Adding upstream version 18.2.2.upstream/18.2.2

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

11 files changed, 846 insertions, 0 deletions

diff --git a/src/spdk/dpdk/kernel/freebsd/BSDmakefile.meson b/src/spdk/dpdk/kernel/freebsd/BSDmakefile.meson
new file mode 100644
index 000000000..6839ac01c
--- /dev/null
+++ b/src/spdk/dpdk/kernel/freebsd/BSDmakefile.meson
@@ -0,0 +1,18 @@
+# SPDX-License-Identifier: BSD-3-Clause
+# Copyright(c) 2017 Intel Corporation
+
+# makefile for building kernel modules using meson
+# takes parameters from the environment
+
+# source file is passed via KMOD_SRC as relative path, we only use final
+# (tail) component of it (:T), as VPATH is used to find actual file. The
+# VPATH is similarly extracted from the non-final (head) portion of the
+# path (:H) converted to absolute path (:tA).  This use of VPATH is to have
+# the .o files placed in the build, not source directory
+
+VPATH := ${KMOD_SRC:H:tA}
+SRCS := ${KMOD_SRC:T} device_if.h bus_if.h pci_if.h
+CFLAGS += $(KMOD_CFLAGS)
+.OBJDIR: ${KMOD_OBJDIR}
+
+.include <bsd.kmod.mk>
diff --git a/src/spdk/dpdk/kernel/freebsd/Makefile b/src/spdk/dpdk/kernel/freebsd/Makefile
new file mode 100644
index 000000000..522d3f68f
--- /dev/null
+++ b/src/spdk/dpdk/kernel/freebsd/Makefile
@@ -0,0 +1,9 @@
+# SPDX-License-Identifier: BSD-3-Clause
+# Copyright 2017 NXP
+
+include $(RTE_SDK)/mk/rte.vars.mk
+
+DIRS-$(CONFIG_RTE_EXEC_ENV_FREEBSD) += contigmem
+DIRS-$(CONFIG_RTE_EXEC_ENV_FREEBSD) += nic_uio
+
+include $(RTE_SDK)/mk/rte.subdir.mk
diff --git a/src/spdk/dpdk/kernel/freebsd/contigmem/BSDmakefile b/src/spdk/dpdk/kernel/freebsd/contigmem/BSDmakefile
new file mode 100644
index 000000000..33ce83ee2
--- /dev/null
+++ b/src/spdk/dpdk/kernel/freebsd/contigmem/BSDmakefile
@@ -0,0 +1,8 @@
+# SPDX-License-Identifier: BSD-3-Clause
+# Copyright(c) 2010-2014 Intel Corporation
+#
+
+KMOD=	contigmem
+SRCS=	contigmem.c device_if.h bus_if.h
+
+.include <bsd.kmod.mk>
diff --git a/src/spdk/dpdk/kernel/freebsd/contigmem/Makefile b/src/spdk/dpdk/kernel/freebsd/contigmem/Makefile
new file mode 100644
index 000000000..428a7edee
--- /dev/null
+++ b/src/spdk/dpdk/kernel/freebsd/contigmem/Makefile
@@ -0,0 +1,24 @@
+# SPDX-License-Identifier: BSD-3-Clause
+# Copyright(c) 2010-2014 Intel Corporation
+
+include $(RTE_SDK)/mk/rte.vars.mk
+
+#
+# module name and path
+#
+MODULE = contigmem
+
+#
+# CFLAGS
+#
+MODULE_CFLAGS += -I$(SRCDIR)
+MODULE_CFLAGS += -I$(RTE_OUTPUT)/include
+MODULE_CFLAGS += -Winline -Wall -Werror
+MODULE_CFLAGS += -include $(RTE_OUTPUT)/include/rte_config.h
+
+#
+# all source are stored in SRCS-y
+#
+SRCS-y := contigmem.c
+
+include $(RTE_SDK)/mk/rte.bsdmodule.mk
diff --git a/src/spdk/dpdk/kernel/freebsd/contigmem/contigmem.c b/src/spdk/dpdk/kernel/freebsd/contigmem/contigmem.c
new file mode 100644
index 000000000..bd72f4d62
--- /dev/null
+++ b/src/spdk/dpdk/kernel/freebsd/contigmem/contigmem.c
@@ -0,0 +1,361 @@
