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: 2022-10-07T01:00:14Z
Lists: pgsql-hackers 
David Rowley <dgrowleyml@gmail.com> writes:
> On Fri, 7 Oct 2022 at 12:35, Tom Lane <tgl@sss.pgh.pa.us> wrote:
>> Which leaves me with the attached proposed wording.

> No objections here.

Cool, I'll push in a little bit.

> With these comments I'd be using slot MCTX_UNUSED4_ID first, then I'd
> probably be looking at MCTX_UNUSED5_ID after adjusting wipe_mem to do
> something other than setting bytes to 0x7F.

Well, the only way that you could free up a bitpattern that way is
to make wipe_mem use something ending in 000 or 001.  I'd be against
using 000 because then wiped memory might appear to contain valid
(aligned) pointers.  But perhaps 001 would be ok.

> I'd then use
> MCTX_UNUSED3_ID since that pattern is only used for larger chunks with
> glibc (per your findings).  After that, I'd probably start looking
> into making more than 3 bits available. If that wasn't possible, I'd
> be using MCTX_UNUSED2_ID and at last resort MCTX_UNUSED1_ID.

If we get to having three-quarters or seven-eighths of the bitpatterns
being valid IDs, we'll have precious little ability to detect garbage.
So personally I'd put "find a fourth bit" higher on the priority list.

In any case, we needn't invest more effort here until someone comes
with a fifth context method ... and I don't recall hearing discussions
of even a fourth one yet.

			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