Movatterモバイル変換

[0]ホーム
URL:

Skip to content

Navigation Menu

Sign in
Appearance settings

Search code, repositories, users, issues, pull requests...

Provide feedback

We read every piece of feedback, and take your input very seriously.

Saved searches

Use saved searches to filter your results more quickly

 Sign up
Appearance settings

Commit586250c

Browse files
committed
Fix hash_search to avoid corruption of the hash table on out-of-memory.
An out-of-memory error during expand_table() on a palloc-based hash tablewould leave a partially-initialized entry in the table. This would not beharmful for transient hash tables, since they'd get thrown away anyway attransaction abort. But for long-lived hash tables, such as the relcachehash, this would effectively corrupt the table, leading to crash or othermisbehavior later.To fix, rearrange the order of operations so that table enlargement isattempted before we insert a new entry, rather than after adding itto the hash table.Problem discovered by Hitoshi Harada, though this is a bit differentfrom his proposed patch.
1 parentc47b408 commit586250c

File tree

1 file changed

+25
-16
lines changed

1 file changed

+25
-16
lines changed

‎src/backend/utils/hash/dynahash.c

Lines changed: 25 additions & 16 deletions
Original file line numberDiff line numberDiff line change
@@ -817,6 +817,27 @@ hash_search_with_hash_value(HTAB *hashp,
817817
hctl->accesses++;
818818
#endif
819819

820+
/*
821+
* If inserting, check if it is time to split a bucket.
822+
*
823+
* NOTE: failure to expand table is not a fatal error, it just means we
824+
* have to run at higher fill factor than we wanted. However, if we're
825+
* using the palloc allocator then it will throw error anyway on
826+
* out-of-memory, so we must do this before modifying the table.
827+
*/
828+
if (action==HASH_ENTER||action==HASH_ENTER_NULL)
829+
{
830+
/*
831+
* Can't split if running in partitioned mode, nor if frozen, nor if
832+
* table is the subject of any active hash_seq_search scans. Strange
833+
* order of these tests is to try to check cheaper conditions first.
834+
*/
835+
if (!IS_PARTITIONED(hctl)&& !hashp->frozen&&
836+
hctl->nentries / (long) (hctl->max_bucket+1) >=hctl->ffactor&&
837+
!has_seq_scans(hashp))
838+
(void)expand_table(hashp);
839+
}
840+
820841
/*
821842
* Do the initial lookup
822843
*/
@@ -937,24 +958,12 @@ hash_search_with_hash_value(HTAB *hashp,
937958
currBucket->hashvalue=hashvalue;
938959
hashp->keycopy(ELEMENTKEY(currBucket),keyPtr,keysize);
939960

940-
/* caller is expected to fill the data field on return */
941-
942961
/*
943-
*Check if itistime tosplit a bucket. Can't split if running
944-
*in partitioned mode, nor if table is the subject of any active
945-
*hash_seq_search scans. Strange order of these tests is to try
946-
*to check cheaper conditions first.
962+
*Callerisexpected tofill the data field on return. DO NOT
963+
*insert any code that could possibly throw error here, as doing
964+
*so would leave the table entry incomplete and hence corrupt the
965+
*caller's data structure.
947966
*/
948-
if (!IS_PARTITIONED(hctl)&&
949-
hctl->nentries / (long) (hctl->max_bucket+1) >=hctl->ffactor&&
950-
!has_seq_scans(hashp))
951-
{
952-
/*
953-
* NOTE: failure to expand table is not a fatal error, it just
954-
* means we have to run at higher fill factor than we wanted.
955-
*/
956-
expand_table(hashp);
957-
}
958967

959968
return (void*)ELEMENTKEY(currBucket);
960969
}

0 commit comments

Comments
 (0)
[8]ページ先頭
©2009-2025 Movatter.jp