[Bug target/55565] New: [4.8 regression] FAIL: gcc.target/powerpc/ppc-mov-1.c scan-assembler-not fmr [0-9]+,[0-9]+

[Bug target/55565] New: [4.8 regression] FAIL: gcc.target/powerpc/ppc-mov-1.c scan-assembler-not fmr [0-9]+,[0-9]+

[schwab@linux-m68k.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=55565

             Bug #: 55565
           Summary: [4.8 regression] FAIL: gcc.target/powerpc/ppc-mov-1.c
                    scan-assembler-not fmr [0-9]+,[0-9]+
    Classification: Unclassified
           Product: gcc
           Version: 4.8.0
            Status: UNCONFIRMED
          Severity: normal
          Priority: P3
         Component: target
        AssignedTo: unassigned@gcc.gnu.org
        ReportedBy: schwab@linux-m68k.org
            Target: powerpc64-*-*


Between r190317 and r191083 the test gcc.target/powerpc/ppc-mov-1.c started to
fail on powerpc64.

[Bug target/55565] [4.8 regression] FAIL: gcc.target/powerpc/ppc-mov-1.c scan-assembler-not fmr [0-9]+,[0-9]+

[schwab@linux-m68k.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=55565

Andreas Schwab <schwab@linux-m68k.org> changed:

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

[Bug target/55565] [4.8 regression] FAIL: gcc.target/powerpc/ppc-mov-1.c scan-assembler-not fmr [0-9]+,[0-9]+

[rguenth at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=55565

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

           What    |Removed                     |Added
----------------------------------------------------------------------------
           Priority|P3                          |P1

[Bug target/55565] [4.8 regression] FAIL: gcc.target/powerpc/ppc-mov-1.c scan-assembler-not fmr [0-9]+,[0-9]+

[jakub at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=55565

Jakub Jelinek <jakub at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |jakub at gcc dot gnu.org

--- Comment #1 from Jakub Jelinek <jakub at gcc dot gnu.org> 2012-12-18 14:53:52 UTC ---
Started with http://gcc.gnu.org/viewcvs?root=gcc&view=rev&rev=190339
which changed *.optimized dump tiny bit.
I'd say the testcase is questionable, on such large routine assuming there
can't be any floating point register moves is dubious.

[Bug target/55565] [4.8 regression] FAIL: gcc.target/powerpc/ppc-mov-1.c scan-assembler-not fmr [0-9]+,[0-9]+

[dominiq at lps dot ens.fr]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=55565

Dominique d'Humieres <dominiq at lps dot ens.fr> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|UNCONFIRMED                 |NEW
   Last reconfirmed|                            |2012-12-20
     Ever Confirmed|0                           |1

--- Comment #2 from Dominique d'Humieres <dominiq at lps dot ens.fr> 2012-12-20 22:57:37 UTC ---
BTW the test passes on powerpc-apple-darwin9 because the regexpr does not catch

    fmr f1,f0

[Bug target/55565] [4.8 regression] FAIL: gcc.target/powerpc/ppc-mov-1.c scan-assembler-not fmr [0-9]+,[0-9]+

[aldyh at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=55565

Aldy Hernandez <aldyh at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |aldyh at gcc dot gnu.org
         AssignedTo|unassigned at gcc dot       |aldyh at gcc dot gnu.org
                   |gnu.org                     |

--- Comment #3 from Aldy Hernandez <aldyh at gcc dot gnu.org> 2013-01-09 19:53:13 UTC ---
I'll take a look.

[Bug target/55565] [4.8 regression] FAIL: gcc.target/powerpc/ppc-mov-1.c scan-assembler-not fmr [0-9]+,[0-9]+

[aldyh at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=55565

--- Comment #4 from Aldy Hernandez <aldyh at gcc dot gnu.org> 2013-01-10 17:46:39 UTC ---
I compared the code generated by trunk with the generated code in rev 190339
which broke the test.  The trunk code is more optimal than when the test
"passed", so I suggest either removing the dg-final in the test, or modifying
the regex in the dg-final.

In the testcase we have something functionally equivalent to (where x is the
original function argument passed in f1):

  if (huge + x > 0.0) {
    if ((int)x > 0)
      return 0.0;
    ...
  } else
    return x;

When the test passed, we used to generate:

.L17:
        ld %r8,.LC0@toc(%r2)
        ld %r10,.LC2@toc(%r2)
        lfd %f13,0(%r8)
        lfd %f0,0(%r10)
        fadd %f1,%f1,%f13
        fcmpu %cr7,%f1,%f0
        bng %cr7,.L4           /* if (!(huge + x > 0.0)) return x; */
        ...
        ...

.L4:                           /* return x; */
        std %r9,-8(%r1)
        ori 2,2,0
        lfd %f1,-8(%r1)
        blr

Notice that since we perform the fadd onto %f1, we need to reload %f1 from
memory.

On trunk though, we keep x/%f1 and 0.0/%f0 around so we can return them
directly.  We can return x/%f1 directly in one case, or copy 0.0/%f0 to the
return register (%f1) in the another case.

.L18:
        ld %r8,.LC6@toc(%r2)
        ld %r9,.LC1@toc(%r2)
        lfd %f13,0(%r8)
        lfd %f0,0(%r9)
        fadd %f13,%f1,%f13
        fcmpu %cr7,%f13,%f0
        bnglr %cr7            /* if (!(huge + x > 0.0)) return x; */
        cmpdi %cr7,%r10,0
        fmr %f1,%f0
        bgelr %cr7            /* if (huge+x>0.0 && (int)x >0) return 0.0 */

Bottom line, on trunk we avoid a branch and memory load/stores.

Testing a patch.

[Bug target/55565] [4.8 regression] FAIL: gcc.target/powerpc/ppc-mov-1.c scan-assembler-not fmr [0-9]+,[0-9]+

[pinskia at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=55565

--- Comment #5 from Andrew Pinski <pinskia at gcc dot gnu.org> 2013-01-10 17:53:24 UTC ---
(In reply to comment #4)
> Bottom line, on trunk we avoid a branch and memory load/stores.

I agree the code is much better now.  Only moving between the fpr and gpr where
needed instead of keeping x into the GPRs.

[Bug target/55565] [4.8 regression] FAIL: gcc.target/powerpc/ppc-mov-1.c scan-assembler-not fmr [0-9]+,[0-9]+

[aldyh at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=55565

--- Comment #6 from Aldy Hernandez <aldyh at gcc dot gnu.org> 2013-01-10 20:28:33 UTC ---
Author: aldyh
Date: Thu Jan 10 20:28:26 2013
New Revision: 195097

URL: http://gcc.gnu.org/viewcvs?root=gcc&view=rev&rev=195097
Log:
    PR target/55565
    * gcc.target/powerpc/ppc-mov-1.c: Update scan-assembler-not
    regex.

Modified:
    trunk/gcc/testsuite/ChangeLog
    trunk/gcc/testsuite/gcc.target/powerpc/ppc-mov-1.c

[Bug target/55565] [4.8 regression] FAIL: gcc.target/powerpc/ppc-mov-1.c scan-assembler-not fmr [0-9]+,[0-9]+

[aldyh at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=55565

Aldy Hernandez <aldyh at gcc dot gnu.org> changed:

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

--- Comment #7 from Aldy Hernandez <aldyh at gcc dot gnu.org> 2013-01-10 20:29:41 UTC ---
Fixed in trunk.