BUG #13487: GetBufferFromRing code bug

From: jingwei_5107(at)qq(dot)com
To: pgsql-bugs(at)postgresql(dot)org
Subject: BUG #13487: GetBufferFromRing code bug
Date: 2015-07-04 15:12:03
Message-ID: 20150704151203.1190.44251@wrigleys.postgresql.org
Views: Raw Message | Whole Thread | Download mbox | Resend email
Thread:
Lists: pgsql-bugs

The following bug has been logged on the website:

Bug reference: 13487
Logged by: will
Email address: jingwei_5107(at)qq(dot)com
PostgreSQL version: 9.4.4
Operating system: It does not matter
Description:

In GetBufferFromRing method(/src/backend/storage/buffer/freelist.c),I think
the following code:
506 buf = &BufferDescriptors[bufnum - 1];
507 LockBufHdr(buf);
508 if (buf->refcount == 0 && buf->usage_count <= 1)
509 {
510 strategy->current_was_in_ring = true;
511 return buf;
512 }
513 UnlockBufHdr(buf);

should be changed to(Adding line 510):

506 buf = &BufferDescriptors[bufnum - 1];
507 LockBufHdr(buf);
508 if (buf->refcount == 0 && buf->usage_count <= 1)
509 {
510 UnlockBufHdr(buf);// add this line
511 strategy->current_was_in_ring = true;
512 return buf;
513 }
514 UnlockBufHdr(buf);

It should Unlock the buf before returning buf, after exclusively accessing
the buf's refcount&usage_count member.

Am I right? Any reply is appreciated if I'm wrong.

Responses

Browse pgsql-bugs by date

  From Date Subject
Next Message Julien Rouhaud 2015-07-04 20:21:26 Re: BUG #12379: pgbench should hint to pgbench -i
Previous Message Michael Paquier 2015-07-04 12:40:31 Re: PQexec() hangs on OOM