[illumos-gate merge]

commit 3038a2b421b40dc5ac11cd88423696618584f85a
    4897 Space accounting mismatch in L2ARC/zpool
commit df15e419cb7359ba56ddddab9045e438d89e7cbc
    4891 want zdb option to dump all metadata
commit 05d57413471eaaa425913a06edc2ab33ad9b05bc
    1667 pkcs11 may deadlock when multi-threaded consumers fork

10 files changed, 132 insertions, 27 deletions

diff --git a/usr/src/cmd/zdb/zdb.c b/usr/src/cmd/zdb/zdb.c
index 281426bb5d..5827623747 100644
--- a/usr/src/cmd/zdb/zdb.c
+++ b/usr/src/cmd/zdb/zdb.c
@@ -112,7 +112,7 @@ usage(void)
 {
 	(void) fprintf(stderr,
 	    "Usage: %s [-CumdibcsDvhLXFPA] [-t txg] [-e [-p path...]] "
-	    "[-U config] [-M inflight I/Os] poolname [object...]\n"
+	    "[-U config] [-M inflight I/Os] [-x dumpdir] poolname [object...]\n"
 	    "       %s [-divPA] [-e -p path...] [-U config] dataset "
 	    "[object...]\n"
 	    "       %s -m [-LXFPA] [-t txg] [-e [-p path...]] [-U config] "
@@ -150,7 +150,7 @@ usage(void)
 	(void) fprintf(stderr, "        -R read and display block from a "
 	    "device\n\n");
 	(void) fprintf(stderr, "    Below options are intended for use "
-	    "with other options (except -l):\n");
+	    "with other options:\n");
 	(void) fprintf(stderr, "        -A ignore assertions (-A), enable "
 	    "panic recovery (-AA) or both (-AAA)\n");
 	(void) fprintf(stderr, "        -F attempt automatic rewind within "
@@ -163,11 +163,14 @@ usage(void)
 	    "has altroot/not in a cachefile\n");
 	(void) fprintf(stderr, "        -p <path> -- use one or more with "
 	    "-e to specify path to vdev dir\n");
+	(void) fprintf(stderr, "        -x <dumpdir> -- "
+	    "dump all read blocks into specified directory\n");
 	(void) fprintf(stderr, "        -P print numbers in parseable form\n");
 	(void) fprintf(stderr, "        -t <txg> -- highest txg to use when "
 	    "searching for uberblocks\n");
 	(void) fprintf(stderr, "        -M <number of inflight I/Os> -- "
-	    "specify the maximum number of checksumming I/Os [default is 200]");
+	    "specify the maximum number of "
+	    "checksumming I/Os [default is 200]\n");
 	(void) fprintf(stderr, "Specify an option more than once (e.g. -bb) "
 	    "to make only that option verbose\n");
 	(void) fprintf(stderr, "Default is to dump everything non-verbosely\n");
@@ -3325,7 +3328,8 @@ main(int argc, char **argv)
 
 	dprintf_setup(&argc, argv);
 
-	while ((c = getopt(argc, argv, "bcdhilmM:suCDRSAFLXevp:t:U:P")) != -1) {
+	while ((c = getopt(argc, argv,
+	    "bcdhilmM:suCDRSAFLXx:evp:t:U:P")) != -1) {
 		switch (c) {
 		case 'b':
 		case 'c':
@@ -3378,6 +3382,9 @@ main(int argc, char **argv)
 			}
 			searchdirs[nsearch++] = optarg;
 			break;
+		case 'x':
+			vn_dumpdir = optarg;
+			break;
 		case 't':
 			max_txg = strtoull(optarg, NULL, 0);
 			if (max_txg < TXG_INITIAL) {
diff --git a/usr/src/lib/libcryptoutil/common/cryptoutil.h b/usr/src/lib/libcryptoutil/common/cryptoutil.h
index b5aad833f7..63a3df665f 100644
--- a/usr/src/lib/libcryptoutil/common/cryptoutil.h
+++ b/usr/src/lib/libcryptoutil/common/cryptoutil.h
@@ -22,6 +22,7 @@
  */
 /*
  * Copyright 2010 Nexenta Systems, Inc.  All rights reserved.
+ * Copyright 2014, OmniTI Computer Consulting, Inc. All rights reserved.
  */
 
