORE |	\
+				ARG_MESS_WITH_MSR_AT)
+
+static int args;
+static int nthread = THREADS;
+static int count_max = COUNT_MAX;
+
+/* checkpoint context */
+static ucontext_t *tmp_uc;
+
+/* Return true with 1/x probability */
+static int one_in_chance(int x)
+{
+	return rand() % x == 0;
+}
+
+/* Change TM states */
+static void mess_with_tm(void)
+{
+	/* Starts a transaction 33% of the time */
+	if (one_in_chance(3)) {
+		asm ("tbegin.	;"
+		     "beq 8	;");
+
+		/* And suspended half of them */
+		if (one_in_chance(2))
+			asm("tsuspend.	;");
+	}
+
+	/* Call 'tend' in 5% of the runs */
+	if (one_in_chance(20))
+		asm("tend.	;");
+}
+
+/* Signal handler that will be invoked with raise() */
+static void trap_signal_handler(int signo, siginfo_t *si, void *uc)
+{
+	ucontext_t *ucp = uc;
+
+	ucp->uc_link = tmp_uc;
+
+	/*
+	 * Set uc_link in three possible ways:
+	 *  - Setting a single 'int' in the whole chunk
+	 *  - Cloning ucp into uc_link
+	 *  - Allocating a new memory chunk
+	 */
+	if (one_in_chance(3)) {
+		memset(ucp->uc_link, rand(), sizeof(ucontext_t));
+	} else if (one_in_chance(2)) {
+		memcpy(ucp->uc_link, uc, sizeof(ucontext_t));
+	} else if (one_in_chance(2)) {
+		if (tmp_uc) {
+			free(tmp_uc);
+			tmp_uc = NULL;
+		}
+		tmp_uc = malloc(sizeof(ucontext_t));
+		ucp->uc_link = tmp_uc;
+		/* Trying to cause a major page fault at Kernel level */
+		madvise(ucp->uc_link, sizeof(ucontext_t), MADV_DONTNEED);
+	}
+
+	if (args & ARG_MESS_WITH_MSR_AT) {
+		/* Changing the checkpointed registers */
+		if (one_in_chance(4)) {
+			ucp->uc_link->uc_mcontext.gp_regs[PT_MSR] |= MSR_TS_S;
+		} else {
+			if (one_in_chance(2)) {
+				ucp->uc_link->uc_mcontext.gp_regs[PT_MSR] |=
+						 MSR_TS_T;
+			} else if (one_in_chance(2)) {
+				ucp->uc_link->uc_mcontext.gp_regs[PT_MSR] |=
+						MSR_TS_T | MSR_TS_S;
+			}
+		}
+
+		/* Checking the current register context */
+		if (one_in_chance(2)) {
+			ucp->uc_mcontext.gp_regs[PT_MSR] |= MSR_TS_S;
+		} else if (one_in_chance(2)) {
+			if (one_in_chance(2))
+				ucp->uc_mcontext.gp_regs[PT_MSR] |=
+					MSR_TS_T;
+			else if (one_in_chance(2))
+				ucp->uc_mcontext.gp_regs[PT_MSR] |=
+					MSR_TS_T | MSR_TS_S;
+		}
+	}
+
+	if (one_in_chance(20)) {
+		/* Nested transaction start */
+		if (one_in_chance(5))
+			mess_with_tm();
+
+		/* Return without changing any other context info */
+		return;
+	}
+
+	if (one_in_chance(10))
+		ucp->uc_mcontext.gp_regs[PT_MSR] = random();
+	if (one_in_chance(10))
+		ucp->uc_mcontext.gp_regs[PT_NIP] = random();
+	if (one_in_chance(10))
+		ucp->uc_link->uc_mcontext.gp_regs[PT_MSR] = random();
+	if (one_in_chance(10))
+		ucp->uc_link->uc_mcontext.gp_regs[PT_NIP] = random();
+
+	ucp->uc_mcontext.gp_regs[PT_TRAP] = random();
+	ucp->uc_mcontext.gp_regs[PT_DSISR] = random();
+	ucp->uc_mcontext.gp_regs[PT_DAR] = random();
+	ucp->uc_mcontext.gp_regs[PT_ORIG_R3] = random();
+	ucp->uc_mcontext.gp_regs[PT_XER] = random();
+	ucp->uc_mcontext.gp_regs[PT_RESULT] = random();
+	ucp->uc_mcontext.gp_regs[PT_SOFTE] = random();
+	ucp->uc_mcontext.gp_regs[PT_DSCR] = random();
+	ucp->uc_mcontext.gp_regs[PT_CTR] = random();
+	ucp->uc_mcontext.gp_regs[PT_LNK] = random();
+	ucp->uc_mcontext.gp_regs[PT_CCR] = random();
+	ucp->uc_mcontext.gp_regs[PT_REGS_COUNT] = random();
+
+	ucp->uc_link->uc_mcontext.gp_regs[PT_TRAP] = random();
+	ucp->uc_link->uc_mcontext.gp_regs[PT_DSISR] = random();
+	ucp->uc_link->uc_mcontext.gp_regs[PT_DAR] = random();
+	ucp->uc_link->uc_mcontext.gp_regs[PT_ORIG_R3] = random();
+	ucp->uc_link->uc_mcontext.gp_regs[PT_XER] = random();
+	ucp->uc_link->uc_mcontext.gp_regs[PT_RESULT] = random();
+	ucp->uc_link->uc_mcontext.gp_regs[PT_SOFTE] = random();
+	ucp->uc_link->uc_mcontext.gp_regs[PT_DSCR] = random();
+	ucp->uc_link->uc_mcontext.gp_regs[PT_CTR] = random();
+	ucp->uc_link->uc_mcontext.gp_regs[PT_LNK] = random();
+	ucp->uc_link->uc_mcontext.gp_regs[PT_CCR] = random();
+	ucp->uc_link->uc_mcontext.gp_regs[PT_REGS_COUNT] = random();
+
+	if (args & ARG_MESS_WITH_TM_BEFORE) {
+		if (one_in_chance(2))
+			mess_with_tm();
+	}
+}
+
+static void seg_signal_handler(int signo, siginfo_t *si, void *uc)
+{
+	/* Clear exit for process that segfaults */
+	exit(0);
+}
+
+static void *sigfuz_test(void *thrid)
+{
+	struct sigaction trap_sa, seg_sa;
+	int ret, i = 0;
+	pid_t t;
+
+	tmp_uc = malloc(sizeof(ucontext_t));
+
+	/* Main signal handler */
+	trap_sa.sa_flags = SA_SIGINFO;
+	trap_sa.sa_sigaction = trap_signal_handler;
+
+	/* SIGSEGV signal handler */
+	seg_sa.sa_flags = SA_SIGINFO;
+	seg_sa.sa_sigaction = seg_signal_handler;
+
+	/* The signal handler will enable MSR_TS */
+	sigaction(SIGUSR1, &trap_sa, NULL);
+
+	/* If it does not crash, it will segfault, avoid it to retest */
+	sigaction(SIGSEGV, &seg_sa, NULL);
+
+	while (i < count_max) {
+		t = fork();
+
+		if (t == 0) {
+			/* Once seed per process */
+			srand(time(NULL) + getpid());
+			if (args & ARG_MESS_WITH_TM_AT) {
+				if (one_in_chance(2))
+					mess_with_tm();
+			}
+			raise(SIGUSR1);
+			exit(0);
+		} else {
+			waitpid(t, &ret, 0);
+		}
+		if (!(args & ARG_FOREVER))
+			i++;
+	}
+
+	/* If not freed already, free now */
+	if (tmp_uc) {
+		free(tmp_uc);
+		tmp_uc = NULL;
+	}
+
+	return NULL;
+}
+
+static int signal_fuzzer(void)
+{
+	int t, rc;
+	pthread_t *threads;
+
+	threads = malloc(nthread * sizeof(pthread_t));
+
+	for (t = 0; t < nthread; t++) {
+		rc = pthread_create(&threads[t], NULL, sigfuz_test,
+				    (void *)&t);
+		if (rc)
+			perror("Thread creation error\n");
+	}
+
+	for (t = 0; t < nthread; t++) {
+		rc = pthread_join(threads[t], NULL);
+		if (rc)
+			perror("Thread join error\n");
+	}
+
+	free(threads);
+
+	return EXIT_SUCCESS;
+}
+
+static void show_help(char *name)
+{
+	printf("%s: Sigfuzzer for powerpc\n", name);
+	printf("Usage:\n");
+	printf("\t-b\t Mess with TM before raising a SIGUSR1 signal\n");
+	printf("\t-a\t Mess with TM after raising a SIGUSR1 signal\n");
+	printf("\t-m\t Mess with MSR[TS] bits at mcontext\n");
+	printf("\t-x\t Mess with everything above\n");
+	printf("\t-f\t Run forever (Press ^C to Quit)\n");
+	printf("\t-i\t Amount of interactions.	(Default = %d)\n", COUNT_MAX);
+	printf("\t-t\t Amount of threads.	(Default = %d)\n", THREADS);
+	exit(-1);
+}
+
+int main(int argc, char **argv)
+{
+	int opt;
+
+	while ((opt = getopt(argc, argv, "bamxt:fi:h")) != -1) {
+		if (opt == 'b') {
+			printf("Mess with TM before signal\n");
+			args |= ARG_MESS_WITH_TM_BEFORE;
+		} else if (opt == 'a') {
+			printf("Mess with TM at signal handler\n");
+			args |= ARG_MESS_WITH_TM_AT;
+		} else if (opt == 'm') {
+			printf("Mess with MSR[TS] bits in mcontext\n");
+			args |= ARG_MESS_WITH_MSR_AT;
+		} else if (opt == 'x') {
+			printf("Running with all options enabled\n");
+			args |= ARG_COMPLETE;
+		} else if (opt == 't') {
+			nthread = atoi(optarg);
+			printf("Threads = %d\n", nthread);
+		} else if (opt == 'f') {
+			args |= ARG_FOREVER;
+			printf("Press ^C to stop\n");
+			test_harness_set_timeout(-1);
+		} else if (opt == 'i') {
+			count_max = atoi(optarg);
+			printf("Running for %d interactions\n", count_max);
+		} else if (opt == 'h') {
+			show_help(argv[0]);
+		}
+	}
+
+	/* Default test suite */
+	if (!args)
+		args = ARG_COMPLETE;
+
+	test_harness(signal_fuzzer, "signal_fuzzer");
+}
diff --git a/tools/testing/selftests/powerpc/signal/signal.S b/tools/testing/selftests/powerpc/signal/signal.S
new file mode 100644
index 000000000..228fba499
--- /dev/null
+++ b/tools/testing/selftests/powerpc/signal/signal.S
@@ -0,0 +1,46 @@
+/* SPDX-License-Identifier: GPL-2.0-or-later */
+/*
+ * Copyright 2015, Cyril Bur, IBM Corp.
+ */
+
+#include "basic_asm.h"
+
+/* long signal_self(pid_t pid, int sig); */
+FUNC_START(signal_self)
+	li	r0,37 /* sys_kill */
+	/* r3 already has our pid in it */
+	/* r4 already has signal type in it */
+	sc
+	bc	4,3,1f
+	subfze	r3,r3
+1:	blr
+FUNC_END(signal_self)
+
+/* long tm_signal_self(pid_t pid, int sig, int *ret); */
+FUNC_START(tm_signal_self)
+	PUSH_BASIC_STACK(8)
+	std	r5,STACK_FRAME_PARAM(0)(sp) /* ret */
+	tbegin.
+	beq	1f
+	tsuspend.
+	li	r0,37 /* sys_kill */
+	/* r3 already has our pid in it */
+	/* r4 already has signal type in it */
+	sc
+	ld	r5,STACK_FRAME_PARAM(0)(sp) /* ret */
+	bc	4,3,2f
+	subfze	r3,r3
+2:	std	r3,0(r5)
+	tabort. 0
+	tresume. /* Be nice to some cleanup, jumps back to tbegin then to 1: */
+	/*
+	 * Transaction should be proper doomed and we should never get
+	 * here
+	 */
+	li	r3,1
+	POP_BASIC_STACK(8)
+	blr
+1:	li	r3,0
+	POP_BASIC_STACK(8)
+	blr
+FUNC_END(tm_signal_self)
diff --git a/tools/testing/selftests/powerpc/signal/signal.c b/tools/testing/selftests/powerpc/signal/signal.c
new file mode 100644
index 000000000..766e484d9
--- /dev/null
+++ b/tools/testing/selftests/powerpc/signal/signal.c
@@ -0,0 +1,107 @@
+// SPDX-License-Identifier: GPL-2.0-or-later
+/*
+ * Copyright 2016, Cyril Bur, IBM Corp.
+ *
+ * Sending one self a signal should always get delivered.
+ */
+
+#include <signal.h>
+#include <stdio.h>
+#include <stdlib.h>
+#include <string.h>
+#include <sys/types.h>
+#include <sys/wait.h>
+#include <unistd.h>
+
+#include <altivec.h>
+
+#include "utils.h"
+
+#define MAX_ATTEMPT 500000
+#define TIMEOUT 5
+
+extern long signal_self(pid_t pid, int sig);
+
+static sig_atomic_t signaled;
+static sig_atomic_t fail;
+
+static void signal_handler(int sig)
+{
+	if (sig == SIGUSR1)
+		signaled = 1;
+	else
+		fail = 1;
+}
+
+static int test_signal()
+{
+	int i;
+	struct sigaction act;
+	pid_t ppid = getpid();
+	pid_t pid;
+
+	act.sa_handler = signal_handler;
+	act.sa_flags = 0;
+	sigemptyset(&act.sa_mask);
+	if (sigaction(SIGUSR1, &act, NULL) < 0) {
+		perror("sigaction SIGUSR1");
+		exit(1);
+	}
+	if (sigaction(SIGALRM, &act, NULL) < 0) {
+		perror("sigaction SIGALRM");
+		exit(1);
+	}
+
+	/* Don't do this for MAX_ATTEMPT, its simply too long */
+	for(i  = 0; i < 1000; i++) {
+		pid = fork();
+		if (pid == -1) {
+			perror("fork");
+			exit(1);
+		}
+		if (pid == 0) {
+			signal_self(ppid, SIGUSR1);
+			exit(1);
+		} else {
+			alarm(0); /* Disable any pending */
+			alarm(2);
+			while (!signaled && !fail)
+				asm volatile("": : :"memory");
+			if (!signaled) {
+				fprintf(stderr, "Didn't get signal from child\n");
+				FAIL_IF(1); /* For the line number */
+			}
+			/* Otherwise we'll loop too fast and fork() will eventually fail */
+			waitpid(pid, NULL, 0);
+		}
+	}
+
+	for (i = 0; i < MAX_ATTEMPT; i++) {
+		long rc;
+
+		alarm(0); /* Disable any pending */
+		signaled = 0;
+		alarm(TIMEOUT);
+		rc = signal_self(ppid, SIGUSR1);
+		if (rc) {
+			fprintf(stderr, "(%d) Fail reason: %d rc=0x%lx",
+					i, fail, rc);
+			FAIL_IF(1); /* For the line number */
+		}
+		while (!signaled && !fail)
+			asm volatile("": : :"memory");
+		if (!signaled) {
+			fprintf(stderr, "(%d) Fail reason: %d rc=0x%lx",
+					i, fail, rc);
+			FAIL_IF(1); /* For the line number */
+		}
+	}
+
+	return 0;
+}
+
+int main(void)
+{
+	test_harness_set_timeout(300);
+	return test_harness(test_signal, "signal");
+}
diff --git a/tools/testing/selftests/powerpc/signal/signal_tm.c b/tools/testing/selftests/powerpc/signal/signal_tm.c
new file mode 100644
index 000000000..5bf2224ef
--- /dev/null
+++ b/tools/testing/selftests/powerpc/signal/signal_tm.c
@@ -0,0 +1,106 @@
+// SPDX-License-Identifier: GPL-2.0-or-later
+/*
+ * Copyright 2016, Cyril Bur, IBM Corp.
+ *
+ * Sending one self a signal should always get delivered.
+ */
+
+#include <errno.h>
+#include <stdlib.h>
+#include <stdio.h>
+#include <string.h>
+#include <signal.h>
+#include <unistd.h>
+
+#include <altivec.h>
+
+#include "utils.h"
+#include "../tm/tm.h"
+
+#define MAX_ATTEMPT 500000
+#define TIMEOUT 10
+
+extern long tm_signal_self(pid_t pid, int sig, long *ret);
+
+static sig_atomic_t signaled;
+static sig_atomic_t fail;
+
+static void signal_handler(int sig)
+{
+	if (tcheck_active()) {
+		fail = 2;
+		return;
+	}
+
+	if (sig == SIGUSR1)
+		signaled = 1;
+	else
+		fail = 1;
+}
+
+static int test_signal_tm()
+{
+	int i;
+	struct sigaction act;
+
+	act.sa_handler = signal_handler;
+	act.sa_flags = 0;
+	sigemptyset(&act.sa_mask);
+	if (sigaction(SIGUSR1, &act, NULL) < 0) {
+		perror("sigaction SIGUSR1");
+		exit(1);
+	}
+	if (sigaction(SIGALRM, &act, NULL) < 0) {
+		perror("sigaction SIGALRM");
+		exit(1);
+	}
+
+	SKIP_IF(!have_htm());
+
+	for (i = 0; i < MAX_ATTEMPT; i++) {
+		/*
+		 * If anything bad happens in ASM and we fail to set ret
+		 * because *handwave* TM this will cause failure
+		 */
+		long ret = 0xdead;
+		long rc = 0xbeef;
+
+		alarm(0); /* Disable any pending */
+		signaled = 0;
+		alarm(TIMEOUT);
+		FAIL_IF(tcheck_transactional());
+		rc = tm_signal_self(getpid(), SIGUSR1, &ret);
+		if (ret == 0xdead)
+			/*
+			 * This basically means the transaction aborted before we
+			 * even got to the suspend... this is crazy but it
+			 * happens.
+			 * Yes this also means we might never make forward
+			 * progress... the alarm() will trip eventually...
+			 */
+			continue;
+
+		if (rc || ret) {
+			/* Ret is actually an errno */
+			printf("TEXASR 0x%016lx, TFIAR 0x%016lx\n",
+					__builtin_get_texasr(), __builtin_get_tfiar());
+			fprintf(stderr, "(%d) Fail reason: %d rc=0x%lx ret=0x%lx\n",
+					i, fail, rc, ret);
+			FAIL_IF(ret);
+		}
+		while(!signaled && !fail)
+			asm volatile("": : :"memory");
+		if (!signaled) {
+			fprintf(stderr, "(%d) Fail reason: %d rc=0x%lx ret=0x%lx\n",
+					i, fail, rc, ret);
+			FAIL_IF(fail); /* For the line number */
+		}
+	}
+
+	return 0;
+}
+
+int main(void)
+{
+	return test_harness(test_signal_tm, "signal_tm");
+}
diff --git a/tools/testing/selftests/powerpc/signal/sigreturn_vdso.c b/tools/testing/selftests/powerpc/signal/sigreturn_vdso.c
new file mode 100644
index 000000000..e282fff0f
--- /dev/null
+++ b/tools/testing/selftests/powerpc/signal/sigreturn_vdso.c
@@ -0,0 +1,127 @@
+// SPDX-License-Identifier: GPL-2.0
+/*
+ * Test that we can take signals with and without the VDSO mapped, which trigger
+ * different paths in the signal handling code.
+ *
+ * See handle_rt_signal64() and setup_trampoline() in signal_64.c
+ */
+
+#define _GNU_SOURCE
+
+#include <errno.h>
+#include <stdio.h>
+#include <signal.h>
+#include <stdlib.h>
+#include <string.h>
+#include <sys/mman.h>
+#include <sys/types.h>
+#include <unistd.h>
+
+// Ensure assert() is not compiled out
+#undef NDEBUG
+#include <assert.h>
+
+#include "utils.h"
+
+static int search_proc_maps(char *needle, unsigned long *low, unsigned long *high)
+{
+	unsigned long start, end;
+	static char buf[4096];
+	char name[128];
+	FILE *f;
+	int rc = -1;
+
+	f = fopen("/proc/self/maps", "r");
+	if (!f) {
+		perror("fopen");
+		return -1;
+	}
+
+	while (fgets(buf, sizeof(buf), f)) {
+		rc = sscanf(buf, "%lx-%lx %*c%*c%*c%*c %*x %*d:%*d %*d %127s\n",
+			    &start, &end, name);
+		if (rc == 2)
+			continue;
+
+		if (rc != 3) {
+			printf("sscanf errored\n");
+			rc = -1;
+			break;
+		}
+
+		if (strstr(name, needle)) {
+			*low = start;
+			*high = end - 1;
+			rc = 0;
+			break;
+		}
+	}
+
+	fclose(f);
+
+	return rc;
+}
+
+static volatile sig_atomic_t took_signal = 0;
+
+static void sigusr1_handler(int sig)
+{
+	took_signal++;
+}
+
+int test_sigreturn_vdso(void)
+{
+	unsigned long low, high, size;
+	struct sigaction act;
+	char *p;
+
+	act.sa_handler = sigusr1_handler;
+	act.sa_flags = 0;
+	sigemptyset(&act.sa_mask);
+
+	assert(sigaction(SIGUSR1, &act, NULL) == 0);
+
+	// Confirm the VDSO is mapped, and work out where it is
+	assert(search_proc_maps("[vdso]", &low, &high) == 0);
+	size = high - low + 1;
+	printf("VDSO is at 0x%lx-0x%lx (%lu bytes)\n", low, high, size);
+
+	kill(getpid(), SIGUSR1);
+	assert(took_signal == 1);
+	printf("Signal delivered OK with VDSO mapped\n");
+
+	// Remap the VDSO somewhere else
+	p = mmap(NULL, size, PROT_READ|PROT_WRITE, MAP_ANONYMOUS|MAP_PRIVATE, -1, 0);
+	assert(p != MAP_FAILED);
+	assert(mremap((void *)low, size, size, MREMAP_MAYMOVE|MREMAP_FIXED, p) != MAP_FAILED);
+	assert(search_proc_maps("[vdso]", &low, &high) == 0);
+	size = high - low + 1;
+	printf("VDSO moved to 0x%lx-0x%lx (%lu bytes)\n", low, high, size);
+
+	kill(getpid(), SIGUSR1);
+	assert(took_signal == 2);
+	printf("Signal delivered OK with VDSO moved\n");
+
+	assert(munmap((void *)low, size) == 0);
+	printf("Unmapped VDSO\n");
+
+	// Confirm the VDSO is not mapped anymore
+	assert(search_proc_maps("[vdso]", &low, &high) != 0);
+
+	// Make the stack executable
+	assert(search_proc_maps("[stack]", &low, &high) == 0);
+	size = high - low + 1;
+	mprotect((void *)low, size, PROT_READ|PROT_WRITE|PROT_EXEC);
+	printf("Remapped the stack executable\n");
+
+	kill(getpid(), SIGUSR1);
+	assert(took_signal == 3);
+	printf("Signal delivered OK with VDSO unmapped\n");
+
+	return 0;
+}
+
+int main(void)
+{
+	return test_harness(test_sigreturn_vdso, "sigreturn_vdso");
+}
diff --git a/tools/testing/selftests/powerpc/stringloops/.gitignore b/tools/testing/selftests/powerpc/stringloops/.gitignore
new file mode 100644
index 000000000..b0dfc74aa
--- /dev/null
+++ b/tools/testing/selftests/powerpc/stringloops/.gitignore
@@ -0,0 +1,5 @@
+# SPDX-License-Identifier: GPL-2.0-only
+memcmp_64
+memcmp_32
+strlen
+strlen_32
diff --git a/tools/testing/selftests/powerpc/stringloops/Makefile b/tools/testing/selftests/powerpc/stringloops/Makefile
new file mode 100644
index 000000000..9c39f55a5
--- /dev/null
+++ b/tools/testing/selftests/powerpc/stringloops/Makefile
@@ -0,0 +1,35 @@
+# SPDX-License-Identifier: GPL-2.0
+# The loops are all 64-bit code
+CFLAGS += -I$(CURDIR)
+
+EXTRA_SOURCES := ../harness.c
+
+build_32bit = $(shell if ($(CC) $(CFLAGS) -m32 -o /dev/null memcmp.c >/dev/null 2>&1) then echo "1"; fi)
+
+TEST_GEN_PROGS := memcmp_64 strlen
+
+$(OUTPUT)/memcmp_64: memcmp.c ../utils.c
+$(OUTPUT)/memcmp_64: CFLAGS += -m64 -maltivec
+
+ifeq ($(build_32bit),1)
+$(OUTPUT)/memcmp_32: memcmp.c
+$(OUTPUT)/memcmp_32: CFLAGS += -m32
+
+TEST_GEN_PROGS += memcmp_32
+endif
+
+$(OUTPUT)/strlen: strlen.c string.c
+
+ifeq ($(build_32bit),1)
+$(OUTPUT)/strlen_32: strlen.c
+$(OUTPUT)/strlen_32: CFLAGS += -m32
+
+TEST_GEN_PROGS += strlen_32
+endif
+
+ASFLAGS = $(CFLAGS)
+
+top_srcdir = ../../../../..