+/* SPDX-License-Identifier: BSD-3-Clause
+ * Copyright(c) 2010-2014 Intel Corporation
+ */
+
+#include <sys/cdefs.h>
+__FBSDID("$FreeBSD$");
+
+#include <sys/param.h>
+#include <sys/bio.h>
+#include <sys/bus.h>
+#include <sys/conf.h>
+#include <sys/kernel.h>
+#include <sys/malloc.h>
+#include <sys/module.h>
+#include <sys/proc.h>
+#include <sys/lock.h>
+#include <sys/rwlock.h>
+#include <sys/mutex.h>
+#include <sys/systm.h>
+#include <sys/sysctl.h>
+#include <sys/vmmeter.h>
+#include <sys/eventhandler.h>
+
+#include <machine/bus.h>
+
+#include <vm/vm.h>
+#include <vm/pmap.h>
+#include <vm/vm_param.h>
+#include <vm/vm_object.h>
+#include <vm/vm_page.h>
+#include <vm/vm_pager.h>
+#include <vm/vm_phys.h>
+
+struct contigmem_buffer {
+	void           *addr;
+	int             refcnt;
+	struct mtx      mtx;
+};
+
+struct contigmem_vm_handle {
+	int             buffer_index;
+};
+
+static int              contigmem_load(void);
+static int              contigmem_unload(void);
+static int              contigmem_physaddr(SYSCTL_HANDLER_ARGS);
+
+static d_mmap_single_t  contigmem_mmap_single;
+static d_open_t         contigmem_open;
+static d_close_t        contigmem_close;
+
+static int              contigmem_num_buffers = RTE_CONTIGMEM_DEFAULT_NUM_BUFS;
+static int64_t          contigmem_buffer_size = RTE_CONTIGMEM_DEFAULT_BUF_SIZE;
+
+static eventhandler_tag contigmem_eh_tag;
+static struct contigmem_buffer contigmem_buffers[RTE_CONTIGMEM_MAX_NUM_BUFS];
+static struct cdev     *contigmem_cdev = NULL;
+static int              contigmem_refcnt;
+
+TUNABLE_INT("hw.contigmem.num_buffers", &contigmem_num_buffers);
+TUNABLE_QUAD("hw.contigmem.buffer_size", &contigmem_buffer_size);
+
+static SYSCTL_NODE(_hw, OID_AUTO, contigmem, CTLFLAG_RD, 0, "contigmem");
+
+SYSCTL_INT(_hw_contigmem, OID_AUTO, num_buffers, CTLFLAG_RD,
+	&contigmem_num_buffers, 0, "Number of contigmem buffers allocated");
+SYSCTL_QUAD(_hw_contigmem, OID_AUTO, buffer_size, CTLFLAG_RD,
+	&contigmem_buffer_size, 0, "Size of each contiguous buffer");
+SYSCTL_INT(_hw_contigmem, OID_AUTO, num_references, CTLFLAG_RD,
+	&contigmem_refcnt, 0, "Number of references to contigmem");
+
+static SYSCTL_NODE(_hw_contigmem, OID_AUTO, physaddr, CTLFLAG_RD, 0,
+	"physaddr");
+
+MALLOC_DEFINE(M_CONTIGMEM, "contigmem", "contigmem(4) allocations");
+
+static int contigmem_modevent(module_t mod, int type, void *arg)
+{
+	int error = 0;
+
+	switch (type) {
+	case MOD_LOAD:
+		error = contigmem_load();
+		break;
+	case MOD_UNLOAD:
+		error = contigmem_unload();
+		break;
+	default:
+		break;
+	}
+
+	return error;
+}
+
+moduledata_t contigmem_mod = {
+	"contigmem",
+	(modeventhand_t)contigmem_modevent,
+	0
+};
+
+DECLARE_MODULE(contigmem, contigmem_mod, SI_SUB_DRIVERS, SI_ORDER_ANY);
+MODULE_VERSION(contigmem, 1);
+
+static struct cdevsw contigmem_ops = {
+	.d_name         = "contigmem",
+	.d_version      = D_VERSION,
+	.d_flags        = D_TRACKCLOSE,
