4891 want zdb option to dump all metadata

Reviewed by: Sonu Pillai <sonu.pillai@delphix.com>
Reviewed by: George Wilson <george.wilson@delphix.com>
Reviewed by: Christopher Siden <christopher.siden@delphix.com>
Reviewed by: Dan McDonald <danmcd@omniti.com>
Reviewed by: Richard Lowe <richlowe@richlowe.net>
Approved by: Garrett D'Amore <garrett@damore.org>

illumos/illumos-gate@df15e419cb

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) {

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));
 }

lib/libzpool/common/sys/zfs_context.h

@@ -389,8 +389,10 @@ 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 {

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]
@@ -352,6 +352,20 @@ Operate on an exported pool, not present in \fB/etc/zfs/zpool.cache\fR. The
 \fB-p\fR flag specifies the path under which devices are to be searched.
 .RE
 
+.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