On Fri, May 09, 2014 at 04:58:54PM +0200, Andres Freund wrote:
> On 2014-05-09 10:49:09 -0400, Robert Haas wrote:
> > > Patch 03: Add valgrind suppression for writing out padding bytes. That's
> > > better than zeroing the data from the get go because unitialized
> > > accesses are still detected.
> >
> > I have no understanding of valgrind suppressions. I can commit this
> > blindly if nobody else wants to pick it up.
>
> I think the only committer with previous experience in that area is
> Noah. Noah?
I can pick up that patch.
Static functions having only one call site are especially vulnerable to
inlining, so avoid naming them in the suppressions file. I do see
ReorderBufferSerializeChange() inlined away at -O2 and higher. Is it fair to
tie the suppression to ReorderBufferSerializeTXN() instead?
Do you happen to have a self-contained procedure for causing the server to
reach the code in question?
--
Noah Misch
EnterpriseDB http://www.enterprisedb.com