Re: [HACKERS] v6.3 snapshot core dump - Mailing list pgsql-hackers

From Bruce Momjian
Subject Re: [HACKERS] v6.3 snapshot core dump
Date
Msg-id 199802131825.NAA14770@candle.pha.pa.us
Whole thread Raw
In response to v6.3 snapshot core dump  (Brett McCormick <brett@techno.chicken.org>)
List pgsql-hackers
I will look into this.  I made many fixes in 6.3 for aggregates and
views, but I did not test views of aggregates.  I will do that soon.

Good test case.

>
>
> this is the latest snapshot under linux 2.0.33
>
> when I create a view with an aggregate such as count (with a group by
> clause) I get a core dump.  The first three lines of the gdb backtrace
> follows:
>
> #0  0x80a9982 in ExecEvalAggreg (agg=0x820fa20, econtext=0x8213a48,
>     isNull=0xbfffd6cf "") at execQual.c:201
> #1  0x80ab271 in ExecEvalExpr (expression=0x820fa20, econtext=0x8213a48,
>     isNull=0xbfffd6cf "", isDone=0xbfffd7c3 "\001\fZ\031\b\030K!\b")
>     at execQual.c:1199
> #2  0x80ab69b in ExecTargetList (targetlist=0x8213498, nodomains=2,
>     targettype=0x8214990, values=0x8214b48, econtext=0x8213a48,
>     isDone=0xbfffd7c3 "\001\fZ\031\b\030K!\b") at execQual.c:1484
>
> code at point:
>
> 201             *isNull = econtext->ecxt_nulls[agg->aggno];
>
> econtext->ecxt_nulls, is, well, a NULL pointer and hard to index :)
> i'll take a look but I don't think I'll find it -- i notice that 6.2.1
> pretends to make a view with aggregates but then it comes up empty
> when selected.
>
> --brett
> let me know if you need more info
>
>


--
Bruce Momjian
maillist@candle.pha.pa.us

pgsql-hackers by date:

Previous
From: ocie@paracel.com
Date:
Subject: Re: [HACKERS] Subselects are in CVS...
Next
From: Bruce Momjian
Date:
Subject: Re: [HACKERS] CREATE FUNCTION broken