+include ../../lib.mk
+
+$(TEST_GEN_PROGS): $(EXTRA_SOURCES)
diff --git a/tools/testing/selftests/powerpc/stringloops/asm/cache.h b/tools/testing/selftests/powerpc/stringloops/asm/cache.h
new file mode 100644
index 000000000..8a2840831
--- /dev/null
+++ b/tools/testing/selftests/powerpc/stringloops/asm/cache.h
@@ -0,0 +1 @@
+#define	IFETCH_ALIGN_BYTES 4
diff --git a/tools/testing/selftests/powerpc/stringloops/asm/export.h b/tools/testing/selftests/powerpc/stringloops/asm/export.h
new file mode 100644
index 000000000..2d14a9b42
--- /dev/null
+++ b/tools/testing/selftests/powerpc/stringloops/asm/export.h
@@ -0,0 +1 @@
+#define EXPORT_SYMBOL(x)
diff --git a/tools/testing/selftests/powerpc/stringloops/asm/ppc-opcode.h b/tools/testing/selftests/powerpc/stringloops/asm/ppc-opcode.h
new file mode 100644
index 000000000..3edd1a1d9
--- /dev/null
+++ b/tools/testing/selftests/powerpc/stringloops/asm/ppc-opcode.h
@@ -0,0 +1,35 @@
+/* SPDX-License-Identifier: GPL-2.0-or-later */
+/*
+ * Copyright 2009 Freescale Semiconductor, Inc.
+ *
+ * provides masks and opcode images for use by code generation, emulation
+ * and for instructions that older assemblers might not know about
+ */
+#ifndef _ASM_POWERPC_PPC_OPCODE_H
+#define _ASM_POWERPC_PPC_OPCODE_H
+
+
+#  define stringify_in_c(...)	__VA_ARGS__
+#  define ASM_CONST(x)		x
+
+
+#define PPC_INST_VCMPEQUD_RC		0x100000c7
+#define PPC_INST_VCMPEQUB_RC		0x10000006
+
+#define __PPC_RC21     (0x1 << 10)
+
+/* macros to insert fields into opcodes */
+#define ___PPC_RA(a)	(((a) & 0x1f) << 16)
+#define ___PPC_RB(b)	(((b) & 0x1f) << 11)
+#define ___PPC_RS(s)	(((s) & 0x1f) << 21)
+#define ___PPC_RT(t)	___PPC_RS(t)
+
+#define VCMPEQUD_RC(vrt, vra, vrb)	stringify_in_c(.long PPC_INST_VCMPEQUD_RC | \
+			      ___PPC_RT(vrt) | ___PPC_RA(vra) | \
+			      ___PPC_RB(vrb) | __PPC_RC21)
+
+#define VCMPEQUB_RC(vrt, vra, vrb)	stringify_in_c(.long PPC_INST_VCMPEQUB_RC | \
+			      ___PPC_RT(vrt) | ___PPC_RA(vra) | \
+			      ___PPC_RB(vrb) | __PPC_RC21)
+
+#endif /* _ASM_POWERPC_PPC_OPCODE_H */
diff --git a/tools/testing/selftests/powerpc/stringloops/asm/ppc_asm.h b/tools/testing/selftests/powerpc/stringloops/asm/ppc_asm.h
new file mode 100644
index 000000000..2b488b78c
--- /dev/null
+++ b/tools/testing/selftests/powerpc/stringloops/asm/ppc_asm.h
@@ -0,0 +1,33 @@
+/* SPDX-License-Identifier: GPL-2.0 */
+#ifndef _PPC_ASM_H
+#define _PPC_ASM_H
+#include <ppc-asm.h>
+
+#ifndef r1
+#define r1 sp
+#endif
+
+#define _GLOBAL(A) FUNC_START(test_ ## A)
+#define _GLOBAL_TOC(A) FUNC_START(test_ ## A)
+
+#define CONFIG_ALTIVEC
+
+#define R14 r14
+#define R15 r15
+#define R16 r16
+#define R17 r17
+#define R18 r18
+#define R19 r19
+#define R20 r20
+#define R21 r21
+#define R22 r22
+#define R29 r29
+#define R30 r30
+#define R31 r31
+
+#define STACKFRAMESIZE	256
+#define STK_REG(i)	(112 + ((i)-14)*8)
+
+#define BEGIN_FTR_SECTION
+#define END_FTR_SECTION_IFSET(val)
+#endif
diff --git a/tools/testing/selftests/powerpc/stringloops/memcmp.c b/tools/testing/selftests/powerpc/stringloops/memcmp.c
new file mode 100644
index 000000000..cb2f18855
--- /dev/null
+++ b/tools/testing/selftests/powerpc/stringloops/memcmp.c
@@ -0,0 +1,169 @@
+// SPDX-License-Identifier: GPL-2.0
+#include <malloc.h>
+#include <stdlib.h>
+#include <string.h>
+#include <sys/mman.h>
+#include <time.h>
+
+#include "utils.h"
+
+#define SIZE 256
+#define ITERATIONS 10000
+
+#define LARGE_SIZE (5 * 1024)
+#define LARGE_ITERATIONS 1000
+#define LARGE_MAX_OFFSET 32
+#define LARGE_SIZE_START 4096
+
+/* This is big enough to fit LARGE_SIZE and works on 4K & 64K kernels */
+#define MAP_SIZE (64 * 1024)
+
+#define MAX_OFFSET_DIFF_S1_S2 48
+
+int vmx_count;
+int enter_vmx_ops(void)
+{
+	vmx_count++;
+	return 1;
+}
+
+void exit_vmx_ops(void)
+{
+	vmx_count--;
+}
+int test_memcmp(const void *s1, const void *s2, size_t n);
+
+/* test all offsets and lengths */
+static void test_one(char *s1, char *s2, unsigned long max_offset,
+		unsigned long size_start, unsigned long max_size)
+{
+	unsigned long offset, size;
+
+	for (offset = 0; offset < max_offset; offset++) {
+		for (size = size_start; size < (max_size - offset); size++) {
+			int x, y;
+			unsigned long i;
+
+			y = memcmp(s1+offset, s2+offset, size);
+			x = test_memcmp(s1+offset, s2+offset, size);
+
+			if (((x ^ y) < 0) &&	/* Trick to compare sign */
+				((x | y) != 0)) { /* check for zero */
+				printf("memcmp returned %d, should have returned %d (offset %ld size %ld)\n", x, y, offset, size);
+
+				for (i = offset; i < offset+size; i++)
+					printf("%02x ", s1[i]);
+				printf("\n");
+
+				for (i = offset; i < offset+size; i++)
+					printf("%02x ", s2[i]);
+				printf("\n");
+				abort();
+			}
+
+			if (vmx_count != 0) {
+				printf("vmx enter/exit not paired.(offset:%ld size:%ld s1:%p s2:%p vc:%d\n",
+					offset, size, s1, s2, vmx_count);
+				printf("\n");
+				abort();
+			}
+		}
+	}
+}
+
+static int testcase(bool islarge)
+{
+	unsigned long i, comp_size, alloc_size;
+	char *p, *s1, *s2;
+	int iterations;
+
+	comp_size = (islarge ? LARGE_SIZE : SIZE);
+	alloc_size = comp_size + MAX_OFFSET_DIFF_S1_S2;
+	iterations = islarge ? LARGE_ITERATIONS : ITERATIONS;
+
+	p = mmap(NULL, 4 * MAP_SIZE, PROT_READ | PROT_WRITE,
+		 MAP_ANONYMOUS | MAP_PRIVATE, -1, 0);
+	FAIL_IF(p == MAP_FAILED);
+
+	/* Put s1/s2 at the end of a page */
+	s1 = p + MAP_SIZE - alloc_size;
+	s2 = p + 3 * MAP_SIZE - alloc_size;
+
+	/* And unmap the subsequent page to force a fault if we overread */
+	munmap(p + MAP_SIZE, MAP_SIZE);
+	munmap(p + 3 * MAP_SIZE, MAP_SIZE);
+
+	srandom(time(0));
+
+	for (i = 0; i < iterations; i++) {
+		unsigned long j;
+		unsigned long change;
+		char *rand_s1 = s1;
+		char *rand_s2 = s2;
+
+		for (j = 0; j < alloc_size; j++)
+			s1[j] = random();
+
+		rand_s1 += random() % MAX_OFFSET_DIFF_S1_S2;
+		rand_s2 += random() % MAX_OFFSET_DIFF_S1_S2;
+		memcpy(rand_s2, rand_s1, comp_size);
+
+		/* change one byte */
+		change = random() % comp_size;
+		rand_s2[change] = random() & 0xff;
+
+		if (islarge)
+			test_one(rand_s1, rand_s2, LARGE_MAX_OFFSET,
+					LARGE_SIZE_START, comp_size);
+		else
+			test_one(rand_s1, rand_s2, SIZE, 0, comp_size);
+	}
+
+	srandom(time(0));
+
+	for (i = 0; i < iterations; i++) {
+		unsigned long j;
+		unsigned long change;
+		char *rand_s1 = s1;
+		char *rand_s2 = s2;
+
+		for (j = 0; j < alloc_size; j++)
+			s1[j] = random();
+
+		rand_s1 += random() % MAX_OFFSET_DIFF_S1_S2;
+		rand_s2 += random() % MAX_OFFSET_DIFF_S1_S2;
+		memcpy(rand_s2, rand_s1, comp_size);
+
+		/* change multiple bytes, 1/8 of total */
+		for (j = 0; j < comp_size / 8; j++) {
+			change = random() % comp_size;
+			s2[change] = random() & 0xff;
+		}
+
+		if (islarge)
+			test_one(rand_s1, rand_s2, LARGE_MAX_OFFSET,
+					LARGE_SIZE_START, comp_size);
+		else
+			test_one(rand_s1, rand_s2, SIZE, 0, comp_size);
+	}
+
+	return 0;
+}
+
+static int testcases(void)
+{
+#ifdef __powerpc64__
+	// vcmpequd used in memcmp_64.S is v2.07
+	SKIP_IF(!have_hwcap2(PPC_FEATURE2_ARCH_2_07));
+#endif
+
+	testcase(0);
+	testcase(1);
+	return 0;
+}
+
+int main(void)
+{
+	test_harness_set_timeout(300);
+	return test_harness(testcases, "memcmp");
+}
diff --git a/tools/testing/selftests/powerpc/stringloops/memcmp_32.S b/tools/testing/selftests/powerpc/stringloops/memcmp_32.S
new file mode 120000
index 000000000..056f2b3af
--- /dev/null
+++ b/tools/testing/selftests/powerpc/stringloops/memcmp_32.S
@@ -0,0 +1 @@
+../../../../../arch/powerpc/lib/memcmp_32.S
\ No newline at end of file
diff --git a/tools/testing/selftests/powerpc/stringloops/memcmp_64.S b/tools/testing/selftests/powerpc/stringloops/memcmp_64.S
new file mode 120000
index 000000000..9bc87e438
--- /dev/null
+++ b/tools/testing/selftests/powerpc/stringloops/memcmp_64.S
@@ -0,0 +1 @@
+../../../../../arch/powerpc/lib/memcmp_64.S
\ No newline at end of file
diff --git a/tools/testing/selftests/powerpc/stringloops/string.c b/tools/testing/selftests/powerpc/stringloops/string.c
new file mode 100644
index 000000000..45e777541
--- /dev/null
+++ b/tools/testing/selftests/powerpc/stringloops/string.c
@@ -0,0 +1,21 @@
+// SPDX-License-Identifier: GPL-2.0
+/*
+ * Copied from linux/lib/string.c
+ *
+ *  Copyright (C) 1991, 1992  Linus Torvalds
+ */
+
+#include <stddef.h>
+
+/**
+ * strlen - Find the length of a string
+ * @s: The string to be sized
+ */
+size_t test_strlen(const char *s)
+{
+	const char *sc;
+
+	for (sc = s; *sc != '\0'; ++sc)
+		/* nothing */;
+	return sc - s;
+}
diff --git a/tools/testing/selftests/powerpc/stringloops/strlen.c b/tools/testing/selftests/powerpc/stringloops/strlen.c
new file mode 100644
index 000000000..9055ebc48
--- /dev/null
+++ b/tools/testing/selftests/powerpc/stringloops/strlen.c
@@ -0,0 +1,127 @@
+// SPDX-License-Identifier: GPL-2.0
+#include <malloc.h>
+#include <stdlib.h>
+#include <string.h>
+#include <time.h>
+#include "utils.h"
+
+#define SIZE 256
+#define ITERATIONS 1000
+#define ITERATIONS_BENCH 100000
+
+int test_strlen(const void *s);
+
+/* test all offsets and lengths */
+static void test_one(char *s)
+{
+	unsigned long offset;
+
+	for (offset = 0; offset < SIZE; offset++) {
+		int x, y;
+		unsigned long i;
+
+		y = strlen(s + offset);
+		x = test_strlen(s + offset);
+
+		if (x != y) {
+			printf("strlen() returned %d, should have returned %d (%p offset %ld)\n", x, y, s, offset);
+
+			for (i = offset; i < SIZE; i++)
+				printf("%02x ", s[i]);
+			printf("\n");
+		}
+	}
+}
+
+static void bench_test(char *s)
+{
+	struct timespec ts_start, ts_end;
+	int i;
+
+	clock_gettime(CLOCK_MONOTONIC, &ts_start);
+
+	for (i = 0; i < ITERATIONS_BENCH; i++)
+		test_strlen(s);
+
+	clock_gettime(CLOCK_MONOTONIC, &ts_end);
+
+	printf("len %3.3d : time = %.6f\n", test_strlen(s), ts_end.tv_sec - ts_start.tv_sec + (ts_end.tv_nsec - ts_start.tv_nsec) / 1e9);
+}
+
+static int testcase(void)
+{
+	char *s;
+	unsigned long i;
+
+	s = memalign(128, SIZE);
+	if (!s) {
+		perror("memalign");
+		exit(1);
+	}
+
+	srandom(1);
+
+	memset(s, 0, SIZE);
+	for (i = 0; i < SIZE; i++) {
+		char c;
+
+		do {
+			c = random() & 0x7f;
+		} while (!c);
+		s[i] = c;
+		test_one(s);
+	}
+
+	for (i = 0; i < ITERATIONS; i++) {
+		unsigned long j;
+
+		for (j = 0; j < SIZE; j++) {
+			char c;
+
+			do {
+				c = random() & 0x7f;
+			} while (!c);
+			s[j] = c;
+		}
+		for (j = 0; j < sizeof(long); j++) {
+			s[SIZE - 1 - j] = 0;
+			test_one(s);
+		}
+	}
+
+	for (i = 0; i < SIZE; i++) {
+		char c;
+
+		do {
+			c = random() & 0x7f;
+		} while (!c);
+		s[i] = c;
+	}
+
+	bench_test(s);
+
+	s[16] = 0;
+	bench_test(s);
+
+	s[8] = 0;
+	bench_test(s);
+
+	s[4] = 0;
+	bench_test(s);
+
+	s[3] = 0;
+	bench_test(s);
+
+	s[2] = 0;
+	bench_test(s);
+
+	s[1] = 0;
+	bench_test(s);
+
+	return 0;
+}
+
+int main(void)
+{
+	return test_harness(testcase, "strlen");
+}
diff --git a/tools/testing/selftests/powerpc/stringloops/strlen_32.S b/tools/testing/selftests/powerpc/stringloops/strlen_32.S
new file mode 120000
index 000000000..72b13731b
--- /dev/null
+++ b/tools/testing/selftests/powerpc/stringloops/strlen_32.S
@@ -0,0 +1 @@
+../../../../../arch/powerpc/lib/strlen_32.S
\ No newline at end of file
diff --git a/tools/testing/selftests/powerpc/switch_endian/.gitignore b/tools/testing/selftests/powerpc/switch_endian/.gitignore
new file mode 100644
index 000000000..30e962cf8
--- /dev/null
+++ b/tools/testing/selftests/powerpc/switch_endian/.gitignore
@@ -0,0 +1,3 @@
+# SPDX-License-Identifier: GPL-2.0-only
+switch_endian_test
+check-reversed.S
diff --git a/tools/testing/selftests/powerpc/switch_endian/Makefile b/tools/testing/selftests/powerpc/switch_endian/Makefile
new file mode 100644
index 000000000..bdc081afe
--- /dev/null
+++ b/tools/testing/selftests/powerpc/switch_endian/Makefile
@@ -0,0 +1,18 @@
+# SPDX-License-Identifier: GPL-2.0
+TEST_GEN_PROGS := switch_endian_test
+
+ASFLAGS += -O2 -Wall -g -nostdlib -m64
+
+EXTRA_CLEAN = $(OUTPUT)/*.o $(OUTPUT)/check-reversed.S
+
+top_srcdir = ../../../../..
+include ../../lib.mk
+
+$(OUTPUT)/switch_endian_test: ASFLAGS += -I $(OUTPUT)
+$(OUTPUT)/switch_endian_test: $(OUTPUT)/check-reversed.S
+
+$(OUTPUT)/check-reversed.o: $(OUTPUT)/check.o
+	$(CROSS_COMPILE)objcopy -j .text --reverse-bytes=4 -O binary $< $@
+
+$(OUTPUT)/check-reversed.S: $(OUTPUT)/check-reversed.o
+	hexdump -v -e '/1 ".byte 0x%02X\n"' $< > $@
diff --git a/tools/testing/selftests/powerpc/switch_endian/check.S b/tools/testing/selftests/powerpc/switch_endian/check.S
new file mode 100644
index 000000000..927a5c675
--- /dev/null
+++ b/tools/testing/selftests/powerpc/switch_endian/check.S
@@ -0,0 +1,101 @@
+/* SPDX-License-Identifier: GPL-2.0 */
+#include "common.h"
+
+/*
+ * Checks that registers contain what we expect, ie. they were not clobbered by
+ * the syscall.
+ *
+ * r15: pattern to check registers against.
+ *
+ * At the end r3 == 0 if everything's OK.
+ */
+	nop			# guaranteed to be illegal in reverse-endian
+	mr	r9,r15
+	cmpd    r9,r3		# check r3
+	bne     1f
+	addi    r9,r15,4	# check r4
+	cmpd    r9,r4
+	bne     1f
+	lis     r9,0x00FF	# check CR
+	ori     r9,r9,0xF000
+	mfcr    r10
+	and     r10,r10,r9
+	cmpw    r9,r10
+	addi    r9,r15,34
+	bne     1f
+	addi    r9,r15,32	# check LR
+	mflr    r10
+	cmpd    r9,r10
+	bne     1f
+	addi    r9,r15,5	# check r5
+	cmpd    r9,r5
+	bne     1f
+	addi    r9,r15,6	# check r6
+	cmpd    r9,r6
+	bne     1f
+	addi    r9,r15,7	# check r7
+	cmpd    r9,r7
+	bne     1f
+	addi    r9,r15,8	# check r8
+	cmpd    r9,r8
+	bne     1f
+	addi    r9,r15,13	# check r13
+	cmpd    r9,r13
+	bne     1f
+	addi    r9,r15,14	# check r14
+	cmpd    r9,r14
+	bne     1f
+	addi    r9,r15,16	# check r16
+	cmpd    r9,r16
+	bne     1f
+	addi    r9,r15,17	# check r17
+	cmpd    r9,r17
+	bne     1f
+	addi    r9,r15,18	# check r18
+	cmpd    r9,r18
+	bne     1f
+	addi    r9,r15,19	# check r19
+	cmpd    r9,r19
+	bne     1f
+	addi    r9,r15,20	# check r20
+	cmpd    r9,r20
+	bne     1f
+	addi    r9,r15,21	# check r21
+	cmpd    r9,r21
+	bne     1f
+	addi    r9,r15,22	# check r22
+	cmpd    r9,r22
+	bne     1f
+	addi    r9,r15,23	# check r23
+	cmpd    r9,r23
+	bne     1f
+	addi    r9,r15,24	# check r24
+	cmpd    r9,r24
+	bne     1f
+	addi    r9,r15,25	# check r25
+	cmpd    r9,r25
+	bne     1f
+	addi    r9,r15,26	# check r26
+	cmpd    r9,r26
+	bne     1f
+	addi    r9,r15,27	# check r27
+	cmpd    r9,r27
+	bne     1f
+	addi    r9,r15,28	# check r28
+	cmpd    r9,r28
+	bne     1f
+	addi    r9,r15,29	# check r29
+	cmpd    r9,r29
+	bne     1f
+	addi    r9,r15,30	# check r30
+	cmpd    r9,r30
+	bne     1f
+	addi    r9,r15,31	# check r31
+	cmpd    r9,r31
+	bne     1f
+	b	2f
+1:	mr	r3, r9
+	li	r0, __NR_exit
+	sc
+2:	li	r0, __NR_switch_endian
+	nop
diff --git a/tools/testing/selftests/powerpc/switch_endian/common.h b/tools/testing/selftests/powerpc/switch_endian/common.h
new file mode 100644
index 000000000..1434cbc2a
--- /dev/null
+++ b/tools/testing/selftests/powerpc/switch_endian/common.h
@@ -0,0 +1,7 @@
+/* SPDX-License-Identifier: GPL-2.0 */
+#include <ppc-asm.h>
+#include <asm/unistd.h>
+
+#ifndef __NR_switch_endian
+#define __NR_switch_endian 363
+#endif
diff --git a/tools/testing/selftests/powerpc/switch_endian/switch_endian_test.S b/tools/testing/selftests/powerpc/switch_endian/switch_endian_test.S
new file mode 100644
index 000000000..7887f78cf
--- /dev/null
+++ b/tools/testing/selftests/powerpc/switch_endian/switch_endian_test.S
@@ -0,0 +1,97 @@
+/* SPDX-License-Identifier: GPL-2.0 */
+#include "common.h"
+
+	.data
+	.balign 8
+success_message:
+	.ascii "success: switch_endian_test\n\0"
+
+	.balign 8
+failure_message:
+	.ascii "failure: switch_endian_test\n\0"
+
+	.section ".toc"
+	.balign 8
+pattern:
+	.8byte 0x5555AAAA5555AAAA
+
+	.text
+FUNC_START(_start)
+	/* Load the pattern */
+	ld	r15, pattern@TOC(%r2)
+
+	/* Setup CR, only CR2-CR4 are maintained */
+	lis	r3, 0x00FF
+	ori	r3, r3, 0xF000
+	mtcr	r3
+
+	/* Load the pattern slightly modified into the registers */
+	mr	r3, r15
+	addi	r4, r15, 4
+
+	addi	r5, r15, 32
+	mtlr	r5
+
+	addi	r5, r15, 5
+	addi	r6, r15, 6
+	addi	r7, r15, 7
+	addi	r8, r15, 8
+
+	/* r9 - r12 are clobbered */
+
+	addi	r13, r15, 13
+	addi	r14, r15, 14
+
+	/* Skip r15 we're using it */
+
+	addi	r16, r15, 16
+	addi	r17, r15, 17
+	addi	r18, r15, 18
+	addi	r19, r15, 19
+	addi	r20, r15, 20
+	addi	r21, r15, 21
+	addi	r22, r15, 22
+	addi	r23, r15, 23
+	addi	r24, r15, 24
+	addi	r25, r15, 25
+	addi	r26, r15, 26
+	addi	r27, r15, 27
+	addi	r28, r15, 28
+	addi	r29, r15, 29
+	addi	r30, r15, 30
+	addi	r31, r15, 31
+
+	/*
+	 * Call the syscall to switch endian.
+	 * It clobbers r9-r12, XER, CTR and CR0-1,5-7.
+	 */
+	li r0, __NR_switch_endian
+	sc
+
+	tdi   0, 0, 0x48	// b +8 if the endian was switched
+	b     .Lfail	  	// exit if endian didn't switch
+
+#include "check-reversed.S"
+
+	/* Flip back, r0 already has the switch syscall number */
+	.long	0x02000044	/* sc */
+
+#include "check.S"
+
+	ld	r4, success_message@got(%r2)
+	li	r5, 28	// strlen(success_message)
+	li	r14, 0	// exit status
+.Lout:
+	li	r0, __NR_write
+	li	r3, 1	/* stdout */
+	sc
+	li      r0, __NR_exit
+	mr	r3, r14
+	sc
+	b       .
+
+.Lfail:
+	ld	r4, failure_message@got(%r2)
+	li	r5, 28	// strlen(failure_message)
+	li	r14, 1
+	b	.Lout
diff --git a/tools/testing/selftests/powerpc/syscalls/.gitignore b/tools/testing/selftests/powerpc/syscalls/.gitignore
new file mode 100644
index 000000000..b00cab225
--- /dev/null
+++ b/tools/testing/selftests/powerpc/syscalls/.gitignore
@@ -0,0 +1,2 @@
+# SPDX-License-Identifier: GPL-2.0-only
+ipc_unmuxed
diff --git a/tools/testing/selftests/powerpc/syscalls/Makefile b/tools/testing/selftests/powerpc/syscalls/Makefile
new file mode 100644
index 000000000..b63f8459c
--- /dev/null
+++ b/tools/testing/selftests/powerpc/syscalls/Makefile
@@ -0,0 +1,9 @@
+# SPDX-License-Identifier: GPL-2.0-only
+TEST_GEN_PROGS := ipc_unmuxed rtas_filter
+
+CFLAGS += -I../../../../../usr/include
+
+top_srcdir = ../../../../..
