database/postgres
database/postgres
1
0
Fork 0
mirror of https://github.com/postgres/postgres.git synced 2025-05-12 16:21:30 +03:00
Code

Plug race in dsa_attach.

Browse Source 
With sufficiently bad luck, it was possible for a parallel worker to
attempt attach to a DSA area after all other backends have detached
from it, which is not legal.  If the worker had waited a little longer
to get started, the DSM itself would have been destroyed, which is why
this wasn't noticed before.

Thomas Munro, per a report from Andreas Seltenreich

Discussion: http://postgr.es/m/87h92g83t3.fsf@credativ.de
This commit is contained in:
Robert Haas 2017-03-29 09:44:29 -04:00
parent 3582b223d4
commit fddf45b380
1 changed files with 7 additions and 0 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

7
src/backend/utils/mmgr/dsa.c
View File

@ -1314,6 +1314,13 @@ attach_internal(void *place, dsm_segment *segment, dsa_handle handle)
/* Bump the reference count. */ /* Bump the reference count. */
LWLockAcquire(DSA_AREA_LOCK(area), LW_EXCLUSIVE); LWLockAcquire(DSA_AREA_LOCK(area), LW_EXCLUSIVE);
if (control->refcnt == 0)
{
/* We can't attach to a DSA area that has already been destroyed. */
ereport(ERROR,
(errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
errmsg("could not attach to dsa_area")));
}
++control->refcnt; ++control->refcnt;
LWLockRelease(DSA_AREA_LOCK(area)); LWLockRelease(DSA_AREA_LOCK(area));