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

Fix scn_queue races on very old pools

Browse Source 
Code for pools before version 11 uses dmu_objset_find_dp() to scan
for children datasets/clones.  It calls enqueue_clones_cb() and
enqueue_cb() callbacks in parallel from multiple taskq threads.
It ends up bad for scan_ds_queue_insert(), corrupting scn_queue
AVL-tree.  Fix it by introducing a mutex to protect those two
scan_ds_queue_insert() calls.  All other calls are done from the
sync thread and so serialized.

Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Reviewed-by: Brian Atkinson <batkinson@lanl.gov>
Signed-off-by: Alexander Motin <mav@FreeBSD.org>
Sponsored by:	iXsystems, Inc.
Closes #16162
This commit is contained in:
Alexander Motin 2024-05-09 10:32:59 -04:00 committed by GitHub
parent a0f3c8aaf1
commit af5dbed319
No known key found for this signature in database
GPG Key ID: B5690EEEBB952194
2 changed files with 7 additions and 0 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

1
include/sys/dsl_scan.h
View File

@ -173,6 +173,7 @@ typedef struct dsl_scan {
dsl_scan_phys_t scn_phys; /* on disk representation of scan */ dsl_scan_phys_t scn_phys; /* on disk representation of scan */
dsl_scan_phys_t scn_phys_cached; dsl_scan_phys_t scn_phys_cached;
avl_tree_t scn_queue; /* queue of datasets to scan */ avl_tree_t scn_queue; /* queue of datasets to scan */
kmutex_t scn_queue_lock; /* serializes scn_queue inserts */
uint64_t scn_queues_pending; /* outstanding data to issue */ uint64_t scn_queues_pending; /* outstanding data to issue */
/* members needed for syncing error scrub status to disk */ /* members needed for syncing error scrub status to disk */
dsl_errorscrub_phys_t errorscrub_phys; dsl_errorscrub_phys_t errorscrub_phys;

6
module/zfs/dsl_scan.c
View File

@ -491,6 +491,7 @@ dsl_scan_init(dsl_pool_t *dp, uint64_t txg)
avl_create(&scn->scn_queue, scan_ds_queue_compare, sizeof (scan_ds_t), avl_create(&scn->scn_queue, scan_ds_queue_compare, sizeof (scan_ds_t),
offsetof(scan_ds_t, sds_node)); offsetof(scan_ds_t, sds_node));
mutex_init(&scn->scn_queue_lock, NULL, MUTEX_DEFAULT, NULL);
avl_create(&scn->scn_prefetch_queue, scan_prefetch_queue_compare, avl_create(&scn->scn_prefetch_queue, scan_prefetch_queue_compare,
sizeof (scan_prefetch_issue_ctx_t), sizeof (scan_prefetch_issue_ctx_t),
offsetof(scan_prefetch_issue_ctx_t, spic_avl_node)); offsetof(scan_prefetch_issue_ctx_t, spic_avl_node));
@ -646,6 +647,7 @@ dsl_scan_fini(dsl_pool_t *dp)
scan_ds_queue_clear(scn); scan_ds_queue_clear(scn);
avl_destroy(&scn->scn_queue); avl_destroy(&scn->scn_queue);
mutex_destroy(&scn->scn_queue_lock);
scan_ds_prefetch_queue_clear(scn); scan_ds_prefetch_queue_clear(scn);
avl_destroy(&scn->scn_prefetch_queue); avl_destroy(&scn->scn_prefetch_queue);
@ -2723,8 +2725,10 @@ enqueue_clones_cb(dsl_pool_t *dp, dsl_dataset_t *hds, void *arg)
return (err); return (err);
ds = prev; ds = prev;
} }
mutex_enter(&scn->scn_queue_lock);
scan_ds_queue_insert(scn, ds->ds_object, scan_ds_queue_insert(scn, ds->ds_object,
dsl_dataset_phys(ds)->ds_prev_snap_txg); dsl_dataset_phys(ds)->ds_prev_snap_txg);
mutex_exit(&scn->scn_queue_lock);
dsl_dataset_rele(ds, FTAG); dsl_dataset_rele(ds, FTAG);
return (0); return (0);
} }
@ -2915,8 +2919,10 @@ enqueue_cb(dsl_pool_t *dp, dsl_dataset_t *hds, void *arg)
ds = prev; ds = prev;
} }
mutex_enter(&scn->scn_queue_lock);
scan_ds_queue_insert(scn, ds->ds_object, scan_ds_queue_insert(scn, ds->ds_object,
dsl_dataset_phys(ds)->ds_prev_snap_txg); dsl_dataset_phys(ds)->ds_prev_snap_txg);
mutex_exit(&scn->scn_queue_lock);
dsl_dataset_rele(ds, FTAG); dsl_dataset_rele(ds, FTAG);
return (0); return (0);
} }