+	.d_mmap_single  = contigmem_mmap_single,
+	.d_open         = contigmem_open,
+	.d_close        = contigmem_close,
+};
+
+static int
+contigmem_load()
+{
+	char index_string[8], description[32];
+	int  i, error = 0;
+	void *addr;
+
+	if (contigmem_num_buffers > RTE_CONTIGMEM_MAX_NUM_BUFS) {
+		printf("%d buffers requested is greater than %d allowed\n",
+				contigmem_num_buffers, RTE_CONTIGMEM_MAX_NUM_BUFS);
+		error = EINVAL;
+		goto error;
+	}
+
+	if (contigmem_buffer_size < PAGE_SIZE ||
+			(contigmem_buffer_size & (contigmem_buffer_size - 1)) != 0) {
+		printf("buffer size 0x%lx is not greater than PAGE_SIZE and "
+				"power of two\n", contigmem_buffer_size);
+		error = EINVAL;
+		goto error;
+	}
+
+	for (i = 0; i < contigmem_num_buffers; i++) {
+		addr = contigmalloc(contigmem_buffer_size, M_CONTIGMEM, M_ZERO,
+			0, BUS_SPACE_MAXADDR, contigmem_buffer_size, 0);
+		if (addr == NULL) {
+			printf("contigmalloc failed for buffer %d\n", i);
+			error = ENOMEM;
+			goto error;
+		}
+
+		printf("%2u: virt=%p phys=%p\n", i, addr,
+			(void *)pmap_kextract((vm_offset_t)addr));
+
+		mtx_init(&contigmem_buffers[i].mtx, "contigmem", NULL, MTX_DEF);
+		contigmem_buffers[i].addr = addr;
+		contigmem_buffers[i].refcnt = 0;
+
+		snprintf(index_string, sizeof(index_string), "%d", i);
+		snprintf(description, sizeof(description),
+				"phys addr for buffer %d", i);
+		SYSCTL_ADD_PROC(NULL,
+				&SYSCTL_NODE_CHILDREN(_hw_contigmem, physaddr), OID_AUTO,
+				index_string, CTLTYPE_U64 | CTLFLAG_RD,
+				(void *)(uintptr_t)i, 0, contigmem_physaddr, "LU",
+				description);
+	}
+
+	contigmem_cdev = make_dev_credf(0, &contigmem_ops, 0, NULL, UID_ROOT,
+			GID_WHEEL, 0600, "contigmem");
+
+	return 0;
+
+error:
+	for (i = 0; i < contigmem_num_buffers; i++) {
+		if (contigmem_buffers[i].addr != NULL) {
+			contigfree(contigmem_buffers[i].addr,
+				contigmem_buffer_size, M_CONTIGMEM);
+			contigmem_buffers[i].addr = NULL;
+		}
+		if (mtx_initialized(&contigmem_buffers[i].mtx))
+			mtx_destroy(&contigmem_buffers[i].mtx);
+	}
+
+	return error;
+}
+
+static int
+contigmem_unload()
+{
+	int i;
+
+	if (contigmem_refcnt > 0)
+		return EBUSY;
+
+	if (contigmem_cdev != NULL)
+		destroy_dev(contigmem_cdev);
+
+	if (contigmem_eh_tag != NULL)
+		EVENTHANDLER_DEREGISTER(process_exit, contigmem_eh_tag);
+
+	for (i = 0; i < RTE_CONTIGMEM_MAX_NUM_BUFS; i++) {
+		if (contigmem_buffers[i].addr != NULL)
+			contigfree(contigmem_buffers[i].addr,
+				contigmem_buffer_size, M_CONTIGMEM);
+		if (mtx_initialized(&contigmem_buffers[i].mtx))
+			mtx_destroy(&contigmem_buffers[i].mtx);
+	}
+
+	return 0;
+}
+
+static int
+contigmem_physaddr(SYSCTL_HANDLER_ARGS)
+{
+	uint64_t	physaddr;
+	int		index = (int)(uintptr_t)arg1;
+
+	physaddr = (uint64_t)vtophys(contigmem_buffers[index].addr);
+	return sysctl_handle_64(oidp, &physaddr, 0, req);
+}
+
+static int
+contigmem_open(struct cdev *cdev, int fflags, int devtype,
