uint64_t *zopt_object = NULL;
int zopt_objects = 0;
libzfs_handle_t *g_zfs;
+uint64_t max_inflight = 200;
/*
* These libumem hooks provide a reasonable set of defaults for the allocator's
{
(void) fprintf(stderr,
"Usage: %s [-CumdibcsDvhLXFPA] [-t txg] [-e [-p path...]] "
- "poolname [object...]\n"
- " %s [-divPA] [-e -p path...] dataset [object...]\n"
- " %s -m [-LXFPA] [-t txg] [-e [-p path...]] "
+ "[-U config] [-M inflight I/Os] poolname [object...]\n"
+ " %s [-divPA] [-e -p path...] [-U config] dataset "
+ "[object...]\n"
+ " %s -m [-LXFPA] [-t txg] [-e [-p path...]] [-U config] "
"poolname [vdev [metaslab...]]\n"
" %s -R [-A] [-e [-p path...]] poolname "
"vdev:offset:size[:flags]\n"
- " %s -S [-PA] [-e [-p path...]] poolname\n"
+ " %s -S [-PA] [-e [-p path...]] [-U config] poolname\n"
" %s -l [-uA] device\n"
" %s -C [-A] [-U config]\n\n",
cmdname, cmdname, cmdname, cmdname, cmdname, cmdname, cmdname);
(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]");
(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");
bp, NULL, NULL, ZIO_FLAG_CANFAIL)), ==, 0);
}
+static void
+zdb_blkptr_done(zio_t *zio)
+{
+ spa_t *spa = zio->io_spa;
+ blkptr_t *bp = zio->io_bp;
+ int ioerr = zio->io_error;
+ zdb_cb_t *zcb = zio->io_private;
+ zbookmark_t *zb = &zio->io_bookmark;
+
+ zio_data_buf_free(zio->io_data, zio->io_size);
+
+ mutex_enter(&spa->spa_scrub_lock);
+ spa->spa_scrub_inflight--;
+ cv_broadcast(&spa->spa_scrub_io_cv);
+
+ if (ioerr && !(zio->io_flags & ZIO_FLAG_SPECULATIVE)) {
+ char blkbuf[BP_SPRINTF_LEN];
+
+ zcb->zcb_haderrors = 1;
+ zcb->zcb_errors[ioerr]++;
+
+ if (dump_opt['b'] >= 2)
+ sprintf_blkptr(blkbuf, bp);
+ else
+ blkbuf[0] = '\0';
+
+ (void) printf("zdb_blkptr_cb: "
+ "Got error %d reading "
+ "<%llu, %llu, %lld, %llx> %s -- skipping\n",
+ ioerr,
+ (u_longlong_t)zb->zb_objset,
+ (u_longlong_t)zb->zb_object,
+ (u_longlong_t)zb->zb_level,
+ (u_longlong_t)zb->zb_blkid,
+ blkbuf);
+ }
+ mutex_exit(&spa->spa_scrub_lock);
+}
+
/* ARGSUSED */
static int
zdb_blkptr_cb(spa_t *spa, zilog_t *zilog, const blkptr_t *bp, arc_buf_t *pbuf,
is_metadata = (BP_GET_LEVEL(bp) != 0 || DMU_OT_IS_METADATA(type));
if (dump_opt['c'] > 1 || (dump_opt['c'] && is_metadata)) {
- int ioerr;
size_t size = BP_GET_PSIZE(bp);
- void *data = malloc(size);
+ void *data = zio_data_buf_alloc(size);
int flags = ZIO_FLAG_CANFAIL | ZIO_FLAG_SCRUB | ZIO_FLAG_RAW;
/* If it's an intent log block, failure is expected. */
if (zb->zb_level == ZB_ZIL_LEVEL)
flags |= ZIO_FLAG_SPECULATIVE;
- ioerr = zio_wait(zio_read(NULL, spa, bp, data, size,
- NULL, NULL, ZIO_PRIORITY_ASYNC_READ, flags, zb));
-
- free(data);
