This is the mail archive of the
gcc-patches@gcc.gnu.org
mailing list for the GCC project.
Re: PING: PATCH: middle-end/36701: [4.4 Regression] unaligned access in gcc.c-torture/execute/complex-7.c
- From: "H.J. Lu" <hjl dot tools at gmail dot com>
- To: "Joseph S. Myers" <joseph at codesourcery dot com>
- Cc: "Andrew Pinski" <pinskia at gmail dot com>, sje at cup dot hp dot com, gcc-patches at gcc dot gnu dot org
- Date: Tue, 2 Sep 2008 15:47:26 -0700
- Subject: Re: PING: PATCH: middle-end/36701: [4.4 Regression] unaligned access in gcc.c-torture/execute/complex-7.c
- References: <6dc9ffc80808060840m3cd0337bx4533b7d0f179d0fa@mail.gmail.com> <1218841014.744023.20838.nullmailer@hpsje.cup.hp.com> <de8d50360808151637g54c8b0ecu6d31e1f1880f30f@mail.gmail.com> <20080816034130.GA30155@lucon.org> <1219075713.440.1.camel@hpsje.cup.hp.com> <de8d50360809021535y7578bdb9r2e859f5816d4b203@mail.gmail.com> <Pine.LNX.4.64.0809022238360.2433@digraph.polyomino.org.uk>
On Tue, Sep 2, 2008 at 3:40 PM, Joseph S. Myers <joseph@codesourcery.com> wrote:
> On Tue, 2 Sep 2008, Andrew Pinski wrote:
>
>> On Mon, Aug 18, 2008 at 9:08 AM, Steve Ellcey <sje@cup.hp.com> wrote:
>> >
>> >>
>> >> If the stack temp size doesnt matter, this patch should work.
>> >>
>> >>
>> >> H.J.
>> >> ----
>> >> 2008-08-15 H.J. Lu <hongjiu.lu@intel.com>
>> >>
>> >> * expr.c (emit_group_store): Don't assert stack temp mode size.
>> >
>> > This patch worked for me. It fixed the problem and caused no
>> > regressions on ia64-hp-hpux11.23.
>>
>> This patch just changed the failure from being an ICE to wrong code on spu-elf:
>> FAIL: gcc.dg/compat/struct-by-value-11
>> c_compat_x_tst.o-c_compat_y_tst.o execute
>
> On SPARC64 we had wrong code before HJ's first patch, then ICEs, so if
> it's back to wrong code (and I haven't tested trunk for SPARC64 since this
> patch went in) then that's not a regression from HJ's patches, just a
> wrong code failure that still needs fixing.
>
That is what I was afraid of. I added the assert to prevent wrong
code. But my processors never trigger that assert. When looking
at the code, remove assert seems OK. But I can't verify it myself.
I think something is still missing and it is back-end specific.
--
H.J.