This is the mail archive of the
gcc-bugs@gcc.gnu.org
mailing list for the GCC project.
Re: fold_truthop patch causes sparcv9 testsuite regressions
- From: Roger Sayle <roger at eyesopen dot com>
- To: Rainer Orth <ro at TechFak dot Uni-Bielefeld dot DE>
- Cc: "David S. Miller" <davem at redhat dot com>, <gcc-bugs at gcc dot gnu dot org>
- Date: Fri, 21 Jun 2002 16:41:53 -0600 (MDT)
- Subject: Re: fold_truthop patch causes sparcv9 testsuite regressions
On Fri, 21 Jun 2002, Rainer Orth wrote:
> Roger,
> this patch
>
> 2002-06-01 Roger Sayle <roger@eyesopen.com>
>
> * fold-const.c (fold_truthop): Transform "a || b" into "(a|b) != 0"
> and "!p && !q" into "(p|q) == 0" under suitable conditions.
>
> causes the sparc-sun-solaris2.8 -m64 (i.e. sparcv9) testsuite regressions
> described in http://gcc.gnu.org/ml/gcc-bugs/2002-06/msg00602.html
Thanks for bringing this to my attention. I've managed to get as
far as building the cc1 from a cross-compiler from i686-pc-linux-gnu
to sparc-sun-solaris2.8 and can confirm that 20000511-1.c when
compiled with "-m64" does mysteriously contain the line
".global .rem" with even though that symbol is never used.
Without "-m64" the assembler output also contains ".global .umul"
and ".global .div" which also aren't used.
I'll look into it further, but its difficult to believe that the
patch above was responsible. But stranger things have happened...
Roger
--
Roger Sayle, E-mail: roger@eyesopen.com
OpenEye Scientific Software, WWW: http://www.eyesopen.com/
Suite 1107, 3600 Cerrillos Road, Tel: (+1) 505-473-7385
Santa Fe, New Mexico, 87507. Fax: (+1) 505-473-0833