+include ../../lib.mk
+
+$(TEST_GEN_PROGS): ../harness.c
diff --git a/tools/testing/selftests/powerpc/syscalls/ipc.h b/tools/testing/selftests/powerpc/syscalls/ipc.h
new file mode 100644
index 000000000..26a20682c
--- /dev/null
+++ b/tools/testing/selftests/powerpc/syscalls/ipc.h
@@ -0,0 +1,48 @@
+/* SPDX-License-Identifier: GPL-2.0 */
+#ifdef __NR_semop
+DO_TEST(semop, __NR_semop)
+#endif
+
+#ifdef __NR_semget
+DO_TEST(semget, __NR_semget)
+#endif
+
+#ifdef __NR_semctl
+DO_TEST(semctl, __NR_semctl)
+#endif
+
+#ifdef __NR_semtimedop
+DO_TEST(semtimedop, __NR_semtimedop)
+#endif
+
+#ifdef __NR_msgsnd
+DO_TEST(msgsnd, __NR_msgsnd)
+#endif
+
+#ifdef __NR_msgrcv
+DO_TEST(msgrcv, __NR_msgrcv)
+#endif
+
+#ifdef __NR_msgget
+DO_TEST(msgget, __NR_msgget)
+#endif
+
+#ifdef __NR_msgctl
+DO_TEST(msgctl, __NR_msgctl)
+#endif
+
+#ifdef __NR_shmat
+DO_TEST(shmat, __NR_shmat)
+#endif
+
+#ifdef __NR_shmdt
+DO_TEST(shmdt, __NR_shmdt)
+#endif
+
+#ifdef __NR_shmget
+DO_TEST(shmget, __NR_shmget)
+#endif
+
+#ifdef __NR_shmctl
+DO_TEST(shmctl, __NR_shmctl)
+#endif
diff --git a/tools/testing/selftests/powerpc/syscalls/ipc_unmuxed.c b/tools/testing/selftests/powerpc/syscalls/ipc_unmuxed.c
new file mode 100644
index 000000000..4c582524a
--- /dev/null
+++ b/tools/testing/selftests/powerpc/syscalls/ipc_unmuxed.c
@@ -0,0 +1,57 @@
+// SPDX-License-Identifier: GPL-2.0-or-later
+/*
+ * Copyright 2015, Michael Ellerman, IBM Corp.
+ *
+ * This test simply tests that certain syscalls are implemented. It doesn't
+ * actually exercise their logic in any way.
+ */
+
+#define _GNU_SOURCE
+#include <errno.h>
+#include <stdio.h>
+#include <unistd.h>
+#include <sys/syscall.h>
+
+#include "utils.h"
+
+
+#define DO_TEST(_name, _num)	\
+static int test_##_name(void)			\
+{						\
+	int rc;					\
+	printf("Testing " #_name);		\
+	errno = 0;				\
+	rc = syscall(_num, -1, 0, 0, 0, 0, 0);	\
+	printf("\treturned %d, errno %d\n", rc, errno); \
+	return errno == ENOSYS;			\
+}
+
+#include "ipc.h"
+#undef DO_TEST
+
+static int ipc_unmuxed(void)
+{
+	int tests_done = 0;
+
+#define DO_TEST(_name, _num)		\
+	FAIL_IF(test_##_name());	\
+	tests_done++;
+
+#include "ipc.h"
+#undef DO_TEST
+
+	/*
+	 * If we ran no tests then it means none of the syscall numbers were
+	 * defined, possibly because we were built against old headers. But it
+	 * means we didn't really test anything, so instead of passing mark it
+	 * as a skip to give the user a clue.
+	 */
+	SKIP_IF(tests_done == 0);
+
+	return 0;
+}
+
+int main(void)
+{
+	return test_harness(ipc_unmuxed, "ipc_unmuxed");
+}
diff --git a/tools/testing/selftests/powerpc/syscalls/rtas_filter.c b/tools/testing/selftests/powerpc/syscalls/rtas_filter.c
new file mode 100644
index 000000000..03b487f18
--- /dev/null
+++ b/tools/testing/selftests/powerpc/syscalls/rtas_filter.c
@@ -0,0 +1,285 @@
+// SPDX-License-Identifier: GPL-2.0+
+/*
+ * Copyright 2005-2020 IBM Corporation.
+ *
+ * Includes code from librtas (https://github.com/ibm-power-utilities/librtas/)
+ */
+
+#include <byteswap.h>
+#include <stdint.h>
+#include <inttypes.h>
+#include <stdio.h>
+#include <string.h>
+#include <sys/syscall.h>
+#include <sys/types.h>
+#include <unistd.h>
+#include <stdarg.h>
+#include <stdlib.h>
+#include <fcntl.h>
+#include <errno.h>
+#include "utils.h"
+
+#if __BYTE_ORDER__ == __ORDER_LITTLE_ENDIAN__
+#define cpu_to_be32(x)		bswap_32(x)
+#define be32_to_cpu(x)		bswap_32(x)
+#else
+#define cpu_to_be32(x)		(x)
+#define be32_to_cpu(x)		(x)
+#endif
+
+#define RTAS_IO_ASSERT	-1098	/* Unexpected I/O Error */
+#define RTAS_UNKNOWN_OP -1099	/* No Firmware Implementation of Function */
+#define BLOCK_SIZE 4096
+#define PAGE_SIZE 4096
+#define MAX_PAGES 64
+
+static const char *ofdt_rtas_path = "/proc/device-tree/rtas";
+
+typedef __be32 uint32_t;
+struct rtas_args {
+	__be32 token;
+	__be32 nargs;
+	__be32 nret;
+	__be32 args[16];
+	__be32 *rets;	  /* Pointer to return values in args[]. */
+};
+
+struct region {
+	uint64_t addr;
+	uint32_t size;
+	struct region *next;
+};
+
+int read_entire_file(int fd, char **buf, size_t *len)
+{
+	size_t buf_size = 0;
+	size_t off = 0;
+	int rc;
+
+	*buf = NULL;
+	do {
+		buf_size += BLOCK_SIZE;
+		if (*buf == NULL)
+			*buf = malloc(buf_size);
+		else
+			*buf = realloc(*buf, buf_size);
+
+		if (*buf == NULL)
+			return -ENOMEM;
+
+		rc = read(fd, *buf + off, BLOCK_SIZE);
+		if (rc < 0)
+			return -EIO;
+
+		off += rc;
+	} while (rc == BLOCK_SIZE);
+
+	if (len)
+		*len = off;
+
+	return 0;
+}
+
+static int open_prop_file(const char *prop_path, const char *prop_name, int *fd)
+{
+	char *path;
+	int len;
+
+	/* allocate enough for two string, a slash and trailing NULL */
+	len = strlen(prop_path) + strlen(prop_name) + 1 + 1;
+	path = malloc(len);
+	if (path == NULL)
+		return -ENOMEM;
+
+	snprintf(path, len, "%s/%s", prop_path, prop_name);
+
+	*fd = open(path, O_RDONLY);
+	free(path);
+	if (*fd < 0)
+		return -errno;
+
+	return 0;
+}
+
+static int get_property(const char *prop_path, const char *prop_name,
+			char **prop_val, size_t *prop_len)
+{
+	int rc, fd;
+
+	rc = open_prop_file(prop_path, prop_name, &fd);
+	if (rc)
+		return rc;
+
+	rc = read_entire_file(fd, prop_val, prop_len);
+	close(fd);
+
+	return rc;
+}
+
+int rtas_token(const char *call_name)
+{
+	char *prop_buf = NULL;
+	size_t len;
+	int rc;
+
+	rc = get_property(ofdt_rtas_path, call_name, &prop_buf, &len);
+	if (rc < 0) {
+		rc = RTAS_UNKNOWN_OP;
+		goto err;
+	}
+
+	rc = be32_to_cpu(*(int *)prop_buf);
+
+err:
+	free(prop_buf);
+	return rc;
+}
+
+static int read_kregion_bounds(struct region *kregion)
+{
+	char *buf;
+	int fd;
+	int rc;
+
+	fd = open("/proc/ppc64/rtas/rmo_buffer", O_RDONLY);
+	if (fd < 0) {
+		printf("Could not open rmo_buffer file\n");
+		return RTAS_IO_ASSERT;
+	}
+
+	rc = read_entire_file(fd, &buf, NULL);
+	close(fd);
+	if (rc) {
+		free(buf);
+		return rc;
+	}
+
+	sscanf(buf, "%" SCNx64 " %x", &kregion->addr, &kregion->size);
+	free(buf);
+
+	if (!(kregion->size && kregion->addr) ||
+	    (kregion->size > (PAGE_SIZE * MAX_PAGES))) {
+		printf("Unexpected kregion bounds\n");
+		return RTAS_IO_ASSERT;
+	}
+
+	return 0;
+}
+
+static int rtas_call(const char *name, int nargs,
+		     int nrets, ...)
+{
+	struct rtas_args args;
+	__be32 *rets[16];
+	int i, rc, token;
+	va_list ap;
+
+	va_start(ap, nrets);
+
+	token = rtas_token(name);
+	if (token == RTAS_UNKNOWN_OP) {
+		// We don't care if the call doesn't exist
+		printf("call '%s' not available, skipping...", name);
+		rc = RTAS_UNKNOWN_OP;
+		goto err;
+	}
+
+	args.token = cpu_to_be32(token);
+	args.nargs = cpu_to_be32(nargs);
+	args.nret = cpu_to_be32(nrets);
+
+	for (i = 0; i < nargs; i++)
+		args.args[i] = (__be32) va_arg(ap, unsigned long);
+
+	for (i = 0; i < nrets; i++)
+		rets[i] = (__be32 *) va_arg(ap, unsigned long);
+
+	rc = syscall(__NR_rtas, &args);
+	if (rc) {
+		rc = -errno;
+		goto err;
+	}
+
+	if (nrets) {
+		*(rets[0]) = be32_to_cpu(args.args[nargs]);
+
+		for (i = 1; i < nrets; i++) {
+			*(rets[i]) = args.args[nargs + i];
+		}
+	}
+
+err:
+	va_end(ap);
+	return rc;
+}
+
+static int test(void)
+{
+	struct region rmo_region;
+	uint32_t rmo_start;
+	uint32_t rmo_end;
+	__be32 rets[1];
+	int rc;
+
+	// Test a legitimate harmless call
+	// Expected: call succeeds
+	printf("Test a permitted call, no parameters... ");
+	rc = rtas_call("get-time-of-day", 0, 1, rets);
+	printf("rc: %d\n", rc);
+	FAIL_IF(rc != 0 && rc != RTAS_UNKNOWN_OP);
+
+	// Test a prohibited call
+	// Expected: call returns -EINVAL
+	printf("Test a prohibited call... ");
+	rc = rtas_call("nvram-fetch", 0, 1, rets);
+	printf("rc: %d\n", rc);
+	FAIL_IF(rc != -EINVAL && rc != RTAS_UNKNOWN_OP);
+
+	// Get RMO
+	rc = read_kregion_bounds(&rmo_region);
+	if (rc) {
+		printf("Couldn't read RMO region bounds, skipping remaining cases\n");
+		return 0;
+	}
+	rmo_start = rmo_region.addr;
+	rmo_end = rmo_start + rmo_region.size - 1;
+	printf("RMO range: %08x - %08x\n", rmo_start, rmo_end);
+
+	// Test a permitted call, user-supplied size, buffer inside RMO
+	// Expected: call succeeds
+	printf("Test a permitted call, user-supplied size, buffer inside RMO... ");
+	rc = rtas_call("ibm,get-system-parameter", 3, 1, 0, cpu_to_be32(rmo_start),
+		       cpu_to_be32(rmo_end - rmo_start + 1), rets);
+	printf("rc: %d\n", rc);
+	FAIL_IF(rc != 0 && rc != RTAS_UNKNOWN_OP);
+
+	// Test a permitted call, user-supplied size, buffer start outside RMO
+	// Expected: call returns -EINVAL
+	printf("Test a permitted call, user-supplied size, buffer start outside RMO... ");
+	rc = rtas_call("ibm,get-system-parameter", 3, 1, 0, cpu_to_be32(rmo_end + 1),
+		       cpu_to_be32(4000), rets);
+	printf("rc: %d\n", rc);
+	FAIL_IF(rc != -EINVAL && rc != RTAS_UNKNOWN_OP);
+
+	// Test a permitted call, user-supplied size, buffer end outside RMO
+	// Expected: call returns -EINVAL
+	printf("Test a permitted call, user-supplied size, buffer end outside RMO... ");
+	rc = rtas_call("ibm,get-system-parameter", 3, 1, 0, cpu_to_be32(rmo_start),
+		       cpu_to_be32(rmo_end - rmo_start + 2), rets);
+	printf("rc: %d\n", rc);
+	FAIL_IF(rc != -EINVAL && rc != RTAS_UNKNOWN_OP);
+
+	// Test a permitted call, fixed size, buffer end outside RMO
+	// Expected: call returns -EINVAL
+	printf("Test a permitted call, fixed size, buffer end outside RMO... ");
+	rc = rtas_call("ibm,configure-connector", 2, 1, cpu_to_be32(rmo_end - 4000), 0, rets);
+	printf("rc: %d\n", rc);
+	FAIL_IF(rc != -EINVAL && rc != RTAS_UNKNOWN_OP);
+
+	return 0;
+}
+
+int main(int argc, char *argv[])
+{
+	return test_harness(test, "rtas_filter");
+}
diff --git a/tools/testing/selftests/powerpc/tm/.gitignore b/tools/testing/selftests/powerpc/tm/.gitignore
new file mode 100644
index 000000000..d8900a0c4
--- /dev/null
+++ b/tools/testing/selftests/powerpc/tm/.gitignore
@@ -0,0 +1,22 @@
+# SPDX-License-Identifier: GPL-2.0-only
+tm-resched-dscr
+tm-syscall
+tm-signal-msr-resv
+tm-signal-stack
+tm-vmxcopy
+tm-fork
+tm-tar
+tm-tmspr
+tm-exec
+tm-signal-context-chk-fpu
+tm-signal-context-chk-gpr
+tm-signal-context-chk-vmx
+tm-signal-context-chk-vsx
+tm-signal-context-force-tm
+tm-signal-sigreturn-nt
+tm-signal-pagefault
+tm-vmx-unavail
+tm-unavailable
+tm-trap
+tm-sigreturn
+tm-poison
diff --git a/tools/testing/selftests/powerpc/tm/Makefile b/tools/testing/selftests/powerpc/tm/Makefile
new file mode 100644
index 000000000..5881e97c7
--- /dev/null
+++ b/tools/testing/selftests/powerpc/tm/Makefile
@@ -0,0 +1,32 @@
+# SPDX-License-Identifier: GPL-2.0
+SIGNAL_CONTEXT_CHK_TESTS := tm-signal-context-chk-gpr tm-signal-context-chk-fpu \
+	tm-signal-context-chk-vmx tm-signal-context-chk-vsx
+
+TEST_GEN_PROGS := tm-resched-dscr tm-syscall tm-signal-msr-resv tm-signal-stack \
+	tm-vmxcopy tm-fork tm-tar tm-tmspr tm-vmx-unavail tm-unavailable tm-trap \
+	$(SIGNAL_CONTEXT_CHK_TESTS) tm-sigreturn tm-signal-sigreturn-nt \
+	tm-signal-context-force-tm tm-poison tm-signal-pagefault
+
+TEST_FILES := settings
+
+top_srcdir = ../../../../..
+include ../../lib.mk
+
+$(TEST_GEN_PROGS): ../harness.c ../utils.c
+
+CFLAGS += -mhtm
+
+$(OUTPUT)/tm-syscall: tm-syscall-asm.S
+$(OUTPUT)/tm-syscall: CFLAGS += -I../../../../../usr/include
+$(OUTPUT)/tm-tmspr: CFLAGS += -pthread
+$(OUTPUT)/tm-vmx-unavail: CFLAGS += -pthread -m64
+$(OUTPUT)/tm-resched-dscr: ../pmu/lib.c
+$(OUTPUT)/tm-unavailable: CFLAGS += -O0 -pthread -m64 -Wno-error=uninitialized -mvsx
+$(OUTPUT)/tm-trap: CFLAGS += -O0 -pthread -m64
+$(OUTPUT)/tm-signal-context-force-tm: CFLAGS += -pthread -m64
+$(OUTPUT)/tm-signal-pagefault: CFLAGS += -pthread -m64
+$(OUTPUT)/tm-poison: CFLAGS += -m64
+
+SIGNAL_CONTEXT_CHK_TESTS := $(patsubst %,$(OUTPUT)/%,$(SIGNAL_CONTEXT_CHK_TESTS))
+$(SIGNAL_CONTEXT_CHK_TESTS): tm-signal.S
+$(SIGNAL_CONTEXT_CHK_TESTS): CFLAGS += -mhtm -m64 -mvsx
diff --git a/tools/testing/selftests/powerpc/tm/settings b/tools/testing/selftests/powerpc/tm/settings
new file mode 100644
index 000000000..e7b941753
--- /dev/null
+++ b/tools/testing/selftests/powerpc/tm/settings
@@ -0,0 +1 @@
+timeout=0
diff --git a/tools/testing/selftests/powerpc/tm/tm-exec.c b/tools/testing/selftests/powerpc/tm/tm-exec.c
new file mode 100644
index 000000000..260cfdb97
--- /dev/null
+++ b/tools/testing/selftests/powerpc/tm/tm-exec.c
@@ -0,0 +1,66 @@
+// SPDX-License-Identifier: GPL-2.0-or-later
+/*
+ * Copyright 2016, Cyril Bur, IBM Corp.
+ *
+ * Syscalls can be performed provided the transactions are suspended.
+ * The exec() class of syscall is unique as a new process is loaded.
+ *
+ * It makes little sense for after an exec() call for the previously
+ * suspended transaction to still exist.
+ */
+
+#define _GNU_SOURCE
+#include <errno.h>
+#include <inttypes.h>
+#include <libgen.h>
+#include <pthread.h>
+#include <stdio.h>
+#include <stdlib.h>
+#include <string.h>
+#include <unistd.h>
+
+#include "utils.h"
+#include "tm.h"
+
+static char *path;
+
+static int test_exec(void)
+{
+	SKIP_IF(!have_htm());
+
+	asm __volatile__(
+		"tbegin.;"
+		"blt    1f; "
+		"tsuspend.;"
+		"1: ;"
+		: : : "memory");
+
+	execl(path, "tm-exec", "--child", NULL);
+
+	/* Shouldn't get here */
+	perror("execl() failed");
+	return 1;
+}
+
+static int after_exec(void)
+{
+	asm __volatile__(
+		"tbegin.;"
+		"blt    1f;"
+		"tsuspend.;"
+		"1: ;"
+		: : : "memory");
+
+	FAIL_IF(failure_is_nesting());
+	return 0;
+}
+
+int main(int argc, char *argv[])
+{
+	path = argv[0];
+
+	if (argc > 1 && strcmp(argv[1], "--child") == 0)
+		return after_exec();
+
+	return test_harness(test_exec, "tm_exec");
+}
diff --git a/tools/testing/selftests/powerpc/tm/tm-fork.c b/tools/testing/selftests/powerpc/tm/tm-fork.c
new file mode 100644
index 000000000..6efa5a685
--- /dev/null
+++ b/tools/testing/selftests/powerpc/tm/tm-fork.c
@@ -0,0 +1,42 @@
+// SPDX-License-Identifier: GPL-2.0-only
+/*
+ * Copyright 2015, Michael Neuling, IBM Corp.
+ *
+ * Edited: Rashmica Gupta, Nov 2015
+ *
+ * This test does a fork syscall inside a transaction. Basic sniff test
+ * to see if we can enter the kernel during a transaction.
+ */
+
+#include <errno.h>
+#include <inttypes.h>
+#include <pthread.h>
+#include <stdio.h>
+#include <stdlib.h>
+#include <unistd.h>
+
+#include "utils.h"
+#include "tm.h"
+
+int test_fork(void)
+{
+	SKIP_IF(!have_htm());
+
+	asm __volatile__(
+		"tbegin.;"
+		"blt    1f; "
+		"li     0, 2;"  /* fork syscall */
+		"sc  ;"
+		"tend.;"
+		"1: ;"
+		: : : "memory", "r0");
+	/* If we reach here, we've passed.  Otherwise we've probably crashed
+	 * the kernel */
+
+	return 0;
+}
+
+int main(int argc, char *argv[])
+{
+	return test_harness(test_fork, "tm_fork");
+}
diff --git a/tools/testing/selftests/powerpc/tm/tm-poison.c b/tools/testing/selftests/powerpc/tm/tm-poison.c
new file mode 100644
index 000000000..29e5f26af
--- /dev/null
+++ b/tools/testing/selftests/powerpc/tm/tm-poison.c
@@ -0,0 +1,182 @@
+// SPDX-License-Identifier: GPL-2.0-only
+/*
+ * Copyright 2019, Gustavo Romero, Michael Neuling, IBM Corp.
+ *
+ * This test will spawn two processes. Both will be attached to the same
+ * CPU (CPU 0). The child will be in a loop writing to FP register f31 and
+ * VMX/VEC/Altivec register vr31 a known value, called poison, calling
+ * sched_yield syscall after to allow the parent to switch on the CPU.
+ * Parent will set f31 and vr31 to 1 and in a loop will check if f31 and
+ * vr31 remain 1 as expected until a given timeout (2m). If the issue is
+ * present child's poison will leak into parent's f31 or vr31 registers,
+ * otherwise, poison will never leak into parent's f31 and vr31 registers.
+ */
+
+#define _GNU_SOURCE
+#include <stdio.h>
+#include <stdlib.h>
+#include <unistd.h>
+#include <inttypes.h>
+#include <sched.h>
+#include <sys/types.h>
+#include <signal.h>
+#include <inttypes.h>
+
+#include "tm.h"
+
+int tm_poison_test(void)
+{
+	int cpu, pid;
+	cpu_set_t cpuset;
+	uint64_t poison = 0xdeadbeefc0dec0fe;
+	uint64_t unknown = 0;
+	bool fail_fp = false;
+	bool fail_vr = false;
+
+	SKIP_IF(!have_htm());
+
+	cpu = pick_online_cpu();
+	FAIL_IF(cpu < 0);
+
+	// Attach both Child and Parent to the same CPU
+	CPU_ZERO(&cpuset);
+	CPU_SET(cpu, &cpuset);
+	FAIL_IF(sched_setaffinity(0, sizeof(cpuset), &cpuset) != 0);
+
+	pid = fork();
+	if (!pid) {
+		/**
+		 * child
+		 */
+		while (1) {
+			sched_yield();
+			asm (
+				"mtvsrd 31, %[poison];" // f31 = poison
+				"mtvsrd 63, %[poison];" // vr31 = poison
+
+				: : [poison] "r" (poison) : );
+		}
+	}
+
+	/**
+	 * parent
+	 */
+	asm (
+		/*
+		 * Set r3, r4, and f31 to known value 1 before entering
+		 * in transaction. They won't be written after that.
+		 */
+		"       li      3, 0x1          ;"
+		"       li      4, 0x1          ;"
+		"       mtvsrd  31, 4           ;"
+
+		/*
+		 * The Time Base (TB) is a 64-bit counter register that is
+		 * independent of the CPU clock and which is incremented
+		 * at a frequency of 512000000 Hz, so every 1.953125ns.
+		 * So it's necessary 120s/0.000000001953125s = 61440000000
+		 * increments to get a 2 minutes timeout. Below we set that
+		 * value in r5 and then use r6 to track initial TB value,
+		 * updating TB values in r7 at every iteration and comparing it
+		 * to r6. When r7 (current) - r6 (initial) > 61440000000 we bail
+		 * out since for sure we spent already 2 minutes in the loop.
+		 * SPR 268 is the TB register.
+		 */
+		"       lis     5, 14           ;"
+		"       ori     5, 5, 19996     ;"
+		"       sldi    5, 5, 16        ;" // r5 = 61440000000
+
+		"       mfspr   6, 268          ;" // r6 (TB initial)
+		"1:     mfspr   7, 268          ;" // r7 (TB current)
+		"       subf    7, 6, 7         ;" // r7 - r6 > 61440000000 ?
+		"       cmpd    7, 5            ;"
+		"       bgt     3f              ;" // yes, exit
+
+		/*
+		 * Main loop to check f31
+		 */
+		"       tbegin.                 ;" // no, try again
+		"       beq     1b              ;" // restart if no timeout
+		"       mfvsrd  3, 31           ;" // read f31
+		"       cmpd    3, 4            ;" // f31 == 1 ?
+		"       bne     2f              ;" // broken :-(
+		"       tabort. 3               ;" // try another transaction
+		"2:     tend.                   ;" // commit transaction
+		"3:     mr    %[unknown], 3     ;" // record r3
+
+		: [unknown] "=r" (unknown)
+		:
+		: "cr0", "r3", "r4", "r5", "r6", "r7", "vs31"
+
+		);
+
+	/*
+	 * On leak 'unknown' will contain 'poison' value from child,
+	 * otherwise (no leak) 'unknown' will contain the same value
+	 * as r3 before entering in transactional mode, i.e. 0x1.
+	 */
+	fail_fp = unknown != 0x1;
+	if (fail_fp)
+		printf("Unknown value %#"PRIx64" leaked into f31!\n", unknown);
+	else
+		printf("Good, no poison or leaked value into FP registers\n");
+
+	asm (
+		/*
+		 * Set r3, r4, and vr31 to known value 1 before entering
+		 * in transaction. They won't be written after that.
+		 */
+		"       li      3, 0x1          ;"
+		"       li      4, 0x1          ;"
+		"       mtvsrd  63, 4           ;"
+
+		"       lis     5, 14           ;"
+		"       ori     5, 5, 19996     ;"
+		"       sldi    5, 5, 16        ;" // r5 = 61440000000
+
+		"       mfspr   6, 268          ;" // r6 (TB initial)
+		"1:     mfspr   7, 268          ;" // r7 (TB current)
+		"       subf    7, 6, 7         ;" // r7 - r6 > 61440000000 ?
+		"       cmpd    7, 5            ;"
+		"       bgt     3f              ;" // yes, exit
+
+		/*
+		 * Main loop to check vr31
+		 */
+		"       tbegin.                 ;" // no, try again
+		"       beq     1b              ;" // restart if no timeout
+		"       mfvsrd  3, 63           ;" // read vr31
+		"       cmpd    3, 4            ;" // vr31 == 1 ?
+		"       bne     2f              ;" // broken :-(
+		"       tabort. 3               ;" // try another transaction
+		"2:     tend.                   ;" // commit transaction
+		"3:     mr    %[unknown], 3     ;" // record r3
+
+		: [unknown] "=r" (unknown)
+		:
+		: "cr0", "r3", "r4", "r5", "r6", "r7", "vs63"
+
+		);
+
+	/*
+	 * On leak 'unknown' will contain 'poison' value from child,
+	 * otherwise (no leak) 'unknown' will contain the same value
+	 * as r3 before entering in transactional mode, i.e. 0x1.
