[Bug tree-optimization/49175] New: vect/pr48172.c execution failure for ARM GNU/Linux

[Bug tree-optimization/49175] New: vect/pr48172.c execution failure for ARM GNU/Linux

[janis at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=49175

           Summary: vect/pr48172.c execution failure for ARM GNU/Linux
           Product: gcc
           Version: unknown
            Status: UNCONFIRMED
          Severity: normal
          Priority: P3
         Component: tree-optimization
        AssignedTo: unassigned@gcc.gnu.org
        ReportedBy: janis@gcc.gnu.org


Test gcc.dg/vect/pr48172.c fails at execution for arm-none-linux-gnueabi with
"Illegal Instruction", with the test run on QEMU for -march=armv4t.  The test
also fails execution for armv5tel-brewer-linux-gnueabi, a native configuration:
<http://gcc.gnu.org/ml/gcc-testresults/2011-05/msg02524.html>.

The test was added on 2011-05-12, but the test program also fails with GCC 4.6
and 4.5.

[Bug tree-optimization/49175] vect/pr48172.c execution failure for ARM GNU/Linux

[mikpe at it dot uu.se]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=49175

Mikael Pettersson <mikpe at it dot uu.se> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |mikpe at it dot uu.se

--- Comment #1 from Mikael Pettersson <mikpe at it dot uu.se> 2011-05-26 17:11:12 UTC ---
I checked the failure on armv5tel-linux-gnueabi: it's due to the test case
containing VFP (Neon?) instructions, so it gets a SIGILL at runtime.

According to gcc/testsuite/gcc/gcc.log it's compiled with -mfpu=neon, which
explains the failure.  Does it need a

{ dg-require-effective-target vect_int }

or something like that?

[Bug testsuite/49175] vect/pr48172.c execution failure for ARM GNU/Linux

[ramana at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=49175

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

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|UNCONFIRMED                 |NEW
   Last reconfirmed|                            |2011.05.31 21:08:33
                 CC|                            |ramana at gcc dot gnu.org
          Component|tree-optimization           |testsuite
     Ever Confirmed|0                           |1

--- Comment #2 from Ramana Radhakrishnan <ramana at gcc dot gnu.org> 2011-05-31 21:08:33 UTC ---
This is more a testism rather than anything else and I think the correct option
is to add a dg-require-effective-target "vect_int" . 

It's also the case in gcc 4.6 . 

cheers
Ramana

[Bug testsuite/49175] vect/pr48172.c execution failure for ARM GNU/Linux

[janis at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=49175

--- Comment #3 from Janis Johnson <janis at gcc dot gnu.org> 2011-05-31 22:43:42 UTC ---
Usually the vect tests only require effective targets like vect_int if they are
checking that particular optimizations took place.  There are no such checks in
this test.

The vector tests use neon options if check_effective_target_arm_neon_hw (from
lib/target-supports.exp) returns true.  The test the proc executes is:

        int
        main (void)
        {
          long long a = 0, b = 1;
          asm ("vorr %P0, %P1, %P2"
               : "=w" (a)
               : "0" (a), "w" (b));
          return (a != 1);
        }

compiled with neon options. That test apparently passes, but there's a neon
instruction generated for the test that fails.  Should the test used for
arm_neon_hw be different?

[Bug testsuite/49175] vect/pr48172.c execution failure for ARM GNU/Linux

[pinskia at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=49175

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

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|NEW                         |RESOLVED
   Target Milestone|---                         |4.7.0
         Resolution|---                         |FIXED

--- Comment #4 from Andrew Pinski <pinskia at gcc dot gnu.org> ---
Fixed by r0-108958 . The bug was only there for a few weeks and was fixed by
the time this bug report was filed even :).  The problem was "dg-do run"
overode what vect.exp figured out.


https://gcc.gnu.org/pipermail/gcc-patches/2011-May/313648.html