[Bug target/105122] New: [12 Regression] Testsuite failures since r12-7931 on i686-linux

[Bug target/105122] New: [12 Regression] Testsuite failures since r12-7931 on i686-linux

[jakub at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=105122

            Bug ID: 105122
           Summary: [12 Regression] Testsuite failures since r12-7931 on
                    i686-linux
           Product: gcc
           Version: 12.0
            Status: UNCONFIRMED
          Severity: normal
          Priority: P3
         Component: target
          Assignee: unassigned at gcc dot gnu.org
          Reporter: jakub at gcc dot gnu.org
  Target Milestone: ---

Starting with r12-7931-gb8207ad367174df5f2e2fdf3305c97ed227d8f78 I'm seeing on
i686-linux (or x86_64-linux with
RUNTESTFLAGS='--target_board=unix/-m32/-mno-sse'
testing:
+FAIL: gcc.dg/memcpy-6.c  (test for warnings, line 25)
+FAIL: gcc.dg/memcpy-6.c scan-tree-dump-not optimized "memcpy"
+FAIL: gcc.dg/memcpy-6.c scan-tree-dump-not optimized "memmove"
+FAIL: gcc.dg/strlenopt-73.c scan-tree-dump-times optimized "_not_eliminated_"
0
+FAIL: gcc.dg/strlenopt-73.c scan-tree-dump-times optimized "strlen" 0
+FAIL: gcc.dg/strlenopt-80.c scan-tree-dump-times optimized "failure_on_line
\\\\(" 0
+FAIL: gcc.dg/ipa/remref-7.c scan-ipa-dump inline "Removed a reference"
Haven't checked if those tests existed before the
r12-3482-g5f6a6c91d7c592cb49f7c519f289777eac09bb74 commit.

[Bug target/105122] [12 Regression] Testsuite failures since r12-7931 on i686-linux

[jakub at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=105122

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

           What    |Removed                     |Added
----------------------------------------------------------------------------
           Priority|P3                          |P1
   Target Milestone|---                         |12.0
                 CC|                            |rguenth at gcc dot gnu.org

[Bug target/105122] [12 Regression] Testsuite failures since r12-7931 on i686-linux

[rguenth at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=105122

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

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|UNCONFIRMED                 |ASSIGNED
   Last reconfirmed|                            |2022-04-01
     Ever confirmed|0                           |1
           Assignee|unassigned at gcc dot gnu.org      |rguenth at gcc dot gnu.org
           Keywords|                            |missed-optimization
             Target|                            |i?86-*-*

--- Comment #1 from Richard Biener <rguenth at gcc dot gnu.org> ---
I will have a look.

+FAIL: gcc.dg/memcpy-6.c  (test for warnings, line 25)
+FAIL: gcc.dg/memcpy-6.c scan-tree-dump-not optimized "memcpy"
+FAIL: gcc.dg/memcpy-6.c scan-tree-dump-not optimized "memmove"

fold_copy_8 and fold_move_8 are no longer folded, that's to be expected with
MOVE_MAX and DImode types not being involved directly.  The r12-3482 change
would have newly folded those.  fold_move_8 is also not expanded inline
by RTL expansion.  GCC 11 seems to fold both.

The gcc.dg/ipa/remref-7.c one is probably spurious (it's a new testcase).

[Bug target/105122] [12 Regression] Testsuite failures since r12-7931 on i686-linux

[rguenth at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=105122

--- Comment #2 from Richard Biener <rguenth at gcc dot gnu.org> ---
Ah, in GCC 11 we had

/* Max number of bytes we can move from memory to memory
   in one reasonably fast instruction.  */
#define MOVE_MAX 16

while in GCC 12 it is now

/* Max number of bytes we can move from memory to memory in one
   reasonably fast instruction, as opposed to MOVE_MAX_PIECES which
   is the number of bytes at a time which we can move efficiently.
   MOVE_MAX_PIECES defaults to MOVE_MAX.  */

#define MOVE_MAX \
  ((TARGET_AVX512F \
    && (ix86_move_max == PVW_AVX512 \
        || ix86_store_max == PVW_AVX512)) \
   ? 64 \
   : ((TARGET_AVX \
       && (ix86_move_max >= PVW_AVX256 \
           || ix86_store_max >= PVW_AVX256)) \
      ? 32 \
      : ((TARGET_SSE2 \
          && TARGET_SSE_UNALIGNED_LOAD_OPTIMAL \
          && TARGET_SSE_UNALIGNED_STORE_OPTIMAL) \
         ? 16 : UNITS_PER_WORD)))

and UNITS_PER_WORD is now 4.  Not sure if that was a concious decision?

