http://gcc.gnu.org/bugzilla/show_bug.cgi?id=60116
--- Comment #11 from Uroš Bizjak ---
OK, confirmed using "-Os -m32" with
gcc version 4.9.0 20140207 (experimental) [trunk revision 207597] (GCC)
on CentOS 5.10.
It looks like a fault in the combine pass.
Before combine, we have:
...
(insn 14
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=60116
Marc Glisse changed:
What|Removed |Added
Attachment #32088|0 |1
is obsolete|
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=60116
--- Comment #9 from Marc Glisse ---
Extract from a gdb session, showing 2 wrong branches taken:
(gdb) p f
$15 = 1
(gdb) step
15 j = k * 54;
(gdb)
16 g = j * 147;
(gdb)
18 if (d)
(gdb) p d
$16 = 1
(gdb) step
21
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=60116
--- Comment #8 from Marc Glisse ---
Er, right, my attachment is useless, it is the same as Uros' (but I can't see
how to remove it). If it matters:
GNU assembler version 2.24 (x86_64-linux-gnu) using BFD version (GNU Binutils
for Debian) 2.24
GNU
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=60116
--- Comment #7 from Uroš Bizjak ---
(In reply to Marc Glisse from comment #5)
> Created attachment 32088 [details]
> asm that prints 0
Well,
$ gcc -m32 u.s && ./a.out
1
as --version
GNU assembler version 2.23.2
Copyright 2012 Free Software Foun
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=60116
--- Comment #6 from Marek Polacek ---
I can reproduce 0 even with the assembly by Uros:
$ ./xgcc -B./ -m32 -xassembler Uok.s; ./a.out
0
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=60116
--- Comment #5 from Marc Glisse ---
Created attachment 32088
--> http://gcc.gnu.org/bugzilla/attachment.cgi?id=32088&action=edit
asm that prints 0
Strange, I can reproduce it easily with r207643 on Debian testing:
stedding /tmp $ /tmp/gcc-mult
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=60116
Jakub Jelinek changed:
What|Removed |Added
CC||jakub at gcc dot gnu.org
--- Comment #4 f
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=60116
--- Comment #3 from Uroš Bizjak ---
Created attachment 32086
--> http://gcc.gnu.org/bugzilla/attachment.cgi?id=32086&action=edit
assembly dump from trunk revision 207643
I'm not able to trigger this problem with trunk revision 207643:
$ /ssd/u
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=60116
Marek Polacek changed:
What|Removed |Added
CC||mpolacek at gcc dot gnu.org,
10 matches
Mail list logo
