On Fri, 9 Sept 2022 at 10:53, Tom Lane <tgl@sss.pgh.pa.us> wrote:
> I hate to give up MemoryContextContains altogether. The assertions
> that David nuked in b76fb6c2a had some value I think,
Those can be put back if we decide to keep MemoryContextContains.
Those newly added Asserts just temporarily had to go due to b76fb6c2a
making MemoryContextContains temporarily always return false.
> The implementation I suggested upthread would reliably distinguish
> malloc from palloc, and while it is potentially a tad expensive
> I don't think it's too much so for Assert checks. I don't have an
> objection to trying to get to a place where we only use it in
> Assert, though.
I really think the Assert only form of MemoryContextContains() is the
best move, and if it's doing Assert only, then we can do the
loop-over-the-blocks idea as you described and I drafted in [1].
If the need comes up that we're certain we always have a pointer to
some allocated chunk, but need to know if it's in some memory context,
then the proper form of expressing that, I think, should be:
if (GetMemoryChunkContext(pointer) == somecontext)
If we're worried about getting that wrong, we can beef up the
MemoryChunk struct with a magic_number field in
MEMORY_CONTEXT_CHECKING builds to ensure we catch any code which
passes invalid pointers.
David
[1] https://www.postgresql.org/message-id/CAApHDvoKjOmPQeokicwDuO-_Edh=tKp23-=jskYcyKfw5QuDhA@mail.gmail.com