+		struct thread *td)
+{
+
+	atomic_add_int(&contigmem_refcnt, 1);
+
+	return 0;
+}
+
+static int
+contigmem_close(struct cdev *cdev, int fflags, int devtype,
+		struct thread *td)
+{
+
+	atomic_subtract_int(&contigmem_refcnt, 1);
+
+	return 0;
+}
+
+static int
+contigmem_cdev_pager_ctor(void *handle, vm_ooffset_t size, vm_prot_t prot,
+		vm_ooffset_t foff, struct ucred *cred, u_short *color)
+{
+	struct contigmem_vm_handle *vmh = handle;
+	struct contigmem_buffer *buf;
+
+	buf = &contigmem_buffers[vmh->buffer_index];
+
+	atomic_add_int(&contigmem_refcnt, 1);
+
+	mtx_lock(&buf->mtx);
+	if (buf->refcnt == 0)
+		memset(buf->addr, 0, contigmem_buffer_size);
+	buf->refcnt++;
+	mtx_unlock(&buf->mtx);
+
+	return 0;
+}
+
+static void
+contigmem_cdev_pager_dtor(void *handle)
+{
+	struct contigmem_vm_handle *vmh = handle;
+	struct contigmem_buffer *buf;
+
+	buf = &contigmem_buffers[vmh->buffer_index];
+
+	mtx_lock(&buf->mtx);
+	buf->refcnt--;
+	mtx_unlock(&buf->mtx);
+
+	free(vmh, M_CONTIGMEM);
+
+	atomic_subtract_int(&contigmem_refcnt, 1);
+}
+
+static int
+contigmem_cdev_pager_fault(vm_object_t object, vm_ooffset_t offset, int prot,
+		vm_page_t *mres)
+{
+	vm_paddr_t paddr;
+	vm_page_t m_paddr, page;
+	vm_memattr_t memattr, memattr1;
+
+	memattr = object->memattr;
+
+	VM_OBJECT_WUNLOCK(object);
+
+	paddr = offset;
+
+	m_paddr = vm_phys_paddr_to_vm_page(paddr);
+	if (m_paddr != NULL) {
+		memattr1 = pmap_page_get_memattr(m_paddr);
+		if (memattr1 != memattr)
+			memattr = memattr1;
+	}
+
+	if (((*mres)->flags & PG_FICTITIOUS) != 0) {
+		/*
+		 * If the passed in result page is a fake page, update it with
+		 * the new physical address.
+		 */
+		page = *mres;
+		VM_OBJECT_WLOCK(object);
+		vm_page_updatefake(page, paddr, memattr);
+	} else {
+		/*
+		 * Replace the passed in reqpage page with our own fake page and
+		 * free up the original page.
+		 */
+		page = vm_page_getfake(paddr, memattr);
+		VM_OBJECT_WLOCK(object);
+#if __FreeBSD__ >= 13
+		vm_page_replace(page, object, (*mres)->pindex, *mres);
+#else
+		vm_page_t mret = vm_page_replace(page, object, (*mres)->pindex);
+		KASSERT(mret == *mres,
+		    ("invalid page replacement, old=%p, ret=%p", *mres, mret));
+		vm_page_lock(mret);
+		vm_page_free(mret);
+		vm_page_unlock(mret);
+#endif
+		*mres = page;
+	}
+
+	page->valid = VM_PAGE_BITS_ALL;
+
+	return VM_PAGER_OK;
+}
+
+static struct cdev_pager_ops contigmem_cdev_pager_ops = {
+	.cdev_pg_ctor = contigmem_cdev_pager_ctor,
+	.cdev_pg_dtor = contigmem_cdev_pager_dtor,
+	.cdev_pg_fault = contigmem_cdev_pager_fault,
+};
+
+static int
+contigmem_mmap_single(struct cdev *cdev, vm_ooffset_t *offset, vm_size_t size,
+		struct vm_object **obj, int nprot)
+{
+	struct contigmem_vm_handle *vmh;
+	uint64_t buffer_index;
+
+	/*
+	 * The buffer index is encoded in the offset.  Divide the offset by
+	 *  PAGE_SIZE to get the index of the buffer requested by the user
+	 *  app.
