Re: [PATCH] Fix -ftree-ter -O0 (PR tree-optimization/43655)

[Jeff Law <law at redhat dot com>]

On 12/14/10 02:41, Jakub Jelinek wrote:
> Hi!
>
> In 4.4 gimple_references_memory_p checked a bit in the stmt, but starting
> with 4.5 that inline just looks at vuses.  Unfortunately, at -O0 there
> are no vuses and thus is_replaceable_p at -O0 -ftree-ter happily moves
> memory loads around.
>
> Fixed thusly, bootstrapped/regtested on x86_64-linux and i686-linux, ok for
> trunk?
>
> 2010-12-14  Jakub Jelinek<jakub@redhat.com>
>
> 	PR tree-optimization/43655
> 	* tree-ssa-ter.c (is_replaceable_p): Don't use
> 	gimple_references_memory_p for -O0, instead check for load
> 	by looking at rhs.
>
> 	* g++.dg/opt/pr43655.C: New test.
I'm a little confused as to why TER is running when we're not 
optimizing.    There's no doubt more bugs of this nature lying around if 
we allow specific optimizers to be turned on at -O0.

If we're going to allow optimizers to run at -O0, wouldn't it be better 
to fix gimple_references_memory_p rather than have to twiddle each of 
its callers to DTRT at -O0.  At least one of the other users of 
gimple_references_memory_p seems to be prone to a similar bug 
(tree-tailcall.c).

Jeff