+	 */
+	fail_vr = unknown != 0x1;
+	if (fail_vr)
+		printf("Unknown value %#"PRIx64" leaked into vr31!\n", unknown);
+	else
+		printf("Good, no poison or leaked value into VEC registers\n");
+
+	kill(pid, SIGKILL);
+
+	return (fail_fp | fail_vr);
+}
+
+int main(int argc, char *argv[])
+{
+	/* Test completes in about 4m */
+	test_harness_set_timeout(250);
+	return test_harness(tm_poison_test, "tm_poison_test");
+}
diff --git a/tools/testing/selftests/powerpc/tm/tm-resched-dscr.c b/tools/testing/selftests/powerpc/tm/tm-resched-dscr.c
new file mode 100644
index 000000000..4cdb83964
--- /dev/null
+++ b/tools/testing/selftests/powerpc/tm/tm-resched-dscr.c
@@ -0,0 +1,100 @@
+// SPDX-License-Identifier: GPL-2.0
+/* Test context switching to see if the DSCR SPR is correctly preserved
+ * when within a transaction.
+ *
+ * Note: We assume that the DSCR has been left at the default value (0)
+ * for all CPUs.
+ *
+ * Method:
+ *
+ * Set a value into the DSCR.
+ *
+ * Start a transaction, and suspend it (*).
+ *
+ * Hard loop checking to see if the transaction has become doomed.
+ *
+ * Now that we *may* have been preempted, record the DSCR and TEXASR SPRS.
+ *
+ * If the abort was because of a context switch, check the DSCR value.
+ * Otherwise, try again.
+ *
+ * (*) If the transaction is not suspended we can't see the problem because
+ * the transaction abort handler will restore the DSCR to it's checkpointed
+ * value before we regain control.
+ */
+
+#include <inttypes.h>
+#include <stdio.h>
+#include <stdlib.h>
+#include <assert.h>
+#include <asm/tm.h>
+
+#include "utils.h"
+#include "tm.h"
+#include "../pmu/lib.h"
+
+#define SPRN_DSCR       0x03
+
+int test_body(void)
+{
+	uint64_t rv, dscr1 = 1, dscr2, texasr;
+
+	SKIP_IF(!have_htm());
+
+	printf("Check DSCR TM context switch: ");
+	fflush(stdout);
+	for (;;) {
+		asm __volatile__ (
+			/* set a known value into the DSCR */
+			"ld      3, %[dscr1];"
+			"mtspr   %[sprn_dscr], 3;"
+
+			"li      %[rv], 1;"
+			/* start and suspend a transaction */
+			"tbegin.;"
+			"beq     1f;"
+			"tsuspend.;"
+
+			/* hard loop until the transaction becomes doomed */
+			"2: ;"
+			"tcheck 0;"
+			"bc      4, 0, 2b;"
+
+			/* record DSCR and TEXASR */
+			"mfspr   3, %[sprn_dscr];"
+			"std     3, %[dscr2];"
+			"mfspr   3, %[sprn_texasr];"
+			"std     3, %[texasr];"
+
+			"tresume.;"
+			"tend.;"
+			"li      %[rv], 0;"
+			"1: ;"
+			: [rv]"=r"(rv), [dscr2]"=m"(dscr2), [texasr]"=m"(texasr)
+			: [dscr1]"m"(dscr1)
+			, [sprn_dscr]"i"(SPRN_DSCR), [sprn_texasr]"i"(SPRN_TEXASR)
+			: "memory", "r3"
+		);
+		assert(rv); /* make sure the transaction aborted */
+		if ((texasr >> 56) != TM_CAUSE_RESCHED) {
+			continue;
+		}
+		if (dscr2 != dscr1) {
+			printf(" FAIL\n");
+			return 1;
+		} else {
+			printf(" OK\n");
+			return 0;
+		}
+	}
+}
+
+static int tm_resched_dscr(void)
+{
+	return eat_cpu(test_body);
+}
+
+int main(int argc, const char *argv[])
+{
+	return test_harness(tm_resched_dscr, "tm_resched_dscr");
+}
diff --git a/tools/testing/selftests/powerpc/tm/tm-signal-context-chk-fpu.c b/tools/testing/selftests/powerpc/tm/tm-signal-context-chk-fpu.c
new file mode 100644
index 000000000..254f912ad
--- /dev/null
+++ b/tools/testing/selftests/powerpc/tm/tm-signal-context-chk-fpu.c
@@ -0,0 +1,110 @@
+// SPDX-License-Identifier: GPL-2.0-or-later
+/*
+ * Copyright 2016, Cyril Bur, IBM Corp.
+ *
+ * Test the kernel's signal frame code.
+ *
+ * The kernel sets up two sets of ucontexts if the signal was to be
+ * delivered while the thread was in a transaction (referred too as
+ * first and second contexts).
+ * Expected behaviour is that the checkpointed state is in the user
+ * context passed to the signal handler (first context). The speculated
+ * state can be accessed with the uc_link pointer (second context).
+ *
+ * The rationale for this is that if TM unaware code (which linked
+ * against TM libs) installs a signal handler it will not know of the
+ * speculative nature of the 'live' registers and may infer the wrong
+ * thing.
+ */
+
+#include <stdlib.h>
+#include <stdio.h>
+#include <signal.h>
+#include <unistd.h>
+
+#include <altivec.h>
+
+#include "utils.h"
+#include "tm.h"
+
+#define MAX_ATTEMPT 500000
+
+#define NV_FPU_REGS 18 /* Number of non-volatile FP registers */
+#define FPR14 14 /* First non-volatile FP register to check in f14-31 subset */
+
+long tm_signal_self_context_load(pid_t pid, long *gprs, double *fps, vector int *vms, vector int *vss);
+
+/* Test only non-volatile registers, i.e. 18 fpr registers from f14 to f31 */
+static double fps[] = {
+	/* First context will be set with these values, i.e. non-speculative */
+	 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18,
+	/* Second context will be set with these values, i.e. speculative */
+	-1,-2,-3,-4,-5,-6,-7,-8,-9,-10,-11,-12,-13,-14,-15,-16,-17,-18
+};
+
+static sig_atomic_t fail, broken;
+
+static void signal_usr1(int signum, siginfo_t *info, void *uc)
+{
+	int i;
+	ucontext_t *ucp = uc;
+	ucontext_t *tm_ucp = ucp->uc_link;
+
+	for (i = 0; i < NV_FPU_REGS; i++) {
+		/* Check first context. Print all mismatches. */
+		fail = (ucp->uc_mcontext.fp_regs[FPR14 + i] != fps[i]);
+		if (fail) {
+			broken = 1;
+			printf("FPR%d (1st context) == %g instead of %g (expected)\n",
+				FPR14 + i, ucp->uc_mcontext.fp_regs[FPR14 + i], fps[i]);
+		}
+	}
+
+	for (i = 0; i < NV_FPU_REGS; i++) {
+		/* Check second context. Print all mismatches. */
+		fail = (tm_ucp->uc_mcontext.fp_regs[FPR14 + i] != fps[NV_FPU_REGS + i]);
+		if (fail) {
+			broken = 1;
+			printf("FPR%d (2nd context) == %g instead of %g (expected)\n",
+				FPR14 + i, tm_ucp->uc_mcontext.fp_regs[FPR14 + i], fps[NV_FPU_REGS + i]);
+		}
+	}
+}
+
+static int tm_signal_context_chk_fpu()
+{
+	struct sigaction act;
+	int i;
+	long rc;
+	pid_t pid = getpid();
+
+	SKIP_IF(!have_htm());
+
+	act.sa_sigaction = signal_usr1;
+	sigemptyset(&act.sa_mask);
+	act.sa_flags = SA_SIGINFO;
+	if (sigaction(SIGUSR1, &act, NULL) < 0) {
+		perror("sigaction sigusr1");
+		exit(1);
+	}
+
+	i = 0;
+	while (i < MAX_ATTEMPT && !broken) {
+		/*
+		 * tm_signal_self_context_load will set both first and second
+		 * contexts accordingly to the values passed through non-NULL
+		 * array pointers to it, in that case 'fps', and invoke the
+		 * signal handler installed for SIGUSR1.
+		 */
+		rc = tm_signal_self_context_load(pid, NULL, fps, NULL, NULL);
+		FAIL_IF(rc != pid);
+		i++;
+	}
+
+	return (broken);
+}
+
+int main(void)
+{
+	return test_harness(tm_signal_context_chk_fpu, "tm_signal_context_chk_fpu");
+}
diff --git a/tools/testing/selftests/powerpc/tm/tm-signal-context-chk-gpr.c b/tools/testing/selftests/powerpc/tm/tm-signal-context-chk-gpr.c
new file mode 100644
index 000000000..0cc680f61
--- /dev/null
+++ b/tools/testing/selftests/powerpc/tm/tm-signal-context-chk-gpr.c
@@ -0,0 +1,112 @@
+// SPDX-License-Identifier: GPL-2.0-or-later
+/*
+ * Copyright 2016, Cyril Bur, IBM Corp.
+ *
+ * Test the kernel's signal frame code.
+ *
+ * The kernel sets up two sets of ucontexts if the signal was to be
+ * delivered while the thread was in a transaction (referred too as
+ * first and second contexts).
+ * Expected behaviour is that the checkpointed state is in the user
+ * context passed to the signal handler (first context). The speculated
+ * state can be accessed with the uc_link pointer (second context).
+ *
+ * The rationale for this is that if TM unaware code (which linked
+ * against TM libs) installs a signal handler it will not know of the
+ * speculative nature of the 'live' registers and may infer the wrong
+ * thing.
+ */
+
+#include <stdlib.h>
+#include <stdio.h>
+#include <signal.h>
+#include <unistd.h>
+
+#include <altivec.h>
+
+#include "utils.h"
+#include "tm.h"
+
+#define MAX_ATTEMPT 500000
+
+#define NV_GPR_REGS 18 /* Number of non-volatile GPR registers */
+#define R14 14 /* First non-volatile register to check in r14-r31 subset */
+
+long tm_signal_self_context_load(pid_t pid, long *gprs, double *fps, vector int *vms, vector int *vss);
+
+static sig_atomic_t fail, broken;
+
+/* Test only non-volatile general purpose registers, i.e. r14-r31 */
+static long gprs[] = {
+	/* First context will be set with these values, i.e. non-speculative */
+	/* R14, R15, ... */
+	 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18,
+	/* Second context will be set with these values, i.e. speculative */
+	/* R14, R15, ... */
+	-1,-2,-3,-4,-5,-6,-7,-8,-9,-10,-11,-12,-13,-14,-15,-16,-17,-18
+};
+
+static void signal_usr1(int signum, siginfo_t *info, void *uc)
+{
+	int i;
+	ucontext_t *ucp = uc;
+	ucontext_t *tm_ucp = ucp->uc_link;
+
+	/* Check first context. Print all mismatches. */
+	for (i = 0; i < NV_GPR_REGS; i++) {
+		fail = (ucp->uc_mcontext.gp_regs[R14 + i] != gprs[i]);
+		if (fail) {
+			broken = 1;
+			printf("GPR%d (1st context) == %lu instead of %lu (expected)\n",
+				R14 + i, ucp->uc_mcontext.gp_regs[R14 + i], gprs[i]);
+		}
+	}
+
+	/* Check second context. Print all mismatches. */
+	for (i = 0; i < NV_GPR_REGS; i++) {
+		fail = (tm_ucp->uc_mcontext.gp_regs[R14 + i] != gprs[NV_GPR_REGS + i]);
+		if (fail) {
+			broken = 1;
+			printf("GPR%d (2nd context) == %lu instead of %lu (expected)\n",
+				R14 + i, tm_ucp->uc_mcontext.gp_regs[R14 + i], gprs[NV_GPR_REGS + i]);
+		}
+	}
+}
+
+static int tm_signal_context_chk_gpr()
+{
+	struct sigaction act;
+	int i;
+	long rc;
+	pid_t pid = getpid();
+
+	SKIP_IF(!have_htm());
+
+	act.sa_sigaction = signal_usr1;
+	sigemptyset(&act.sa_mask);
+	act.sa_flags = SA_SIGINFO;
+	if (sigaction(SIGUSR1, &act, NULL) < 0) {
+		perror("sigaction sigusr1");
+		exit(1);
+	}
+
+	i = 0;
+	while (i < MAX_ATTEMPT && !broken) {
+                /*
+                 * tm_signal_self_context_load will set both first and second
+                 * contexts accordingly to the values passed through non-NULL
+                 * array pointers to it, in that case 'gprs', and invoke the
+                 * signal handler installed for SIGUSR1.
+                 */
+		rc = tm_signal_self_context_load(pid, gprs, NULL, NULL, NULL);
+		FAIL_IF(rc != pid);
+		i++;
+	}
+
+	return broken;
+}
+
+int main(void)
+{
+	return test_harness(tm_signal_context_chk_gpr, "tm_signal_context_chk_gpr");
+}
diff --git a/tools/testing/selftests/powerpc/tm/tm-signal-context-chk-vmx.c b/tools/testing/selftests/powerpc/tm/tm-signal-context-chk-vmx.c
new file mode 100644
index 000000000..b6d52730a
--- /dev/null
+++ b/tools/testing/selftests/powerpc/tm/tm-signal-context-chk-vmx.c
@@ -0,0 +1,135 @@
+// SPDX-License-Identifier: GPL-2.0-or-later
+/*
+ * Copyright 2016, Cyril Bur, IBM Corp.
+ *
+ * Test the kernel's signal frame code.
+ *
+ * The kernel sets up two sets of ucontexts if the signal was to be
+ * delivered while the thread was in a transaction (referred too as
+ * first and second contexts).
+ * Expected behaviour is that the checkpointed state is in the user
+ * context passed to the signal handler (first context). The speculated
+ * state can be accessed with the uc_link pointer (second context).
+ *
+ * The rationale for this is that if TM unaware code (which linked
+ * against TM libs) installs a signal handler it will not know of the
+ * speculative nature of the 'live' registers and may infer the wrong
+ * thing.
+ */
+
+#include <stdlib.h>
+#include <stdio.h>
+#include <string.h>
+#include <signal.h>
+#include <unistd.h>
+
+#include <altivec.h>
+
+#include "utils.h"
+#include "tm.h"
+
+#define MAX_ATTEMPT 500000
+
+#define NV_VMX_REGS 12 /* Number of non-volatile VMX registers */
+#define VMX20 20 /* First non-volatile register to check in vr20-31 subset */
+
+long tm_signal_self_context_load(pid_t pid, long *gprs, double *fps, vector int *vms, vector int *vss);
+
+static sig_atomic_t fail, broken;
+
+/* Test only non-volatile registers, i.e. 12 vmx registers from vr20 to vr31 */
+vector int vms[] = {
+	/* First context will be set with these values, i.e. non-speculative */
+	/* VMX20     ,  VMX21      , ... */
+	{ 1, 2, 3, 4},{ 5, 6, 7, 8},{ 9,10,11,12},
+	{13,14,15,16},{17,18,19,20},{21,22,23,24},
+	{25,26,27,28},{29,30,31,32},{33,34,35,36},
+	{37,38,39,40},{41,42,43,44},{45,46,47,48},
+	/* Second context will be set with these values, i.e. speculative */
+	/* VMX20        , VMX21            , ... */
+	{ -1, -2, -3, -4},{ -5, -6, -7, -8},{ -9,-10,-11,-12},
+	{-13,-14,-15,-16},{-17,-18,-19,-20},{-21,-22,-23,-24},
+	{-25,-26,-27,-28},{-29,-30,-31,-32},{-33,-34,-35,-36},
+	{-37,-38,-39,-40},{-41,-42,-43,-44},{-45,-46,-47,-48}
+};
+
+static void signal_usr1(int signum, siginfo_t *info, void *uc)
+{
+	int i, j;
+	ucontext_t *ucp = uc;
+	ucontext_t *tm_ucp = ucp->uc_link;
+
+	for (i = 0; i < NV_VMX_REGS; i++) {
+		/* Check first context. Print all mismatches. */
+		fail = memcmp(ucp->uc_mcontext.v_regs->vrregs[VMX20 + i],
+				&vms[i], sizeof(vector int));
+		if (fail) {
+			broken = 1;
+			printf("VMX%d (1st context) == 0x", VMX20 + i);
+			/* Print actual value in first context. */
+			for (j = 0; j < 4; j++)
+				printf("%08x", ucp->uc_mcontext.v_regs->vrregs[VMX20 + i][j]);
+			printf(" instead of 0x");
+			/* Print expected value. */
+			for (j = 0; j < 4; j++)
+				printf("%08x", vms[i][j]);
+			printf(" (expected)\n");
+		}
+	}
+
+	for (i = 0; i < NV_VMX_REGS; i++)  {
+		/* Check second context. Print all mismatches. */
+		fail = memcmp(tm_ucp->uc_mcontext.v_regs->vrregs[VMX20 + i],
+				&vms[NV_VMX_REGS + i], sizeof (vector int));
+		if (fail) {
+			broken = 1;
+			printf("VMX%d (2nd context) == 0x", NV_VMX_REGS + i);
+			/* Print actual value in second context. */
+			for (j = 0; j < 4; j++)
+				printf("%08x", tm_ucp->uc_mcontext.v_regs->vrregs[VMX20 + i][j]);
+			printf(" instead of 0x");
+			/* Print expected value. */
+			for (j = 0; j < 4; j++)
+				printf("%08x", vms[NV_VMX_REGS + i][j]);
+			printf(" (expected)\n");
+		}
+	}
+}
+
+static int tm_signal_context_chk()
+{
+	struct sigaction act;
+	int i;
+	long rc;
+	pid_t pid = getpid();
+
+	SKIP_IF(!have_htm());
+
+	act.sa_sigaction = signal_usr1;
+	sigemptyset(&act.sa_mask);
+	act.sa_flags = SA_SIGINFO;
+	if (sigaction(SIGUSR1, &act, NULL) < 0) {
+		perror("sigaction sigusr1");
+		exit(1);
+	}
+
+	i = 0;
+	while (i < MAX_ATTEMPT && !broken) {
+		/*
+		 * tm_signal_self_context_load will set both first and second
+		 * contexts accordingly to the values passed through non-NULL
+		 * array pointers to it, in that case 'vms', and invoke the
+		 * signal handler installed for SIGUSR1.
+		 */
+		rc = tm_signal_self_context_load(pid, NULL, NULL, vms, NULL);
+		FAIL_IF(rc != pid);
+		i++;
+	}
+
+	return (broken);
+}
+
+int main(void)
+{
+	return test_harness(tm_signal_context_chk, "tm_signal_context_chk_vmx");
+}
diff --git a/tools/testing/selftests/powerpc/tm/tm-signal-context-chk-vsx.c b/tools/testing/selftests/powerpc/tm/tm-signal-context-chk-vsx.c
new file mode 100644
index 000000000..8e25e2072
--- /dev/null
+++ b/tools/testing/selftests/powerpc/tm/tm-signal-context-chk-vsx.c
@@ -0,0 +1,184 @@
+// SPDX-License-Identifier: GPL-2.0-or-later
+/*
+ * Copyright 2016, Cyril Bur, IBM Corp.
+ *
+ * Test the kernel's signal frame code.
+ *
+ * The kernel sets up two sets of ucontexts if the signal was to be
+ * delivered while the thread was in a transaction (referred too as
+ * first and second contexts).
+ * Expected behaviour is that the checkpointed state is in the user
+ * context passed to the signal handler (first context). The speculated
+ * state can be accessed with the uc_link pointer (second context).
+ *
+ * The rationale for this is that if TM unaware code (which linked
+ * against TM libs) installs a signal handler it will not know of the
+ * speculative nature of the 'live' registers and may infer the wrong
+ * thing.
+ */
+
+#include <stdlib.h>
+#include <stdio.h>
+#include <string.h>
+#include <signal.h>
+#include <unistd.h>
+
+#include <altivec.h>
+
+#include "utils.h"
+#include "tm.h"
+
+#define MAX_ATTEMPT 500000
+
+#define NV_VSX_REGS 12 /* Number of VSX registers to check. */
+#define VSX20 20 /* First VSX register to check in vsr20-vsr31 subset */
+#define FPR20 20 /* FPR20 overlaps VSX20 most significant doubleword */
+
+long tm_signal_self_context_load(pid_t pid, long *gprs, double *fps, vector int *vms, vector int *vss);
+
+static sig_atomic_t fail, broken;
+
+/* Test only 12 vsx registers from vsr20 to vsr31 */
+vector int vsxs[] = {
+	/* First context will be set with these values, i.e. non-speculative */
+	/* VSX20     ,  VSX21      , ... */
+	{ 1, 2, 3, 4},{ 5, 6, 7, 8},{ 9,10,11,12},
+	{13,14,15,16},{17,18,19,20},{21,22,23,24},
+	{25,26,27,28},{29,30,31,32},{33,34,35,36},
+	{37,38,39,40},{41,42,43,44},{45,46,47,48},
+	/* Second context will be set with these values, i.e. speculative */
+	/* VSX20         ,  VSX21          , ... */
+	{-1, -2, -3, -4 },{-5, -6, -7, -8 },{-9, -10,-11,-12},
+	{-13,-14,-15,-16},{-17,-18,-19,-20},{-21,-22,-23,-24},
+	{-25,-26,-27,-28},{-29,-30,-31,-32},{-33,-34,-35,-36},
+	{-37,-38,-39,-40},{-41,-42,-43,-44},{-45,-46,-47,-48}
+};
+
+static void signal_usr1(int signum, siginfo_t *info, void *uc)
+{
+	int i, j;
+	uint8_t vsx[sizeof(vector int)];
+	uint8_t vsx_tm[sizeof(vector int)];
+	ucontext_t *ucp = uc;
+	ucontext_t *tm_ucp = ucp->uc_link;
+
+	/*
+	 * FP registers and VMX registers overlap the VSX registers.
+	 *
+	 * FP registers (f0-31) overlap the most significant 64 bits of VSX
+	 * registers vsr0-31, whilst VMX registers vr0-31, being 128-bit like
+	 * the VSX registers, overlap fully the other half of VSX registers,
+	 * i.e. vr0-31 overlaps fully vsr32-63.
+	 *
+	 * Due to compatibility and historical reasons (VMX/Altivec support
+	 * appeared first on the architecture), VMX registers vr0-31 (so VSX
+	 * half vsr32-63 too) are stored right after the v_regs pointer, in an
+	 * area allocated for 'vmx_reverse' array (please see
+	 * arch/powerpc/include/uapi/asm/sigcontext.h for details about the
+	 * mcontext_t structure on Power).
+	 *
+	 * The other VSX half (vsr0-31) is hence stored below vr0-31/vsr32-63
+	 * registers, but only the least significant 64 bits of vsr0-31. The
+	 * most significant 64 bits of vsr0-31 (f0-31), as it overlaps the FP
+	 * registers, is kept in fp_regs.
+	 *
+	 * v_regs is a 16 byte aligned pointer at the start of vmx_reserve
+	 * (vmx_reserve may or may not be 16 aligned) where the v_regs structure
+	 * exists, so v_regs points to where vr0-31 / vsr32-63 registers are
+	 * fully stored. Since v_regs type is elf_vrregset_t, v_regs + 1
+	 * skips all the slots used to store vr0-31 / vsr32-64 and points to
+	 * part of one VSX half, i.e. v_regs + 1 points to the least significant
+	 * 64 bits of vsr0-31. The other part of this half (the most significant
+	 * part of vsr0-31) is stored in fp_regs.
+	 *
+	 */
+	/* Get pointer to least significant doubleword of vsr0-31 */
+	long *vsx_ptr = (long *)(ucp->uc_mcontext.v_regs + 1);
+	long *tm_vsx_ptr = (long *)(tm_ucp->uc_mcontext.v_regs + 1);
+
+	/* Check first context. Print all mismatches. */
+	for (i = 0; i < NV_VSX_REGS; i++) {
+		/*
+		 * Copy VSX most significant doubleword from fp_regs and
+		 * copy VSX least significant one from 64-bit slots below
+		 * saved VMX registers.
+		 */
+		memcpy(vsx, &ucp->uc_mcontext.fp_regs[FPR20 + i], 8);
+		memcpy(vsx + 8, &vsx_ptr[VSX20 + i], 8);
+
+		fail = memcmp(vsx, &vsxs[i], sizeof(vector int));
+
+		if (fail) {
+			broken = 1;
+			printf("VSX%d (1st context) == 0x", VSX20 + i);
+			for (j = 0; j < 16; j++)
+				printf("%02x", vsx[j]);
+			printf(" instead of 0x");
+			for (j = 0; j < 4; j++)
+				printf("%08x", vsxs[i][j]);
+			printf(" (expected)\n");
+		}
+	}
+
+	/* Check second context. Print all mismatches. */
+	for (i = 0; i < NV_VSX_REGS; i++) {
+		/*
+		 * Copy VSX most significant doubleword from fp_regs and
+		 * copy VSX least significant one from 64-bit slots below
+		 * saved VMX registers.
