This is the mail archive of the
gcc@gcc.gnu.org
mailing list for the GCC project.
Re: The reincarnation of PR15242
- From: Andrew Haley <aph at redhat dot com>
- To: Ian Lance Taylor <iant at google dot com>
- Cc: gcc at gcc dot gnu dot org, pinskia at gcc dot gnu dot org
- Date: Wed, 09 Sep 2009 10:17:46 +0100
- Subject: Re: The reincarnation of PR15242
- References: <4AA530D2.8070706@redhat.com> <m3bpllxhj6.fsf@google.com>
Ian Lance Taylor wrote:
> Andrew Haley <aph@redhat.com> writes:
>
>> This seems to be an an old bug that has come back. We're generating
>>
>> L1210:
>> jmp *%eax
>> .L4:
>> .L5:
>> ...
>> jmp .L1210
>> .L1171:
>> .L1172:
>> ...
>> jmp .L1210
>> .L1168:
>> .L1169:
>> ...
>> jmp .L1210
>>
>>
>> instead of
>>
>> .L1210:
>> jmp *%eax
>> .L4:
>> .L5:
>> ...
>> jmp *%eax
>> .L1171:
>> .L1172:
>> ...
>> jmp *%eax
>> .L1168:
>> .L1169:
>> ...
>> jmp *%eax
>>
>> Current gcc trunk, x86, gcc -O3 ef.i -S -m32.
>>
>> Attachment in old Bugzilla entry at
>> http://gcc.gnu.org/bugzilla/attachment.cgi?id=6206&action=view
>
>
> We should have something like this in the testsuite.
I've been thinking about that. How would one go about writing a testcase?
I guess we're looking for a lot of jmp instrs all bracnhing to the same
label.
Andrew.