[Bug tree-optimization/95581] New: [11 Regression] ICE in gimple_call_arg, at gimple.h:3260 since r11-959-gb825a22890740f341eae566af27e18e528cd29a7

[Bug tree-optimization/95581] New: [11 Regression] ICE in gimple_call_arg, at gimple.h:3260 since r11-959-gb825a22890740f341eae566af27e18e528cd29a7

[marxin at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=95581

            Bug ID: 95581
           Summary: [11 Regression] ICE in gimple_call_arg, at
                    gimple.h:3260 since
                    r11-959-gb825a22890740f341eae566af27e18e528cd29a7
           Product: gcc
           Version: 11.0
            Status: UNCONFIRMED
          Keywords: ice-on-valid-code
          Severity: normal
          Priority: P3
         Component: tree-optimization
          Assignee: unassigned at gcc dot gnu.org
          Reporter: marxin at gcc dot gnu.org
                CC: msebor at gcc dot gnu.org
  Target Milestone: ---
              Host: x86_64-linux-gnu
            Target: ppc64le-linux-gnu

Since the revision I see:

$ ./xgcc -B.
/home/marxin/Programming/gcc/gcc/testsuite/gcc.c-torture/compile/pr34458.c
-ftree-loop-vectorize -mno-allow-movmisalign -Wextra -O1 -mpower8-vector -S
during GIMPLE pass: uninit
/home/marxin/Programming/gcc/gcc/testsuite/gcc.c-torture/compile/pr34458.c: In
function ‘lint_operate’:
/home/marxin/Programming/gcc/gcc/testsuite/gcc.c-torture/compile/pr34458.c:9:6:
internal compiler error: in gimple_call_arg, at gimple.h:3260
    9 | Lint lint_operate (Lint a, long long ammount)
      |      ^~~~~~~~~~~~
0x147f267 gimple_call_arg
        /home/marxin/Programming/gcc2/gcc/gimple.h:3260
0x147f2bd gimple_call_arg
        /home/marxin/Programming/gcc2/gcc/gimple.h:3268
0x1481038 maybe_warn_pass_by_reference
        /home/marxin/Programming/gcc2/gcc/tree-ssa-uninit.c:517
0x14816ad warn_uninitialized_vars
        /home/marxin/Programming/gcc2/gcc/tree-ssa-uninit.c:631
0x14863a3 execute
        /home/marxin/Programming/gcc2/gcc/tree-ssa-uninit.c:2936
Please submit a full bug report,
with preprocessed source if appropriate.
Please include the complete backtrace with any bug report.
See <https://gcc.gnu.org/bugs/> for instructions.

[Bug tree-optimization/95581] [11 Regression] ICE in gimple_call_arg, at gimple.h:3260 since r11-959-gb825a22890740f341eae566af27e18e528cd29a7

[marxin at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=95581

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

           What    |Removed                     |Added
----------------------------------------------------------------------------
     Ever confirmed|0                           |1
   Last reconfirmed|                            |2020-06-08
             Status|UNCONFIRMED                 |NEW

[Bug tree-optimization/95581] [11 Regression] ICE in gimple_call_arg, at gimple.h:3260 since r11-959-gb825a22890740f341eae566af27e18e528cd29a7

[msebor at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=95581

--- Comment #1 from Martin Sebor <msebor at gcc dot gnu.org> ---
gimple_call_arg(stmt, 2) asserts for this GIMPLE_CALL statement:

  vect__5.6_24 = __builtin_altivec_mask_for_load (vectp_a.5_8);

