[Bug target/103200] New: FAIL: gcc.target/i386/pr91333.c scan-assembler-times vmovapd|vmovsd 3

[Bug target/103200] New: FAIL: gcc.target/i386/pr91333.c scan-assembler-times vmovapd|vmovsd 3

[msebor at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=103200

            Bug ID: 103200
           Summary: FAIL: gcc.target/i386/pr91333.c scan-assembler-times
                    vmovapd|vmovsd 3
           Product: gcc
           Version: 12.0
            Status: UNCONFIRMED
          Severity: normal
          Priority: P3
         Component: target
          Assignee: unassigned at gcc dot gnu.org
          Reporter: msebor at gcc dot gnu.org
  Target Milestone: ---

I see the following failure in an x86_64-linux build with today's top of trunk:

Executing on host: /build/gcc-master/gcc/xgcc -B/build/gcc-master/gcc/
/src/gcc/master/gcc/testsuite/gcc.target/i386/pr91333.c   
-fdiagnostics-plain-output  -O2 -mavx -ffat-lto-objects -fno-ident -S -o
pr91333.s    (timeout = 300)
spawn -ignore SIGHUP /build/gcc-master/gcc/xgcc -B/build/gcc-master/gcc/
/src/gcc/master/gcc/testsuite/gcc.target/i386/pr91333.c
-fdiagnostics-plain-output -O2 -mavx -ffat-lto-objects -fno-ident -S -o
pr91333.s
PASS: gcc.target/i386/pr91333.c (test for excess errors)
gcc.target/i386/pr91333.c: vmovapd|vmovsd found 1 times
FAIL: gcc.target/i386/pr91333.c scan-assembler-times vmovapd|vmovsd 3

[Bug target/103200] FAIL: gcc.target/i386/pr91333.c scan-assembler-times vmovapd|vmovsd 3

[rguenth at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=103200

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

           What    |Removed                     |Added
----------------------------------------------------------------------------
   Last reconfirmed|                            |2021-11-12
           Keywords|                            |missed-optimization,
                   |                            |needs-bisection
             Status|UNCONFIRMED                 |NEW
     Ever confirmed|0                           |1
             Target|                            |x86_64-*-* i?86-*-*

--- Comment #1 from Richard Biener <rguenth at gcc dot gnu.org> ---
Confirmed.  We now "spill"(?) to %rax:

        vmovq   %xmm0, %rax
        vmovsd  %xmm1, %xmm1, %xmm2
        vaddsd  %xmm1, %xmm2, %xmm0
        vmovq   %rax, %xmm1
        vmovq   %rax, %xmm2
        vaddsd  %xmm2, %xmm1, %xmm1
        vaddsd  %xmm1, %xmm0, %xmm0

compared to gcc 11:

        vmovsd  %xmm1, %xmm1, %xmm2
        vaddsd  %xmm1, %xmm2, %xmm2
        vmovsd  %xmm0, %xmm0, %xmm1
        vaddsd  %xmm0, %xmm1, %xmm0
        vmovsd  %xmm2, %xmm2, %xmm1
        vaddsd  %xmm1, %xmm0, %xmm0

[Bug target/103200] [12 Regression] FAIL: gcc.target/i386/pr91333.c scan-assembler-times vmovapd|vmovsd 3

[rguenth at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=103200

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

           What    |Removed                     |Added
----------------------------------------------------------------------------
   Target Milestone|---                         |12.0
           Priority|P3                          |P1
            Summary|FAIL:                       |[12 Regression] FAIL:
                   |gcc.target/i386/pr91333.c   |gcc.target/i386/pr91333.c
                   |scan-assembler-times        |scan-assembler-times
                   |vmovapd|vmovsd 3            |vmovapd|vmovsd 3

[Bug target/103200] [12 Regression] FAIL: gcc.target/i386/pr91333.c scan-assembler-times vmovapd|vmovsd 3 since r12-5177-g494bdadf28d0fb35