I'm not sure we want to "cheat" here.  For memcpy-6.c we have sth like

char a[32];
void fold_copy_8 (void)
{
  __builtin_memcpy (a + 3, a, 8);
}

where if we'd try to use 'long long' we'd succeed (since 'a' is properly
aligned).  We'd have to use lang_hook.types.types_for_mode to not get
too large types but even with that we'll happily use __int128_t on
i?86 with -mno-sse when copying 16 bytes.  The idea of using larger
types than MOVE_MAX was to restrict that to the cases where we do say

 __int128_t tem;
 memcpy (&tem, a, 16);

and thus the large type is used in the source already (and is > MOVE_MAX).
Similarly for 'double' on i?86 where we'd use DImode.  For the sake
of removing abstraction.  But the gcc.dg/memcpy-6.c testcase should be
about RTL expansion (with the known lack of handling of memmmove).

Btw, lang_hooks.types.type_for_mode reveals __int128_t via

#if HOST_BITS_PER_WIDE_INT >= 64
  if (mode == TYPE_MODE (intTI_type_node))
    return unsignedp ? unsigned_intTI_type_node : intTI_type_node;
#endif

even though we have

(gdb) p int_n_enabled_p
$2 = {false}
(gdb) p int_n_data[0]
$3 = {bitsize = 128, m = TImode}

of course __int128_t != unsigned __attribute__((mode(TI))), but ...

[Bug target/105122] [12 Regression] Testsuite failures since r12-7931 on i686-linux

[rguenth at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=105122

--- Comment #3 from Richard Biener <rguenth at gcc dot gnu.org> ---
PASS: gcc.dg/strlenopt-73.c (test for excess errors)
gcc.dg/strlenopt-73.c: pattern found 4 times
FAIL: gcc.dg/strlenopt-73.c scan-tree-dump-times optimized "strlen" 0
gcc.dg/strlenopt-73.c: pattern found 6 times
FAIL: gcc.dg/strlenopt-73.c scan-tree-dump-times optimized "_not_eliminated_" 0

#if defined(__i386__) || defined(__x86_64__) || defined(__aarch64__) \
    || defined(__s390__) || defined(__powerpc64__)

/* The following tests assume GCC transforms the memcpy calls into
   long long assignments which it does only on targets that define
   the MOVE_MAX macro to 8 or higher.  Enable on a set of targets
   known to do that.  */

that's no longer true for i?86 w/o SSE (and now memcpy folding honors that).
I'll adjust this testcase.

PASS: gcc.dg/strlenopt-80.c (test for excess errors)
gcc.dg/strlenopt-80.c: pattern found 5 times
FAIL: gcc.dg/strlenopt-80.c scan-tree-dump-times optimized "failure_on_line
\\(" 0

similar, the testcase says

   The optimization is only implemented for MEM_REF stores and other
   targets than those below may not transform the memcpy call into
   such a store.
   { dg-do compile { target { { aarch64*-*-* i?86-*-* x86_64-*-* } || { {
powerpc*-*-* } && lp64 } } } }

one way to fix this testcase would be to explicitely enable SSE for i?86-*-*.

[Bug target/105122] [12 Regression] Testsuite failures since r12-7931 on i686-linux

[cvs-commit at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=105122

--- Comment #4 from CVS Commits <cvs-commit at gcc dot gnu.org> ---
The master branch has been updated by Richard Biener <rguenth@gcc.gnu.org>:

https://gcc.gnu.org/g:e2a818641ba5d07ebe2c241906896c4886910d18

commit r12-8011-ge2a818641ba5d07ebe2c241906896c4886910d18
Author: Richard Biener <rguenther@suse.de>
Date:   Mon Apr 4 08:41:59 2022 +0200

    testsuite/105122 - adjust testcases after memcpy folding changes

    After r12-7931 we again honor MOVE_MAX when folding memcpy to
    a load/store pair.  On i?86-*-* without SSE this now exposes the
    change done in r12-2666-g29f0e955c97da0 which adjusts MOVE_MAX
    from 16 to 4 on those targets.  This makes adjusting testcases
    necessary that assume that we transform memcpy to load/store pairs
    on GIMPLE for sizes larger or equal to 8.

    2022-04-04  Richard Biener  <rguenther@suse.de>

            PR testsuite/105122
            * gcc.dg/memcpy-6.c: Adjust.
            * gcc.dg/strlenopt-73.c: Likewise.
            * gcc.dg/strlenopt-80.c: Likewise.

[Bug target/105122] [12 Regression] Testsuite failures since r12-7931 on i686-linux

[rguenth at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=105122

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

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

--- Comment #5 from Richard Biener <rguenth at gcc dot gnu.org> ---
Fixed.