 #ifndef _CRYPTOUTIL_H
@@ -216,9 +217,6 @@ extern int pkcs11_seed_urandom(void *sbuf, size_t slen);
 extern int pkcs11_get_random(void *dbuf, size_t dlen);
 extern int pkcs11_get_urandom(void *dbuf, size_t dlen);
 extern int pkcs11_get_nzero_urandom(void *dbuf, size_t dlen);
-extern void pkcs11_close_random(void);
-extern void pkcs11_close_urandom(void);
-extern void pkcs11_close_urandom_seed(void);
 extern int pkcs11_read_data(char *filename, void **dbuf, size_t *dlen);
 
 extern int open_nointr(const char *path, int oflag, ...);
diff --git a/usr/src/lib/libcryptoutil/common/mapfile-vers b/usr/src/lib/libcryptoutil/common/mapfile-vers
index 5d3c214b55..c7f2576f37 100644
--- a/usr/src/lib/libcryptoutil/common/mapfile-vers
+++ b/usr/src/lib/libcryptoutil/common/mapfile-vers
@@ -19,6 +19,7 @@
 # CDDL HEADER END
 #
 # Copyright (c) 2006, 2010, Oracle and/or its affiliates. All rights reserved.
+# Copyright 2014, OmniTI Computer Consulting Inc. All rights reserved.
 #
 
 #
@@ -54,9 +55,6 @@ SYMBOL_VERSION SUNWprivate {
 	get_pkcs11conf_info;
 	hexstr_to_bytes;
 	open_nointr;
-	pkcs11_close_random;
-	pkcs11_close_urandom;
-	pkcs11_close_urandom_seed;
 	pkcs11_default_token;
 	pkcs11_free_uri;
 	pkcs11_get_nzero_urandom;
diff --git a/usr/src/lib/libcryptoutil/common/random.c b/usr/src/lib/libcryptoutil/common/random.c
index 771112850a..ab07168409 100644
--- a/usr/src/lib/libcryptoutil/common/random.c
+++ b/usr/src/lib/libcryptoutil/common/random.c
@@ -21,6 +21,7 @@
 
 /*
  * Copyright (c) 2003, 2010, Oracle and/or its affiliates. All rights reserved.
+ * Copyright 2014, OmniTI Computer Consulting, Inc. All rights reserved.
  */
 
 #include <stdio.h>
@@ -33,6 +34,7 @@
 #include <cryptoutil.h>
 #include <pthread.h>
 
+#pragma init(pkcs11_random_init)
 
 static pthread_mutex_t	random_mutex = PTHREAD_MUTEX_INITIALIZER;
 static pthread_mutex_t	urandom_mutex = PTHREAD_MUTEX_INITIALIZER;
@@ -181,13 +183,13 @@ pkcs11_close_common(int *fd, pthread_mutex_t *mtx)
 	(void) pthread_mutex_unlock(mtx);
 }
 
-void
+static void
 pkcs11_close_random(void)
 {
 	pkcs11_close_common(&random_fd, &random_mutex);
 }
 
-void
+static void
 pkcs11_close_urandom(void)
 {
 	pkcs11_close_common(&urandom_fd, &urandom_mutex);
@@ -199,7 +201,7 @@ pkcs11_close_random_seed(void)
 	pkcs11_close_common(&random_seed_fd, &random_seed_mutex);
 }
 
