Panic in bookmark deletion
The problem is that when dsl_bookmark_destroy_check() is executed from open context (the pre-check), it fills in dbda_success based on the existence of the bookmark.
But the bookmark (or containing filesystem as in this case) can be destroyed before we get to syncing context. When we re-run dsl_bookmark_destroy_check() in syncing
context, it will not add the deleted bookmark to dbda_success, intending for dsl_bookmark_destroy_sync() to not process it. But because the bookmark is still in dbda_success
from the open-context call, we do try to destroy it.
The fix is that dsl_bookmark_destroy_check() should not modify dbda_success when called from open context.
Updated by Electric Monk about 2 years ago
- Status changed from New to Closed
- % Done changed from 0 to 100
commit 42418f9e73f0d007aa87675ecc206c26fc8e073e Author: Matthew Ahrens <email@example.com> Date: 2017-06-29T23:10:42.000Z 8377 Panic in bookmark deletion Reviewed by: Paul Dagnelie <firstname.lastname@example.org> Reviewed by: Pavel Zakharov <email@example.com> Reviewed by: George Wilson <firstname.lastname@example.org> Approved by: Robert Mustacchi <email@example.com>
Updated by Serapheim Dimitropoulos over 1 year ago
For completeness, here is the stack of the failure for this issue:
panic message: assertion failed: dsl_bookmark_hold_ds(dp, nvpair_name(pair), &ds, ((char *)__func__), &shortname) 0 (0x2 0x0), file: ../../common/fs/zfs/dsl_bookmark.c, line: 903
dump content: kernel pages only
ffffff000cf599c0 dsl_bookmark_destroy_sync+0x12f(ffffff000db74b50, ffffff03d9b24c80)
ffffff000cf59a00 dsl_sync_task_sync+0x10a(ffffff000db74a60, ffffff03d9b24c80)
ffffff000cf59a90 dsl_pool_sync+0x26b(ffffff03d314f000, 237f)
ffffff000cf59b70 spa_sync+0x3ff(ffffff03317f4000, 237f)