1 files changed, 278 insertions, 0 deletions
diff --git a/test/run-marklock-deadlock.c b/test/run-marklock-deadlock.c
new file mode 100644
index 0000000..37e959f
--- /dev/null
+++ b/test/run-marklock-deadlock.c
@@ -0,0 +1,278 @@
+#include "../common/tdb_private.h"
+#include "../common/io.c"
+#include "../common/tdb.c"
+#include "../common/lock.c"
+#include "../common/freelist.c"
+#include "../common/traverse.c"
+#include "../common/transaction.c"
+#include "../common/error.c"
+#include "../common/open.c"
+#include "../common/check.c"
+#include "../common/hash.c"
+#include "../common/mutex.c"
+#include "tap-interface.h"
+#include <stdlib.h>
+#include <sys/types.h>
+#include <sys/wait.h>
+#include <stdarg.h>
+#include "logging.h"
+
+static TDB_DATA key, data;
+
+static void do_chainlock(const char *name, int tdb_flags, int up, int down)
+{
+	struct tdb_context *tdb;
+	int ret;
+	ssize_t nread, nwritten;
+	char c = 0;
+
+	tdb = tdb_open_ex(name, 3, tdb_flags,
+			  O_RDWR|O_CREAT, 0755, &taplogctx, NULL);
+	ok(tdb, "tdb_open_ex should succeed");
+
+	ret = tdb_chainlock(tdb, key);
+	ok(ret == 0, "tdb_chainlock should succeed");
+
+	nwritten = write(up, &c, sizeof(c));
+	ok(nwritten == sizeof(c), "write should succeed");
+
+	nread = read(down, &c, sizeof(c));
+	ok(nread == sizeof(c), "read should succeed");
+
+	exit(0);
+}
+
+static void do_allrecord_lock(const char *name, int tdb_flags, int up, int down)
+{
+	struct tdb_context *tdb;
+	int ret;
+	ssize_t nread, nwritten;
+	char c = 0;
+
+	tdb = tdb_open_ex(name, 3, tdb_flags,
+			  O_RDWR|O_CREAT, 0755, &taplogctx, NULL);
+	ok(tdb, "tdb_open_ex should succeed");
+
+	ret = tdb_allrecord_lock(tdb, F_WRLCK, TDB_LOCK_WAIT, false);
+	ok(ret == 0, "tdb_allrecord_lock should succeed");
+
+	nwritten = write(up, &c, sizeof(c));
+	ok(nwritten == sizeof(c), "write should succeed");
+
+	nread = read(down, &c, sizeof(c));
+	ok(nread == sizeof(c), "read should succeed");
+
+	exit(0);
+}
+
+/* The code should barf on TDBs created with rwlocks. */
+static int do_tests(const char *name, int tdb_flags)
+{
+	struct tdb_context *tdb;
+	int ret;
+	pid_t chainlock_child, allrecord_child;
+	int chainlock_down[2];
+	int chainlock_up[2];
+	int allrecord_down[2];
+	int allrecord_up[2];
+	char c;
+	ssize_t nread, nwritten;
+
+	key.dsize = strlen("hi");
+	key.dptr = discard_const_p(uint8_t, "hi");
+	data.dsize = strlen("world");
+	data.dptr = discard_const_p(uint8_t, "world");
+
+	ret = pipe(chainlock_down);
+	ok(ret == 0, "pipe should succeed");
+
+	ret = pipe(chainlock_up);
+	ok(ret == 0, "pipe should succeed");
+
+	ret = pipe(allrecord_down);
+	ok(ret == 0, "pipe should succeed");
+
+	ret = pipe(allrecord_up);
+	ok(ret == 0, "pipe should succeed");
+
+	chainlock_child = fork();
+	ok(chainlock_child != -1, "fork should succeed");
+
+	if (chainlock_child == 0) {
+		close(chainlock_up[0]);
+		close(chainlock_down[1]);
+		close(allrecord_up[0]);
+		close(allrecord_up[1]);
+		close(allrecord_down[0]);
+		close(allrecord_down[1]);
+		do_chainlock(name, tdb_flags,
+			     chainlock_up[1], chainlock_down[0]);
+		exit(0);
+	}
+	close(chainlock_up[1]);
+	close(chainlock_down[0]);
+
+	nread = read(chainlock_up[0], &c, sizeof(c));
+	ok(nread == sizeof(c), "read should succeed");
+
+	/*
+	 * Now we have a process holding a chainlock. Start another process
+	 * trying the allrecord lock. This will block.
+	 */
+
+	allrecord_child = fork();
+	ok(allrecord_child != -1, "fork should succeed");
+
+	if (allrecord_child == 0) {
+		close(chainlock_up[0]);
+		close(chainlock_up[1]);
+		close(chainlock_down[0]);
+		close(chainlock_down[1]);
+		close(allrecord_up[0]);
+		close(allrecord_down[1]);
+		do_allrecord_lock(name, tdb_flags,
+				  allrecord_up[1], allrecord_down[0]);
+		exit(0);
+	}
+	close(allrecord_up[1]);
+	close(allrecord_down[0]);
+
+	poll(NULL, 0, 500);
+
+	tdb = tdb_open_ex(name, 3, tdb_flags,
+			  O_RDWR|O_CREAT, 0755, &taplogctx, NULL);
+	ok(tdb, "tdb_open_ex should succeed");
+
+	/*
+	 * Someone already holds a chainlock, but we're able to get the
+	 * freelist lock.