+	 */
+	buffer_index = *offset / PAGE_SIZE;
+	if (buffer_index >= contigmem_num_buffers)
+		return EINVAL;
+
+	if (size > contigmem_buffer_size)
+		return EINVAL;
+
+	vmh = malloc(sizeof(*vmh), M_CONTIGMEM, M_NOWAIT | M_ZERO);
+	if (vmh == NULL)
+		return ENOMEM;
+	vmh->buffer_index = buffer_index;
+
+	*offset = (vm_ooffset_t)vtophys(contigmem_buffers[buffer_index].addr);
+	*obj = cdev_pager_allocate(vmh, OBJT_DEVICE, &contigmem_cdev_pager_ops,
+			size, nprot, *offset, curthread->td_ucred);
+
+	return 0;
+}
diff --git a/src/spdk/dpdk/kernel/freebsd/contigmem/meson.build b/src/spdk/dpdk/kernel/freebsd/contigmem/meson.build
new file mode 100644
index 000000000..8fb2ab786
--- /dev/null
+++ b/src/spdk/dpdk/kernel/freebsd/contigmem/meson.build
@@ -0,0 +1,4 @@
+# SPDX-License-Identifier: BSD-3-Clause
+# Copyright(c) 2017 Intel Corporation
+
+sources = files('contigmem.c')
diff --git a/src/spdk/dpdk/kernel/freebsd/meson.build b/src/spdk/dpdk/kernel/freebsd/meson.build
new file mode 100644
index 000000000..dc156a43f
--- /dev/null
+++ b/src/spdk/dpdk/kernel/freebsd/meson.build
@@ -0,0 +1,35 @@
+# SPDX-License-Identifier: BSD-3-Clause
+# Copyright(c) 2018 Intel Corporation
+
+kmods = ['contigmem', 'nic_uio']
+
+# for building kernel modules, we use kernel build system using make, as
+# with Linux. We have a skeleton BSDmakefile, which pulls many of its
+# values from the environment. Each module only has a single source file
+# right now, which allows us to simplify things. We pull in the sourcer
+# files from the individual meson.build files, and then use a custom
+# target to call make, passing in the values as env parameters.
+kmod_cflags = ['-I' + meson.build_root(),
+		'-I' + join_paths(meson.source_root(), 'config'),
+		'-include rte_config.h']
+
+# to avoid warnings due to race conditions with creating the dev_if.h, etc.
+# files, serialize the kernel module builds. Each module will depend on
+# previous ones
+built_kmods = []
+foreach k:kmods
+	subdir(k)
+	built_kmods += custom_target(k,
+		input: [files('BSDmakefile.meson'), sources],
+		output: k + '.ko',
+		command: ['make', '-f', '@INPUT0@',
+			'KMOD_OBJDIR=@OUTDIR@',
+			'KMOD_SRC=@INPUT1@',
+			'KMOD=' + k,
+			'KMOD_CFLAGS=' + ' '.join(kmod_cflags),
+			'CC=clang'],
+		depends: built_kmods, # make each module depend on prev
+		build_by_default: get_option('enable_kmods'),
+		install: get_option('enable_kmods'),
+		install_dir: '/boot/modules/')
+endforeach
diff --git a/src/spdk/dpdk/kernel/freebsd/nic_uio/BSDmakefile b/src/spdk/dpdk/kernel/freebsd/nic_uio/BSDmakefile
new file mode 100644
index 000000000..b6f92d55e
--- /dev/null
+++ b/src/spdk/dpdk/kernel/freebsd/nic_uio/BSDmakefile
@@ -0,0 +1,8 @@
+# SPDX-License-Identifier: BSD-3-Clause
+# Copyright(c) 2010-2014 Intel Corporation
+#
+
+KMOD=	nic_uio
+SRCS=	nic_uio.c device_if.h bus_if.h pci_if.h
+
+.include <bsd.kmod.mk>
diff --git a/src/spdk/dpdk/kernel/freebsd/nic_uio/Makefile b/src/spdk/dpdk/kernel/freebsd/nic_uio/Makefile
new file mode 100644
index 000000000..376ef3a3a
--- /dev/null
+++ b/src/spdk/dpdk/kernel/freebsd/nic_uio/Makefile
@@ -0,0 +1,24 @@
+# SPDX-License-Identifier: BSD-3-Clause
+# Copyright(c) 2010-2014 Intel Corporation
+
+include $(RTE_SDK)/mk/rte.vars.mk
+
+#
+# module name and path
+#
+MODULE = nic_uio
+
+#
+# CFLAGS
+#
+MODULE_CFLAGS += -I$(SRCDIR)
+MODULE_CFLAGS += -I$(RTE_OUTPUT)/include
+MODULE_CFLAGS += -Winline -Wall -Werror
+MODULE_CFLAGS += -include $(RTE_OUTPUT)/include/rte_config.h
+
+#
+# all source are stored in SRCS-y
+#
+SRCS-y := nic_uio.c
+
