[Bug target/64358] New: Wrong code for __int128 operations in powerpc64le

[Bug target/64358] New: Wrong code for __int128 operations in powerpc64le

[roger.ferrer at bsc dot es]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=64358

            Bug ID: 64358
           Summary: Wrong code for __int128 operations in powerpc64le
           Product: gcc
           Version: 4.9.1
            Status: UNCONFIRMED
          Severity: normal
          Priority: P3
         Component: target
          Assignee: unassigned at gcc dot gnu.org
          Reporter: roger.ferrer at bsc dot es

Hi,

I'm observing a weird behaviour in PowerPC64 Little Endian that does not seem
to occur on other architectures supporting __int128. The following code, when
compiled with -O1 generates wrong output in gcc 4.9.1.

-- test.c
#include <stdio.h>

typedef unsigned __int128 uint128_t;

#define PRINT(value) \
    { union u { uint128_t i; unsigned long long l[2]; } _t = { .i = value }; \
        fprintf(stderr, "%s => <%016llx, %016llx>\n", #value, _t.l[1],
_t.l[0]); }

__attribute__((noinline))
uint128_t get_int(uint128_t value, unsigned int num_bytes)
{
    uint128_t mask = ~(uint128_t)0;
    mask <<= (uint128_t)(8 * num_bytes); /* assuming 1 byte = 8 bits */
    mask = ~mask;
    value &= mask;

    return value;
}

int main(int argc, char* argv[])
{
    uint128_t x = 0;
    x = get_int(10, /* num_bytes */ 1);

    PRINT(x);

    return 0;
}
-- end of test.c

The problem is still present in a recent version from the svn:

$ gcc -v
Using built-in specs.
COLLECT_GCC=/home/Computational/rferrer/gcc/install/bin/gcc
COLLECT_LTO_WRAPPER=/home/Computational/rferrer/gcc/install/libexec/gcc/powerpc64le-unknown-linux-gnu/5.0.0/lto-wrapper
Target: powerpc64le-unknown-linux-gnu
Configured with: ../gcc-src/configure
--prefix=/home/Computational/rferrer/gcc/install
--enable-languages=c,c++,fortran
--with-gmp=/home/Computational/rferrer/gcc/install
--with-mpfr=/home/Computational/rferrer/gcc/install
--with-mpc=/home/Computational/rferrer/gcc/install --enable-multiarch
--disable-multilib
Thread model: posix
gcc version 5.0.0 20141218 (experimental) (GCC)

$ make
gcc  -O0 -o test.O0 test.c
./test.O0
x => <0000000000000000, 000000000000000a>
gcc  -O1 -o test.O1 test.c
./test.O1
x => <ffffffffffffffff, ffffffffffffff00>

Kind regards,

[Bug target/64358] Wrong code for __int128 operations in powerpc64le

[pthaugen at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=64358

Pat Haugen <pthaugen at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Target|powerpc64le                 |powerpc64le, powerpc64
             Status|UNCONFIRMED                 |ASSIGNED
   Last reconfirmed|                            |2014-12-19
                 CC|                            |dje at gcc dot gnu.org
           Assignee|unassigned at gcc dot gnu.org      |pthaugen at gcc dot gnu.org
     Ever confirmed|0                           |1

--- Comment #1 from Pat Haugen <pthaugen at gcc dot gnu.org> ---
This occurs on powerpc64 Big Endian too and is fixed by a patch I just
submitted yesterday. Talk about timing...

https://gcc.gnu.org/ml/gcc-patches/2014-12/msg01585.html

[Bug target/64358] Wrong code for __int128 operations in powerpc64le

[roger.ferrer at bsc dot es]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=64358

--- Comment #2 from Roger Ferrer Ibanez <roger.ferrer at bsc dot es> ---
(In reply to Pat Haugen from comment #1)
> This occurs on powerpc64 Big Endian too and is fixed by a patch I just
> submitted yesterday. Talk about timing...

Fantastic

> 
> https://gcc.gnu.org/ml/gcc-patches/2014-12/msg01585.html

I applied the patch in my tree and it fixes this issue.

Thanks a lot!

[Bug target/64358] Wrong code for __int128 operations in powerpc64le

[segher at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=64358

--- Comment #3 from Segher Boessenkool <segher at gcc dot gnu.org> ---
Author: segher
Date: Sat Dec 20 15:37:06 2014
New Revision: 218989

URL: https://gcc.gnu.org/viewcvs?rev=218989&root=gcc&view=rev
Log:
    PR target/64358
    * config/rs6000/rs6000.c (rs6000_split_logical_inner): Swap the
    input operands if only the second is inverted.
    * config/rs6000/rs6000.md (*boolc<mode>3_internal1 for BOOL_128):
    Swap BOOL_REGS_OP1 and BOOL_REGS_OP2.  Correct arguments to
    rs6000_split_logical.
    (*boolc<mode>3_internal2 for TI2): Swap operands[1] and operands[2].

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

[Bug target/64358] Wrong code for __int128 operations in powerpc64le

[pthaugen at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=64358

--- Comment #4 from Pat Haugen <pthaugen at gcc dot gnu.org> ---
Author: pthaugen
Date: Tue Jan 13 19:08:49 2015
New Revision: 219548

URL: https://gcc.gnu.org/viewcvs?rev=219548&root=gcc&view=rev
Log:
    Backport from mainline
    2014-12-20  Segher Boessenkool  <segher@kernel.crashing.org>

    PR target/64358
    * config/rs6000/rs6000.c (rs6000_split_logical_inner): Swap the
    input operands if only the second is inverted.
    * config/rs6000/rs6000.md (*boolc<mode>3_internal1 for BOOL_128):
    Swap BOOL_REGS_OP1 and BOOL_REGS_OP2.  Correct arguments to
    rs6000_split_logical.
    (*boolc<mode>3_internal2 for TI2): Swap operands[1] and operands[2].


Modified:
    branches/gcc-4_8-branch/gcc/ChangeLog
    branches/gcc-4_8-branch/gcc/config/rs6000/rs6000.c
    branches/gcc-4_8-branch/gcc/config/rs6000/rs6000.md

[Bug target/64358] Wrong code for __int128 operations in powerpc64le

[pthaugen at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=64358

--- Comment #5 from Pat Haugen <pthaugen at gcc dot gnu.org> ---
Author: pthaugen
Date: Tue Jan 13 19:17:42 2015
New Revision: 219551

URL: https://gcc.gnu.org/viewcvs?rev=219551&root=gcc&view=rev
Log:
    Backport from mainline
    2014-12-20  Segher Boessenkool  <segher@kernel.crashing.org>

    PR target/64358
    * config/rs6000/rs6000.c (rs6000_split_logical_inner): Swap the
    input operands if only the second is inverted.
    * config/rs6000/rs6000.md (*boolc<mode>3_internal1 for BOOL_128):
    Swap BOOL_REGS_OP1 and BOOL_REGS_OP2.  Correct arguments to
    rs6000_split_logical.
    (*boolc<mode>3_internal2 for TI2): Swap operands[1] and operands[2].


Modified:
    branches/gcc-4_9-branch/gcc/ChangeLog
    branches/gcc-4_9-branch/gcc/config/rs6000/rs6000.c
    branches/gcc-4_9-branch/gcc/config/rs6000/rs6000.md