-void
+static void
 pkcs11_close_urandom_seed(void)
 {
 	pkcs11_close_common(&urandom_seed_fd, &urandom_seed_mutex);
@@ -377,3 +379,45 @@ pkcs11_get_nzero_urandom(void *dbuf, size_t dlen)
 	}
 	return (0);
 }
+
+static void
+pkcs11_random_prepare(void)
+{
+	/*
+	 * NOTE - None of these are acquired more than one at a time.
+	 * I can therefore acquire all four without fear of deadlock.
+	 */
+	(void) pthread_mutex_lock(&random_mutex);
+	(void) pthread_mutex_lock(&urandom_mutex);
+	(void) pthread_mutex_lock(&random_seed_mutex);
+	(void) pthread_mutex_lock(&urandom_seed_mutex);
+}
+
+static void
+pkcs11_random_parent_post(void)
+{
+	/* Drop the mutexes and get back to work! */
+	(void) pthread_mutex_unlock(&urandom_seed_mutex);
+	(void) pthread_mutex_unlock(&random_seed_mutex);
+	(void) pthread_mutex_unlock(&urandom_mutex);
+	(void) pthread_mutex_unlock(&random_mutex);
+}
+
+static void
+pkcs11_random_child_post(void)
+{
+	pkcs11_random_parent_post();
+
+	/* Also, close the FDs, just in case. */
+	pkcs11_close_random();
+	pkcs11_close_urandom();
+	pkcs11_close_random_seed();
+	pkcs11_close_urandom_seed();
+}
+
+static void
+pkcs11_random_init(void)
+{
+	(void) pthread_atfork(pkcs11_random_prepare, pkcs11_random_parent_post,
+	    pkcs11_random_child_post);
+}
diff --git a/usr/src/lib/libzpool/common/kernel.c b/usr/src/lib/libzpool/common/kernel.c
index d030700415..418ef435b0 100644
--- a/usr/src/lib/libzpool/common/kernel.c
+++ b/usr/src/lib/libzpool/common/kernel.c
@@ -31,6 +31,7 @@
 #include <stdlib.h>
 #include <string.h>
 #include <zlib.h>
+#include <libgen.h>
 #include <sys/spa.h>
 #include <sys/stat.h>
 #include <sys/processor.h>
@@ -51,6 +52,9 @@ char hw_serial[HW_HOSTID_LEN];
 kmutex_t cpu_lock;
 vmem_t *zio_arena = NULL;
 
+/* If set, all blocks read will be copied to the specified directory. */
+char *vn_dumpdir = NULL;
+
 struct utsname utsname = {
 	"userland", "libzpool", "1", "1", "na"
 };
@@ -394,6 +398,7 @@ int
 vn_open(char *path, int x1, int flags, int mode, vnode_t **vpp, int x2, int x3)
 {
 	int fd;
+	int dump_fd;
 	vnode_t *vp;
 	int old_umask;
 	char realpath[MAXPATHLEN];
@@ -442,6 +447,17 @@ vn_open(char *path, int x1, int flags, int mode, vnode_t **vpp, int x2, int x3)
 	if (flags & FCREAT)
 		(void) umask(old_umask);
 
+	if (vn_dumpdir != NULL) {
+		char dumppath[MAXPATHLEN];
+		(void) snprintf(dumppath, sizeof (dumppath),
+		    "%s/%s", vn_dumpdir, basename(realpath));
+		dump_fd = open64(dumppath, O_CREAT | O_WRONLY, 0666);
+		if (dump_fd == -1)
+			return (errno);
+	} else {
+		dump_fd = -1;
+	}
+
 	if (fd == -1)
 		return (errno);
 
@@ -457,6 +473,7 @@ vn_open(char *path, int x1, int flags, int mode, vnode_t **vpp, int x2, int x3)
 	vp->v_fd = fd;
 	vp->v_size = st.st_size;
 	vp->v_path = spa_strdup(path);
+	vp->v_dump_fd = dump_fd;
 
 	return (0);
 }
@@ -489,6 +506,11 @@ vn_rdwr(int uio, vnode_t *vp, void *addr, ssize_t len, offset_t offset,
 