+	 *
+	 * The freelist lock/mutex is independent from the allrecord lock/mutex.
+	 */
+
+	ret = tdb_chainlock_nonblock(tdb, key);
+	ok(ret == -1, "tdb_chainlock_nonblock should not succeed");
+
+	ret = tdb_lock_nonblock(tdb, -1, F_WRLCK);
+	ok(ret == 0, "tdb_lock_nonblock should succeed");
+
+	ret = tdb_unlock(tdb, -1, F_WRLCK);
+	ok(ret == 0, "tdb_unlock should succeed");
+
+	/*
+	 * We have someone else having done the lock for us. Just mark it.
+	 */
+
+	ret = tdb_chainlock_mark(tdb, key);
+	ok(ret == 0, "tdb_chainlock_mark should succeed");
+
+	/*
+	 * The tdb_store below will block the freelist. In one version of the
+	 * mutex patches, the freelist was already blocked here by the
+	 * allrecord child, which was waiting for the chainlock child to give
+	 * up its chainlock. Make sure that we don't run into this
+	 * deadlock. To exercise the deadlock, just comment out the "ok"
+	 * line.
+	 *
+	 * The freelist lock/mutex is independent from the allrecord lock/mutex.
+	 */
+
+	ret = tdb_lock_nonblock(tdb, -1, F_WRLCK);
+	ok(ret == 0, "tdb_lock_nonblock should succeed");
+
+	ret = tdb_unlock(tdb, -1, F_WRLCK);
+	ok(ret == 0, "tdb_unlock should succeed");
+
+	ret = tdb_store(tdb, key, data, TDB_INSERT);
+	ok(ret == 0, "tdb_store should succeed");
+
+	ret = tdb_chainlock_unmark(tdb, key);
+	ok(ret == 0, "tdb_chainlock_unmark should succeed");
+
+	nwritten = write(chainlock_down[1], &c, sizeof(c));
+	ok(nwritten == sizeof(c), "write should succeed");
+
+	nread = read(chainlock_up[0], &c, sizeof(c));
+	ok(nread == 0, "read should succeed");
+
+	nread = read(allrecord_up[0], &c, sizeof(c));
+	ok(nread == sizeof(c), "read should succeed");
+
+	/*
+	 * Someone already holds the allrecord lock, but we're able to get the
+	 * freelist lock.
+	 *
+	 * The freelist lock/mutex is independent from the allrecord lock/mutex.
+	 */
+
+	ret = tdb_chainlock_nonblock(tdb, key);
+	ok(ret == -1, "tdb_chainlock_nonblock should not succeed");
+
+	ret = tdb_lockall_nonblock(tdb);
+	ok(ret == -1, "tdb_lockall_nonblock should not succeed");
+
+	ret = tdb_lock_nonblock(tdb, -1, F_WRLCK);
+	ok(ret == 0, "tdb_lock_nonblock should succeed");
+
+	ret = tdb_unlock(tdb, -1, F_WRLCK);
+	ok(ret == 0, "tdb_unlock should succeed");
+
+	/*
+	 * We have someone else having done the lock for us. Just mark it.
+	 */
+
+	ret = tdb_lockall_mark(tdb);
+	ok(ret == 0, "tdb_lockall_mark should succeed");
+
+	ret = tdb_lock_nonblock(tdb, -1, F_WRLCK);
+	ok(ret == 0, "tdb_lock_nonblock should succeed");
+
+	ret = tdb_unlock(tdb, -1, F_WRLCK);
+	ok(ret == 0, "tdb_unlock should succeed");
+
+	ret = tdb_store(tdb, key, data, TDB_REPLACE);
+	ok(ret == 0, "tdb_store should succeed");
+
+	ret = tdb_lockall_unmark(tdb);
+	ok(ret == 0, "tdb_lockall_unmark should succeed");
+
+	nwritten = write(allrecord_down[1], &c, sizeof(c));
+	ok(nwritten == sizeof(c), "write should succeed");
+
+	nread = read(allrecord_up[0], &c, sizeof(c));
+	ok(nread == 0, "read should succeed");
+
+	close(chainlock_up[0]);
+	close(chainlock_down[1]);
+	close(allrecord_up[0]);
+	close(allrecord_down[1]);
+	diag("%s tests done", name);
+	return exit_status();
+}
+
+int main(int argc, char *argv[])
+{
+	int ret;
+	bool mutex_support;
+
+	mutex_support = tdb_runtime_check_for_robust_mutexes();
+
+	ret = do_tests("marklock-deadlock-fcntl.tdb",
+		       TDB_CLEAR_IF_FIRST |
+		       TDB_INCOMPATIBLE_HASH);
+	ok(ret == 0, "marklock-deadlock-fcntl.tdb tests should succeed");
+
+	if (!mutex_support) {
+		skip(1, "No robust mutex support, "
+			"skipping marklock-deadlock-mutex.tdb tests");
+		return exit_status();
+	}
+
+	ret = do_tests("marklock-deadlock-mutex.tdb",
+		       TDB_CLEAR_IF_FIRST |
+		       TDB_MUTEX_LOCKING |
+		       TDB_INCOMPATIBLE_HASH);
+	ok(ret == 0, "marklock-deadlock-mutex.tdb tests should succeed");
+
+	return exit_status();
+}