Non-emergency patch for bug #17679

From: Tom Lane <tgl(at)sss(dot)pgh(dot)pa(dot)us>
To: pgsql-hackers(at)lists(dot)postgresql(dot)org
Cc: Robert Haas <robertmhaas(at)gmail(dot)com>, Andres Freund <andres(at)anarazel(dot)de>
Subject: Non-emergency patch for bug #17679
Date: 2022-11-08 16:28:08
Message-ID: 3797575.1667924888@sss.pgh.pa.us
Views: Raw Message | Whole Thread | Download mbox | Resend email
Thread:
Lists: pgsql-hackers

In the release team's discussion leading up to commit 0e758ae89,
Andres opined that what commit 4ab5dae94 had done to mdunlinkfork
was a mess, and I concur. It invented an entirely new code path
through that function, and required two different behaviors from the
segment-deletion loop. I think a very straight line can be drawn
between that extra complexity and the introduction of a nasty bug.
It's all unnecessary too, because AFAICS all we really need is to
apply the pre-existing behavior for temp tables and REDO mode
to binary-upgrade mode as well.

Hence, the attached reverts everything 4ab5dae94 did to this function,
and most of 0e758ae89 too, and instead makes IsBinaryUpgrade an
additional reason to take the immediate-unlink path.

Barring objections, I'll push this after the release freeze lifts.

regards, tom lane

Attachment Content-Type Size
fix-mdunlinkfork-more-cleanly.patch text/x-diff 3.4 KB

Responses

Browse pgsql-hackers by date

  From Date Subject
Next Message Peter Geoghegan 2022-11-08 17:57:42 Re: [PoC] Improve dead tuple storage for lazy vacuum
Previous Message David Christensen 2022-11-08 16:17:56 Re: [PATCHES] Post-special page storage TDE support