[Bug target/55276] New: [4.8 regression] ppc: callee-saved vector registers not preserved

[Bug target/55276] New: [4.8 regression] ppc: callee-saved vector registers not preserved

[mans at mansr dot com]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=55276

             Bug #: 55276
           Summary: [4.8 regression] ppc: callee-saved vector registers
                    not preserved
    Classification: Unclassified
           Product: gcc
           Version: 4.8.0
            Status: UNCONFIRMED
          Severity: normal
          Priority: P3
         Component: target
        AssignedTo: unassigned@gcc.gnu.org
        ReportedBy: mans@mansr.com


Created attachment 28662
  --> http://gcc.gnu.org/bugzilla/attachment.cgi?id=28662
Preserve callee-saved vector registers even when vrsave is not in use

Since r191986 the callee-saved vector registers v20-v31 are not preserved when
TARGET_ALTIVEC_VRSAVE is unset.  The attached patch fixes this.

[Bug target/55276] [4.8 regression] ppc: callee-saved vector registers not preserved

[mans at mansr dot com]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=55276

--- Comment #1 from Mans Rullgard <mans at mansr dot com> 2012-11-11 21:07:12 UTC ---
Created attachment 28663
  --> http://gcc.gnu.org/bugzilla/attachment.cgi?id=28663
Test case

[Bug target/55276] [4.8 regression] ppc: callee-saved vector registers not preserved

[dje at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=55276

David Edelsohn <dje at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|UNCONFIRMED                 |NEW
   Last reconfirmed|                            |2012-11-19
     Ever Confirmed|0                           |1

--- Comment #2 from David Edelsohn <dje at gcc dot gnu.org> 2012-11-19 15:40:05 UTC ---
The patch looks correct, but the testcase only demonstrates the problem by
looking at assembly, not by checking for correct results.

[Bug target/55276] [4.8 regression] ppc: callee-saved vector registers not preserved

[dje at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=55276

--- Comment #3 from David Edelsohn <dje at gcc dot gnu.org> 2012-11-19 16:58:42 UTC ---
Author: dje
Date: Mon Nov 19 16:58:31 2012
New Revision: 193628

URL: http://gcc.gnu.org/viewcvs?root=gcc&view=rev&rev=193628
Log:
2012-11-19  Mans Rullgard  <mans@mansr.com>

        PR target/55276
        * config/rs6000/rs6000.c (rs6000_stack_info): Always set vrsave_mask
        for TARGET_ALTIVEC_ABI.  Zero vrsave_save_offset if
        !TARGET_ALTIVEC_VRSAVE.
        (rs6000_emit_prologue): For SAVE_INLINE_VLRs, check vrsave_size
        not vrsave_mask.

Modified:
    trunk/gcc/ChangeLog

[Bug target/55276] [4.8 regression] ppc: callee-saved vector registers not preserved

[dje at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=55276

David Edelsohn <dje at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|NEW                         |WAITING

--- Comment #4 from David Edelsohn <dje at gcc dot gnu.org> 2012-11-19 18:16:32 UTC ---
I committed the patch, but this should have a testcase as well.

[Bug target/55276] [4.8 regression] ppc: callee-saved vector registers not preserved

[rguenth at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=55276

Richard Biener <rguenth at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
   Target Milestone|---                         |4.8.0

[Bug target/55276] [4.8 regression] ppc: callee-saved vector registers not preserved

[rguenth at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=55276

Richard Biener <rguenth at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|WAITING                     |RESOLVED
         Resolution|                            |FIXED

--- Comment #5 from Richard Biener <rguenth at gcc dot gnu.org> 2012-12-07 09:43:36 UTC ---
Fixed anyway.