Re: Buildfarm failures for hash indexes: buffer leaks

On 2018-11-19 14:38:04 -0800, Andres Freund wrote:
> Hi,
> 
> 
> On 2018-11-19 17:32:37 -0500, Tom Lane wrote:
> > Andres Freund <andres@anarazel.de> writes:
> > > FWIW, it seems that gcc's trunk works again. But I'm not sure this isn't
> > > just an accident and the optimization's introduced in the above revision
> > > aren't still broken.
> > > https://buildfarm.postgresql.org/cgi-bin/show_history.pl?nm=moonjelly&br=HEAD
> > 
> > Yeah, I saw that moonjelly had gone green.  Are you going to update
> > serinus' compiler soon?
> 
> I'm using debian's gcc-snapshot package, which updates a bit less
> frequently than what Fabien does. It'll auto-update as soon as that has
> happened.
> 
> What disconcerts me a bit is that afaict there wasn't an intentional fix
> in the relevant area (disabling value range propagation fixes the
> miscompilation), so I'm not quite sure.

Ooops, somehow deleted too much when editing the sentence. I was trying
to say that I'm not sure whether this is actually fixed, or a bug was
just hidden.

The issue is "fixed" in :
https://github.com/gcc-mirror/gcc/commit/48625f587d786089ce1e245f192bcf706f8fc1fc
and had been "caused" (or surfaced) in:
https://github.com/gcc-mirror/gcc/commit/be44111ed7ea7611d324d4d570e2f2d380f784b4

Greetings,

Andres Freund