+		 */
+		memcpy(vsx_tm, &tm_ucp->uc_mcontext.fp_regs[FPR20 + i], 8);
+		memcpy(vsx_tm + 8, &tm_vsx_ptr[VSX20 + i], 8);
+
+		fail = memcmp(vsx_tm, &vsxs[NV_VSX_REGS + i], sizeof(vector int));
+
+		if (fail) {
+			broken = 1;
+			printf("VSX%d (2nd context) == 0x", VSX20 + i);
+			for (j = 0; j < 16; j++)
+				printf("%02x", vsx_tm[j]);
+			printf(" instead of 0x");
+			for (j = 0; j < 4; j++)
+				printf("%08x", vsxs[NV_VSX_REGS + i][j]);
+			printf("(expected)\n");
+		}
+	}
+}
+
+static int tm_signal_context_chk()
+{
+	struct sigaction act;
+	int i;
+	long rc;
+	pid_t pid = getpid();
+
+	SKIP_IF(!have_htm());
+
+	act.sa_sigaction = signal_usr1;
+	sigemptyset(&act.sa_mask);
+	act.sa_flags = SA_SIGINFO;
+	if (sigaction(SIGUSR1, &act, NULL) < 0) {
+		perror("sigaction sigusr1");
+		exit(1);
+	}
+
+	i = 0;
+	while (i < MAX_ATTEMPT && !broken) {
+               /*
+                * tm_signal_self_context_load will set both first and second
+                * contexts accordingly to the values passed through non-NULL
+                * array pointers to it, in that case 'vsxs', and invoke the
+                * signal handler installed for SIGUSR1.
+                */
+		rc = tm_signal_self_context_load(pid, NULL, NULL, NULL, vsxs);
+		FAIL_IF(rc != pid);
+		i++;
+	}
+
+	return (broken);
+}
+
+int main(void)
+{
+	return test_harness(tm_signal_context_chk, "tm_signal_context_chk_vsx");
+}
diff --git a/tools/testing/selftests/powerpc/tm/tm-signal-context-force-tm.c b/tools/testing/selftests/powerpc/tm/tm-signal-context-force-tm.c
new file mode 100644
index 000000000..421cb082f
--- /dev/null
+++ b/tools/testing/selftests/powerpc/tm/tm-signal-context-force-tm.c
@@ -0,0 +1,180 @@
+// SPDX-License-Identifier: GPL-2.0
+/*
+ * Copyright 2018, Breno Leitao, Gustavo Romero, IBM Corp.
+ *
+ * This test raises a SIGUSR1 signal, and toggle the MSR[TS]
+ * fields at the signal handler. With MSR[TS] being set, the kernel will
+ * force a recheckpoint, which may cause a segfault when returning to
+ * user space. Since the test needs to re-run, the segfault needs to be
+ * caught and handled.
+ *
+ * In order to continue the test even after a segfault, the context is
+ * saved prior to the signal being raised, and it is restored when there is
+ * a segmentation fault. This happens for COUNT_MAX times.
+ *
+ * This test never fails (as returning EXIT_FAILURE). It either succeeds,
+ * or crash the kernel (on a buggy kernel).
+ */
+
+#define _GNU_SOURCE
+#include <stdio.h>
+#include <stdlib.h>
+#include <signal.h>
+#include <string.h>
+#include <ucontext.h>
+#include <unistd.h>
+#include <sys/mman.h>
+
+#include "tm.h"
+#include "utils.h"
+#include "reg.h"
+
+#define COUNT_MAX       5000		/* Number of interactions */
+
+/*
+ * This test only runs on 64 bits system. Unsetting MSR_TS_S to avoid
+ * compilation issue on 32 bits system. There is no side effect, since the
+ * whole test will be skipped if it is not running on 64 bits system.
+ */
+#ifndef __powerpc64__
+#undef  MSR_TS_S
+#define MSR_TS_S	0
+#endif
+
+/* Setting contexts because the test will crash and we want to recover */
+ucontext_t init_context;
+
+/* count is changed in the signal handler, so it must be volatile */
+static volatile int count;
+
+void usr_signal_handler(int signo, siginfo_t *si, void *uc)
+{
+	ucontext_t *ucp = uc;
+	int ret;
+
+	/*
+	 * Allocating memory in a signal handler, and never freeing it on
+	 * purpose, forcing the heap increase, so, the memory leak is what
+	 * we want here.
+	 */
+	ucp->uc_link = mmap(NULL, sizeof(ucontext_t),
+			    PROT_READ | PROT_WRITE,
+			    MAP_PRIVATE | MAP_ANONYMOUS, 0, 0);
+	if (ucp->uc_link == (void *)-1) {
+		perror("Mmap failed");
+		exit(-1);
+	}
+
+	/* Forcing the page to be allocated in a page fault */
+	ret = madvise(ucp->uc_link, sizeof(ucontext_t), MADV_DONTNEED);
+	if (ret) {
+		perror("madvise failed");
+		exit(-1);
+	}
+
+	memcpy(&ucp->uc_link->uc_mcontext, &ucp->uc_mcontext,
+		sizeof(ucp->uc_mcontext));
+
+	/* Forcing to enable MSR[TM] */
+	UCONTEXT_MSR(ucp) |= MSR_TS_S;
+
+	/*
+	 * A fork inside a signal handler seems to be more efficient than a
+	 * fork() prior to the signal being raised.
+	 */
+	if (fork() == 0) {
+		/*
+		 * Both child and parent will return, but, child returns
+		 * with count set so it will exit in the next segfault.
+		 * Parent will continue to loop.
+		 */
+		count = COUNT_MAX;
+	}
+
+	/*
+	 * If the change above does not hit the bug, it will cause a
+	 * segmentation fault, since the ck structures are NULL.
+	 */
+}
+
+void seg_signal_handler(int signo, siginfo_t *si, void *uc)
+{
+	count++;
+
+	/* Reexecute the test */
+	setcontext(&init_context);
+}
+
+void tm_trap_test(void)
+{
+	struct sigaction usr_sa, seg_sa;
+	stack_t ss;
+
+	usr_sa.sa_flags = SA_SIGINFO | SA_ONSTACK;
+	usr_sa.sa_sigaction = usr_signal_handler;
+
+	seg_sa.sa_flags = SA_SIGINFO;
+	seg_sa.sa_sigaction = seg_signal_handler;
+
+	/*
+	 * Set initial context. Will get back here from
+	 * seg_signal_handler()
+	 */
+	getcontext(&init_context);
+
+	while (count < COUNT_MAX) {
+		/* Allocated an alternative signal stack area */
+		ss.ss_sp = mmap(NULL, SIGSTKSZ, PROT_READ | PROT_WRITE,
+				MAP_PRIVATE | MAP_ANONYMOUS, 0, 0);
+		ss.ss_size = SIGSTKSZ;
+		ss.ss_flags = 0;
+
+		if (ss.ss_sp == (void *)-1) {
+			perror("mmap error\n");
+			exit(-1);
+		}
+
+		/* Force the allocation through a page fault */
+		if (madvise(ss.ss_sp, SIGSTKSZ, MADV_DONTNEED)) {
+			perror("madvise\n");
+			exit(-1);
+		}
+
+		/*
+		 * Setting an alternative stack to generate a page fault when
+		 * the signal is raised.
+		 */
+		if (sigaltstack(&ss, NULL)) {
+			perror("sigaltstack\n");
+			exit(-1);
+		}
+
+		/* The signal handler will enable MSR_TS */
+		sigaction(SIGUSR1, &usr_sa, NULL);
+		/* If it does not crash, it might segfault, avoid it to retest */
+		sigaction(SIGSEGV, &seg_sa, NULL);
+
+		raise(SIGUSR1);
+		count++;
+	}
+}
+
+int tm_signal_context_force_tm(void)
+{
+	SKIP_IF(!have_htm());
+	/*
+	 * Skipping if not running on 64 bits system, since I think it is
+	 * not possible to set mcontext's [MSR] with TS, due to it being 32
+	 * bits.
+	 */
+	SKIP_IF(!is_ppc64le());
+
+	tm_trap_test();
+
+	return EXIT_SUCCESS;
+}
+
+int main(int argc, char **argv)
+{
+	test_harness(tm_signal_context_force_tm, "tm_signal_context_force_tm");
+}
diff --git a/tools/testing/selftests/powerpc/tm/tm-signal-msr-resv.c b/tools/testing/selftests/powerpc/tm/tm-signal-msr-resv.c
new file mode 100644
index 000000000..4a61e9bd1
--- /dev/null
+++ b/tools/testing/selftests/powerpc/tm/tm-signal-msr-resv.c
@@ -0,0 +1,74 @@
+// SPDX-License-Identifier: GPL-2.0-only
+/*
+ * Copyright 2015, Michael Neuling, IBM Corp.
+ *
+ * Test the kernel's signal return code to ensure that it doesn't
+ * crash when both the transactional and suspend MSR bits are set in
+ * the signal context.
+ *
+ * For this test, we send ourselves a SIGUSR1.  In the SIGUSR1 handler
+ * we modify the signal context to set both MSR TM S and T bits (which
+ * is "reserved" by the PowerISA). When we return from the signal
+ * handler (implicit sigreturn), the kernel should detect reserved MSR
+ * value and send us with a SIGSEGV.
+ */
+
+#include <stdlib.h>
+#include <stdio.h>
+#include <signal.h>
+#include <unistd.h>
+
+#include "utils.h"
+#include "tm.h"
+
+int segv_expected = 0;
+
+void signal_segv(int signum)
+{
+	if (segv_expected && (signum == SIGSEGV))
+		_exit(0);
+	_exit(1);
+}
+
+void signal_usr1(int signum, siginfo_t *info, void *uc)
+{
+	ucontext_t *ucp = uc;
+
+	/* Link tm checkpointed context to normal context */
+	ucp->uc_link = ucp;
+	/* Set all TM bits so that the context is now invalid */
+#ifdef __powerpc64__
+	ucp->uc_mcontext.gp_regs[PT_MSR] |= (7ULL << 32);
+#else
+	ucp->uc_mcontext.uc_regs->gregs[PT_MSR] |= (7ULL);
+#endif
+	/* Should segv on return becuase of invalid context */
+	segv_expected = 1;
+}
+
+int tm_signal_msr_resv()
+{
+	struct sigaction act;
+
+	SKIP_IF(!have_htm());
+
+	act.sa_sigaction = signal_usr1;
+	sigemptyset(&act.sa_mask);
+	act.sa_flags = SA_SIGINFO;
+	if (sigaction(SIGUSR1, &act, NULL) < 0) {
+		perror("sigaction sigusr1");
+		exit(1);
+	}
+	if (signal(SIGSEGV, signal_segv) == SIG_ERR)
+		exit(1);
+
+	raise(SIGUSR1);
+
+	/* We shouldn't get here as we exit in the segv handler */
+	return 1;
+}
+
+int main(void)
+{
+	return test_harness(tm_signal_msr_resv, "tm_signal_msr_resv");
+}
diff --git a/tools/testing/selftests/powerpc/tm/tm-signal-pagefault.c b/tools/testing/selftests/powerpc/tm/tm-signal-pagefault.c
new file mode 100644
index 000000000..5908bc6ab
--- /dev/null
+++ b/tools/testing/selftests/powerpc/tm/tm-signal-pagefault.c
@@ -0,0 +1,284 @@
+// SPDX-License-Identifier: GPL-2.0
+/*
+ * Copyright 2020, Gustavo Luiz Duarte, IBM Corp.
+ *
+ * This test starts a transaction and triggers a signal, forcing a pagefault to
+ * happen when the kernel signal handling code touches the user signal stack.
+ *
+ * In order to avoid pre-faulting the signal stack memory and to force the
+ * pagefault to happen precisely in the kernel signal handling code, the
+ * pagefault handling is done in userspace using the userfaultfd facility.
+ *
+ * Further pagefaults are triggered by crafting the signal handler's ucontext
+ * to point to additional memory regions managed by the userfaultfd, so using
+ * the same mechanism used to avoid pre-faulting the signal stack memory.
+ *
+ * On failure (bug is present) kernel crashes or never returns control back to
+ * userspace. If bug is not present, tests completes almost immediately.
+ */
+
+#include <stdio.h>
+#include <stdlib.h>
+#include <string.h>
+#include <linux/userfaultfd.h>
+#include <poll.h>
+#include <unistd.h>
+#include <sys/ioctl.h>
+#include <sys/syscall.h>
+#include <fcntl.h>
+#include <sys/mman.h>
+#include <pthread.h>
+#include <signal.h>
+#include <errno.h>
+
+#include "tm.h"
+
+
+#define UF_MEM_SIZE 655360	/* 10 x 64k pages */
+
+/* Memory handled by userfaultfd */
+static char *uf_mem;
+static size_t uf_mem_offset = 0;
+
+/*
+ * Data that will be copied into the faulting pages (instead of zero-filled
+ * pages). This is used to make the test more reliable and avoid segfaulting
+ * when we return from the signal handler. Since we are making the signal
+ * handler's ucontext point to newly allocated memory, when that memory is
+ * paged-in it will contain the expected content.
+ */
+static char backing_mem[UF_MEM_SIZE];
+
+static size_t pagesize;
+
+/*
+ * Return a chunk of at least 'size' bytes of memory that will be handled by
+ * userfaultfd. If 'backing_data' is not NULL, its content will be save to
+ * 'backing_mem' and then copied into the faulting pages when the page fault
+ * is handled.
+ */
+void *get_uf_mem(size_t size, void *backing_data)
+{
+	void *ret;
+
+	if (uf_mem_offset + size > UF_MEM_SIZE) {
+		fprintf(stderr, "Requesting more uf_mem than expected!\n");
+		exit(EXIT_FAILURE);
+	}
+
+	ret = &uf_mem[uf_mem_offset];
+
+	/* Save the data that will be copied into the faulting page */
+	if (backing_data != NULL)
+		memcpy(&backing_mem[uf_mem_offset], backing_data, size);
+
+	/* Reserve the requested amount of uf_mem */
+	uf_mem_offset += size;
+	/* Keep uf_mem_offset aligned to the page size (round up) */
+	uf_mem_offset = (uf_mem_offset + pagesize - 1) & ~(pagesize - 1);
+
+	return ret;
+}
+
+void *fault_handler_thread(void *arg)
+{
+	struct uffd_msg msg;	/* Data read from userfaultfd */
+	long uffd;		/* userfaultfd file descriptor */
+	struct uffdio_copy uffdio_copy;
+	struct pollfd pollfd;
+	ssize_t nread, offset;
+
+	uffd = (long) arg;
+
+	for (;;) {
+		pollfd.fd = uffd;
+		pollfd.events = POLLIN;
+		if (poll(&pollfd, 1, -1) == -1) {
+			perror("poll() failed");
+			exit(EXIT_FAILURE);
+		}
+
+		nread = read(uffd, &msg, sizeof(msg));
+		if (nread == 0) {
+			fprintf(stderr, "read(): EOF on userfaultfd\n");
+			exit(EXIT_FAILURE);
+		}
+
+		if (nread == -1) {
+			perror("read() failed");
+			exit(EXIT_FAILURE);
+		}
+
+		/* We expect only one kind of event */
+		if (msg.event != UFFD_EVENT_PAGEFAULT) {
+			fprintf(stderr, "Unexpected event on userfaultfd\n");
+			exit(EXIT_FAILURE);
+		}
+
+		/*
+		 * We need to handle page faults in units of pages(!).
+		 * So, round faulting address down to page boundary.
+		 */
+		uffdio_copy.dst = msg.arg.pagefault.address & ~(pagesize-1);
+
+		offset = (char *) uffdio_copy.dst - uf_mem;
+		uffdio_copy.src = (unsigned long) &backing_mem[offset];
+
+		uffdio_copy.len = pagesize;
+		uffdio_copy.mode = 0;
+		uffdio_copy.copy = 0;
+		if (ioctl(uffd, UFFDIO_COPY, &uffdio_copy) == -1) {
+			perror("ioctl-UFFDIO_COPY failed");
+			exit(EXIT_FAILURE);
+		}
+	}
+}
+
+void setup_uf_mem(void)
+{
+	long uffd;		/* userfaultfd file descriptor */
+	pthread_t thr;
+	struct uffdio_api uffdio_api;
+	struct uffdio_register uffdio_register;
+	int ret;
+
+	pagesize = sysconf(_SC_PAGE_SIZE);
+
+	/* Create and enable userfaultfd object */
+	uffd = syscall(__NR_userfaultfd, O_CLOEXEC | O_NONBLOCK);
+	if (uffd == -1) {
+		perror("userfaultfd() failed");
+		exit(EXIT_FAILURE);
+	}
+	uffdio_api.api = UFFD_API;
+	uffdio_api.features = 0;
+	if (ioctl(uffd, UFFDIO_API, &uffdio_api) == -1) {
+		perror("ioctl-UFFDIO_API failed");
+		exit(EXIT_FAILURE);
+	}
+
+	/*
+	 * Create a private anonymous mapping. The memory will be demand-zero
+	 * paged, that is, not yet allocated. When we actually touch the memory
+	 * the related page will be allocated via the userfaultfd mechanism.
+	 */
+	uf_mem = mmap(NULL, UF_MEM_SIZE, PROT_READ | PROT_WRITE,
+		      MAP_PRIVATE | MAP_ANONYMOUS, -1, 0);
+	if (uf_mem == MAP_FAILED) {
+		perror("mmap() failed");
+		exit(EXIT_FAILURE);
+	}
+
+	/*
+	 * Register the memory range of the mapping we've just mapped to be
+	 * handled by the userfaultfd object. In 'mode' we request to track
+	 * missing pages (i.e. pages that have not yet been faulted-in).
+	 */
+	uffdio_register.range.start = (unsigned long) uf_mem;
+	uffdio_register.range.len = UF_MEM_SIZE;
+	uffdio_register.mode = UFFDIO_REGISTER_MODE_MISSING;
+	if (ioctl(uffd, UFFDIO_REGISTER, &uffdio_register) == -1) {
+		perror("ioctl-UFFDIO_REGISTER");
+		exit(EXIT_FAILURE);
+	}
+
+	/* Create a thread that will process the userfaultfd events */
+	ret = pthread_create(&thr, NULL, fault_handler_thread, (void *) uffd);
+	if (ret != 0) {
+		fprintf(stderr, "pthread_create(): Error. Returned %d\n", ret);
+		exit(EXIT_FAILURE);
+	}
+}
+
+/*
+ * Assumption: the signal was delivered while userspace was in transactional or
+ * suspended state, i.e. uc->uc_link != NULL.
+ */
+void signal_handler(int signo, siginfo_t *si, void *uc)
+{
+	ucontext_t *ucp = uc;
+
+	/* Skip 'trap' after returning, otherwise we get a SIGTRAP again */
+	ucp->uc_link->uc_mcontext.regs->nip += 4;
+
+	ucp->uc_mcontext.v_regs =
+		get_uf_mem(sizeof(elf_vrreg_t), ucp->uc_mcontext.v_regs);
+
+	ucp->uc_link->uc_mcontext.v_regs =
+		get_uf_mem(sizeof(elf_vrreg_t), ucp->uc_link->uc_mcontext.v_regs);
+
+	ucp->uc_link = get_uf_mem(sizeof(ucontext_t), ucp->uc_link);
+}
+
+bool have_userfaultfd(void)
+{
+	long rc;
+
+	errno = 0;
+	rc = syscall(__NR_userfaultfd, -1);
+
+	return rc == 0 || errno != ENOSYS;
+}
+
+int tm_signal_pagefault(void)
+{
+	struct sigaction sa;
+	stack_t ss;
+
+	SKIP_IF(!have_htm());
+	SKIP_IF(!have_userfaultfd());
+
+	setup_uf_mem();
+
+	/*
+	 * Set an alternative stack that will generate a page fault when the
+	 * signal is raised. The page fault will be treated via userfaultfd,
+	 * i.e. via fault_handler_thread.
+	 */
+	ss.ss_sp = get_uf_mem(SIGSTKSZ, NULL);
+	ss.ss_size = SIGSTKSZ;
+	ss.ss_flags = 0;
+	if (sigaltstack(&ss, NULL) == -1) {
+		perror("sigaltstack() failed");
+		exit(EXIT_FAILURE);
+	}
+
+	sa.sa_flags = SA_SIGINFO | SA_ONSTACK;
+	sa.sa_sigaction = signal_handler;
+	if (sigaction(SIGTRAP, &sa, NULL) == -1) {
+		perror("sigaction() failed");
+		exit(EXIT_FAILURE);
+	}
+
+	/* Trigger a SIGTRAP in transactional state */
+	asm __volatile__(
+			"tbegin.;"
+			"beq    1f;"
+			"trap;"
+			"1: ;"
+			: : : "memory");
+
+	/* Trigger a SIGTRAP in suspended state */
+	asm __volatile__(
+			"tbegin.;"
+			"beq    1f;"
+			"tsuspend.;"
+			"trap;"
+			"tresume.;"
+			"1: ;"
+			: : : "memory");
+
+	return EXIT_SUCCESS;
+}
+
+int main(int argc, char **argv)
+{
+	/*
+	 * Depending on kernel config, the TM Bad Thing might not result in a
+	 * crash, instead the kernel never returns control back to userspace, so
+	 * set a tight timeout. If the test passes it completes almost
+	 * immediately.
+	 */
+	test_harness_set_timeout(2);
+	return test_harness(tm_signal_pagefault, "tm_signal_pagefault");
+}
diff --git a/tools/testing/selftests/powerpc/tm/tm-signal-sigreturn-nt.c b/tools/testing/selftests/powerpc/tm/tm-signal-sigreturn-nt.c
new file mode 100644
index 000000000..07c388147
--- /dev/null
+++ b/tools/testing/selftests/powerpc/tm/tm-signal-sigreturn-nt.c
@@ -0,0 +1,50 @@
+// SPDX-License-Identifier: GPL-2.0
+/*
+ * Copyright 2018, Breno Leitao, Gustavo Romero, IBM Corp.
+ *
+ * A test case that creates a signal and starts a suspended transaction
+ * inside the signal handler.
+ *
+ * It returns from the signal handler with the CPU at suspended state, but
+ * without setting usercontext MSR Transaction State (TS) fields.
+ */
+
+#define _GNU_SOURCE
+#include <stdio.h>
+#include <stdlib.h>
+#include <signal.h>
+
+#include "utils.h"
+#include "tm.h"
+
+void trap_signal_handler(int signo, siginfo_t *si, void *uc)
+{
+	ucontext_t *ucp = (ucontext_t *) uc;
+
+	asm("tbegin.; tsuspend.;");
+
+	/* Skip 'trap' instruction if it succeed */
+	ucp->uc_mcontext.regs->nip += 4;
+}
+
+int tm_signal_sigreturn_nt(void)
+{
+	struct sigaction trap_sa;
+
+	SKIP_IF(!have_htm());
+
+	trap_sa.sa_flags = SA_SIGINFO;
+	trap_sa.sa_sigaction = trap_signal_handler;
+
+	sigaction(SIGTRAP, &trap_sa, NULL);
+
+	raise(SIGTRAP);
+
+	return EXIT_SUCCESS;
+}
+
+int main(int argc, char **argv)
+{
+	test_harness(tm_signal_sigreturn_nt, "tm_signal_sigreturn_nt");
+}
+
diff --git a/tools/testing/selftests/powerpc/tm/tm-signal-stack.c b/tools/testing/selftests/powerpc/tm/tm-signal-stack.c
new file mode 100644
index 000000000..cdcf8c5bb
--- /dev/null
+++ b/tools/testing/selftests/powerpc/tm/tm-signal-stack.c
@@ -0,0 +1,76 @@
+// SPDX-License-Identifier: GPL-2.0-only
+/*
+ * Copyright 2015, Michael Neuling, IBM Corp.
+ *
+ * Test the kernel's signal delievery code to ensure that we don't
+ * trelaim twice in the kernel signal delivery code.  This can happen
+ * if we trigger a signal when in a transaction and the stack pointer
+ * is bogus.
+ *
+ * This test case registers a SEGV handler, sets the stack pointer
+ * (r1) to NULL, starts a transaction and then generates a SEGV.  The
+ * SEGV should be handled but we exit here as the stack pointer is
+ * invalid and hance we can't sigreturn.  We only need to check that
+ * this flow doesn't crash the kernel.
+ */
+
+#include <unistd.h>
+#include <sys/types.h>
+#include <sys/wait.h>
+#include <stdlib.h>
+#include <stdio.h>
+#include <signal.h>
+
+#include "utils.h"
+#include "tm.h"
+
+void signal_segv(int signum)
+{
+	/* This should never actually run since stack is foobar */
+	exit(1);
+}
+
+int tm_signal_stack()
+{
+	int pid;
+
+	SKIP_IF(!have_htm());
+
+	pid = fork();
+	if (pid < 0)
+		exit(1);
+
+	if (pid) { /* Parent */
+		/*
+		 * It's likely the whole machine will crash here so if
+		 * the child ever exits, we are good.
+		 */
+		wait(NULL);
+		return 0;
+	}
+
+	/*
+	 * The flow here is:
+	 * 1) register a signal handler (so signal delievery occurs)
+	 * 2) make stack pointer (r1) = NULL
+	 * 3) start transaction
+	 * 4) cause segv
+	 */
+	if (signal(SIGSEGV, signal_segv) == SIG_ERR)
+		exit(1);
+	asm volatile("li 1, 0 ;"		/* stack ptr == NULL */
+		     "1:"
+		     "tbegin.;"
+		     "beq 1b ;"			/* retry forever */
+		     "tsuspend.;"
+		     "ld 2, 0(1) ;"		/* trigger segv" */
+		     : : : "memory");
+
+	/* This should never get here due to above segv */
+	return 1;
+}
+
+int main(void)
+{
+	return test_harness(tm_signal_stack, "tm_signal_stack");
+}
diff --git a/tools/testing/selftests/powerpc/tm/tm-signal.S b/tools/testing/selftests/powerpc/tm/tm-signal.S
new file mode 100644
index 000000000..c80c91366
--- /dev/null
+++ b/tools/testing/selftests/powerpc/tm/tm-signal.S
@@ -0,0 +1,110 @@
+/* SPDX-License-Identifier: GPL-2.0-or-later */
+/*
+ * Copyright 2015, Cyril Bur, IBM Corp.