+ mutex_enter(&spa->spa_scrub_lock);
+ while (spa->spa_scrub_inflight > max_inflight)
+ cv_wait(&spa->spa_scrub_io_cv, &spa->spa_scrub_lock);
+ spa->spa_scrub_inflight++;
+ mutex_exit(&spa->spa_scrub_lock);
- if (ioerr && !(flags & ZIO_FLAG_SPECULATIVE)) {
- zcb->zcb_haderrors = 1;
- zcb->zcb_errors[ioerr]++;
+ zio_nowait(zio_read(NULL, spa, bp, data, size,
+ zdb_blkptr_done, zcb, ZIO_PRIORITY_ASYNC_READ, flags, zb));
- if (dump_opt['b'] >= 2)
- sprintf_blkptr(blkbuf, bp);
- else
- blkbuf[0] = '\0';
-
- (void) printf("zdb_blkptr_cb: "
- "Got error %d reading "
- "<%llu, %llu, %lld, %llx> %s -- skipping\n",
- ioerr,
- (u_longlong_t)zb->zb_objset,
- (u_longlong_t)zb->zb_object,
- (u_longlong_t)zb->zb_level,
- (u_longlong_t)zb->zb_blkid,
- blkbuf);
- }
}
zcb->zcb_readfails = 0;
zcb.zcb_haderrors |= traverse_pool(spa, 0, flags, zdb_blkptr_cb, &zcb);
+ /*
+ * If we've traversed the data blocks then we need to wait for those
+ * I/Os to complete. We leverage "The Godfather" zio to wait on
+ * all async I/Os to complete.
+ */
+ if (dump_opt['c']) {
+ (void) zio_wait(spa->spa_async_zio_root);
+ spa->spa_async_zio_root = zio_root(spa, NULL, NULL,
+ ZIO_FLAG_CANFAIL | ZIO_FLAG_SPECULATIVE |
+ ZIO_FLAG_GODFATHER);
+ }
+
if (zcb.zcb_haderrors) {
(void) printf("\nError counts:\n\n");
(void) printf("\t%5s %s\n", "errno", "count");
dprintf_setup(&argc, argv);
- while ((c = getopt(argc, argv, "bcdhilmsuCDRSAFLXevp:t:U:P")) != -1) {
+ while ((c = getopt(argc, argv, "bcdhilmM:suCDRSAFLXevp:t:U:P")) != -1) {
switch (c) {
case 'b':
case 'c':
case 'v':
verbose++;
break;
+ case 'M':
+ max_inflight = strtoull(optarg, NULL, 0);
+ if (max_inflight == 0) {
+ (void) fprintf(stderr, "maximum number "
+ "of inflight I/Os must be greater "
+ "than 0\n");
+ usage();
+ }
+ break;
case 'p':
if (searchdirs == NULL) {
searchdirs = umem_alloc(sizeof (char *),
.\"
.\"
.\" Copyright 2012, Richard Lowe.
+.\" Copyright (c) 2012 by Delphix. All rights reserved.
.\"
.TH "ZDB" "8" "February 15, 2012" "" ""
.SH "SYNOPSIS"
\fBzdb\fR [-CumdibcsDvhLXFPA] [-e [-p \fIpath\fR...]] [-t \fItxg\fR]
- \fIpoolname\fR [\fIobject\fR ...]
+ [-U \fIcache\fR] [-M \fIinflight I/Os\fR] [\fIpoolname\fR
+ [\fIobject\fR ...]]
.P
-\fBzdb\fR [-divPA] [-e [-p \fIpath\fR...]] \fIdataset\fR [\fIobject\fR ...]
+\fBzdb\fR [-divPA] [-e [-p \fIpath\fR...]] [-U \fIcache\fR]
+ \fIdataset\fR [\fIobject\fR ...]
.P
-\fBzdb\fR -m [-LXFPA] [-t \fItxg\fR] [-e [-p \fIpath\fR...]] \fIpoolname\fR
- [\fIvdev\fR [\fImetaslab\fR ...]]
+\fBzdb\fR -m [-LXFPA] [-t \fItxg\fR] [-e [-p \fIpath\fR...]] [-U \fIcache\fR]
+ \fIpoolname\fR [\fIvdev\fR [\fImetaslab\fR ...]]