+include $(RTE_SDK)/mk/rte.bsdmodule.mk
diff --git a/src/spdk/dpdk/kernel/freebsd/nic_uio/meson.build b/src/spdk/dpdk/kernel/freebsd/nic_uio/meson.build
new file mode 100644
index 000000000..4bdaf9694
--- /dev/null
+++ b/src/spdk/dpdk/kernel/freebsd/nic_uio/meson.build
@@ -0,0 +1,4 @@
+# SPDX-License-Identifier: BSD-3-Clause
+# Copyright(c) 2017 Intel Corporation
+
+sources = files('nic_uio.c')
diff --git a/src/spdk/dpdk/kernel/freebsd/nic_uio/nic_uio.c b/src/spdk/dpdk/kernel/freebsd/nic_uio/nic_uio.c
new file mode 100644
index 000000000..7a81694c9
--- /dev/null
+++ b/src/spdk/dpdk/kernel/freebsd/nic_uio/nic_uio.c
@@ -0,0 +1,351 @@
+/* SPDX-License-Identifier: BSD-3-Clause
+ * Copyright(c) 2010-2014 Intel Corporation
+ */
+#include <sys/cdefs.h>
+__FBSDID("$FreeBSD$");
+
+#include <sys/param.h> /* defines used in kernel.h */
+#include <sys/module.h>
+#include <sys/kernel.h> /* types used in module initialization */
+#include <sys/conf.h> /* cdevsw struct */
+#include <sys/bus.h> /* structs, prototypes for pci bus stuff and DEVMETHOD */
+#include <sys/rman.h>
+#include <sys/systm.h>
+#include <sys/lock.h>
+#include <sys/rwlock.h>
+#include <sys/proc.h>
+
+#include <machine/bus.h>
+#include <dev/pci/pcivar.h> /* For pci_get macros! */
+#include <dev/pci/pcireg.h> /* The softc holds our per-instance data. */
+#include <vm/vm.h>
+#include <vm/uma.h>
+#include <vm/vm_object.h>
+#include <vm/vm_page.h>
+#include <vm/vm_pager.h>
+
+
+#define MAX_BARS (PCIR_MAX_BAR_0 + 1)
+
+#define MAX_DETACHED_DEVICES	128
+static device_t detached_devices[MAX_DETACHED_DEVICES] = {};
+static int num_detached = 0;
+
+struct nic_uio_softc {
+	device_t        dev_t;
+	struct cdev     *my_cdev;
+	int              bar_id[MAX_BARS];
+	struct resource *bar_res[MAX_BARS];
+	u_long           bar_start[MAX_BARS];
+	u_long           bar_size[MAX_BARS];
+};
+
+/* Function prototypes */
+static d_open_t         nic_uio_open;
+static d_close_t        nic_uio_close;
+static d_mmap_t         nic_uio_mmap;
+static d_mmap_single_t  nic_uio_mmap_single;
+static int              nic_uio_probe(device_t dev);
+static int              nic_uio_attach(device_t dev);
+static int              nic_uio_detach(device_t dev);
+static int              nic_uio_shutdown(void);
+static int              nic_uio_modevent(module_t mod, int type, void *arg);
+
+static struct cdevsw uio_cdevsw = {
+		.d_name        = "nic_uio",
+		.d_version     = D_VERSION,
+		.d_open        = nic_uio_open,
+		.d_close       = nic_uio_close,
+		.d_mmap        = nic_uio_mmap,
+		.d_mmap_single = nic_uio_mmap_single,
+};
+
+static device_method_t nic_uio_methods[] = {
+	DEVMETHOD(device_probe,    nic_uio_probe),
+	DEVMETHOD(device_attach,   nic_uio_attach),
+	DEVMETHOD(device_detach,   nic_uio_detach),
+	DEVMETHOD_END
+};
+
+struct device {
+    int vend;
+    int dev;
+};
+
+struct pci_bdf {
+	uint32_t bus;
+	uint32_t devid;
+	uint32_t function;
+};
+
+static devclass_t nic_uio_devclass;
+
+DEFINE_CLASS_0(nic_uio, nic_uio_driver, nic_uio_methods, sizeof(struct nic_uio_softc));
+DRIVER_MODULE(nic_uio, pci, nic_uio_driver, nic_uio_devclass, nic_uio_modevent, 0);
+
+static int
+nic_uio_mmap(struct cdev *cdev, vm_ooffset_t offset, vm_paddr_t *paddr,
+		int prot, vm_memattr_t *memattr)
+{
+	*paddr = offset;
+	return 0;
+}
+
+static int
+nic_uio_mmap_single(struct cdev *cdev, vm_ooffset_t *offset, vm_size_t size,
+		struct vm_object **obj, int nprot)
+{
+	/*
+	 * The BAR index is encoded in the offset.  Divide the offset by
+	 *  PAGE_SIZE to get the index of the bar requested by the user
+	 *  app.