[marxin at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=103200

Martin Liška <marxin at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
           Keywords|needs-bisection             |
            Summary|[12 Regression] FAIL:       |[12 Regression] FAIL:
                   |gcc.target/i386/pr91333.c   |gcc.target/i386/pr91333.c
                   |scan-assembler-times        |scan-assembler-times
                   |vmovapd|vmovsd 3            |vmovapd|vmovsd 3 since
                   |                            |r12-5177-g494bdadf28d0fb35
                 CC|                            |hubicka at gcc dot gnu.org,
                   |                            |marxin at gcc dot gnu.org

--- Comment #2 from Martin Liška <marxin at gcc dot gnu.org> ---
Started with r12-5177-g494bdadf28d0fb35.
@Honza: Did you run tests for the change?

[Bug target/103200] [12 Regression] FAIL: gcc.target/i386/pr91333.c scan-assembler-times vmovapd|vmovsd 3 since r12-5177-g494bdadf28d0fb35

[hubicka at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=103200

--- Comment #3 from Jan Hubicka <hubicka at gcc dot gnu.org> ---
Sorry, I managed to include last minute change that was wrong.  Currently
modref does not track if function is deterministic so we can not detect looping
pure/consts. I am testing.

diff --git a/gcc/ipa-modref.c b/gcc/ipa-modref.c
index 72006251f29..4da5284c94f 100644
--- a/gcc/ipa-modref.c
+++ b/gcc/ipa-modref.c
@@ -2790,9 +2790,10 @@ analyze_function (function *f, bool ipa)

   if (!ipa && flag_ipa_pure_const)
     {
-      if (!summary->stores->every_base && !summary->stores->bases)
+      if (!summary->stores->every_base && !summary->stores->bases
+         && !summary->side_effects)
        {
-         if (!summary->loads->every_base && !summary->loads->bases)
+         if (!summary->loads->every_base && !summary->loads->bases
            fixup_cfg = ipa_make_function_const
                   (cgraph_node::get (current_function_decl),
                    summary->side_effects, true);
@@ -4380,7 +4381,8 @@ modref_propagate_in_scc (cgraph_node *component_node)
        modref_summary_lto *summary_lto = summaries_lto
                                          ? summaries_lto->get (cur)
                                          : NULL;
-       if (summary && !summary->stores->every_base && !summary->stores->bases)
+       if (summary && !summary->stores->every_base && !summary->stores->bases
+           && !summary->side_effects)
          {
            if (!summary->loads->every_base && !summary->loads->bases)
              pureconst |= ipa_make_function_const
@@ -4390,7 +4392,7 @@ modref_propagate_in_scc (cgraph_node *component_node)
                     (cur, summary->side_effects, false);
          }
        if (summary_lto && !summary_lto->stores->every_base
-           && !summary_lto->stores->bases)
+           && !summary_lto->stores->bases && !summary->side_effects)
          {
            if (!summary_lto->loads->every_base && !summary_lto->loads->bases)
              pureconst |= ipa_make_function_const

[Bug target/103200] [12 Regression] FAIL: gcc.target/i386/pr91333.c scan-assembler-times vmovapd|vmovsd 3 since r12-5177-g494bdadf28d0fb35

[cvs-commit at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=103200

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

https://gcc.gnu.org/g:1b62cddcf091fb8cadf575246a7d3ff778650a6b

commit r12-5195-g1b62cddcf091fb8cadf575246a7d3ff778650a6b
Author: Jan Hubicka <jh@suse.cz>
Date:   Fri Nov 12 14:00:47 2021 +0100

    Fix ipa-modref pure/const discovery

            PR ipa/103200
            * ipa-modref.c (analyze_function, modref_propagate_in_scc): Do
            not mark pure/const function if there are side-effects.

[Bug target/103200] [12 Regression] FAIL: gcc.target/i386/pr91333.c scan-assembler-times vmovapd|vmovsd 3 since r12-5177-g494bdadf28d0fb35

[marxin at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=103200

Martin Liška <marxin at gcc dot gnu.org> changed:

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

--- Comment #5 from Martin Liška <marxin at gcc dot gnu.org> ---
Fixed now.