--- Comment #11 from rguenth at gcc dot gnu dot org 2009-03-15 11:55
---
Huh. Weird. I can reproduce this with
typedef int myint __attribute__((may_alias));
int foo(void *p)
{
myint *q = (int *)p;
return *q;
}
happens since 4.0. But only for the C frontend. - PR39464.
--
--- Comment #4 from pinskia at gmail dot com 2009-03-14 06:03 ---
Subject: Re: New: strange aliasing warnings compiling pymol under gcc 4.4
Sent from my iPhone
On Mar 13, 2009, at 9:11 PM, howarth at nitro dot med dot uc dot edu
gcc-bugzi...@gcc.gnu.org wrote:
There are a
--- Comment #5 from howarth at nitro dot med dot uc dot edu 2009-03-14
06:25 ---
Well, these two offending inlined routines do have the comments...
/* Only works with aligned 4-byte quantities, will cause a bus error */
/* on some platforms if used on unaligned data.
--- Comment #6 from rguenth at gcc dot gnu dot org 2009-03-14 09:32 ---
You can work around it in a gcc specific way by doing
typedef int my_aliased_int __attribute__((may_alias));
and using
my_aliased_int *data = (int *) v;
my_aliased_int *N;
This forces TBAA not to be applied
--- Comment #7 from rguenth at gcc dot gnu dot org 2009-03-14 09:32 ---
Invalid.
--
rguenth at gcc dot gnu dot org changed:
What|Removed |Added
--- Comment #8 from howarth at nitro dot med dot uc dot edu 2009-03-14
16:24 ---
Thanks. One last question. Is the presence of the 'does break' warning from
tree-ssa-structalias.c considered to be more likely to indicate actual code
breakage than the 'will break' warning from
--- Comment #9 from rguenth at gcc dot gnu dot org 2009-03-14 17:15 ---
In 4.5 this warning indicates that the load/store will be simply eliminated.
For
4.4 a miscompilation is less likely.
Note that this warning is designed to have zero false positives (so either you
discovered a
--- Comment #10 from howarth at nitro dot med dot uc dot edu 2009-03-14
22:50 ---
Interestingly the suggestion in Comment 6 introduces the new warning...
endianswap.h: In function 'swap4_aligned':
endianswap.h:113: warning: pointer targets in initialization differ in
signedness
--- Comment #1 from howarth at nitro dot med dot uc dot edu 2009-03-14
04:14 ---
Created an attachment (id=17462)
-- (http://gcc.gnu.org/bugzilla/attachment.cgi?id=17462action=view)
preprocessed source file for
pymol/contrib/uiuc/plugins/molfile_plugin/src/gromacsplugin.cpp
--
--- Comment #2 from howarth at nitro dot med dot uc dot edu 2009-03-14
04:15 ---
These odd aliasing warnings appear with...
g++-4 -I../../include -O3 -Wall -c -o gromacsplugin.o gromacsplugin.ii
--
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=39460
--- Comment #3 from howarth at nitro dot med dot uc dot edu 2009-03-14
04:18 ---
I am seeing this on...
Using built-in specs.
Target: i686-apple-darwin10
Configured with: ../gcc-4.4-20090313/configure --prefix=/sw
--prefix=/sw/lib/gcc4.4 --mandir=/sw/share/man --infodir=/sw/share/info
11 matches
Mail list logo