The call is made with just one argument but the type of the function is
actually

 <function_type 0x7ffff789a0a8
    type <vector_type 0x7ffff787b0a8 __vector signed char
        type <integer_type 0x7ffff77b22a0 signed char public QI
            size <integer_cst 0x7ffff77b40c0 constant 8>
            unit-size <integer_cst 0x7ffff77b40d8 constant 1>
            align:8 warn_if_not_align:0 symtab:0 alias-set -1 canonical-type
0x7ffff77b22a0 precision:8 min <integer_cst 0x7ffff77b4078 -128> max
<integer_cst 0x7ffff77b40a8 127>>
        V16QI
        size <integer_cst 0x7ffff77b4018 constant 128>
        unit-size <integer_cst 0x7ffff77b4030 constant 16>
        align:128 warn_if_not_align:0 symtab:0 alias-set -1 canonical-type
0x7ffff787b000 nunits:16>
    SI
    size <integer_cst 0x7ffff77b4210 type <integer_type 0x7ffff77b20a8
bitsizetype> constant 32>
    unit-size <integer_cst 0x7ffff77b4228 type <integer_type 0x7ffff77b2000
sizetype> constant 4>
    align:32 warn_if_not_align:0 symtab:0 alias-set -1 canonical-type
0x7ffff789a888
    arg-types <tree_list 0x7ffff789b0f0
        value <integer_type 0x7ffff77b2738 long int public DI
            size <integer_cst 0x7ffff77a1fc0 constant 64>
            unit-size <integer_cst 0x7ffff77a1fd8 constant 8>
            align:64 warn_if_not_align:0 symtab:0 alias-set -1 canonical-type
0x7ffff77b2738 precision:64 min <integer_cst 0x7ffff77b4258
-9223372036854775808> max <integer_cst 0x7ffff77b4270 9223372036854775807>
            pointer_to_this <pointer_type 0x7ffff77c1dc8>>
        chain <tree_list 0x7ffff789b118 value <pointer_type 0x7ffff77ba150>
            chain <tree_list 0x7ffff77af0a0 value <void_type 0x7ffff77b2f18
void>>>>
    pointer_to_this <pointer_type 0x7ffff794be70>>

I couldn't found the function documented in the manual or exercised by the test
suite but I see it created in gcc/config/rs6000/rs6000-call.c:

  decl = add_builtin_function ("__builtin_altivec_mask_for_load",
                               v16qi_ftype_long_pcvoid,
                               ALTIVEC_BUILTIN_MASK_FOR_LOAD,
                               BUILT_IN_MD, NULL, NULL_TREE);

Is it okay to call it with fewer arguments than its type indicates?

[Bug tree-optimization/95581] [11 Regression] ICE in gimple_call_arg, at gimple.h:3260 since r11-959-gb825a22890740f341eae566af27e18e528cd29a7

[iains at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=95581

--- Comment #2 from Iain Sandoe <iains at gcc dot gnu.org> ---
BTW, this is a bootstrap break for powerpc-darwin9, it fires when building
libgfortran (OK it's an 'all languages' bootstrap break).

[Bug tree-optimization/95581] [11 Regression] ICE in gimple_call_arg, at gimple.h:3260 since r11-959-gb825a22890740f341eae566af27e18e528cd29a7

[msebor at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=95581

--- Comment #3 from Martin Sebor <msebor at gcc dot gnu.org> ---
I posted a patch to avoid the ICE but it's not clear to me that it's the right
fix: https://gcc.gnu.org/pipermail/gcc-patches/2020-June/547548.html

[Bug tree-optimization/95581] [11 Regression] ICE in gimple_call_arg, at gimple.h:3260 since r11-959-gb825a22890740f341eae566af27e18e528cd29a7

[segher at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=95581

Segher Boessenkool <segher at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |segher at gcc dot gnu.org

--- Comment #4 from Segher Boessenkool <segher at gcc dot gnu.org> ---
This is documented at
https://gcc.gnu.org/git/?p=gcc.git;a=blob;f=gcc/target.def;h=07059a87caf7cc0237d583124b476ee45ea41ed5;hb=HEAD#l1744
(the altivec_ function is just our usual implementation of the target hook).

[Bug tree-optimization/95581] [11 Regression] ICE in gimple_call_arg, at gimple.h:3260 since r11-959-gb825a22890740f341eae566af27e18e528cd29a7

[msebor at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=95581

Martin Sebor <msebor at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |seurer at linux dot vnet.ibm.com

--- Comment #5 from Martin Sebor <msebor at gcc dot gnu.org> ---
*** Bug 95595 has been marked as a duplicate of this bug. ***

[Bug tree-optimization/95581] [11 Regression] ICE in gimple_call_arg, at gimple.h:3260 since r11-959-gb825a22890740f341eae566af27e18e528cd29a7

[rguenth at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=95581

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

           What    |Removed                     |Added
----------------------------------------------------------------------------
   Target Milestone|---                         |11.0
           Priority|P3                          |P1

[Bug tree-optimization/95581] [11 Regression] ICE in gimple_call_arg, at gimple.h:3260 since r11-959-gb825a22890740f341eae566af27e18e528cd29a7

[seurer at linux dot vnet.ibm.com]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=95581

--- Comment #6 from Bill Seurer <seurer at linux dot vnet.ibm.com> ---
I tried the patch and it allowed me to build gcc on a power7 system.

