Horton

Re: Reducing the chunk header sizes on all memory context types

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

View thread View on postgr.es
From: Tom Lane <tgl@sss.pgh.pa.us>
To: David Rowley <dgrowleyml@gmail.com>
Cc: Andres Freund <andres@anarazel.de>, Julien Rouhaud <rjuju123@gmail.com>, Robert Haas <robertmhaas@gmail.com>, Yura Sokolov <y.sokolov@postgrespro.ru>, PostgreSQL Developers <pgsql-hackers@lists.postgresql.org>
Date: 2022-10-10T19:35:16Z
Lists: pgsql-hackers

Attachments

I wrote:
> What I am mainly wondering about at this point is whether Asserts
> are good enough or we should use actual test-and-elog checks for
> these things.

Hearing no comments on that, I decided that a good policy would be
to use Asserts in the paths dealing with small chunks but test-and-elog
in the paths dealing with large chunks.  We already had test-and-elog
sanity checks in the latter paths, at least in aset.c, which the new
checks can reasonably be combined with.  It's unlikely that the
large-chunk case is at all performance-critical, too.  But adding
runtime checks in the small-chunk paths would probably risk losing
some performance in production builds, and I'm not currently prepared
to argue that the problem is big enough to justify that.

Hence v2 attached, which cleans things up a tad in aset.c and then
extends similar policy to generation.c and slab.c.  Of note is
that slab.c was doing things like this:

	SlabContext *slab = castNode(SlabContext, context);

	Assert(slab);

which has about the same effect as what I'm proposing with
AllocSetIsValid, but (a) it's randomly different from the
other allocators, and (b) it's probably a hair less efficient,
because I doubt the compiler can optimize away castNode's
special handling of NULL.  So I made these bits follow the
style of aset.c.

			regards, tom lane

Commits

  1. Harden memory context allocators against bogus chunk pointers.

  2. Improve our ability to detect bogus pointers passed to pfree et al.

  3. Remove MemoryContextContains().

  4. Remove uses of MemoryContextContains in nodeAgg.c and nodeWindowAgg.c.

  5. Temporarily make MemoryContextContains return false

  6. Make MemoryContextContains work correctly again

  7. Make more effort to put a sentinel at the end of allocated memory

  8. Fix some possibly latent bugs in slab.c

  9. Various cleanups of the new memory context header code

  10. Revert "Add missing padding from MemoryChunk struct"

  11. Use MAXALIGN() in calculations using sizeof(SlabBlock)

  12. Add missing padding from MemoryChunk struct

  13. Improve performance of and reduce overheads of memory management