Horton

Re: [CLOBBER_CACHE]Server crashed with segfault 11 while executing clusterdb

Tom Lane <tgl@sss.pgh.pa.us>

View thread View on postgr.es
From: Tom Lane <tgl@sss.pgh.pa.us>
To: Amul Sul <sulamul@gmail.com>
Cc: Alvaro Herrera <alvherre@alvh.no-ip.org>, Kyotaro Horiguchi <horikyota.ntt@gmail.com>, Michael Paquier <michael@paquier.xyz>, Noah Misch <noah@leadboat.com>, Neha Sharma <neha.sharma@enterprisedb.com>, Amit Langote <amitlangote09@gmail.com>, PostgreSQL-development <pgsql-hackers@postgresql.org>
Date: 2021-07-12T21:07:46Z
Lists: pgsql-hackers 
Amul Sul <sulamul@gmail.com> writes:
> [ v5_Add-RelationGetSmgr-inline-function.patch ]

Pushed with minor cosmetic adjustments.

RelationCopyStorage() kind of gives me the willies.
It's not really an smgr-level function, but we call it
everywhere with smgr pointers that belong to relcache entries:

 	/* copy main fork */
-	RelationCopyStorage(rel->rd_smgr, dstrel, MAIN_FORKNUM,
+	RelationCopyStorage(RelationGetSmgr(rel), dstrel, MAIN_FORKNUM,
 						rel->rd_rel->relpersistence);

So that would fail hard if a relcache flush could occur inside
that function.  It seems impossible today, so I settled for
just annotating the function to that effect.  But it won't
surprise me a bit if somebody breaks it in future due to not
having read/understood the comment.

			regards, tom lane

Commits

  1. Replace RelationOpenSmgr() with RelationGetSmgr().

  2. Avoid possible crash while finishing up a heap rewrite.

  3. Redefine pg_class.reltuples to be -1 before the first VACUUM or ANALYZE.