[Bug tree-optimization/95581] [11 Regression] ICE in gimple_call_arg, at gimple.h:3260 since r11-959-gb825a22890740f341eae566af27e18e528cd29a7

[iains at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=95581

--- Comment #7 from Iain Sandoe <iains at gcc dot gnu.org> ---
(In reply to Bill Seurer from comment #6)
> I tried the patch and it allowed me to build gcc on a power7 system.

but AFAIU the discussion - the builtin has the wrong signature (and needs to be
fixed)?
or did I misunderstand?

[Bug target/95581] [11 Regression] ICE in gimple_call_arg, at gimple.h:3260 since r11-959-gb825a22890740f341eae566af27e18e528cd29a7

[msebor at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=95581

Martin Sebor <msebor at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
          Component|tree-optimization           |target

--- Comment #8 from Martin Sebor <msebor at gcc dot gnu.org> ---
Yes, the conclusion of the discussion of the patch is that the fix needs to
happen in the back end.

[Bug target/95581] [11 Regression] ICE in gimple_call_arg, at gimple.h:3260 since r11-959-gb825a22890740f341eae566af27e18e528cd29a7

[iains at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=95581

--- Comment #9 from Iain Sandoe <iains at gcc dot gnu.org> ---
(In reply to Martin Sebor from comment #8)
> Yes, the conclusion of the discussion of the patch is that the fix needs to
> happen in the back end.

I still need the patch at comment #3 to bootstrap with Fortran on
powerpc-darwin

Is there any consensus for applying that patch (or some recommendation about
how to fix things 'more correctly')?

[Bug target/95581] [11 Regression] ICE in gimple_call_arg, at gimple.h:3260 since r11-959-gb825a22890740f341eae566af27e18e528cd29a7

[seurer at linux dot vnet.ibm.com]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=95581

--- Comment #10 from Bill Seurer <seurer at linux dot vnet.ibm.com> ---
I think this will work but haven't had a chance to try it except on the one
power 7 system.

diff --git a/gcc/config/rs6000/rs6000-call.c b/gcc/config/rs6000/rs6000-call.c
index 817a14c9c..37eb31222 100644
--- a/gcc/config/rs6000/rs6000-call.c
+++ b/gcc/config/rs6000/rs6000-call.c
@@ -12454,6 +12454,10 @@ altivec_init_builtins (void)
     = build_function_type_list (opaque_V4SI_type_node,
                                long_integer_type_node, pcvoid_type_node,
                                NULL_TREE);
+  tree v16qi_ftype_pcvoid
+    = build_function_type_list (V16QI_type_node,
+                                pcvoid_type_node,
+                                NULL_TREE);
   tree v16qi_ftype_long_pcvoid
     = build_function_type_list (V16QI_type_node,
                                long_integer_type_node, pcvoid_type_node,
@@ -12856,7 +12860,7 @@ altivec_init_builtins (void)
      targetm.vectorize.builtin_mask_for_load.  */
   decl = add_builtin_function ("__builtin_altivec_mask_for_load",
-                              v16qi_ftype_long_pcvoid,
+                              v16qi_ftype_pcvoid,
                               ALTIVEC_BUILTIN_MASK_FOR_LOAD,
                               BUILT_IN_MD, NULL, NULL_TREE);
   TREE_READONLY (decl) = 1;

[Bug target/95581] [11 Regression] ICE in gimple_call_arg, at gimple.h:3260 since r11-959-gb825a22890740f341eae566af27e18e528cd29a7

[iains at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=95581

--- Comment #11 from Iain Sandoe <iains at gcc dot gnu.org> ---
(In reply to Bill Seurer from comment #10)
> I think this will work but haven't had a chance to try it except on the one
> power 7 system.

Hi Bill - thanks for the patch.

