Re: basebackup: do not verify checksums on pages written before enabling checksums

From: Daniel Gustafsson <daniel(at)yesql(dot)se>
To: Bertrand Drouvot <bertranddrouvot(dot)pg(at)gmail(dot)com>
Cc: Zsolt Parragi <zsolt(dot)parragi(at)percona(dot)com>, Yilin Zhang <jiezhilove(at)126(dot)com>, PostgreSQL Hackers <pgsql-hackers(at)lists(dot)postgresql(dot)org>
Subject: Re: basebackup: do not verify checksums on pages written before enabling checksums
Date: 2026-08-18 22:16:55
Message-ID: 04204C86-484F-41FD-A3A0-E100F82313D4@yesql.se
Views: Whole Thread | Raw Message | Download mbox | Resend email
Thread:
Lists: pgsql-hackers

> On 18 Aug 2026, at 14:42, Daniel Gustafsson <daniel(at)yesql(dot)se> wrote:
>
>> On 18 Aug 2026, at 14:39, Bertrand Drouvot <bertranddrouvot(dot)pg(at)gmail(dot)com> wrote:
>
>> Shouldn't hit be incremented only when the injection point is attached?
>
> Good point, will fix.
>
>> + checksum ^= checksum;
>>
>> That's same as "checksum = 0". That works because 0 is not a valid value but
>> was the intend to "checksum ^= 1"? to express that we want a different checksum?
>
> Ugh, sorry, I didn't notice that I hadn't replaced my debugcode before
> generating the patch, I blame the lack of caffeine this morning.

I ended up settling for setting the checksum to zero since it also makes the
checksum predictable, and with that pushed and backpatched.

--
Daniel Gustafsson

In response to

Browse pgsql-hackers by date

  From Date Subject
Next Message Michael Paquier 2026-08-18 22:35:18 Re: Fix a relcache reference leak in reorderbuffer.c
Previous Message Daniel Gustafsson 2026-08-18 22:13:33 Re: pg_control_checkpoint(): add "data_checksum_version" (Pg19)?