Re: SRF memory mgmt patch (was [HACKERS] Concern about

Joe Conway wrote:
> Unfortunately I have not been able to get back to a point where I see a
> coredump :(. But, that did seem to be related to calling the function
> with an inappropriate declaration (now it just gives me garbage instead
> of dumping core, even though I reverted the per_MultiFuncCall changes I
> made earlier). I'll keep messing with this for a while, but I was hoping
> the attached info would lead to some more suggestions of where to be
> looking.

It's back as of cvs tip. This time, it looks like all table functions
are failing in the same manner:

#4  0x081845fb in pfree (pointer=0x7f7f7f7f) at mcxt.c:470
#5  0x0806bdb2 in heap_freetuple (htup=0x82fc7b8) at heaptuple.c:736
#6  0x080e4cbf in ExecClearTuple (slot=0x82f92f4) at execTuples.c:406
#7  0x0817d3ad in per_MultiFuncCall (fcinfo=0xbfffe9e0) at funcapi.c:88
#8  0x0814af25 in pg_lock_status (fcinfo=0xbfffe9e0) at lockfuncs.c:69
#9  0x080e3990 in ExecMakeTableFunctionResult (funcexpr=0x82e9fa0,

#4  0x081845fb in pfree (pointer=0x7f7f7f7f) at mcxt.c:470
#5  0x0806bdb2 in heap_freetuple (htup=0x82f43a4) at heaptuple.c:736
#6  0x080e4cbf in ExecClearTuple (slot=0x82e9f2c) at execTuples.c:406
#7  0x0817d3ad in per_MultiFuncCall (fcinfo=0xbfffe9e0) at funcapi.c:88
#8  0x40016a4b in dblink_record (fcinfo=0xbfffe9e0) at dblink.c:518
#9  0x080e3990 in ExecMakeTableFunctionResult (funcexpr=0x82e8df8,

#4  0x081845fb in pfree (pointer=0x7f7f7f7f) at mcxt.c:470
#5  0x0806bdb2 in heap_freetuple (htup=0x83026f8) at heaptuple.c:736
#6  0x080e4cbf in ExecClearTuple (slot=0x82f71cc) at execTuples.c:406
#7  0x0817d3ad in per_MultiFuncCall (fcinfo=0xbfffe9e0) at funcapi.c:88
#8  0x08181635 in show_all_settings (fcinfo=0xbfffe9e0) at guc.c:2469
#9  0x080e3990 in ExecMakeTableFunctionResult (funcexpr=0x82f64a0,

Currently all C language table functions are broken :(, but all sql
language table functions seem to work -- which is why regression doesn't
fail (pointing out the need to add a select * from pg_settings to a
regression test somewhere).

I'm looking at this now. I suspect the easy fix is to remove
ExecClearTuple from per_MultiFuncCall, but I'll try to understand what's
going on first.

Joe