 	if (uio == UIO_READ) {
 		iolen = pread64(vp->v_fd, addr, len, offset);
+		if (vp->v_dump_fd != -1) {
+			int status =
+			    pwrite64(vp->v_dump_fd, addr, iolen, offset);
+			ASSERT(status != -1);
+		}
 	} else {
 		/*
 		 * To simulate partial disk writes, we split writes into two
@@ -515,6 +537,8 @@ void
 vn_close(vnode_t *vp)
 {
 	close(vp->v_fd);
+	if (vp->v_dump_fd != -1)
+		close(vp->v_dump_fd);
 	spa_strfree(vp->v_path);
 	umem_free(vp, sizeof (vnode_t));
 }
diff --git a/usr/src/lib/libzpool/common/sys/zfs_context.h b/usr/src/lib/libzpool/common/sys/zfs_context.h
index 7ba48fce69..8ee04e7009 100644
--- a/usr/src/lib/libzpool/common/sys/zfs_context.h
+++ b/usr/src/lib/libzpool/common/sys/zfs_context.h
@@ -390,8 +390,10 @@ typedef struct vnode {
 	uint64_t	v_size;
 	int		v_fd;
 	char		*v_path;
+	int		v_dump_fd;
 } vnode_t;
 
+extern char *vn_dumpdir;
 #define	AV_SCANSTAMP_SZ	32		/* length of anti-virus scanstamp */
 
 typedef struct xoptattr {
diff --git a/usr/src/lib/pkcs11/libpkcs11/common/metaGeneral.c b/usr/src/lib/pkcs11/libpkcs11/common/metaGeneral.c
index 5e5c339b03..32b00216da 100644
--- a/usr/src/lib/pkcs11/libpkcs11/common/metaGeneral.c
+++ b/usr/src/lib/pkcs11/libpkcs11/common/metaGeneral.c
@@ -21,8 +21,9 @@
 /*
  * Copyright 2009 Sun Microsystems, Inc.  All rights reserved.
  * Use is subject to license terms.
+ *
+ * Copyright 2014, OmniTI Computer Consulting, Inc. All rights reserved.
  */
-
 /*
  * General-Purpose Functions
  * (as defined in PKCS#11 spec section 11.4)
@@ -196,8 +197,12 @@ meta_Finalize(CK_VOID_PTR pReserved)
 
 	(void) pthread_mutex_lock(&initmutex);
 
-	pkcs11_close_urandom();
-	pkcs11_close_urandom_seed();
+	/*
+	 * There used to be calls to cleanup libcryptoutil here.  Given that
+	 * libcryptoutil can be linked and invoked independently of PKCS#11,
+	 * cleaning up libcryptoutil here makes no sense.  Decoupling these
+	 * two also prevent deadlocks and other artificial dependencies.
+	 */
 
 	meta_objectManager_finalize();
 
diff --git a/usr/src/lib/pkcs11/pkcs11_softtoken/common/softGeneral.c b/usr/src/lib/pkcs11/pkcs11_softtoken/common/softGeneral.c
index 396a3c5bf4..c44cbcb2a2 100644
--- a/usr/src/lib/pkcs11/pkcs11_softtoken/common/softGeneral.c
+++ b/usr/src/lib/pkcs11/pkcs11_softtoken/common/softGeneral.c
@@ -21,6 +21,8 @@
 /*
  * Copyright 2009 Sun Microsystems, Inc.  All rights reserved.
  * Use is subject to license terms.
+ *
+ * Copyright 2014, OmniTI Computer Consulting, Inc. All rights reserved.
  */
 
 #include <strings.h>
@@ -354,9 +356,12 @@ finalize_common(boolean_t force, CK_VOID_PTR pReserved) {
 	softtoken_initialized = B_FALSE;
 	softtoken_pid = 0;
 
-	pkcs11_close_urandom();
-	pkcs11_close_urandom_seed();
-	pkcs11_close_random();
+	/*
+	 * There used to be calls to cleanup libcryptoutil here.  Given that
+	 * libcryptoutil can be linked and invoked independently of PKCS#11,
+	 * cleaning up libcryptoutil here makes no sense.  Decoupling these
+	 * two also prevent deadlocks and other artificial dependencies.
+	 */
 