+ */
+
+#include "basic_asm.h"
+#include "gpr_asm.h"
+#include "fpu_asm.h"
+#include "vmx_asm.h"
+#include "vsx_asm.h"
+
+/*
+ * Large caveat here being that the caller cannot expect the
+ * signal to always be sent! The hardware can (AND WILL!) abort
+ * the transaction between the tbegin and the tsuspend (however
+ * unlikely it seems or infrequently it actually happens).
+ * You have been warned.
+ */
+/* long tm_signal_self(pid_t pid, long *gprs, double *fps, vector *vms, vector *vss); */
+FUNC_START(tm_signal_self_context_load)
+	PUSH_BASIC_STACK(512)
+	/*
+	 * Don't strictly need to save and restore as it depends on if
+	 * we're going to use them, however this reduces messy logic
+	 */
+	PUSH_VMX(STACK_FRAME_LOCAL(5,0),r8)
+	PUSH_FPU(512)
+	PUSH_NVREGS_BELOW_FPU(512)
+	std r3, STACK_FRAME_PARAM(0)(sp) /* pid */
+	std r4, STACK_FRAME_PARAM(1)(sp) /* gps */
+	std r5, STACK_FRAME_PARAM(2)(sp) /* fps */
+	std r6, STACK_FRAME_PARAM(3)(sp) /* vms */
+	std r7, STACK_FRAME_PARAM(4)(sp) /* vss */
+
+	ld r3, STACK_FRAME_PARAM(1)(sp)
+	cmpdi r3, 0
+	beq skip_gpr_lc
+	bl load_gpr
+skip_gpr_lc:
+	ld r3, STACK_FRAME_PARAM(2)(sp)
+	cmpdi	r3, 0
+	beq	skip_fpu_lc
+	bl load_fpu
+skip_fpu_lc:
+	ld r3, STACK_FRAME_PARAM(3)(sp)
+	cmpdi r3, 0
+	beq	skip_vmx_lc
+	bl load_vmx
+skip_vmx_lc:
+	ld r3, STACK_FRAME_PARAM(4)(sp)
+	cmpdi	r3, 0
+	beq	skip_vsx_lc
+	bl load_vsx
+skip_vsx_lc:
+	/*
+	 * Set r3 (return value) before tbegin. Use the pid as a known
+	 * 'all good' return value, zero is used to indicate a non-doomed
+	 * transaction.
+	 */
+	ld	r3, STACK_FRAME_PARAM(0)(sp)
+	tbegin.
+	beq	1f
+	tsuspend. /* Can't enter a syscall transactionally */
+	ld	r3, STACK_FRAME_PARAM(1)(sp)
+	cmpdi	r3, 0
+	beq skip_gpr_lt
+	/* Get the second half of the array */
+	addi	r3, r3, 8 * 18
+	bl load_gpr
+skip_gpr_lt:
+	ld r3, STACK_FRAME_PARAM(2)(sp)
+	cmpdi	r3, 0
+	beq	skip_fpu_lt
+	/* Get the second half of the array */
+	addi	r3, r3, 8 * 18
+	bl load_fpu
+skip_fpu_lt:
+	ld r3, STACK_FRAME_PARAM(3)(sp)
+	cmpdi r3, 0
+	beq	skip_vmx_lt
+	/* Get the second half of the array */
+	addi	r3, r3, 16 * 12
+	bl load_vmx
+skip_vmx_lt:
+	ld r3, STACK_FRAME_PARAM(4)(sp)
+	cmpdi	r3, 0
+	beq	skip_vsx_lt
+	/* Get the second half of the array */
+	addi	r3, r3, 16 * 12
+	bl load_vsx
+skip_vsx_lt:
+	li	r0, 37 /* sys_kill */
+	ld r3, STACK_FRAME_PARAM(0)(sp) /* pid */
+	li r4, 10 /* SIGUSR1 */
+	sc /* Taking the signal will doom the transaction */
+	tabort. 0
+	tresume. /* Be super sure we abort */
+	/*
+	 * This will cause us to resume doomed transaction and cause
+	 * hardware to cleanup, we'll end up at 1: anything between
+	 * tresume. and 1: shouldn't ever run.
+	 */
+	li r3, 0
+	1:
+	POP_VMX(STACK_FRAME_LOCAL(5,0),r4)
+	POP_FPU(512)
+	POP_NVREGS_BELOW_FPU(512)
+	POP_BASIC_STACK(512)
+	blr
+FUNC_END(tm_signal_self_context_load)
diff --git a/tools/testing/selftests/powerpc/tm/tm-sigreturn.c b/tools/testing/selftests/powerpc/tm/tm-sigreturn.c
new file mode 100644
index 000000000..9a6017a1d
--- /dev/null
+++ b/tools/testing/selftests/powerpc/tm/tm-sigreturn.c
@@ -0,0 +1,93 @@
+// SPDX-License-Identifier: GPL-2.0
+
+/*
+ * Copyright 2015, Laurent Dufour, IBM Corp.
+ *
+ * Test the kernel's signal returning code to check reclaim is done if the
+ * sigreturn() is called while in a transaction (suspended since active is
+ * already dropped trough the system call path).
+ *
+ * The kernel must discard the transaction when entering sigreturn, since
+ * restoring the potential TM SPRS from the signal frame is requiring to not be
+ * in a transaction.
+ */
+
+#include <signal.h>
+#include <stdio.h>
+#include <stdlib.h>
+#include <string.h>
+#include <sys/types.h>
+#include <sys/wait.h>
+#include <unistd.h>
+
+#include "tm.h"
+#include "utils.h"
+
+
+void handler(int sig)
+{
+	uint64_t ret;
+
+	asm __volatile__(
+		"li             3,1             ;"
+		"tbegin.                        ;"
+		"beq            1f              ;"
+		"li             3,0             ;"
+		"tsuspend.                      ;"
+		"1:                             ;"
+		"std%X[ret]     3, %[ret]       ;"
+		: [ret] "=m"(ret)
+		:
+		: "memory", "3", "cr0");
+
+	if (ret)
+		exit(1);
+
+	/*
+	 * We return from the signal handle while in a suspended transaction
+	 */
+}
+
+
+int tm_sigreturn(void)
+{
+	struct sigaction sa;
+	uint64_t ret = 0;
+
+	SKIP_IF(!have_htm());
+	SKIP_IF(!is_ppc64le());
+
+	memset(&sa, 0, sizeof(sa));
+	sa.sa_handler = handler;
+	sigemptyset(&sa.sa_mask);
+
+	if (sigaction(SIGSEGV, &sa, NULL))
+		exit(1);
+
+	asm __volatile__(
+		"tbegin.                        ;"
+		"beq            1f              ;"
+		"li             3,0             ;"
+		"std            3,0(3)          ;" /* trigger SEGV */
+		"li             3,1             ;"
+		"std%X[ret]     3,%[ret]        ;"
+		"tend.                          ;"
+		"b              2f              ;"
+		"1:                             ;"
+		"li             3,2             ;"
+		"std%X[ret]     3,%[ret]        ;"
+		"2:                             ;"
+		: [ret] "=m"(ret)
+		:
+		: "memory", "3", "cr0");
+
+	if (ret != 2)
+		exit(1);
+
+	exit(0);
+}
+
+int main(void)
+{
+	return test_harness(tm_sigreturn, "tm_sigreturn");
+}
diff --git a/tools/testing/selftests/powerpc/tm/tm-syscall-asm.S b/tools/testing/selftests/powerpc/tm/tm-syscall-asm.S
new file mode 100644
index 000000000..bd1ca25fe
--- /dev/null
+++ b/tools/testing/selftests/powerpc/tm/tm-syscall-asm.S
@@ -0,0 +1,28 @@
+/* SPDX-License-Identifier: GPL-2.0 */
+#include <ppc-asm.h>
+#include <asm/unistd.h>
+
+	.text
+FUNC_START(getppid_tm_active)
+	tbegin.
+	beq 1f
+	li	r0, __NR_getppid
+	sc
+	tend.
+	blr
+1:
+	li	r3, -1
+	blr
+
+FUNC_START(getppid_tm_suspended)
+	tbegin.
+	beq 1f
+	li	r0, __NR_getppid
+	tsuspend.
+	sc
+	tresume.
+	tend.
+	blr
+1:
+	li	r3, -1
+	blr
diff --git a/tools/testing/selftests/powerpc/tm/tm-syscall.c b/tools/testing/selftests/powerpc/tm/tm-syscall.c
new file mode 100644
index 000000000..becb8207b
--- /dev/null
+++ b/tools/testing/selftests/powerpc/tm/tm-syscall.c
@@ -0,0 +1,106 @@
+// SPDX-License-Identifier: GPL-2.0-only
+/*
+ * Copyright 2015, Sam Bobroff, IBM Corp.
+ *
+ * Test the kernel's system call code to ensure that a system call
+ * made from within an active HTM transaction is aborted with the
+ * correct failure code.
+ * Conversely, ensure that a system call made from within a
+ * suspended transaction can succeed.
+ */
+
+#include <stdio.h>
+#include <unistd.h>
+#include <sys/syscall.h>
+#include <asm/tm.h>
+#include <sys/time.h>
+#include <stdlib.h>
+
+#include "utils.h"
+#include "tm.h"
+
+extern int getppid_tm_active(void);
+extern int getppid_tm_suspended(void);
+
+unsigned retries = 0;
+
+#define TEST_DURATION 10 /* seconds */
+#define TM_RETRIES 100
+
+pid_t getppid_tm(bool suspend)
+{
+	int i;
+	pid_t pid;
+
+	for (i = 0; i < TM_RETRIES; i++) {
+		if (suspend)
+			pid = getppid_tm_suspended();
+		else
+			pid = getppid_tm_active();
+
+		if (pid >= 0)
+			return pid;
+
+		if (failure_is_persistent()) {
+			if (failure_is_syscall())
+				return -1;
+
+			printf("Unexpected persistent transaction failure.\n");
+			printf("TEXASR 0x%016lx, TFIAR 0x%016lx.\n",
+			       __builtin_get_texasr(), __builtin_get_tfiar());
+			exit(-1);
+		}
+
+		retries++;
+	}
+
+	printf("Exceeded limit of %d temporary transaction failures.\n", TM_RETRIES);
+	printf("TEXASR 0x%016lx, TFIAR 0x%016lx.\n",
+	       __builtin_get_texasr(), __builtin_get_tfiar());
+
+	exit(-1);
+}
+
+int tm_syscall(void)
+{
+	unsigned count = 0;
+	struct timeval end, now;
+
+	SKIP_IF(!have_htm_nosc());
+
+	setbuf(stdout, NULL);
+
+	printf("Testing transactional syscalls for %d seconds...\n", TEST_DURATION);
+
+	gettimeofday(&end, NULL);
+	now.tv_sec = TEST_DURATION;
+	now.tv_usec = 0;
+	timeradd(&end, &now, &end);
+
+	for (count = 0; timercmp(&now, &end, <); count++) {
+		/*
+		 * Test a syscall within a suspended transaction and verify
+		 * that it succeeds.
+		 */
+		FAIL_IF(getppid_tm(true) == -1); /* Should succeed. */
+
+		/*
+		 * Test a syscall within an active transaction and verify that
+		 * it fails with the correct failure code.
+		 */
+		FAIL_IF(getppid_tm(false) != -1);  /* Should fail... */
+		FAIL_IF(!failure_is_persistent()); /* ...persistently... */
+		FAIL_IF(!failure_is_syscall());    /* ...with code syscall. */
+		gettimeofday(&now, 0);
+	}
+
+	printf("%d active and suspended transactions behaved correctly.\n", count);
+	printf("(There were %d transaction retries.)\n", retries);
+
+	return 0;
+}
+
+int main(void)
+{
+	return test_harness(tm_syscall, "tm_syscall");
+}
diff --git a/tools/testing/selftests/powerpc/tm/tm-tar.c b/tools/testing/selftests/powerpc/tm/tm-tar.c
new file mode 100644
index 000000000..03be8c472
--- /dev/null
+++ b/tools/testing/selftests/powerpc/tm/tm-tar.c
@@ -0,0 +1,91 @@
+// SPDX-License-Identifier: GPL-2.0-only
+/*
+ * Copyright 2015, Michael Neuling, IBM Corp.
+ * Original: Michael Neuling 19/7/2013
+ * Edited: Rashmica Gupta 01/12/2015
+ *
+ * Do some transactions, see if the tar is corrupted.
+ * If the transaction is aborted, the TAR should be rolled back to the
+ * checkpointed value before the transaction began. The value written to
+ * TAR in suspended mode should only remain in TAR if the transaction
+ * completes.
+ */
+
+#include <stdio.h>
+#include <stdlib.h>
+#include <unistd.h>
+#include <string.h>
+
+#include "tm.h"
+#include "utils.h"
+
+int	num_loops	= 10000;
+
+int test_tar(void)
+{
+	int i;
+
+	SKIP_IF(!have_htm());
+	SKIP_IF(!is_ppc64le());
+
+	for (i = 0; i < num_loops; i++)
+	{
+		uint64_t result = 0;
+		asm __volatile__(
+			"li	7, 1;"
+			"mtspr	%[tar], 7;"	/* tar = 1 */
+			"tbegin.;"
+			"beq	3f;"
+			"li	4, 0x7000;"	/* Loop lots, to use time */
+			"2:;"			/* Start loop */
+			"li	7, 2;"
+			"mtspr	%[tar], 7;"	/* tar = 2 */
+			"tsuspend.;"
+			"li	7, 3;"
+			"mtspr	%[tar], 7;"	/* tar = 3 */
+			"tresume.;"
+			"subi	4, 4, 1;"
+			"cmpdi	4, 0;"
+			"bne	2b;"
+			"tend.;"
+
+			/* Transaction sucess! TAR should be 3 */
+			"mfspr  7, %[tar];"
+			"ori	%[res], 7, 4;"  // res = 3|4 = 7
+			"b	4f;"
+
+			/* Abort handler. TAR should be rolled back to 1 */
+			"3:;"
+			"mfspr  7, %[tar];"
+			"ori	%[res], 7, 8;"	// res = 1|8 = 9
+			"4:;"
+
+			: [res]"=r"(result)
+			: [tar]"i"(SPRN_TAR)
+			   : "memory", "r0", "r4", "r7");
+
+		/* If result is anything else other than 7 or 9, the tar
+		 * value must have been corrupted. */
+		if ((result != 7) && (result != 9))
+			return 1;
+	}
+	return 0;
+}
+
+int main(int argc, char *argv[])
+{
+	/* A low number of iterations (eg 100) can cause a false pass */
+	if (argc > 1) {
+		if (strcmp(argv[1], "-h") == 0) {
+			printf("Syntax:\n\t%s [<num loops>]\n",
+			       argv[0]);
+			return 1;
+		} else {
+			num_loops = atoi(argv[1]);
+		}
+	}
+
+	printf("Starting, %d loops\n", num_loops);
+
+	return test_harness(test_tar, "tm_tar");
+}
diff --git a/tools/testing/selftests/powerpc/tm/tm-tmspr.c b/tools/testing/selftests/powerpc/tm/tm-tmspr.c
new file mode 100644
index 000000000..794d574db
--- /dev/null
+++ b/tools/testing/selftests/powerpc/tm/tm-tmspr.c
@@ -0,0 +1,143 @@
+// SPDX-License-Identifier: GPL-2.0-only
+/*
+ * Copyright 2015, Michael Neuling, IBM Corp.
+ *
+ * Original: Michael Neuling 3/4/2014
+ * Modified: Rashmica Gupta 8/12/2015
+ *
+ * Check if any of the Transaction Memory SPRs get corrupted.
+ * - TFIAR  - stores address of location of transaction failure
+ * - TFHAR  - stores address of software failure handler (if transaction
+ *   fails)
+ * - TEXASR - lots of info about the transacion(s)
+ *
+ * (1) create more threads than cpus
+ * (2) in each thread:
+ * 	(a) set TFIAR and TFHAR a unique value
+ * 	(b) loop for awhile, continually checking to see if
+ * 	either register has been corrupted.
+ *
+ * (3) Loop:
+ * 	(a) begin transaction
+ *    	(b) abort transaction
+ *	(c) check TEXASR to see if FS has been corrupted
+ */
+
+#define _GNU_SOURCE
+#include <stdio.h>
+#include <stdlib.h>
+#include <unistd.h>
+#include <pthread.h>
+#include <string.h>
+
+#include "utils.h"
+#include "tm.h"
+
+int	num_loops	= 1000000;
+int	passed = 1;
+
+void tfiar_tfhar(void *in)
+{
+	unsigned long tfhar, tfhar_rd, tfiar, tfiar_rd;
+	int i;
+
+	/* TFIAR: Last bit has to be high so userspace can read register */
+	tfiar = ((unsigned long)in) + 1;
+	tfiar += 2;
+	mtspr(SPRN_TFIAR, tfiar);
+
+	/* TFHAR: Last two bits are reserved */
+	tfhar = ((unsigned long)in);
+	tfhar &= ~0x3UL;
+	tfhar += 4;
+	mtspr(SPRN_TFHAR, tfhar);
+
+	for (i = 0; i < num_loops; i++)	{
+		tfhar_rd = mfspr(SPRN_TFHAR);
+		tfiar_rd = mfspr(SPRN_TFIAR);
+		if ( (tfhar != tfhar_rd) || (tfiar != tfiar_rd) ) {
+			passed = 0;
+			return;
+		}
+	}
+	return;
+}
+
+void texasr(void *in)
+{
+	unsigned long i;
+	uint64_t result = 0;
+
+	for (i = 0; i < num_loops; i++) {
+		asm __volatile__(
+			"tbegin.;"
+			"beq    3f ;"
+			"tabort. 0 ;"
+			"tend.;"
+
+			/* Abort handler */
+			"3: ;"
+			::: "memory");
+
+                /* Check the TEXASR */
+                result = mfspr(SPRN_TEXASR);
+		if ((result & TEXASR_FS) == 0) {
+			passed = 0;
+			return;
+		}
+	}
+	return;
+}
+
+int test_tmspr()
+{
+	pthread_t	*thread;
+	int	   	thread_num;
+	unsigned long	i;
+
+	SKIP_IF(!have_htm());
+
+	/* To cause some context switching */
+	thread_num = 10 * sysconf(_SC_NPROCESSORS_ONLN);
+
+	thread = malloc(thread_num * sizeof(pthread_t));
+	if (thread == NULL)
+		return EXIT_FAILURE;
+
+	/* Test TFIAR and TFHAR */
+	for (i = 0; i < thread_num; i += 2) {
+		if (pthread_create(&thread[i], NULL, (void *)tfiar_tfhar,
+				   (void *)i))
+			return EXIT_FAILURE;
+	}
+	/* Test TEXASR */
+	for (i = 1; i < thread_num; i += 2) {
+		if (pthread_create(&thread[i], NULL, (void *)texasr, (void *)i))
+			return EXIT_FAILURE;
+	}
+
+	for (i = 0; i < thread_num; i++) {
+		if (pthread_join(thread[i], NULL) != 0)
+			return EXIT_FAILURE;
+	}
+
+	free(thread);
+
+	if (passed)
+		return 0;
+	else
+		return 1;
+}
+
+int main(int argc, char *argv[])
+{
+	if (argc > 1) {
+		if (strcmp(argv[1], "-h") == 0) {
+			printf("Syntax:\t [<num loops>]\n");
+			return 0;
+		} else {
+			num_loops = atoi(argv[1]);
+		}
+	}
+	return test_harness(test_tmspr, "tm_tmspr");
+}
diff --git a/tools/testing/selftests/powerpc/tm/tm-trap.c b/tools/testing/selftests/powerpc/tm/tm-trap.c
new file mode 100644
index 000000000..c75960af8
--- /dev/null
+++ b/tools/testing/selftests/powerpc/tm/tm-trap.c
@@ -0,0 +1,333 @@
+// SPDX-License-Identifier: GPL-2.0-only
+/*
+ * Copyright 2017, Gustavo Romero, IBM Corp.
+ *
+ * Check if thread endianness is flipped inadvertently to BE on trap
+ * caught in TM whilst MSR.FP and MSR.VEC are zero (i.e. just after
+ * load_fp and load_vec overflowed).
+ *
+ * The issue can be checked on LE machines simply by zeroing load_fp
+ * and load_vec and then causing a trap in TM. Since the endianness
+ * changes to BE on return from the signal handler, 'nop' is
+ * thread as an illegal instruction in following sequence:
+ *	tbegin.
+ *	beq 1f
+ *	trap
+ *	tend.
+ * 1:	nop
+ *
+ * However, although the issue is also present on BE machines, it's a
+ * bit trickier to check it on BE machines because MSR.LE bit is set
+ * to zero which determines a BE endianness that is the native
+ * endianness on BE machines, so nothing notably critical happens,
+ * i.e. no illegal instruction is observed immediately after returning
+ * from the signal handler (as it happens on LE machines). Thus to test
+ * it on BE machines LE endianness is forced after a first trap and then
+ * the endianness is verified on subsequent traps to determine if the
+ * endianness "flipped back" to the native endianness (BE).
+ */
+
+#define _GNU_SOURCE
+#include <error.h>
+#include <stdio.h>
+#include <stdlib.h>
+#include <unistd.h>
+#include <htmintrin.h>
+#include <inttypes.h>
+#include <pthread.h>
+#include <sched.h>
+#include <signal.h>
+#include <stdbool.h>
+
+#include "tm.h"
+#include "utils.h"
+
+#define pr_error(error_code, format, ...) \
+	error_at_line(1, error_code, __FILE__, __LINE__, format, ##__VA_ARGS__)
+
+#define MSR_LE 1UL
+#define LE     1UL
+
+pthread_t t0_ping;
+pthread_t t1_pong;
+
+int exit_from_pong;
+
+int trap_event;
+int le;
+
+bool success;
+
+void trap_signal_handler(int signo, siginfo_t *si, void *uc)
+{
+	ucontext_t *ucp = uc;
+	uint64_t thread_endianness;
+
+	/* Get thread endianness: extract bit LE from MSR */
+	thread_endianness = MSR_LE & ucp->uc_mcontext.gp_regs[PT_MSR];
+
+	/***
+	 * Little-Endian Machine
+	 */
+
+	if (le) {
+		/* First trap event */
+		if (trap_event == 0) {
+			/* Do nothing. Since it is returning from this trap
+			 * event that endianness is flipped by the bug, so just
+			 * let the process return from the signal handler and
+			 * check on the second trap event if endianness is
+			 * flipped or not.
+			 */
+		}
+		/* Second trap event */
+		else if (trap_event == 1) {
+			/*
+			 * Since trap was caught in TM on first trap event, if
+			 * endianness was still LE (not flipped inadvertently)
+			 * after returning from the signal handler instruction
+			 * (1) is executed (basically a 'nop'), as it's located
+			 * at address of tbegin. +4 (rollback addr). As (1) on
+			 * LE endianness does in effect nothing, instruction (2)
+			 * is then executed again as 'trap', generating a second
+			 * trap event (note that in that case 'trap' is caught
+			 * not in transacional mode). On te other hand, if after
+			 * the return from the signal handler the endianness in-
+			 * advertently flipped, instruction (1) is tread as a
+			 * branch instruction, i.e. b .+8, hence instruction (3)
+			 * and (4) are executed (tbegin.; trap;) and we get sim-
+			 * ilaly on the trap signal handler, but now in TM mode.
+			 * Either way, it's now possible to check the MSR LE bit
+			 * once in the trap handler to verify if endianness was
+			 * flipped or not after the return from the second trap
+			 * event. If endianness is flipped, the bug is present.
+			 * Finally, getting a trap in TM mode or not is just
+			 * worth noting because it affects the math to determine
+			 * the offset added to the NIP on return: the NIP for a
+			 * trap caught in TM is the rollback address, i.e. the
+			 * next instruction after 'tbegin.', whilst the NIP for
+			 * a trap caught in non-transactional mode is the very
+			 * same address of the 'trap' instruction that generated
+			 * the trap event.
+			 */
+
+			if (thread_endianness == LE) {
+				/* Go to 'success', i.e. instruction (6) */
+				ucp->uc_mcontext.gp_regs[PT_NIP] += 16;
+			} else {
+				/*
+				 * Thread endianness is BE, so it flipped
+				 * inadvertently. Thus we flip back to LE and
+				 * set NIP to go to 'failure', instruction (5).
+				 */
+				ucp->uc_mcontext.gp_regs[PT_MSR] |= 1UL;
+				ucp->uc_mcontext.gp_regs[PT_NIP] += 4;
+			}
+		}
+	}
+
+	/***
+	 * Big-Endian Machine
+	 */
+
+	else {
+		/* First trap event */
+		if (trap_event == 0) {
+			/*
+			 * Force thread endianness to be LE. Instructions (1),
+			 * (3), and (4) will be executed, generating a second
+			 * trap in TM mode.
+			 */
+			ucp->uc_mcontext.gp_regs[PT_MSR] |= 1UL;
+		}
+		/* Second trap event */
+		else if (trap_event == 1) {
+			/*
+			 * Do nothing. If bug is present on return from this
+			 * second trap event endianness will flip back "automat-
+			 * ically" to BE, otherwise thread endianness will
+			 * continue to be LE, just as it was set above.
+			 */
+		}
+		/* A third trap event */
+		else {
+			/*
+			 * Once here it means that after returning from the sec-
+			 * ond trap event instruction (4) (trap) was executed
+			 * as LE, generating a third trap event. In that case
+			 * endianness is still LE as set on return from the
+			 * first trap event, hence no bug. Otherwise, bug
+			 * flipped back to BE on return from the second trap
+			 * event and instruction (4) was executed as 'tdi' (so
+			 * basically a 'nop') and branch to 'failure' in
+			 * instruction (5) was taken to indicate failure and we
+			 * never get here.
+			 */
+
+			/*
+			 * Flip back to BE and go to instruction (6), i.e. go to
+			 * 'success'.
