[Bug target/65569] New: Fix for PR target/65240 causes a regression in building Spec for power4/power5/power6

[Bug target/65569] New: Fix for PR target/65240 causes a regression in building Spec for power4/power5/power6

[meissner at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=65569

            Bug ID: 65569
           Summary: Fix for PR target/65240 causes a regression in
                    building Spec for power4/power5/power6
           Product: gcc
           Version: 5.0
            Status: UNCONFIRMED
          Severity: major
          Priority: P3
         Component: target
          Assignee: unassigned at gcc dot gnu.org
          Reporter: meissner at gcc dot gnu.org
                CC: pthaugen at us dot ibm.com
              Host: powerpc64-unknown-linux-gnu
            Target: powerpc64-unknown-linux-gnu
             Build: powerpc64-unknown-linux-gnu

The fix for PR target/65240 causes a regression in building the dealII
benchmark for 64-bit power4/power5/power6 systems.

The reason is the fix eliminated support for floating point constants being in
RTL other than 0.0 on VSX systems.  The extenddftf2_fprs pattern expects to be
able to create a 0.0 to do the conversion, and for pre-VSX systems, you can't
create a 0.0 in a register without doing a load.

[Bug target/65569] Fix for PR target/65240 causes a regression in building Spec for power4/power5/power6

[meissner at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=65569

Michael Meissner <meissner at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|UNCONFIRMED                 |ASSIGNED
   Last reconfirmed|                            |2015-03-25
           Assignee|unassigned at gcc dot gnu.org      |meissner at gcc dot gnu.org
     Ever confirmed|0                           |1

[Bug target/65569] Fix for PR target/65240 causes a regression in building Spec for power4/power5/power6

[meissner at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=65569

--- Comment #1 from Michael Meissner <meissner at gcc dot gnu.org> ---
Created attachment 35139
  --> https://gcc.gnu.org/bugzilla/attachment.cgi?id=35139&action=edit
Polynomial.ii file to show the problem

[Bug target/65569] Fix for PR target/65240 causes a regression in building Spec for power4/power5/power6

[meissner at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=65569

--- Comment #2 from Michael Meissner <meissner at gcc dot gnu.org> ---
Created attachment 35141
  --> https://gcc.gnu.org/bugzilla/attachment.cgi?id=35141&action=edit
Proposed patch to fix the problem

[Bug target/65569] Fix for PR target/65240 causes a regression in building Spec for power4/power5/power6

[meissner at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=65569

--- Comment #3 from Michael Meissner <meissner at gcc dot gnu.org> ---
Author: meissner
Date: Thu Mar 26 04:07:21 2015
New Revision: 221683

URL: https://gcc.gnu.org/viewcvs?rev=221683&root=gcc&view=rev
Log:
2015-03-25  Michael Meissner  <meissner@linux.vnet.ibm.com>

    PR target/65569
    * config/rs6000/rs6000.md (extenddftf2_fprs): On VSX systems use
    XXLXOR to create 0.0.  On pre-VSX systems make sure the constant
    0.0 is correctly setup.
    (extenddftf2_internal): Likewise.


Modified:
    trunk/gcc/ChangeLog
    trunk/gcc/config/rs6000/rs6000.md

[Bug target/65569] Fix for PR target/65240 causes a regression in building Spec for power4/power5/power6

[meissner at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=65569

Michael Meissner <meissner at gcc dot gnu.org> changed:

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

--- Comment #4 from Michael Meissner <meissner at gcc dot gnu.org> ---
Fixed in subversion revision 221683.

[Bug target/65569] Fix for PR target/65240 causes a regression in building Spec for power4/power5/power6

[meissner at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=65569

--- Comment #5 from Michael Meissner <meissner at gcc dot gnu.org> ---
Author: meissner
Date: Thu Mar 26 04:15:08 2015
New Revision: 221684

URL: https://gcc.gnu.org/viewcvs?rev=221684&root=gcc&view=rev
Log:
Backport PR 65569 fix

Modified:
    branches/ibm/gcc-4_9-addr/gcc/ChangeLog.meissner
    branches/ibm/gcc-4_9-addr/gcc/config/rs6000/rs6000.md

[Bug target/65569] Fix for PR target/65240 causes a regression in building Spec for power4/power5/power6

[meissner at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=65569

Michael Meissner <meissner at gcc dot gnu.org> changed:

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

--- Comment #6 from Michael Meissner <meissner at gcc dot gnu.org> ---
Looking at the code generated for this as part of looking at PR target/67808,
it looks like the XXLXOR is no longer being generated on VSX systems.

[Bug target/65569] Fix for PR target/65240 causes a regression in building Spec for power4/power5/power6

[meissner at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=65569

Michael Meissner <meissner at gcc dot gnu.org> changed:

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

--- Comment #7 from Michael Meissner <meissner at gcc dot gnu.org> ---
I jumped the gun. I can't get it to fail, so this isn't a bug.