.P
-\fBzdb\fR -R [-A] [-e [-p \fIpath\fR...]] \fIpoolname\fR
+\fBzdb\fR -R [-A] [-e [-p \fIpath\fR...]] [-U \fIcache\fR] \fIpoolname\fR
\fIvdev\fR:\fIoffset\fR:\fIsize\fR[:\fIflags\fR]
.P
-\fBzdb\fR -S [-AP] [-e [-p \fIpath\fR...]] \fIpoolname\fR
+\fBzdb\fR -S [-AP] [-e [-p \fIpath\fR...]] [-U \fIcache\fR] \fIpoolname\fR
.P
\fBzdb\fR -l [-uA] \fIdevice\fR
transactions.
.RE
+.sp
+.ne 2
+.na
+\fB-M \fIinflight I/Os\fR \fR
+.ad
+.sp .6
+.RS 4n
+Limit the number of outstanding checksum I/Os to the specified value. The
+default value is 200. This option affects the performance of the \fB-c\fR
+option.
+.RE
+
.sp
.ne 2
.na
.ad
.sp .6
.RS 4n
-Use a cache file other than \fB/etc/zfs/zpool.cache\fR. This option is only
-valid with \fB-C\fR
+Use a cache file other than \fB/etc/zfs/zpool.cache\fR.
.RE
.sp
#include <sys/zfs_context.h>
#include <sys/spa.h>
+#include <sys/spa_impl.h>
#include <sys/vdev_file.h>
#include <sys/vdev_impl.h>
#include <sys/zio.h>
vd->vdev_tsd = NULL;
}
-static int
-vdev_file_io_start(zio_t *zio)
+static void
+vdev_file_io_strategy(void *arg)
{
+ zio_t *zio = (zio_t *)arg;
vdev_t *vd = zio->io_vd;
- vdev_file_t *vf;
- ssize_t resid = 0;
+ vdev_file_t *vf = vd->vdev_tsd;
+ ssize_t resid;
- if (!vdev_readable(vd)) {
- zio->io_error = ENXIO;
- return (ZIO_PIPELINE_CONTINUE);
- }
+ zio->io_error = vn_rdwr(zio->io_type == ZIO_TYPE_READ ?
+ UIO_READ : UIO_WRITE, vf->vf_vnode, zio->io_data,
+ zio->io_size, zio->io_offset, UIO_SYSSPACE,
+ 0, RLIM64_INFINITY, kcred, &resid);
+
+ if (resid != 0 && zio->io_error == 0)
+ zio->io_error = ENOSPC;
- vf = vd->vdev_tsd;
+ zio_interrupt(zio);
+}
+
+static int
+vdev_file_io_start(zio_t *zio)
+{
+ spa_t *spa = zio->io_spa;
+ vdev_t *vd = zio->io_vd;
+ vdev_file_t *vf = vd->vdev_tsd;
if (zio->io_type == ZIO_TYPE_IOCTL) {
+ /* XXPOLICY */
+ if (!vdev_readable(vd)) {
+ zio->io_error = ENXIO;
+ return (ZIO_PIPELINE_CONTINUE);
+ }
+
switch (zio->io_cmd) {
case DKIOCFLUSHWRITECACHE:
zio->io_error = VOP_FSYNC(vf->vf_vnode, FSYNC | FDSYNC,
return (ZIO_PIPELINE_CONTINUE);
}
- zio->io_error = vn_rdwr(zio->io_type == ZIO_TYPE_READ ?
- UIO_READ : UIO_WRITE, vf->vf_vnode, zio->io_data,
- zio->io_size, zio->io_offset, UIO_SYSSPACE,
- 0, RLIM64_INFINITY, kcred, &resid);
-
- if (resid != 0 && zio->io_error == 0)
- zio->io_error = ENOSPC;
-
- zio_interrupt(zio);
+ taskq_dispatch_ent(spa->spa_zio_taskq[ZIO_TYPE_FREE][ZIO_TASKQ_ISSUE],
+ vdev_file_io_strategy, zio, 0, &zio->io_tqent);
return (ZIO_PIPELINE_STOP);
}
projects / zfs / commitdiff