c/mirror_zfs
1
0
Fork 0
mirror of https://git.proxmox.com/git/mirror_zfs.git synced 2024-12-25 18:59:33 +03:00
Code Issues Packages Projects Releases Wiki Activity

26126 panic system rather than corrupting pool if we hit bug 26100

Browse Source 
References:
  delphix/delphix-os@931c8aaab7

Ported-by: Richard Yao <ryao@gentoo.org>
Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
Closes #1650
This commit is contained in:
Matthew Ahrens 2013-08-12 12:53:33 -04:00 committed by Brian Behlendorf
parent 2517c8ee08
commit 78e2739d3c
4 changed files with 26 additions and 10 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

29
module/zfs/bptree.c
View File

@ -20,7 +20,7 @@
*/
/*
* Copyright (c) 2012 by Delphix. All rights reserved.
* Copyright (c) 2013 by Delphix. All rights reserved.
*/
#include <sys/arc.h>
@ -180,6 +180,7 @@ bptree_iterate(objset_t *os, uint64_t obj, boolean_t free, bptree_itor_t func,
err = 0;
for (i = ba.ba_phys->bt_begin; i < ba.ba_phys->bt_end; i++) {
bptree_entry_phys_t bte;
int flags = TRAVERSE_PREFETCH_METADATA | TRAVERSE_POST;
ASSERT(!free || i == ba.ba_phys->bt_begin);
@ -188,13 +189,13 @@ bptree_iterate(objset_t *os, uint64_t obj, boolean_t free, bptree_itor_t func,
if (err != 0)
break;
if (zfs_recover)
flags |= TRAVERSE_HARD;
err = traverse_dataset_destroyed(os->os_spa, &bte.be_bp,
bte.be_birth_txg, &bte.be_zb,
TRAVERSE_PREFETCH_METADATA | TRAVERSE_POST,
bte.be_birth_txg, &bte.be_zb, flags,
bptree_visit_cb, &ba);
if (free) {
ASSERT(err == 0 || err == ERESTART);
if (err != 0) {
if (err == ERESTART) {
/* save bookmark for future resume */
ASSERT3U(bte.be_zb.zb_objset, ==,
ZB_DESTROYED_OBJSET);
@ -202,11 +203,21 @@ bptree_iterate(objset_t *os, uint64_t obj, boolean_t free, bptree_itor_t func,
dmu_write(os, obj, i * sizeof (bte),
sizeof (bte), &bte, tx);
break;
} else {
ba.ba_phys->bt_begin++;
(void) dmu_free_range(os, obj,
i * sizeof (bte), sizeof (bte), tx);
}
if (err != 0) {
/*
* We can not properly handle an i/o
* error, because the traversal code
* does not know how to resume from an
* arbitrary bookmark.
*/
zfs_panic_recover("error %u from "
"traverse_dataset_destroyed()", err);
}
ba.ba_phys->bt_begin++;
(void) dmu_free_range(os, obj,
i * sizeof (bte), sizeof (bte), tx);
}
}

2
module/zfs/dmu_traverse.c
View File

@ -361,7 +361,7 @@ traverse_visitbp(traverse_data_t *td, const dnode_phys_t *dnp,
(void) arc_buf_remove_ref(buf, &buf);
post:
if (err == 0 && lasterr == 0 && (td->td_flags & TRAVERSE_POST)) {
if (err == 0 && (td->td_flags & TRAVERSE_POST)) {
err = td->td_func(td->td_spa, NULL, bp, zb, dnp, td->td_arg);
if (err == ERESTART)
pause = B_TRUE;

3
module/zfs/dsl_scan.c
View File

@ -1334,6 +1334,9 @@ dsl_scan_free_should_pause(dsl_scan_t *scn)
{
uint64_t elapsed_nanosecs;
if (zfs_recover)
return (B_FALSE);
elapsed_nanosecs = gethrtime() - scn->scn_sync_start_time;
return (elapsed_nanosecs / NANOSEC > zfs_txg_timeout ||
(NSEC2MSEC(elapsed_nanosecs) > zfs_free_min_time_ms &&

2
module/zfs/zfs_debug.c
View File

@ -41,6 +41,8 @@ int zfs_flags = 0;
* zfs_recover can be set to nonzero to attempt to recover from
* otherwise-fatal errors, typically caused by on-disk corruption. When
* set, calls to zfs_panic_recover() will turn into warning messages.
* This should only be used as a last resort, as it typically results
* in leaked space, or worse.
*/
int zfs_recover = 0;