 	/* Destroy the session list lock here */
 	(void) pthread_mutex_destroy(&soft_sessionlist_mutex);
diff --git a/usr/src/man/man1m/zdb.1m b/usr/src/man/man1m/zdb.1m
index b4796e19c3..c7d7f38f5b 100644
--- a/usr/src/man/man1m/zdb.1m
+++ b/usr/src/man/man1m/zdb.1m
@@ -11,7 +11,7 @@
 .\"
 .\"
 .\" Copyright 2012, Richard Lowe.
-.\" Copyright (c) 2012 by Delphix. All rights reserved.
+.\" Copyright (c) 2012, 2014 by Delphix. All rights reserved.
 .\"
 .TH "ZDB" "1M" "March 6, 2014" "" ""
 
@@ -20,8 +20,8 @@
 
 .SH "SYNOPSIS"
 \fBzdb\fR [-CumdibcsDvhLXFPA] [-e [-p \fIpath\fR...]] [-t \fItxg\fR]
-    [-U \fIcache\fR] [-M \fIinflight I/Os\fR] [\fIpoolname\fR
-    [\fIobject\fR ...]]
+    [-U \fIcache\fR] [-M \fIinflight I/Os\fR] [-x \fIdumpdir\fR]
+    [\fIpoolname\fR [\fIobject\fR ...]]
 
 .P
 \fBzdb\fR [-divPA] [-e [-p \fIpath\fR...]] [-U \fIcache\fR]
@@ -355,6 +355,20 @@ Operate on an exported pool, not present in \fB/etc/zfs/zpool.cache\fR. The
 .sp
 .ne 2
 .na
+\fB-x\fR \fIdumpdir\fR
+.ad
+.sp .6
+.RS 4n
+All blocks accessed will be copied to files in the specified directory.
+The blocks will be placed in sparse files whose name is the same as
+that of the file or device read.  zdb can be then run on the generated files.
+Note that the \fB-bbc\fR flags are sufficient to access (and thus copy)
+all metadata on the pool.
+.RE
+
+.sp
+.ne 2
+.na
 \fB-F\fR
 .ad
 .sp .6
diff --git a/usr/src/uts/common/fs/zfs/arc.c b/usr/src/uts/common/fs/zfs/arc.c
index 73bc0780ce..e6a0480402 100644
--- a/usr/src/uts/common/fs/zfs/arc.c
+++ b/usr/src/uts/common/fs/zfs/arc.c
@@ -23,7 +23,7 @@
  * Copyright (c) 2012, Joyent, Inc. All rights reserved.
  * Copyright (c) 2013 by Delphix. All rights reserved.
  * Copyright (c) 2014 by Saso Kiselkov. All rights reserved.
- * Copyright 2013 Nexenta Systems, Inc.  All rights reserved.
+ * Copyright 2014 Nexenta Systems, Inc.  All rights reserved.
  */
 
 /*
@@ -1595,6 +1595,8 @@ arc_hdr_destroy(arc_buf_hdr_t *hdr)
 			list_remove(l2hdr->b_dev->l2ad_buflist, hdr);
 			ARCSTAT_INCR(arcstat_l2_size, -hdr->b_size);
 			ARCSTAT_INCR(arcstat_l2_asize, -l2hdr->b_asize);
+			vdev_space_update(l2hdr->b_dev->l2ad_vdev,
+			    -l2hdr->b_asize, 0, 0);
 			kmem_free(l2hdr, sizeof (l2arc_buf_hdr_t));
 			if (hdr->b_state == arc_l2c_only)
 				l2arc_hdr_stat_remove();
@@ -3468,6 +3470,8 @@ arc_release(arc_buf_t *buf, void *tag)
 
