Horton

Re: BUG #14897: Segfault on statitics SQL request

Tom Lane <tgl@sss.pgh.pa.us>

View thread View on postgr.es
From: Tom Lane <tgl@sss.pgh.pa.us>
To: Andres Freund <andres@anarazel.de>
Cc: Dmitry Dolgov <9erthalion6@gmail.com>, Michael Paquier <michael.paquier@gmail.com>, Vincent Lachenal <vincent.lachenal@gmail.com>, PostgreSQL mailing lists <pgsql-bugs@postgresql.org>
Date: 2017-11-12T17:43:02Z
Lists: pgsql-bugs 
I wrote:
> Andres Freund <andres@anarazel.de> writes:
>> Specifying MCXT_ALLOC_ALIGN_CACHELINE rather than CACHELINE_ALIGNMENT
>> doesn't seem to make it meaningfully harder to adjust. Or are you
>> thinking of probing the hardware?

> Yeah, the latter.  The two approaches seem pretty much equivalent if
> you're assuming compile-time decisions, but if we ever wanted a run-
> time decision, I think having a flag bit that's interpreted inside
> MemoryContextAllocExtended would be easier to deal with.

Actually, independently of which way we do that, it seems like we can't
easily hide this inside a palloc wrapper.  If we palloc some extra space
and then adjust the return value to be 16-aligned, what happens when the
caller tries to pfree the block?

			regards, tom lane

Commits

  1. Prevent int128 from requiring more than MAXALIGN alignment.

  2. Rearrange c.h to create a "compiler characteristics" section.