Re: BUG #6222: Segmentation fault on unlogged table

On Mon, Sep 26, 2011 at 11:08 AM, Robert Haas <robertmhaas(at)gmail(dot)com> wrote:
> On Mon, Sep 26, 2011 at 11:40 AM, Robert Haas <robertmhaas(at)gmail(dot)com> wrote:
>> To check my work, I did this:
>>
>> --- a/src/backend/executor/execQual.c
>> +++ b/src/backend/executor/execQual.c
>> @@ -5003,6 +5003,7 @@ ExecQual(List *qual, ExprContext *econtext, bool
>> resultForNull)
>>                Datum           expr_value;
>>                bool            isNull;
>>
>> +               Assert(!IsA(clause, List));
>>                expr_value = ExecEvalExpr(clause, econtext, &isNull, NULL);
>>
>>                if (isNull)
>>
>> And in fact the test case (when run against the unlogged tables) fails
>> that assertion:
>>
>> TRAP: FailedAssertion("!(!((((Node*)(clause))->type) == T_List))",
>> File: "execQual.c", Line: 5006)
>>
>> Now I'm not too sure why that is happening yet, but I'm leaning toward
>> the idea that the bug here is timing-related and that unlogged tables
>> aren't the cause, but rather just make it easier to hit whatever the
>> race condition is by removing some overhead.
>
> I cannot reproduce this on commit
> e6faf910d75027bdce7cd0f2033db4e912592bcc.  But on the very next commit
> I can:
>
> commit e6faf910d75027bdce7cd0f2033db4e912592bcc
> Author: Tom Lane <tgl(at)sss(dot)pgh(dot)pa(dot)us>
> Date:   Fri Sep 16 00:42:53 2011 -0400
>
>    Redesign the plancache mechanism for more flexibility and efficiency.
>
> Tom, any thoughts?

hm. any relation to YAMAMOTO Takashi's recent email, [BUGS] bug in
plancache.c, which hasn't hit the archives yet?

> "GetCachedPlan can pass the 'qlist' to the planner twice.
if i understand the code correctly, it's unsafe because the planner is
destructive wrt the input tree. for my application, it often causes
a crash in executor."

merlin