http://gcc.gnu.org/bugzilla/show_bug.cgi?id=59163
Jakub Jelinek changed:
What|Removed |Added
Status|ASSIGNED|RESOLVED
Resolution|---
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=59163
--- Comment #22 from Jakub Jelinek ---
Fixed.
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=59163
--- Comment #21 from Jakub Jelinek ---
Author: jakub
Date: Wed Dec 4 15:50:02 2013
New Revision: 205671
URL: http://gcc.gnu.org/viewcvs?rev=205671&root=gcc&view=rev
Log:
PR target/59163
* config/i386/i386.c (ix86_legitimate_combined_insn
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=59163
--- Comment #19 from Jakub Jelinek ---
Author: jakub
Date: Wed Dec 4 11:11:24 2013
New Revision: 205661
URL: http://gcc.gnu.org/viewcvs?rev=205661&root=gcc&view=rev
Log:
PR target/59163
* config/i386/i386.c (ix86_legitimate_combined_insn
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=59163
--- Comment #20 from Jakub Jelinek ---
Author: jakub
Date: Wed Dec 4 11:12:04 2013
New Revision: 205663
URL: http://gcc.gnu.org/viewcvs?rev=205663&root=gcc&view=rev
Log:
PR target/59163
* config/i386/i386.c (ix86_legitimate_combined_insn
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=59163
--- Comment #18 from Uroš Bizjak ---
(In reply to Jakub Jelinek from comment #17)
> Perhaps add new attribute ssememalign, with default 0, which would be
> (maximum for all alternatives) required alignment for memory operands in the
> instruction
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=59163
--- Comment #17 from Jakub Jelinek ---
Perhaps add new attribute ssememalign, with default 0, which would be (maximum
for all alternatives) required alignment for memory operands in the instruction
pre-AVX, or 0 for GET_MODE_ALIGNMENT. So, instru
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=59163
--- Comment #16 from Jakub Jelinek ---
Note that (according to my reading of the docs) e.g. movlps/movhps don't allow
unaligned memory, so blindly allow any combine is wrong, but while the MEM
operand in those cases is say V4SFmode, the loads or s
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=59163
--- Comment #15 from Jakub Jelinek ---
(In reply to Uroš Bizjak from comment #14)
> (In reply to Uroš Bizjak from comment #13)
> > (In reply to Jakub Jelinek from comment #12)
> > > Created attachment 31332 [details]
> > > gcc49-pr59163.patch
> >
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=59163
--- Comment #14 from Uroš Bizjak ---
(In reply to Uroš Bizjak from comment #13)
> (In reply to Jakub Jelinek from comment #12)
> > Created attachment 31332 [details]
> > gcc49-pr59163.patch
> >
> > So like this?
>
> Yes, with adjusted comment in
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=59163
--- Comment #13 from Uroš Bizjak ---
(In reply to Jakub Jelinek from comment #12)
> Created attachment 31332 [details]
> gcc49-pr59163.patch
>
> So like this?
Yes, with adjusted comment in ix86_legitimate_combined_insn.
IIRC, unaligned moves wo
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=59163
Jakub Jelinek changed:
What|Removed |Added
Attachment #31331|0 |1
is obsolete|
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=59163
--- Comment #11 from Uroš Bizjak ---
(In reply to Jakub Jelinek from comment #10)
> For stores I think the patch already allows that, that is the
> if (GET_CODE (*x) == SET && &SET_DEST (*x) == data)
> return 1;
> in there (the reason why I'
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=59163
--- Comment #10 from Jakub Jelinek ---
For stores I think the patch already allows that, that is the
if (GET_CODE (*x) == SET && &SET_DEST (*x) == data)
return 1;
in there (the reason why I've added it was that for the misaligned store insns
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=59163
--- Comment #9 from Uroš Bizjak ---
(In reply to Jakub Jelinek from comment #8)
> Created attachment 31331 [details]
> gcc49-pr59163.patch
>
> So like this? Untested...
Yes, but I think that we can also allow simple vector loads and stores - th
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=59163
--- Comment #8 from Jakub Jelinek ---
Created attachment 31331
--> http://gcc.gnu.org/bugzilla/attachment.cgi?id=31331&action=edit
gcc49-pr59163.patch
So like this? Untested...
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=59163
--- Comment #7 from Uroš Bizjak ---
Maybe even better idea is to use ix86_legitimate_combined_insn and reject
combinations that would result in unaligned operands of all but vector move
instructions.
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=59163
--- Comment #6 from Uroš Bizjak ---
I think that we should disallow tie of TImode with 128bit vector modes due to
different alignment requirements. Integer register pairs can load unaligned
TImode without problems, while unaligned TImode will cras
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=59163
Jakub Jelinek changed:
What|Removed |Added
CC||uros at gcc dot gnu.org
--- Comment #5 fr
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=59163
Richard Biener changed:
What|Removed |Added
Keywords||wrong-code
Target|
20 matches
Mail list logo