 	if (l2hdr) {
 		ARCSTAT_INCR(arcstat_l2_asize, -l2hdr->b_asize);
+		vdev_space_update(l2hdr->b_dev->l2ad_vdev,
+		    -l2hdr->b_asize, 0, 0);
 		kmem_free(l2hdr, sizeof (l2arc_buf_hdr_t));
 		ARCSTAT_INCR(arcstat_l2_size, -buf_size);
 		mutex_exit(&l2arc_buflist_mtx);
@@ -4261,6 +4265,7 @@ l2arc_write_done(zio_t *zio)
 	arc_buf_hdr_t *head, *ab, *ab_prev;
 	l2arc_buf_hdr_t *abl2;
 	kmutex_t *hash_lock;
+	int64_t bytes_dropped = 0;
 
 	cb = zio->io_private;
 	ASSERT(cb != NULL);
@@ -4308,6 +4313,7 @@ l2arc_write_done(zio_t *zio)
 			 */
 			list_remove(buflist, ab);
 			ARCSTAT_INCR(arcstat_l2_asize, -abl2->b_asize);
+			bytes_dropped += abl2->b_asize;
 			ab->b_l2hdr = NULL;
 			kmem_free(abl2, sizeof (l2arc_buf_hdr_t));
 			ARCSTAT_INCR(arcstat_l2_size, -ab->b_size);
@@ -4326,6 +4332,8 @@ l2arc_write_done(zio_t *zio)
 	kmem_cache_free(hdr_cache, head);
 	mutex_exit(&l2arc_buflist_mtx);
 
+	vdev_space_update(dev->l2ad_vdev, -bytes_dropped, 0, 0);
+
 	l2arc_do_free_on_write();
 
 	kmem_free(cb, sizeof (l2arc_write_callback_t));
@@ -4464,6 +4472,7 @@ l2arc_evict(l2arc_dev_t *dev, uint64_t distance, boolean_t all)
 	arc_buf_hdr_t *ab, *ab_prev;
 	kmutex_t *hash_lock;
 	uint64_t taddr;
+	int64_t bytes_evicted = 0;
 
 	buflist = dev->l2ad_buflist;
 
@@ -4562,6 +4571,7 @@ top:
 			if (ab->b_l2hdr != NULL) {
 				abl2 = ab->b_l2hdr;
 				ARCSTAT_INCR(arcstat_l2_asize, -abl2->b_asize);
+				bytes_evicted += abl2->b_asize;
 				ab->b_l2hdr = NULL;
 				kmem_free(abl2, sizeof (l2arc_buf_hdr_t));
 				ARCSTAT_INCR(arcstat_l2_size, -ab->b_size);
@@ -4578,7 +4588,7 @@ top:
 	}
 	mutex_exit(&l2arc_buflist_mtx);
 
-	vdev_space_update(dev->l2ad_vdev, -(taddr - dev->l2ad_evict), 0, 0);
+	vdev_space_update(dev->l2ad_vdev, -bytes_evicted, 0, 0);
 	dev->l2ad_evict = taddr;
 }
 
@@ -4821,15 +4831,13 @@ l2arc_write_buffers(spa_t *spa, l2arc_dev_t *dev, uint64_t target_sz,
 	ARCSTAT_INCR(arcstat_l2_write_bytes, write_asize);
 	ARCSTAT_INCR(arcstat_l2_size, write_sz);
 	ARCSTAT_INCR(arcstat_l2_asize, write_asize);
-	vdev_space_update(dev->l2ad_vdev, write_psize, 0, 0);
+	vdev_space_update(dev->l2ad_vdev, write_asize, 0, 0);
 
 	/*
 	 * Bump device hand to the device start if it is approaching the end.
 	 * l2arc_evict() will already have evicted ahead for this case.
 	 */
 	if (dev->l2ad_hand >= (dev->l2ad_end - target_sz)) {
-		vdev_space_update(dev->l2ad_vdev,
-		    dev->l2ad_end - dev->l2ad_hand, 0, 0);
 		dev->l2ad_hand = dev->l2ad_start;
 		dev->l2ad_evict = dev->l2ad_start;
 		dev->l2ad_first = B_FALSE;