Skip to contentSkip to conversation

blockjob: update nodes head while removing all bdrv

PATCH Superseded Sergio Lopez 2 participants 3 messages Sep 10, 2019

v1 of 2 Superseded Latest is v2 →

lore
Share & tools
lore
#0 lore raw

PATCH 1/1 blockjob: update nodes head while removing all bdrv

block_job_remove_all_bdrv() iterates through job->nodes, calling bdrv_root_unref_child() for each entry. The call to the latter may reach child_job_[can_]set_aio_ctx(), which will also attempt to traverse job->nodes, potentially finding entries that where freed on previous iterations.

To avoid this situation, update job->nodes head on each iteration to ensure that already freed entries are no longer linked to the list.

RHBZ: https://bugzilla.redhat.com/show_bug.cgi?id=1746631

---
 blockjob.c | 6 ++++++
 1 file changed, 6 insertions(+)
blockjob.c +6 -0
index 6e32d1a0c0..7b1551d981 100644
@@ -192,6 +192,12 @@ void block_job_remove_all_bdrv(BlockJob *job)
         BdrvChild *c = l->data;
         bdrv_op_unblock_all(c->bs, job->blocker);
         bdrv_root_unref_child(c);
+        /*
+         * The call above may reach child_job_[can_]set_aio_ctx(), which will
+         * also traverse job->nodes, so update the head here to make sure it
+         * doesn't attempt to process an already freed BdrvChild.
+         */
+        job->nodes = l->next;
     }
     g_slist_free(job->nodes);
     job->nodes = NULL;
Signature · 1 line
2.21.0
#1 lore raw
Quoted 21 lines · Sergio Lopez

block_job_remove_all_bdrv() iterates through job->nodes, calling bdrv_root_unref_child() for each entry. The call to the latter may reach child_job_[can_]set_aio_ctx(), which will also attempt to traverse job->nodes, potentially finding entries that where freed on previous iterations.

To avoid this situation, update job->nodes head on each iteration to ensure that already freed entries are no longer linked to the list.

RHBZ: https://bugzilla.redhat.com/show_bug.cgi?id=1746631

---
 blockjob.c | 6 ++++++
 1 file changed, 6 insertions(+)
blockjob.c +0 -0
index 6e32d1a0c0..7b1551d981 100644
@@ -192,6 +192,12 @@ void block_job_remove_all_bdrv(BlockJob *job)
         BdrvChild *c = l->data;

blockjob.c void block_job_remove_all_bdrv(BlockJob *job) · line 202

Quoted 10 lines
blockjob.c +6 -0
         bdrv_op_unblock_all(c->bs, job->blocker);
         bdrv_root_unref_child(c);
+        /*
+         * The call above may reach child_job_[can_]set_aio_ctx(), which will
+         * also traverse job->nodes, so update the head here to make sure it
+         * doesn't attempt to process an already freed BdrvChild.
+         */
+        job->nodes = l->next;
     }
     g_slist_free(job->nodes);

But this will leak the whole list.

Max

Quoted 2 lines
blockjob.c +0 -0
     job->nodes = NULL;
#2 lore raw
Quoted 26 lines · Max Reitz
Quoted 25 lines · Sergio Lopez

block_job_remove_all_bdrv() iterates through job->nodes, calling bdrv_root_unref_child() for each entry. The call to the latter may reach child_job_[can_]set_aio_ctx(), which will also attempt to traverse job->nodes, potentially finding entries that where freed on previous iterations.

To avoid this situation, update job->nodes head on each iteration to ensure that already freed entries are no longer linked to the list.

RHBZ: https://bugzilla.redhat.com/show_bug.cgi?id=1746631

---
 blockjob.c | 6 ++++++
 1 file changed, 6 insertions(+)
blockjob.c +2 -0
index 6e32d1a0c0..7b1551d981 100644
@@ -192,6 +192,12 @@ void block_job_remove_all_bdrv(BlockJob *job)
         BdrvChild *c = l->data;
         bdrv_op_unblock_all(c->bs, job->blocker);
         bdrv_root_unref_child(c);
+        /*
+         * The call above may reach child_job_[can_]set_aio_ctx(), which will
Quoted 8 lines

blockjob.c void block_job_remove_all_bdrv(BlockJob *job) · line 202

Quoted 6 lines
blockjob.c +4 -0
+         * also traverse job->nodes, so update the head here to make sure it
+         * doesn't attempt to process an already freed BdrvChild.
+         */
+        job->nodes = l->next;
     }
     g_slist_free(job->nodes);

But this will leak the whole list.

Ouch. This is what happens when you rush up things. I'll send a v2 ASAP.

Thanks,
Sergio.