Re: pgsql: Generational memory allocator

On 2017-11-23 22:34:57 +0100, Tomas Vondra wrote:
> > I think it's a legitimate complaint that postmaster.log wasn't captured
> > in this failure, but that's a buildfarm script oversight and hardly
> > Andres' fault.
> >
>
> Are the valgrind errors really written to postmaster log? I'm assuming it
> failed because valgrind ran into an issue and killed the process.

Yes. Search e.g. in
https://buildfarm.postgresql.org/cgi-bin/show_log.pl?nm=skink&dt=2017-09-18%2004%3A10%3A01
for VALGRINDERROR-BEGIN.

(You could argue that they're only written there in certain
configurations, because I'd assume it'd not work in e.g. syslog
configured systems, because valgrind just prints to stderr).

> Hmmm, I see. Presumably adding this to GenerationChunk (similarly to what we
> do in AllocChunkData) would address the issue:
>
> #if MAXIMUM_ALIGNOF > 4 && SIZEOF_VOID_P == 4
> Size padding;
> #endif
>
> but I have no way to verify that (no access to such machine). I wonder why
> SlabChunk doesn't need to do that (perhaps a comment explaining that would
> be appropriate?).

Can't you just compile pg on a 32bit system and manually define MAXALIGN
to 8 bytes?

Greetings,

Andres Freund