+			 */
+			ucp->uc_mcontext.gp_regs[PT_MSR] &= ~1UL;
+			ucp->uc_mcontext.gp_regs[PT_NIP] += 8;
+		}
+	}
+
+	trap_event++;
+}
+
+void usr1_signal_handler(int signo, siginfo_t *si, void *not_used)
+{
+	/* Got a USR1 signal from ping(), so just tell pong() to exit */
+	exit_from_pong = 1;
+}
+
+void *ping(void *not_used)
+{
+	uint64_t i;
+
+	trap_event = 0;
+
+	/*
+	 * Wait an amount of context switches so load_fp and load_vec overflows
+	 * and MSR_[FP|VEC|V] is 0.
+	 */
+	for (i = 0; i < 1024*1024*512; i++)
+		;
+
+	asm goto(
+		/*
+		 * [NA] means "Native Endianness", i.e. it tells how a
+		 * instruction is executed on machine's native endianness (in
+		 * other words, native endianness matches kernel endianness).
+		 * [OP] means "Opposite Endianness", i.e. on a BE machine, it
+		 * tells how a instruction is executed as a LE instruction; con-
+		 * versely, on a LE machine, it tells how a instruction is
+		 * executed as a BE instruction. When [NA] is omitted, it means
+		 * that the native interpretation of a given instruction is not
+		 * relevant for the test. Likewise when [OP] is omitted.
+		 */
+
+		" tbegin.        ;" /* (0) tbegin. [NA]                    */
+		" tdi  0, 0, 0x48;" /* (1) nop     [NA]; b (3) [OP]        */
+		" trap           ;" /* (2) trap    [NA]                    */
+		".long 0x1D05007C;" /* (3) tbegin. [OP]                    */
+		".long 0x0800E07F;" /* (4) trap    [OP]; nop   [NA]        */
+		" b %l[failure]  ;" /* (5) b [NA]; MSR.LE flipped (bug)    */
+		" b %l[success]  ;" /* (6) b [NA]; MSR.LE did not flip (ok)*/
+
+		: : : : failure, success);
+
+failure:
+	success = false;
+	goto exit_from_ping;
+
+success:
+	success = true;
+
+exit_from_ping:
+	/* Tell pong() to exit before leaving */
+	pthread_kill(t1_pong, SIGUSR1);
+	return NULL;
+}
+
+void *pong(void *not_used)
+{
+	while (!exit_from_pong)
+		/*
+		 * Induce context switches on ping() thread
+		 * until ping() finishes its job and signs
+		 * to exit from this loop.
+		 */
+		sched_yield();
+
+	return NULL;
+}
+
+int tm_trap_test(void)
+{
+	uint16_t k = 1;
+	int cpu, rc;
+
+	pthread_attr_t attr;
+	cpu_set_t cpuset;
+
+	struct sigaction trap_sa;
+
+	SKIP_IF(!have_htm());
+
+	trap_sa.sa_flags = SA_SIGINFO;
+	trap_sa.sa_sigaction = trap_signal_handler;
+	sigaction(SIGTRAP, &trap_sa, NULL);
+
+	struct sigaction usr1_sa;
+
+	usr1_sa.sa_flags = SA_SIGINFO;
+	usr1_sa.sa_sigaction = usr1_signal_handler;
+	sigaction(SIGUSR1, &usr1_sa, NULL);
+
+	cpu = pick_online_cpu();
+	FAIL_IF(cpu < 0);
+
+	// Set only one CPU in the mask. Both threads will be bound to that CPU.
+	CPU_ZERO(&cpuset);
+	CPU_SET(cpu, &cpuset);
+
+	/* Init pthread attribute */
+	rc = pthread_attr_init(&attr);
+	if (rc)
+		pr_error(rc, "pthread_attr_init()");
+
+	/*
+	 * Bind thread ping() and pong() both to CPU 0 so they ping-pong and
+	 * speed up context switches on ping() thread, speeding up the load_fp
+	 * and load_vec overflow.
+	 */
+	rc = pthread_attr_setaffinity_np(&attr, sizeof(cpu_set_t), &cpuset);
+	if (rc)
+		pr_error(rc, "pthread_attr_setaffinity()");
+
+	/* Figure out the machine endianness */
+	le = (int) *(uint8_t *)&k;
+
+	printf("%s machine detected. Checking if endianness flips %s",
+		le ? "Little-Endian" : "Big-Endian",
+		"inadvertently on trap in TM... ");
+
+	rc = fflush(0);
+	if (rc)
+		pr_error(rc, "fflush()");
+
+	/* Launch ping() */
+	rc = pthread_create(&t0_ping, &attr, ping, NULL);
+	if (rc)
+		pr_error(rc, "pthread_create()");
+
+	exit_from_pong = 0;
+
+	/* Launch pong() */
+	rc = pthread_create(&t1_pong, &attr, pong, NULL);
+	if (rc)
+		pr_error(rc, "pthread_create()");
+
+	rc = pthread_join(t0_ping, NULL);
+	if (rc)
+		pr_error(rc, "pthread_join()");
+
+	rc = pthread_join(t1_pong, NULL);
+	if (rc)
+		pr_error(rc, "pthread_join()");
+
+	if (success) {
+		printf("no.\n"); /* no, endianness did not flip inadvertently */
+		return EXIT_SUCCESS;
+	}
+
+	printf("yes!\n"); /* yes, endianness did flip inadvertently */
+	return EXIT_FAILURE;
+}
+
+int main(int argc, char **argv)
+{
+	return test_harness(tm_trap_test, "tm_trap_test");
+}
diff --git a/tools/testing/selftests/powerpc/tm/tm-unavailable.c b/tools/testing/selftests/powerpc/tm/tm-unavailable.c
new file mode 100644
index 000000000..a1348a5f7
--- /dev/null
+++ b/tools/testing/selftests/powerpc/tm/tm-unavailable.c
@@ -0,0 +1,410 @@
+// SPDX-License-Identifier: GPL-2.0-only
+/*
+ * Copyright 2017, Gustavo Romero, Breno Leitao, Cyril Bur, IBM Corp.
+ *
+ * Force FP, VEC and VSX unavailable exception during transaction in all
+ * possible scenarios regarding the MSR.FP and MSR.VEC state, e.g. when FP
+ * is enable and VEC is disable, when FP is disable and VEC is enable, and
+ * so on. Then we check if the restored state is correctly set for the
+ * FP and VEC registers to the previous state we set just before we entered
+ * in TM, i.e. we check if it corrupts somehow the recheckpointed FP and
+ * VEC/Altivec registers on abortion due to an unavailable exception in TM.
+ * N.B. In this test we do not test all the FP/Altivec/VSX registers for
+ * corruption, but only for registers vs0 and vs32, which are respectively
+ * representatives of FP and VEC/Altivec reg sets.
+ */
+
+#define _GNU_SOURCE
+#include <error.h>
+#include <stdio.h>
+#include <stdlib.h>
+#include <unistd.h>
+#include <inttypes.h>
+#include <stdbool.h>
+#include <pthread.h>
+#include <sched.h>
+
+#include "tm.h"
+
+#define DEBUG 0
+
+/* Unavailable exceptions to test in HTM */
+#define FP_UNA_EXCEPTION	0
+#define VEC_UNA_EXCEPTION	1
+#define VSX_UNA_EXCEPTION	2
+
+#define NUM_EXCEPTIONS		3
+#define err_at_line(status, errnum, format, ...) \
+	error_at_line(status, errnum,  __FILE__, __LINE__, format ##__VA_ARGS__)
+
+#define pr_warn(code, format, ...) err_at_line(0, code, format, ##__VA_ARGS__)
+#define pr_err(code, format, ...) err_at_line(1, code, format, ##__VA_ARGS__)
+
+struct Flags {
+	int touch_fp;
+	int touch_vec;
+	int result;
+	int exception;
+} flags;
+
+bool expecting_failure(void)
+{
+	if (flags.touch_fp && flags.exception == FP_UNA_EXCEPTION)
+		return false;
+
+	if (flags.touch_vec && flags.exception == VEC_UNA_EXCEPTION)
+		return false;
+
+	/*
+	 * If both FP and VEC are touched it does not mean that touching VSX
+	 * won't raise an exception. However since FP and VEC state are already
+	 * correctly loaded, the transaction is not aborted (i.e.
+	 * treclaimed/trecheckpointed) and MSR.VSX is just set as 1, so a TM
+	 * failure is not expected also in this case.
+	 */
+	if ((flags.touch_fp && flags.touch_vec) &&
+	     flags.exception == VSX_UNA_EXCEPTION)
+		return false;
+
+	return true;
+}
+
+/* Check if failure occurred whilst in transaction. */
+bool is_failure(uint64_t condition_reg)
+{
+	/*
+	 * When failure handling occurs, CR0 is set to 0b1010 (0xa). Otherwise
+	 * transaction completes without failure and hence reaches out 'tend.'
+	 * that sets CR0 to 0b0100 (0x4).
+	 */
+	return ((condition_reg >> 28) & 0xa) == 0xa;
+}
+
+void *tm_una_ping(void *input)
+{
+
+	/*
+	 * Expected values for vs0 and vs32 after a TM failure. They must never
+	 * change, otherwise they got corrupted.
+	 */
+	uint64_t high_vs0 = 0x5555555555555555;
+	uint64_t low_vs0 = 0xffffffffffffffff;
+	uint64_t high_vs32 = 0x5555555555555555;
+	uint64_t low_vs32 = 0xffffffffffffffff;
+
+	/* Counter for busy wait */
+	uint64_t counter = 0x1ff000000;
+
+	/*
+	 * Variable to keep a copy of CR register content taken just after we
+	 * leave the transactional state.
+	 */
+	uint64_t cr_ = 0;
+
+	/*
+	 * Wait a bit so thread can get its name "ping". This is not important
+	 * to reproduce the issue but it's nice to have for systemtap debugging.
+	 */
+	if (DEBUG)
+		sleep(1);
+
+	printf("If MSR.FP=%d MSR.VEC=%d: ", flags.touch_fp, flags.touch_vec);
+
+	if (flags.exception != FP_UNA_EXCEPTION &&
+	    flags.exception != VEC_UNA_EXCEPTION &&
+	    flags.exception != VSX_UNA_EXCEPTION) {
+		printf("No valid exception specified to test.\n");
+		return NULL;
+	}
+
+	asm (
+		/* Prepare to merge low and high. */
+		"	mtvsrd		33, %[high_vs0]		;"
+		"	mtvsrd		34, %[low_vs0]		;"
+
+		/*
+		 * Adjust VS0 expected value after an TM failure,
+		 * i.e. vs0 = 0x5555555555555555555FFFFFFFFFFFFFFFF
+		 */
+		"	xxmrghd		0, 33, 34		;"
+
+		/*
+		 * Adjust VS32 expected value after an TM failure,
+		 * i.e. vs32 = 0x5555555555555555555FFFFFFFFFFFFFFFF
+		 */
+		"	xxmrghd		32, 33, 34		;"
+
+		/*
+		 * Wait an amount of context switches so load_fp and load_vec
+		 * overflow and MSR.FP, MSR.VEC, and MSR.VSX become zero (off).
+		 */
+		"	mtctr		%[counter]		;"
+
+		/* Decrement CTR branch if CTR non zero. */
+		"1:	bdnz 1b					;"
+
+		/*
+		 * Check if we want to touch FP prior to the test in order
+		 * to set MSR.FP = 1 before provoking an unavailable
+		 * exception in TM.
+		 */
+		"	cmpldi		%[touch_fp], 0		;"
+		"	beq		no_fp			;"
+		"	fadd		10, 10, 10		;"
+		"no_fp:						;"
+
+		/*
+		 * Check if we want to touch VEC prior to the test in order
+		 * to set MSR.VEC = 1 before provoking an unavailable
+		 * exception in TM.
+		 */
+		"	cmpldi		%[touch_vec], 0		;"
+		"	beq		no_vec			;"
+		"	vaddcuw		10, 10, 10		;"
+		"no_vec:					;"
+
+		/*
+		 * Perhaps it would be a better idea to do the
+		 * compares outside transactional context and simply
+		 * duplicate code.
+		 */
+		"	tbegin.					;"
+		"	beq		trans_fail		;"
+
+		/* Do we do FP Unavailable? */
+		"	cmpldi		%[exception], %[ex_fp]	;"
+		"	bne		1f			;"
+		"	fadd		10, 10, 10		;"
+		"	b		done			;"
+
+		/* Do we do VEC Unavailable? */
+		"1:	cmpldi		%[exception], %[ex_vec]	;"
+		"	bne		2f			;"
+		"	vaddcuw		10, 10, 10		;"
+		"	b		done			;"
+
+		/*
+		 * Not FP or VEC, therefore VSX. Ensure this
+		 * instruction always generates a VSX Unavailable.
+		 * ISA 3.0 is tricky here.
+		 * (xxmrghd will on ISA 2.07 and ISA 3.0)
+		 */
+		"2:	xxmrghd		10, 10, 10		;"
+
+		"done:	tend. ;"
+
+		"trans_fail: ;"
+
+		/* Give values back to C. */
+		"	mfvsrd		%[high_vs0], 0		;"
+		"	xxsldwi		3, 0, 0, 2		;"
+		"	mfvsrd		%[low_vs0], 3		;"
+		"	mfvsrd		%[high_vs32], 32	;"
+		"	xxsldwi		3, 32, 32, 2		;"
+		"	mfvsrd		%[low_vs32], 3		;"
+
+		/* Give CR back to C so that it can check what happened. */
+		"	mfcr		%[cr_]		;"
+
+		: [high_vs0]  "+r" (high_vs0),
+		  [low_vs0]   "+r" (low_vs0),
+		  [high_vs32] "=r" (high_vs32),
+		  [low_vs32]  "=r" (low_vs32),
+		  [cr_]       "+r" (cr_)
+		: [touch_fp]  "r"  (flags.touch_fp),
+		  [touch_vec] "r"  (flags.touch_vec),
+		  [exception] "r"  (flags.exception),
+		  [ex_fp]     "i"  (FP_UNA_EXCEPTION),
+		  [ex_vec]    "i"  (VEC_UNA_EXCEPTION),
+		  [ex_vsx]    "i"  (VSX_UNA_EXCEPTION),
+		  [counter]   "r"  (counter)
+
+		: "cr0", "ctr", "v10", "vs0", "vs10", "vs3", "vs32", "vs33",
+		  "vs34", "fr10"
+
+		);
+
+	/*
+	 * Check if we were expecting a failure and it did not occur by checking
+	 * CR0 state just after we leave the transaction. Either way we check if
+	 * vs0 or vs32 got corrupted.
+	 */
+	if (expecting_failure() && !is_failure(cr_)) {
+		printf("\n\tExpecting the transaction to fail, %s",
+			"but it didn't\n\t");
+		flags.result++;
+	}
+
+	/* Check if we were not expecting a failure and a it occurred. */
+	if (!expecting_failure() && is_failure(cr_) &&
+	    !failure_is_reschedule()) {
+		printf("\n\tUnexpected transaction failure 0x%02lx\n\t",
+			failure_code());
+		return (void *) -1;
+	}
+
+	/*
+	 * Check if TM failed due to the cause we were expecting. 0xda is a
+	 * TM_CAUSE_FAC_UNAV cause, otherwise it's an unexpected cause, unless
+	 * it was caused by a reschedule.
+	 */
+	if (is_failure(cr_) && !failure_is_unavailable() &&
+	    !failure_is_reschedule()) {
+		printf("\n\tUnexpected failure cause 0x%02lx\n\t",
+			failure_code());
+		return (void *) -1;
+	}
+
+	/* 0x4 is a success and 0xa is a fail. See comment in is_failure(). */
+	if (DEBUG)
+		printf("CR0: 0x%1lx ", cr_ >> 28);
+
+	/* Check FP (vs0) for the expected value. */
+	if (high_vs0 != 0x5555555555555555 || low_vs0 != 0xFFFFFFFFFFFFFFFF) {
+		printf("FP corrupted!");
+			printf("  high = %#16" PRIx64 "  low = %#16" PRIx64 " ",
+				high_vs0, low_vs0);
+		flags.result++;
+	} else
+		printf("FP ok ");
+
+	/* Check VEC (vs32) for the expected value. */
+	if (high_vs32 != 0x5555555555555555 || low_vs32 != 0xFFFFFFFFFFFFFFFF) {
+		printf("VEC corrupted!");
+			printf("  high = %#16" PRIx64 "  low = %#16" PRIx64,
+				high_vs32, low_vs32);
+		flags.result++;
+	} else
+		printf("VEC ok");
+
+	putchar('\n');
+
+	return NULL;
+}
+
+/* Thread to force context switch */
+void *tm_una_pong(void *not_used)
+{
+	/* Wait thread get its name "pong". */
+	if (DEBUG)
+		sleep(1);
+
+	/* Classed as an interactive-like thread. */
+	while (1)
+		sched_yield();
+}
+
+/* Function that creates a thread and launches the "ping" task. */
+void test_fp_vec(int fp, int vec, pthread_attr_t *attr)
+{
+	int retries = 2;
+	void *ret_value;
+	pthread_t t0;
+
+	flags.touch_fp = fp;
+	flags.touch_vec = vec;
+
+	/*
+	 * Without luck it's possible that the transaction is aborted not due to
+	 * the unavailable exception caught in the middle as we expect but also,
+	 * for instance, due to a context switch or due to a KVM reschedule (if
+	 * it's running on a VM). Thus we try a few times before giving up,
+	 * checking if the failure cause is the one we expect.
+	 */
+	do {
+		int rc;
+
+		/* Bind to CPU 0, as specified in 'attr'. */
+		rc = pthread_create(&t0, attr, tm_una_ping, (void *) &flags);
+		if (rc)
+			pr_err(rc, "pthread_create()");
+		rc = pthread_setname_np(t0, "tm_una_ping");
+		if (rc)
+			pr_warn(rc, "pthread_setname_np");
+		rc = pthread_join(t0, &ret_value);
+		if (rc)
+			pr_err(rc, "pthread_join");
+
+		retries--;
+	} while (ret_value != NULL && retries);
+
+	if (!retries) {
+		flags.result = 1;
+		if (DEBUG)
+			printf("All transactions failed unexpectedly\n");
+
+	}
+}
+
+int tm_unavailable_test(void)
+{
+	int cpu, rc, exception; /* FP = 0, VEC = 1, VSX = 2 */
+	pthread_t t1;
+	pthread_attr_t attr;
+	cpu_set_t cpuset;
+
+	SKIP_IF(!have_htm());
+
+	cpu = pick_online_cpu();
+	FAIL_IF(cpu < 0);
+
+	// Set only one CPU in the mask. Both threads will be bound to that CPU.
+	CPU_ZERO(&cpuset);
+	CPU_SET(cpu, &cpuset);
+
+	/* Init pthread attribute. */
+	rc = pthread_attr_init(&attr);
+	if (rc)
+		pr_err(rc, "pthread_attr_init()");
+
+	/* Set CPU 0 mask into the pthread attribute. */
+	rc = pthread_attr_setaffinity_np(&attr, sizeof(cpu_set_t), &cpuset);
+	if (rc)
+		pr_err(rc, "pthread_attr_setaffinity_np()");
+
+	rc = pthread_create(&t1, &attr /* Bind to CPU 0 */, tm_una_pong, NULL);
+	if (rc)
+		pr_err(rc, "pthread_create()");
+
+	/* Name it for systemtap convenience */
+	rc = pthread_setname_np(t1, "tm_una_pong");
+	if (rc)
+		pr_warn(rc, "pthread_create()");
+
+	flags.result = 0;
+
+	for (exception = 0; exception < NUM_EXCEPTIONS; exception++) {
+		printf("Checking if FP/VEC registers are sane after");
+
+		if (exception == FP_UNA_EXCEPTION)
+			printf(" a FP unavailable exception...\n");
+
+		else if (exception == VEC_UNA_EXCEPTION)
+			printf(" a VEC unavailable exception...\n");
+
+		else
+			printf(" a VSX unavailable exception...\n");
+
+		flags.exception = exception;
+
+		test_fp_vec(0, 0, &attr);
+		test_fp_vec(1, 0, &attr);
+		test_fp_vec(0, 1, &attr);
+		test_fp_vec(1, 1, &attr);
+
+	}
+
+	if (flags.result > 0) {
+		printf("result: failed!\n");
+		exit(1);
+	} else {
+		printf("result: success\n");
+		exit(0);
+	}
+}
+
+int main(int argc, char **argv)
+{
+	test_harness_set_timeout(220);
+	return test_harness(tm_unavailable_test, "tm_unavailable_test");
+}
diff --git a/tools/testing/selftests/powerpc/tm/tm-vmx-unavail.c b/tools/testing/selftests/powerpc/tm/tm-vmx-unavail.c
new file mode 100644
index 000000000..e2a0c07e8
--- /dev/null
+++ b/tools/testing/selftests/powerpc/tm/tm-vmx-unavail.c
@@ -0,0 +1,118 @@
+// SPDX-License-Identifier: GPL-2.0-only
+/*
+ * Copyright 2017, Michael Neuling, IBM Corp.
+ * Original: Breno Leitao <brenohl@br.ibm.com> &
+ *           Gustavo Bueno Romero <gromero@br.ibm.com>
+ * Edited: Michael Neuling
+ *
+ * Force VMX unavailable during a transaction and see if it corrupts
+ * the checkpointed VMX register state after the abort.
+ */
+
+#include <inttypes.h>
+#include <htmintrin.h>
+#include <string.h>
+#include <stdlib.h>
+#include <stdio.h>
+#include <pthread.h>
+#include <sys/mman.h>
+#include <unistd.h>
+#include <pthread.h>
+
+#include "tm.h"
+#include "utils.h"
+
+int passed;
+
+void *worker(void *unused)
+{
+	__int128 vmx0;
+	uint64_t texasr;
+
+	asm goto (
+		"li       3, 1;"  /* Stick non-zero value in VMX0 */
+		"std      3, 0(%[vmx0_ptr]);"
+		"lvx      0, 0, %[vmx0_ptr];"
+
+		/* Wait here a bit so we get scheduled out 255 times */
+		"lis      3, 0x3fff;"
+		"1: ;"
+		"addi     3, 3, -1;"
+		"cmpdi    3, 0;"
+		"bne      1b;"
+
+		/* Kernel will hopefully turn VMX off now */
+
+		"tbegin. ;"
+		"beq      failure;"
+
+		/* Cause VMX unavail. Any VMX instruction */
+		"vaddcuw  0,0,0;"
+
+		"tend. ;"
+		"b        %l[success];"
+
+		/* Check VMX0 sanity after abort */
+		"failure: ;"
+		"lvx       1,  0, %[vmx0_ptr];"
+		"vcmpequb. 2,  0, 1;"
+		"bc        4, 24, %l[value_mismatch];"
+		"b        %l[value_match];"
+		:
+		: [vmx0_ptr] "r"(&vmx0)
+		: "r3"
+		: success, value_match, value_mismatch
+		);
+
+	/* HTM aborted and VMX0 is corrupted */
+value_mismatch:
+	texasr = __builtin_get_texasr();
+
+	printf("\n\n==============\n\n");
+	printf("Failure with error: %lx\n",   _TEXASR_FAILURE_CODE(texasr));
+	printf("Summary error     : %lx\n",   _TEXASR_FAILURE_SUMMARY(texasr));
+	printf("TFIAR exact       : %lx\n\n", _TEXASR_TFIAR_EXACT(texasr));
+
+	passed = 0;
+	return NULL;
+
+	/* HTM aborted but VMX0 is correct */
+value_match:
+//	printf("!");
+	return NULL;
+
+success:
+//	printf(".");
+	return NULL;
+}
+
+int tm_vmx_unavail_test()
+{
+	int threads;
+	pthread_t *thread;
+
+	SKIP_IF(!have_htm());
+
+	passed = 1;
+
+	threads = sysconf(_SC_NPROCESSORS_ONLN) * 4;
+	thread = malloc(sizeof(pthread_t)*threads);
+	if (!thread)
+		return EXIT_FAILURE;
+
+	for (uint64_t i = 0; i < threads; i++)
+		pthread_create(&thread[i], NULL, &worker, NULL);
+
+	for (uint64_t i = 0; i < threads; i++)
+		pthread_join(thread[i], NULL);
+
+	free(thread);
+
+	return passed ? EXIT_SUCCESS : EXIT_FAILURE;
+}
+
+
+int main(int argc, char **argv)
+{
+	return test_harness(tm_vmx_unavail_test, "tm_vmx_unavail_test");
+}
diff --git a/tools/testing/selftests/powerpc/tm/tm-vmxcopy.c b/tools/testing/selftests/powerpc/tm/tm-vmxcopy.c
new file mode 100644
index 000000000..c1e788a6d
--- /dev/null
+++ b/tools/testing/selftests/powerpc/tm/tm-vmxcopy.c
@@ -0,0 +1,104 @@
+// SPDX-License-Identifier: GPL-2.0-only
+/*
+ * Copyright 2015, Michael Neuling, IBM Corp.
+ *
+ * Original: Michael Neuling 4/12/2013
+ * Edited: Rashmica Gupta 4/12/2015
+ *
+ * See if the altivec state is leaked out of an aborted transaction due to
+ * kernel vmx copy loops.
+ *
+ * When the transaction aborts, VSR values should rollback to the values
+ * they held before the transaction commenced. Using VSRs while transaction
+ * is suspended should not affect the checkpointed values.