+	 */
+	unsigned bar = *offset/PAGE_SIZE;
+	struct nic_uio_softc *sc = cdev->si_drv1;
+
+	if (bar >= MAX_BARS)
+		return EINVAL;
+
+	if (sc->bar_res[bar] == NULL) {
+		sc->bar_id[bar] = PCIR_BAR(bar);
+
+		if (PCI_BAR_IO(pci_read_config(sc->dev_t, sc->bar_id[bar], 4)))
+			sc->bar_res[bar] = bus_alloc_resource_any(sc->dev_t, SYS_RES_IOPORT,
+					&sc->bar_id[bar], RF_ACTIVE);
+		else
+			sc->bar_res[bar] = bus_alloc_resource_any(sc->dev_t, SYS_RES_MEMORY,
+					&sc->bar_id[bar], RF_ACTIVE);
+	}
+	if (sc->bar_res[bar] == NULL)
+		return ENXIO;
+
+	sc->bar_start[bar] = rman_get_start(sc->bar_res[bar]);
+	sc->bar_size[bar] = rman_get_size(sc->bar_res[bar]);
+
+	device_printf(sc->dev_t, "Bar %u @ %lx, size %lx\n", bar,
+			sc->bar_start[bar], sc->bar_size[bar]);
+
+	*offset = sc->bar_start[bar];
+	*obj = vm_pager_allocate(OBJT_DEVICE, cdev, size, nprot, *offset,
+				curthread->td_ucred);
+	return 0;
+}
+
+
+int
+nic_uio_open(struct cdev *dev, int oflags, int devtype, struct thread *td)
+{
+	return 0;
+}
+
+int
+nic_uio_close(struct cdev *dev, int fflag, int devtype, struct thread *td)
+{
+	return 0;
+}
+
+static int
+nic_uio_probe (device_t dev)
+{
+	int i;
+	unsigned int bus = pci_get_bus(dev);
+	unsigned int device = pci_get_slot(dev);
+	unsigned int function = pci_get_function(dev);
+
+	char bdf_str[256];
+	char *token, *remaining;
+
+	/* First check if we found this on load */
+	for (i = 0; i < num_detached; i++)
+		if (bus == pci_get_bus(detached_devices[i]) &&
+		    device == pci_get_slot(detached_devices[i]) &&
+		    function == pci_get_function(detached_devices[i])) {
+			device_set_desc(dev, "DPDK PCI Device");
+			return BUS_PROBE_SPECIFIC;
+		}
+
+	/* otherwise check if it's a new device and if it matches the BDF */
+	memset(bdf_str, 0, sizeof(bdf_str));
+	TUNABLE_STR_FETCH("hw.nic_uio.bdfs", bdf_str, sizeof(bdf_str));
+	remaining = bdf_str;
+	while (1) {
+		if (remaining == NULL || remaining[0] == '\0')
+			break;
+		token = strsep(&remaining, ",:");
+		if (token == NULL)
+			break;
+		bus = strtol(token, NULL, 10);
+		token = strsep(&remaining, ",:");
+		if (token == NULL)
+			break;
+		device = strtol(token, NULL, 10);
+		token = strsep(&remaining, ",:");
+		if (token == NULL)
+			break;
+		function = strtol(token, NULL, 10);
+
+		if (bus == pci_get_bus(dev) &&
+				device == pci_get_slot(dev) &&
+				function == pci_get_function(dev)) {
+
+			if (num_detached < MAX_DETACHED_DEVICES) {
+				printf("%s: probed dev=%p\n",
+					       __func__, dev);
+				detached_devices[num_detached++] = dev;
+				device_set_desc(dev, "DPDK PCI Device");
+				return BUS_PROBE_SPECIFIC;
+			} else {
+				printf("%s: reached MAX_DETACHED_DEVICES=%d. dev=%p won't be reattached\n",
+						__func__, MAX_DETACHED_DEVICES,
+						dev);
+				break;
+			}
+		}
+	}
+
+	return ENXIO;
+}
+
+static int
+nic_uio_attach(device_t dev)
+{
+	int i;
+	struct nic_uio_softc *sc;
+
+	sc = device_get_softc(dev);
+	sc->dev_t = dev;
+	sc->my_cdev = make_dev(&uio_cdevsw, device_get_unit(dev),
+			UID_ROOT, GID_WHEEL, 0600, "uio@pci:%u:%u:%u",
