[Bug middle-end/60192] New: Test case gcc.dg/tree-ssa/sra-12.c fails on ARM

[Bug middle-end/60192] New: Test case gcc.dg/tree-ssa/sra-12.c fails on ARM

[bernd.edlinger at hotmail dot de]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=60192

            Bug ID: 60192
           Summary: Test case gcc.dg/tree-ssa/sra-12.c fails on ARM
           Product: gcc
           Version: 4.9.0
            Status: UNCONFIRMED
          Severity: normal
          Priority: P3
         Component: middle-end
          Assignee: unassigned at gcc dot gnu.org
          Reporter: bernd.edlinger at hotmail dot de

FAIL: gcc.dg/tree-ssa/sra-12.c scan-tree-dump-times release_ssa "l;" 0

$ gcc -v
Using built-in specs.
COLLECT_GCC=gcc
COLLECT_LTO_WRAPPER=/home/ed/gnu/arm-linux-gnueabihf/libexec/gcc/armv7l-unknown-linux-gnueabihf/4.9.0/lto-wrapper
Target: armv7l-unknown-linux-gnueabihf
Configured with: ../gcc-4.9-20140209/configure
--prefix=/home/ed/gnu/arm-linux-gnueabihf
--enable-languages=c,c++,objc,obj-c++,fortran,ada,go --with-arch=armv7-a
--with-tune=cortex-a9 --with-fpu=vfpv3-d16 --with-float=hard
Thread model: posix
gcc version 4.9.0 20140209 (experimental) (GCC)

$ cat ../gcc-4.9-20140209/LAST_UPDATED 
Obtained from SVN: trunk revision 207646

cat  sra-12.c.036t.release_ssa 

;; Function foo (foo, funcdef_no=0, decl_uid=4136, symbol_order=0)

Released 2 names, 22.22%, removed 2 holes
foo (struct S * p)
{
  int l$i;
  struct S l;
  int _4;

  <bb 2>:
  l = *p_2(D);
  l$i_5 = MEM[(struct S *)p_2(D)];
  _4 = l$i_5 + 1;
  MEM[(struct S *)&l] = _4;
  *p_2(D) = l;
  l ={v} {CLOBBER};
  return;

}


Test fails because "l;" is found in above file.

[Bug middle-end/60192] Test case gcc.dg/tree-ssa/sra-12.c fails on ARM

[rguenth at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=60192

--- Comment #1 from Richard Biener <rguenth at gcc dot gnu.org> ---
Well, then SRA didn't work.  The aggregate copy should have been elided and
the local 'l' should have been removed:

Released 11 names, 183.33%, removed 11 holes
foo (struct S * p)
{
  int l$i;
  int _3;

  <bb 2>:
  l$i_4 = MEM[(struct S *)p_2(D)];
  _3 = l$i_4 + 1;
  MEM[(struct S *)p_2(D)] = _3;
  return;

}

(from x86_64)

Maybe some cost oddity on arm prevents this?

[Bug middle-end/60192] Test case gcc.dg/tree-ssa/sra-12.c fails on ARM

[ramana at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=60192

Ramana Radhakrishnan <ramana at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|UNCONFIRMED                 |NEW
   Last reconfirmed|                            |2014-03-13
                 CC|                            |ramana at gcc dot gnu.org
     Ever confirmed|0                           |1

--- Comment #2 from Ramana Radhakrishnan <ramana at gcc dot gnu.org> ---
Confirmed - Kyrill worked out that it was one of the MOVE_COSTS IIRC for this
purpose.

[Bug testsuite/60192] Test case gcc.dg/tree-ssa/sra-12.c fails on ARM

[pinskia at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=60192

Andrew Pinski <pinskia at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
         Resolution|---                         |FIXED
             Status|NEW                         |RESOLVED
   Target Milestone|---                         |6.0
          Component|middle-end                  |testsuite

--- Comment #3 from Andrew Pinski <pinskia at gcc dot gnu.org> ---
Fixed by r6-4105.