+ *
+ * (1) write A to a VSR
+ * (2) start transaction
+ * (3) suspend transaction
+ * (4) change the VSR to B
+ * (5) trigger kernel vmx copy loop
+ * (6) abort transaction
+ * (7) check that the VSR value is A
+ */
+
+#include <inttypes.h>
+#include <stdio.h>
+#include <stdlib.h>
+#include <unistd.h>
+#include <sys/mman.h>
+#include <string.h>
+#include <assert.h>
+
+#include "tm.h"
+#include "utils.h"
+
+int test_vmxcopy()
+{
+	long double vecin = 1.3;
+	long double vecout;
+	unsigned long pgsize = getpagesize();
+	int i;
+	int fd;
+	int size = pgsize*16;
+	char tmpfile[] = "/tmp/page_faultXXXXXX";
+	char buf[pgsize];
+	char *a;
+	uint64_t aborted = 0;
+
+	SKIP_IF(!have_htm());
+	SKIP_IF(!is_ppc64le());
+
+	fd = mkstemp(tmpfile);
+	assert(fd >= 0);
+
+	memset(buf, 0, pgsize);
+	for (i = 0; i < size; i += pgsize)
+		assert(write(fd, buf, pgsize) == pgsize);
+
+	unlink(tmpfile);
+
+	a = mmap(NULL, size, PROT_READ|PROT_WRITE, MAP_PRIVATE, fd, 0);
+	assert(a != MAP_FAILED);
+
+	asm __volatile__(
+		"lxvd2x 40,0,%[vecinptr];"	/* set 40 to initial value*/
+		"tbegin.;"
+		"beq	3f;"
+		"tsuspend.;"
+		"xxlxor 40,40,40;"		/* set 40 to 0 */
+		"std	5, 0(%[map]);"		/* cause kernel vmx copy page */
+		"tabort. 0;"
+		"tresume.;"
+		"tend.;"
+		"li	%[res], 0;"
+		"b	5f;"
+
+		/* Abort handler */
+		"3:;"
+		"li	%[res], 1;"
+
+		"5:;"
+		"stxvd2x 40,0,%[vecoutptr];"
+		: [res]"=&r"(aborted)
+		: [vecinptr]"r"(&vecin),
+		  [vecoutptr]"r"(&vecout),
+		  [map]"r"(a)
+		: "memory", "r0", "r3", "r4", "r5", "r6", "r7");
+
+	if (aborted && (vecin != vecout)){
+		printf("FAILED: vector state leaked on abort %f != %f\n",
+		       (double)vecin, (double)vecout);
+		return 1;
+	}
+
+	munmap(a, size);
+
+	close(fd);
+
+	return 0;
+}
+
+int main(void)
+{
+	return test_harness(test_vmxcopy, "tm_vmxcopy");
+}
diff --git a/tools/testing/selftests/powerpc/tm/tm.h b/tools/testing/selftests/powerpc/tm/tm.h
new file mode 100644
index 000000000..c5a1e5c16
--- /dev/null
+++ b/tools/testing/selftests/powerpc/tm/tm.h
@@ -0,0 +1,96 @@
+/* SPDX-License-Identifier: GPL-2.0-only */
+/*
+ * Copyright 2015, Michael Ellerman, IBM Corp.
+ */
+
+#ifndef _SELFTESTS_POWERPC_TM_TM_H
+#define _SELFTESTS_POWERPC_TM_TM_H
+
+#include <stdbool.h>
+#include <asm/tm.h>
+
+#include "utils.h"
+
+static inline bool have_htm(void)
+{
+#ifdef PPC_FEATURE2_HTM
+	return have_hwcap2(PPC_FEATURE2_HTM);
+#else
+	printf("PPC_FEATURE2_HTM not defined, can't check AT_HWCAP2\n");
+	return false;
+#endif
+}
+
+static inline bool have_htm_nosc(void)
+{
+#ifdef PPC_FEATURE2_HTM_NOSC
+	return have_hwcap2(PPC_FEATURE2_HTM_NOSC);
+#else
+	printf("PPC_FEATURE2_HTM_NOSC not defined, can't check AT_HWCAP2\n");
+	return false;
+#endif
+}
+
+static inline long failure_code(void)
+{
+	return __builtin_get_texasru() >> 24;
+}
+
+static inline bool failure_is_persistent(void)
+{
+	return (failure_code() & TM_CAUSE_PERSISTENT) == TM_CAUSE_PERSISTENT;
+}
+
+static inline bool failure_is_syscall(void)
+{
+	return (failure_code() & TM_CAUSE_SYSCALL) == TM_CAUSE_SYSCALL;
+}
+
+static inline bool failure_is_unavailable(void)
+{
+	return (failure_code() & TM_CAUSE_FAC_UNAV) == TM_CAUSE_FAC_UNAV;
+}
+
+static inline bool failure_is_reschedule(void)
+{
+	if ((failure_code() & TM_CAUSE_RESCHED) == TM_CAUSE_RESCHED ||
+	    (failure_code() & TM_CAUSE_KVM_RESCHED) == TM_CAUSE_KVM_RESCHED ||
+	    (failure_code() & TM_CAUSE_KVM_FAC_UNAV) == TM_CAUSE_KVM_FAC_UNAV)
+		return true;
+
+	return false;
+}
+
+static inline bool failure_is_nesting(void)
+{
+	return (__builtin_get_texasru() & 0x400000);
+}
+
+static inline int tcheck(void)
+{
+	long cr;
+	asm volatile ("tcheck 0" : "=r"(cr) : : "cr0");
+	return (cr >> 28) & 4;
+}
+
+static inline bool tcheck_doomed(void)
+{
+	return tcheck() & 8;
+}
+
+static inline bool tcheck_active(void)
+{
+	return tcheck() & 4;
+}
+
+static inline bool tcheck_suspended(void)
+{
+	return tcheck() & 2;
+}
+
+static inline bool tcheck_transactional(void)
+{
+	return tcheck() & 6;
+}
+
+#endif /* _SELFTESTS_POWERPC_TM_TM_H */
diff --git a/tools/testing/selftests/powerpc/utils.c b/tools/testing/selftests/powerpc/utils.c
new file mode 100644
index 000000000..1f36ee1a9
--- /dev/null
+++ b/tools/testing/selftests/powerpc/utils.c
@@ -0,0 +1,303 @@
+// SPDX-License-Identifier: GPL-2.0-only
+/*
+ * Copyright 2013-2015, Michael Ellerman, IBM Corp.
+ */
+
+#define _GNU_SOURCE	/* For CPU_ZERO etc. */
+
+#include <elf.h>
+#include <errno.h>
+#include <fcntl.h>
+#include <link.h>
+#include <sched.h>
+#include <stdio.h>
+#include <stdlib.h>
+#include <string.h>
+#include <sys/ioctl.h>
+#include <sys/stat.h>
+#include <sys/sysinfo.h>
+#include <sys/types.h>
+#include <sys/utsname.h>
+#include <unistd.h>
+#include <asm/unistd.h>
+#include <linux/limits.h>
+
+#include "utils.h"
+
+static char auxv[4096];
+
+int read_auxv(char *buf, ssize_t buf_size)
+{
+	ssize_t num;
+	int rc, fd;
+
+	fd = open("/proc/self/auxv", O_RDONLY);
+	if (fd == -1) {
+		perror("open");
+		return -errno;
+	}
+
+	num = read(fd, buf, buf_size);
+	if (num < 0) {
+		perror("read");
+		rc = -EIO;
+		goto out;
+	}
+
+	if (num > buf_size) {
+		printf("overflowed auxv buffer\n");
+		rc = -EOVERFLOW;
+		goto out;
+	}
+
+	rc = 0;
+out:
+	close(fd);
+	return rc;
+}
+
+void *find_auxv_entry(int type, char *auxv)
+{
+	ElfW(auxv_t) *p;
+
+	p = (ElfW(auxv_t) *)auxv;
+
+	while (p->a_type != AT_NULL) {
+		if (p->a_type == type)
+			return p;
+
+		p++;
+	}
+
+	return NULL;
+}
+
+void *get_auxv_entry(int type)
+{
+	ElfW(auxv_t) *p;
+
+	if (read_auxv(auxv, sizeof(auxv)))
+		return NULL;
+
+	p = find_auxv_entry(type, auxv);
+	if (p)
+		return (void *)p->a_un.a_val;
+
+	return NULL;
+}
+
+int pick_online_cpu(void)
+{
+	int ncpus, cpu = -1;
+	cpu_set_t *mask;
+	size_t size;
+
+	ncpus = get_nprocs_conf();
+	size = CPU_ALLOC_SIZE(ncpus);
+	mask = CPU_ALLOC(ncpus);
+	if (!mask) {
+		perror("malloc");
+		return -1;
+	}
+
+	CPU_ZERO_S(size, mask);
+
+	if (sched_getaffinity(0, size, mask)) {
+		perror("sched_getaffinity");
+		goto done;
+	}
+
+	/* We prefer a primary thread, but skip 0 */
+	for (cpu = 8; cpu < ncpus; cpu += 8)
+		if (CPU_ISSET_S(cpu, size, mask))
+			goto done;
+
+	/* Search for anything, but in reverse */
+	for (cpu = ncpus - 1; cpu >= 0; cpu--)
+		if (CPU_ISSET_S(cpu, size, mask))
+			goto done;
+
+	printf("No cpus in affinity mask?!\n");
+
+done:
+	CPU_FREE(mask);
+	return cpu;
+}
+
+bool is_ppc64le(void)
+{
+	struct utsname uts;
+	int rc;
+
+	errno = 0;
+	rc = uname(&uts);
+	if (rc) {
+		perror("uname");
+		return false;
+	}
+
+	return strcmp(uts.machine, "ppc64le") == 0;
+}
+
+int read_sysfs_file(char *fpath, char *result, size_t result_size)
+{
+	char path[PATH_MAX] = "/sys/";
+	int rc = -1, fd;
+
+	strncat(path, fpath, PATH_MAX - strlen(path) - 1);
+
+	if ((fd = open(path, O_RDONLY)) < 0)
+		return rc;
+
+	rc = read(fd, result, result_size);
+
+	close(fd);
+
+	if (rc < 0)
+		return rc;
+
+	return 0;
+}
+
+int read_debugfs_file(char *debugfs_file, int *result)
+{
+	int rc = -1, fd;
+	char path[PATH_MAX];
+	char value[16];
+
+	strcpy(path, "/sys/kernel/debug/");
+	strncat(path, debugfs_file, PATH_MAX - strlen(path) - 1);
+
+	if ((fd = open(path, O_RDONLY)) < 0)
+		return rc;
+
+	if ((rc = read(fd, value, sizeof(value))) < 0)
+		return rc;
+
+	value[15] = 0;
+	*result = atoi(value);
+	close(fd);
+
+	return 0;
+}
+
+int write_debugfs_file(char *debugfs_file, int result)
+{
+	int rc = -1, fd;
+	char path[PATH_MAX];
+	char value[16];
+
+	strcpy(path, "/sys/kernel/debug/");
+	strncat(path, debugfs_file, PATH_MAX - strlen(path) - 1);
+
+	if ((fd = open(path, O_WRONLY)) < 0)
+		return rc;
+
+	snprintf(value, 16, "%d", result);
+
+	if ((rc = write(fd, value, strlen(value))) < 0)
+		return rc;
+
+	close(fd);
+
+	return 0;
+}
+
+static long perf_event_open(struct perf_event_attr *hw_event, pid_t pid,
+		int cpu, int group_fd, unsigned long flags)
+{
+	return syscall(__NR_perf_event_open, hw_event, pid, cpu,
+		      group_fd, flags);
+}
+
+static void perf_event_attr_init(struct perf_event_attr *event_attr,
+					unsigned int type,
+					unsigned long config)
+{
+	memset(event_attr, 0, sizeof(*event_attr));
+
+	event_attr->type = type;
+	event_attr->size = sizeof(struct perf_event_attr);
+	event_attr->config = config;
+	event_attr->read_format = PERF_FORMAT_GROUP;
+	event_attr->disabled = 1;
+	event_attr->exclude_kernel = 1;
+	event_attr->exclude_hv = 1;
+	event_attr->exclude_guest = 1;
+}
+
+int perf_event_open_counter(unsigned int type,
+			    unsigned long config, int group_fd)
+{
+	int fd;
+	struct perf_event_attr event_attr;
+
+	perf_event_attr_init(&event_attr, type, config);
+
+	fd = perf_event_open(&event_attr, 0, -1, group_fd, 0);
+
+	if (fd < 0)
+		perror("perf_event_open() failed");
+
+	return fd;
+}
+
+int perf_event_enable(int fd)
+{
+	if (ioctl(fd, PERF_EVENT_IOC_ENABLE, PERF_IOC_FLAG_GROUP) == -1) {
+		perror("error while enabling perf events");
+		return -1;
+	}
+
+	return 0;
+}
+
+int perf_event_disable(int fd)
+{
+	if (ioctl(fd, PERF_EVENT_IOC_DISABLE, PERF_IOC_FLAG_GROUP) == -1) {
+		perror("error disabling perf events");
+		return -1;
+	}
+
+	return 0;
+}
+
+int perf_event_reset(int fd)
+{
+	if (ioctl(fd, PERF_EVENT_IOC_RESET, PERF_IOC_FLAG_GROUP) == -1) {
+		perror("error resetting perf events");
+		return -1;
+	}
+
+	return 0;
+}
+
+int using_hash_mmu(bool *using_hash)
+{
+	char line[128];
+	FILE *f;
+	int rc;
+
+	f = fopen("/proc/cpuinfo", "r");
+	FAIL_IF(!f);
+
+	rc = 0;
+	while (fgets(line, sizeof(line), f) != NULL) {
+		if (!strcmp(line, "MMU		: Hash\n") ||
+		    !strcmp(line, "platform	: Cell\n") ||
+		    !strcmp(line, "platform	: PowerMac\n")) {
+			*using_hash = true;
+			goto out;
+		}
+
+		if (strcmp(line, "MMU		: Radix\n") == 0) {
+			*using_hash = false;
+			goto out;
+		}
+	}
+
+	rc = -1;
+out:
+	fclose(f);
+	return rc;
+}
diff --git a/tools/testing/selftests/powerpc/vphn/.gitignore b/tools/testing/selftests/powerpc/vphn/.gitignore
new file mode 100644
index 000000000..b744aedfd
--- /dev/null
+++ b/tools/testing/selftests/powerpc/vphn/.gitignore
@@ -0,0 +1,2 @@
+# SPDX-License-Identifier: GPL-2.0-only
+test-vphn
diff --git a/tools/testing/selftests/powerpc/vphn/Makefile b/tools/testing/selftests/powerpc/vphn/Makefile
new file mode 100644
index 000000000..cf65cbf33
--- /dev/null
+++ b/tools/testing/selftests/powerpc/vphn/Makefile
@@ -0,0 +1,10 @@
+# SPDX-License-Identifier: GPL-2.0-only
+TEST_GEN_PROGS := test-vphn
+
+CFLAGS += -m64 -I$(CURDIR)
+
+top_srcdir = ../../../../..
+include ../../lib.mk
+
+$(TEST_GEN_PROGS): ../harness.c
+
diff --git a/tools/testing/selftests/powerpc/vphn/asm/lppaca.h b/tools/testing/selftests/powerpc/vphn/asm/lppaca.h
new file mode 120000
index 000000000..942b1d009
--- /dev/null
+++ b/tools/testing/selftests/powerpc/vphn/asm/lppaca.h
@@ -0,0 +1 @@
+../../../../../../arch/powerpc/include/asm/lppaca.h
\ No newline at end of file
diff --git a/tools/testing/selftests/powerpc/vphn/test-vphn.c b/tools/testing/selftests/powerpc/vphn/test-vphn.c
new file mode 100644
index 000000000..81d3069ff
--- /dev/null
+++ b/tools/testing/selftests/powerpc/vphn/test-vphn.c
@@ -0,0 +1,411 @@
+// SPDX-License-Identifier: GPL-2.0
+#include <stdio.h>
+#include <byteswap.h>
+#include "utils.h"
+#include "subunit.h"
+
+#if __BYTE_ORDER__ == __ORDER_LITTLE_ENDIAN__
+#define cpu_to_be32(x)		bswap_32(x)
+#define be32_to_cpu(x)		bswap_32(x)
+#define be16_to_cpup(x)		bswap_16(*x)
+#define cpu_to_be64(x)		bswap_64(x)
+#else
+#define cpu_to_be32(x)		(x)
+#define be32_to_cpu(x)		(x)
+#define be16_to_cpup(x)		(*x)
+#define cpu_to_be64(x)		(x)
+#endif
+
+#include "vphn.c"
+
+static struct test {
+	char *descr;
+	long input[VPHN_REGISTER_COUNT];
+	u32 expected[VPHN_ASSOC_BUFSIZE];
+} all_tests[] = {
+	{
+		"vphn: no data",
+		{
+			0xffffffffffffffff,
+			0xffffffffffffffff,
+			0xffffffffffffffff,
+			0xffffffffffffffff,
+			0xffffffffffffffff,
+			0xffffffffffffffff,
+		},
+		{
+			0x00000000
+		}
+	},
+	{
+		"vphn: 1 x 16-bit value",
+		{
+			0x8001ffffffffffff,
+			0xffffffffffffffff,
+			0xffffffffffffffff,
+			0xffffffffffffffff,
+			0xffffffffffffffff,
+			0xffffffffffffffff,
+		},
+		{
+			0x00000001,
+			0x00000001
+		}
+	},
+	{
+		"vphn: 2 x 16-bit values",
+		{
+			0x80018002ffffffff,
+			0xffffffffffffffff,
+			0xffffffffffffffff,
+			0xffffffffffffffff,
+			0xffffffffffffffff,
+			0xffffffffffffffff,
+		},
+		{
+			0x00000002,
+			0x00000001,
+			0x00000002
+		}
+	},
+	{
+		"vphn: 3 x 16-bit values",
+		{
+			0x800180028003ffff,
+			0xffffffffffffffff,
+			0xffffffffffffffff,
+			0xffffffffffffffff,
+			0xffffffffffffffff,
+			0xffffffffffffffff,
+		},
+		{
+			0x00000003,
+			0x00000001,
+			0x00000002,
+			0x00000003
+		}
+	},
+	{
+		"vphn: 4 x 16-bit values",
+		{
+			0x8001800280038004,
+			0xffffffffffffffff,
+			0xffffffffffffffff,
+			0xffffffffffffffff,
+			0xffffffffffffffff,
+			0xffffffffffffffff,
+		},
+		{
+			0x00000004,
+			0x00000001,
+			0x00000002,
+			0x00000003,
+			0x00000004
+		}
+	},
+	{
+		/* Parsing the next 16-bit value out of the next 64-bit input
+		 * value.
+		 */
+		"vphn: 5 x 16-bit values",
+		{
+			0x8001800280038004,
+			0x8005ffffffffffff,
+			0xffffffffffffffff,
+			0xffffffffffffffff,
+			0xffffffffffffffff,
+			0xffffffffffffffff,
+		},
+		{
+			0x00000005,
+			0x00000001,
+			0x00000002,
+			0x00000003,
+			0x00000004,
+			0x00000005
+		}
+	},
+	{
+		/* Parse at most 6 x 64-bit input values */
+		"vphn: 24 x 16-bit values",
+		{
+			0x8001800280038004,
+			0x8005800680078008,
+			0x8009800a800b800c,
+			0x800d800e800f8010,
+			0x8011801280138014,
+			0x8015801680178018
+		},
+		{
+			0x00000018,
+			0x00000001,
+			0x00000002,
+			0x00000003,
+			0x00000004,
+			0x00000005,
+			0x00000006,
+			0x00000007,
+			0x00000008,
+			0x00000009,
+			0x0000000a,
+			0x0000000b,
+			0x0000000c,
+			0x0000000d,
+			0x0000000e,
+			0x0000000f,
+			0x00000010,
+			0x00000011,
+			0x00000012,
+			0x00000013,
+			0x00000014,
+			0x00000015,
+			0x00000016,
+			0x00000017,
+			0x00000018
+		}
+	},
+	{
+		"vphn: 1 x 32-bit value",
+		{
+			0x00000001ffffffff,
+			0xffffffffffffffff,
+			0xffffffffffffffff,
+			0xffffffffffffffff,
+			0xffffffffffffffff,
+			0xffffffffffffffff
+		},
+		{
+			0x00000001,
+			0x00000001
+		}
+	},
+	{
+		"vphn: 2 x 32-bit values",
+		{
+			0x0000000100000002,
+			0xffffffffffffffff,
+			0xffffffffffffffff,
+			0xffffffffffffffff,
+			0xffffffffffffffff,
+			0xffffffffffffffff
+		},
+		{
+			0x00000002,
+			0x00000001,
+			0x00000002
+		}
+	},
+	{
+		/* Parsing the next 32-bit value out of the next 64-bit input
+		 * value.
+		 */
+		"vphn: 3 x 32-bit values",
+		{
+			0x0000000100000002,
+			0x00000003ffffffff,
+			0xffffffffffffffff,
+			0xffffffffffffffff,
+			0xffffffffffffffff,
+			0xffffffffffffffff
+		},
+		{
+			0x00000003,
+			0x00000001,
+			0x00000002,
+			0x00000003
+		}
+	},
+	{
+		/* Parse at most 6 x 64-bit input values */
+		"vphn: 12 x 32-bit values",
+		{
+			0x0000000100000002,
+			0x0000000300000004,
+			0x0000000500000006,
+			0x0000000700000008,
+			0x000000090000000a,
+			0x0000000b0000000c
+		},
+		{
+			0x0000000c,
+			0x00000001,
+			0x00000002,
+			0x00000003,
+			0x00000004,
+			0x00000005,
+			0x00000006,
+			0x00000007,
+			0x00000008,
+			0x00000009,
+			0x0000000a,
+			0x0000000b,
+			0x0000000c
+		}
+	},
+	{
+		"vphn: 16-bit value followed by 32-bit value",
+		{
+			0x800100000002ffff,
+			0xffffffffffffffff,
+			0xffffffffffffffff,
+			0xffffffffffffffff,
+			0xffffffffffffffff,
+			0xffffffffffffffff
+		},
+		{
+			0x00000002,
+			0x00000001,
+			0x00000002
+		}
+	},
+	{
+		"vphn: 32-bit value followed by 16-bit value",
+		{
+			0x000000018002ffff,
+			0xffffffffffffffff,
+			0xffffffffffffffff,
+			0xffffffffffffffff,
+			0xffffffffffffffff,
+			0xffffffffffffffff
+		},
+		{
+			0x00000002,
+			0x00000001,
+			0x00000002
+		}
+	},
+	{
+		/* Parse a 32-bit value split accross two consecutives 64-bit
+		 * input values.
+		 */
+		"vphn: 16-bit value followed by 2 x 32-bit values",
+		{
+			0x8001000000020000,
+			0x0003ffffffffffff,
+			0xffffffffffffffff,
+			0xffffffffffffffff,
+			0xffffffffffffffff,
+			0xffffffffffffffff
+		},
+		{
+			0x00000003,
+			0x00000001,
+			0x00000002,
+			0x00000003,
+			0x00000004,
+			0x00000005
+		}
+	},
+	{
+		/* The lower bits in 0x0001ffff don't get mixed up with the
+		 * 0xffff terminator.
+		 */
+		"vphn: 32-bit value has all ones in 16 lower bits",
+		{
+			0x0001ffff80028003,
+			0xffffffffffffffff,
+			0xffffffffffffffff,
+			0xffffffffffffffff,
+			0xffffffffffffffff,
+			0xffffffffffffffff
+		},
+		{
+			0x00000003,
+			0x0001ffff,
+			0x00000002,
+			0x00000003
+		}
+	},
+	{
+		/* The following input doesn't follow the specification.
+		 */
+		"vphn: last 32-bit value is truncated",
+		{
+			0x0000000100000002,
+			0x0000000300000004,
+			0x0000000500000006,
+			0x0000000700000008,
+			0x000000090000000a,
+			0x0000000b800c2bad
+		},
+		{
+			0x0000000c,
+			0x00000001,
+			0x00000002,
+			0x00000003,
+			0x00000004,
+			0x00000005,
+			0x00000006,
+			0x00000007,
+			0x00000008,
+			0x00000009,
+			0x0000000a,
+			0x0000000b,
+			0x0000000c
+		}
+	},
+	{
+		"vphn: garbage after terminator",
+		{
+			0xffff2bad2bad2bad,
+			0x2bad2bad2bad2bad,
+			0x2bad2bad2bad2bad,
+			0x2bad2bad2bad2bad,
+			0x2bad2bad2bad2bad,
+			0x2bad2bad2bad2bad
+		},
+		{
+			0x00000000
+		}
+	},
+	{
+		NULL
+	}
+};
+
+static int test_one(struct test *test)
+{
+	__be32 output[VPHN_ASSOC_BUFSIZE] = { 0 };
+	int i, len;
+
+	vphn_unpack_associativity(test->input, output);
+
+	len = be32_to_cpu(output[0]);
+	if (len != test->expected[0]) {
+		printf("expected %d elements, got %d\n", test->expected[0],
+		       len);
+		return 1;
+	}
+
+	for (i = 1; i < len; i++) {
+		u32 val = be32_to_cpu(output[i]);
+		if (val != test->expected[i]) {
+			printf("element #%d is 0x%x, should be 0x%x\n", i, val,
+			       test->expected[i]);
+			return 1;
+		}
+	}
+
+	return 0;
+}
+
+static int test_vphn(void)
+{
+	static struct test *test;
+
+	for (test = all_tests; test->descr; test++) {
+		int ret;
+
+		ret = test_one(test);
+		test_finish(test->descr, ret);
+		if (ret)
+			return ret;
+	}
+
+	return 0;
+}
+
+int main(int argc, char **argv)
+{
+	return test_harness(test_vphn, "test-vphn");
+}
diff --git a/tools/testing/selftests/powerpc/vphn/vphn.c b/tools/testing/selftests/powerpc/vphn/vphn.c
new file mode 120000
index 000000000..5b5fbddcc
--- /dev/null
+++ b/tools/testing/selftests/powerpc/vphn/vphn.c
@@ -0,0 +1 @@
+../../../../../arch/powerpc/platforms/pseries/vphn.c
\ No newline at end of file