+			pci_get_bus(dev), pci_get_slot(dev), pci_get_function(dev));
+	if (sc->my_cdev == NULL)
+		return ENXIO;
+	sc->my_cdev->si_drv1 = sc;
+
+	for (i = 0; i < MAX_BARS; i++)
+		sc->bar_res[i] = NULL;
+
+	pci_enable_busmaster(dev);
+
+	return 0;
+}
+
+static int
+nic_uio_detach(device_t dev)
+{
+	int i;
+	struct nic_uio_softc *sc;
+	sc = device_get_softc(dev);
+
+	for (i = 0; i < MAX_BARS; i++)
+		if (sc->bar_res[i] != NULL) {
+
+			if (PCI_BAR_IO(pci_read_config(dev, sc->bar_id[i], 4)))
+				bus_release_resource(dev, SYS_RES_IOPORT, sc->bar_id[i],
+						sc->bar_res[i]);
+			else
+				bus_release_resource(dev, SYS_RES_MEMORY, sc->bar_id[i],
+						sc->bar_res[i]);
+		}
+
+	if (sc->my_cdev != NULL)
+		destroy_dev(sc->my_cdev);
+	return 0;
+}
+
+static void
+nic_uio_load(void)
+{
+	uint32_t bus, device, function;
+	device_t dev;
+	char bdf_str[256];
+	char *token, *remaining;
+
+	memset(bdf_str, 0, sizeof(bdf_str));
+	TUNABLE_STR_FETCH("hw.nic_uio.bdfs", bdf_str, sizeof(bdf_str));
+	remaining = bdf_str;
+	printf("nic_uio: hw.nic_uio.bdfs = '%s'\n", bdf_str);
+	/*
+	 * Users should specify PCI BDFs in the format "b:d:f,b:d:f,b:d:f".
+	 *  But the code below does not try differentiate between : and ,
+	 *  and just blindly uses 3 tokens at a time to construct a
+	 *  bus/device/function tuple.
+	 *
+	 * There is no checking on strtol() return values, but this should
+	 *  be OK.  Worst case is it cannot convert and returns 0.  This
+	 *  could give us a different BDF than intended, but as long as the
+	 *  PCI device/vendor ID does not match it will not matter.
+	 */
+	while (1) {
+		if (remaining == NULL || remaining[0] == '\0')
+			break;
+		token = strsep(&remaining, ",:");
+		if (token == NULL)
+			break;
+		bus = strtol(token, NULL, 10);
+		token = strsep(&remaining, ",:");
+		if (token == NULL)
+			break;
+		device = strtol(token, NULL, 10);
+		token = strsep(&remaining, ",:");
+		if (token == NULL)
+			break;
+		function = strtol(token, NULL, 10);
+
+		dev = pci_find_bsf(bus, device, function);
+		if (dev == NULL)
+			continue;
+
+		if (num_detached < MAX_DETACHED_DEVICES) {
+			printf("nic_uio_load: detaching and storing dev=%p\n",
+			       dev);
+			detached_devices[num_detached++] = dev;
+		} else {
+			printf("nic_uio_load: reached MAX_DETACHED_DEVICES=%d. dev=%p won't be reattached\n",
+			       MAX_DETACHED_DEVICES, dev);
+		}
+		device_detach(dev);
+	}
+}
+
+static void
+nic_uio_unload(void)
+{
+	int i;
+	printf("nic_uio_unload: entered...\n");
+
+	for (i = 0; i < num_detached; i++) {
+		printf("nic_uio_unload: calling to device_probe_and_attach for dev=%p...\n",
+			detached_devices[i]);
+		device_probe_and_attach(detached_devices[i]);
+		printf("nic_uio_unload: done.\n");
+	}
+
+	printf("nic_uio_unload: leaving...\n");
+}
+
+static int
+nic_uio_shutdown(void)
+{
+	return 0;
+}
+
+static int
+nic_uio_modevent(module_t mod, int type, void *arg)
+{
+
+	switch (type) {
+	case MOD_LOAD:
+		nic_uio_load();
+		break;
+	case MOD_UNLOAD:
+		nic_uio_unload();
+		break;
+	case MOD_SHUTDOWN:
+		nic_uio_shutdown();
+		break;
+	default:
+		break;
+	}
+
+	return 0;
+}