Horton

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

David Rowley <dgrowleyml@gmail.com>

View thread View on postgr.es
From: David Rowley <dgrowleyml@gmail.com>
To: Tom Lane <tgl@sss.pgh.pa.us>
Cc: Robert Haas <robertmhaas@gmail.com>, Amit Kapila <amit.kapila16@gmail.com>, Andres Freund <andres@anarazel.de>, Yura Sokolov <y.sokolov@postgrespro.ru>, PostgreSQL Developers <pgsql-hackers@lists.postgresql.org>
Date: 2022-08-30T15:31:24Z
Lists: pgsql-hackers 
On Wed, 31 Aug 2022 at 01:36, Tom Lane <tgl@sss.pgh.pa.us> wrote:
>
> David Rowley <dgrowleyml@gmail.com> writes:
> > I think the Assert is useful as if we were ever to add an enum member
> > with the value of 8 and forgot to adjust MEMORY_CONTEXT_METHODID_BITS
> > then bad things would happen inside MemoryChunkSetHdrMask() and
> > MemoryChunkSetHdrMaskExternal().  I think it's unlikely we'll ever get
> > that many MemoryContext types, but I don't know for sure and would
> > rather the person who adds the 9th one get alerted to the lack of bit
> > space in MemoryChunk as soon as possible.
>
> I think that's a weak argument, so I don't mind dropping this Assert.
> What would be far more useful is a comment inside the
> MemoryContextMethodID enum pointing out that we can support at most
> 8 values because XYZ.

I'd just sleep better knowing that we have some test coverage to
ensure that MemoryChunkSetHdrMask() and
MemoryChunkSetHdrMaskExternal() have some verification that we don't
end up with future code that will cause the hdrmask to be invalid.  I
tried to make those functions as lightweight as possible. Without the
Assert, I just feel that there's a bit too much trust that none of the
bits overlap.  I've no objections to adding a comment to the enum to
explain to future devs. My vote would be for that and add the (int)
cast as proposed by Robert.

David

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