This allowed bootstrap to complete on powerpc-darwin for r11-1811 and testing
looks nominal.

[Bug target/95581] [11 Regression] ICE in gimple_call_arg, at gimple.h:3260 since r11-959-gb825a22890740f341eae566af27e18e528cd29a7

[segher at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=95581

--- Comment #12 from Segher Boessenkool <segher at gcc dot gnu.org> ---
This is pre-approved with a testcase (if it fixes that testcase, etc.,
yadda yadda).

Thanks!

[Bug target/95581] [11 Regression] ICE in gimple_call_arg, at gimple.h:3260 since r11-959-gb825a22890740f341eae566af27e18e528cd29a7

[rguenth at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=95581

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

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|NEW                         |ASSIGNED
           Assignee|unassigned at gcc dot gnu.org      |seurer at linux dot vnet.ibm.com

--- Comment #13 from Richard Biener <rguenth at gcc dot gnu.org> ---
Still broken.  Can you fix it please?

[Bug target/95581] [11 Regression] ICE in gimple_call_arg, at gimple.h:3260 since r11-959-gb825a22890740f341eae566af27e18e528cd29a7

[cvs-commit at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=95581

--- Comment #14 from CVS Commits <cvs-commit at gcc dot gnu.org> ---
The master branch has been updated by Bill Seurer <seurer@gcc.gnu.org>:

https://gcc.gnu.org/g:c6b7ba5de624f2a17d799bac5ff017cd065ce035

commit r11-2018-gc6b7ba5de624f2a17d799bac5ff017cd065ce035
Author: Bill Seurer, 507-253-3502, seurer@us.ibm.com <(no_default)>
Date:   Thu Jul 9 16:41:38 2020 -0500

    rs6000: Fix __builtin_altivec_mask_for_load to use correct type

    gcc/ChangeLog:

            PR target/95581
            * config/rs6000/rs6000-call.c: Add new type v16qi_ftype_pcvoid.
            (altivec_init_builtins) Change __builtin_altivec_mask_for_load to
use
            v16qi_ftype_pcvoid with correct number of parameters.

[Bug target/95581] [11 Regression] ICE in gimple_call_arg, at gimple.h:3260 since r11-959-gb825a22890740f341eae566af27e18e528cd29a7

[cvs-commit at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=95581

--- Comment #15 from CVS Commits <cvs-commit at gcc dot gnu.org> ---
The releases/gcc-10 branch has been updated by Bill Seurer
<seurer@gcc.gnu.org>:

https://gcc.gnu.org/g:21d3be2159f1b55e903caa2b1f80369d3681902b

commit r10-8463-g21d3be2159f1b55e903caa2b1f80369d3681902b
Author: Bill Seurer <seurer@linux.vnet.ibm.com>
Date:   Fri Jul 10 17:30:26 2020 -0500

    rs6000: Fix __builtin_altivec_mask_for_load to use correct type

    gcc/ChangeLog:

            PR target/95581
            * config/rs6000/rs6000-call.c: Add new type v16qi_ftype_pcvoid.
            (altivec_init_builtins) Change __builtin_altivec_mask_for_load to
use
            v16qi_ftype_pcvoid with correct number of parameters.

    (cherry picked from commit c6b7ba5de624f2a17d799bac5ff017cd065ce035)

[Bug target/95581] [11 Regression] ICE in gimple_call_arg, at gimple.h:3260 since r11-959-gb825a22890740f341eae566af27e18e528cd29a7

[bergner at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=95581

Peter Bergner <bergner at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |bergner at gcc dot gnu.org

--- Comment #16 from Peter Bergner <bergner at gcc dot gnu.org> ---
Is there a reason we have left this in ASSIGNED state?  ...or can we mark this
as FIXED now?

[Bug target/95581] [11 Regression] ICE in gimple_call_arg, at gimple.h:3260 since r11-959-gb825a22890740f341eae566af27e18e528cd29a7

[seurer at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=95581

seurer at gcc dot gnu.org changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
         Resolution|---                         |FIXED
                 CC|                            |seurer at gcc dot gnu.org
             Status|ASSIGNED                    |RESOLVED

--- Comment #17 from seurer at gcc dot gnu.org ---
I will close it as everything we are going to do is done.