Josh Kupershmidt <schmiddy@gmail.com> writes:
> On Wed, Nov 30, 2011 at 9:17 PM, Tom Lane <tgl@sss.pgh.pa.us> wrote:
>> Can you try that on 9.1 branch tip to see if it's already fixed?
> Hrm, don't think that helped - I get the same error in the logs using
> a checkout of branch REL9_1_STABLE.
Well, was worth a try, although the stack trace didn't look very close
to any of the GIN or hot-standby fixes that I can remember offhand.
Not sure what to make of it. If the stack trace can be believed,
it's crashing in one of the memsets in ReadBuffer_common() ... which
ought to be impossible, because they're fixed-length and always pointed
at a buffer.
It might be worth recompiling at -O0, first to see if that changes the
behavior and second to see if it changes the reported stack trace.
regards, tom lane