Possible data race on Windows (src/bin/pg_dump/parallel.c)

From: Ranier Vilela <ranier(dot)vf(at)gmail(dot)com>
To: Pg Hackers <pgsql-hackers(at)postgresql(dot)org>
Subject: Possible data race on Windows (src/bin/pg_dump/parallel.c)
Date: 2024-04-29 13:16:33
Message-ID: CAEudQAqDPXE0imkFrE-YRWr0ezzbZEYQW-dD77eq97Qk+zGesA@mail.gmail.com
Views: Raw Message | Whole Thread | Download mbox | Resend email
Thread:
Lists: pgsql-hackers

Hi,

Per Coverity.

CID 1542943: (#1 of 1): Data race condition (MISSING_LOCK)
3. missing_lock: Accessing slot->AH without holding lock signal_info_lock.
Elsewhere, ParallelSlot.AH is written to with signal_info_lock held 1 out
of 1 times (1 of these accesses strongly imply that it is necessary).

The function DisconnectDatabase effectively writes the ParallelSlot.AH.
So the call in the function archive_close_connection:

if (slot->AH)
DisconnectDatabase(&(slot->AH->public));

It should also be protected on Windows, correct?

Patch attached.

best regards,
Ranier Vilela

Attachment Content-Type Size
fix-possible-data-race-windows-parallel.patch application/octet-stream 598 bytes

Browse pgsql-hackers by date

  From Date Subject
Next Message Tom Lane 2024-04-29 13:45:16 Re: A failure in prepared_xacts test
Previous Message Robert Haas 2024-04-29 13:15:52 Re: tablecmds.c/MergeAttributes() cleanup