Re: changeset generation v5-01 - Patches & git tree

On 2013-07-05 14:03:56 +0200, Andres Freund wrote:
> On 2013-06-27 21:52:03 -0700, Kevin Grittner wrote:
> > Andres Freund  wrote:
> > 
> > > Hm. There were some issues with the test_logical_decoding
> > > Makefile not cleaning up the regression installation properly.
> > > Which might have caused the issue.
> > >
> > > Could you try after applying the patches and executing a clean
> > > and then rebuild?
> > 
> > Tried, and problem persists.
> > 
> > > Otherwise, could you try applying my git tree so we are sure we
> > > test the same thing?
> > >
> > > $ git remote add af git://git.postgresql.org/git/users/andresfreund/postgres.git
> > > $ git fetch af
> > > $ git checkout -b xlog-decoding af/xlog-decoding-rebasing-cf4
> > > $ ./configure ...
> > > $ make
> > 
> > Tried that, too, and problem persists.  The log shows the last
> > commit on your branch as 022c2da1873de2fbc93ae524819932719ca41bdb.
> 
> Ok. I think I have a slight idea what's going on. Could you check
> whether recompiling with -O0 "fixes" the issue?
> 
> There's something strange going on here, not sure whether it's just a
> bug that's hidden, by either not doing optimizations or by adding more
> elog()s, or wheter it's a compiler bug.

Ok. It was supreme stupidity on my end. Sorry for the time you spent on
it.

Some versions of gcc (and probably other compilers) were removing
sections of code when optimizing because the code was doing undefined
things. Parts of the rdata chain were allocated locally in an
if (needs_key). Which obviously is utterly bogus... A warning would have
been nice though.

Fix pushed and attached.

Greetings,

Andres Freund

-- 
 Andres Freund	                   http://www.2ndQuadrant.com/
 PostgreSQL Development, 24x7 Support, Training & Services