[Bug target/102477] New: ICE: in emit_move_insn, at expr.c:4026 at -Og with __builtin_shufflevector()

[Bug target/102477] New: ICE: in emit_move_insn, at expr.c:4026 at -Og with __builtin_shufflevector()

[zsojka at seznam dot cz]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=102477

            Bug ID: 102477
           Summary: ICE: in emit_move_insn, at expr.c:4026 at -Og with
                    __builtin_shufflevector()
           Product: gcc
           Version: 12.0
            Status: UNCONFIRMED
          Keywords: ice-on-valid-code
          Severity: normal
          Priority: P3
         Component: target
          Assignee: unassigned at gcc dot gnu.org
          Reporter: zsojka at seznam dot cz
  Target Milestone: ---
              Host: x86_64-pc-linux-gnu
            Target: x86_64-pc-linux-gnu

Created attachment 51507
  --> https://gcc.gnu.org/bugzilla/attachment.cgi?id=51507&action=edit
reduced testcase

Compiler output:
$ x86_64-pc-linux-gnu-gcc -Og testcase.c 
testcase.c: In function 'foo':
testcase.c:10:1: warning: AVX512F vector return without AVX512F enabled changes
the ABI [-Wpsabi]
   10 | {
      | ^
during RTL pass: expand
testcase.c:9:1: internal compiler error: in emit_move_insn, at expr.c:4026
    9 | foo (void)
      | ^~~
0x6be049 emit_move_insn(rtx_def*, rtx_def*)
        /repo/gcc-trunk/gcc/expr.c:4026
0xe3f43a expand_gimple_stmt_1
        /repo/gcc-trunk/gcc/cfgexpand.c:4006
0xe3f43a expand_gimple_stmt
        /repo/gcc-trunk/gcc/cfgexpand.c:4040
0xe4555a expand_gimple_basic_block
        /repo/gcc-trunk/gcc/cfgexpand.c:6082
0xe47207 execute
        /repo/gcc-trunk/gcc/cfgexpand.c:6808
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.

$ x86_64-pc-linux-gnu-gcc -v
Using built-in specs.
COLLECT_GCC=/repo/gcc-trunk/binary-latest-amd64/bin/x86_64-pc-linux-gnu-gcc
COLLECT_LTO_WRAPPER=/repo/gcc-trunk/binary-trunk-r12-3878-20210924101619-g710c6ab4ad5-checking-yes-rtl-df-extra-nobootstrap-amd64/bin/../libexec/gcc/x86_64-pc-linux-gnu/12.0.0/lto-wrapper
Target: x86_64-pc-linux-gnu
Configured with: /repo/gcc-trunk//configure --enable-languages=c,c++
--enable-valgrind-annotations --disable-nls --enable-checking=yes,rtl,df,extra
--disable-bootstrap --with-cloog --with-ppl --with-isl
--build=x86_64-pc-linux-gnu --host=x86_64-pc-linux-gnu
--target=x86_64-pc-linux-gnu --with-ld=/usr/bin/x86_64-pc-linux-gnu-ld
--with-as=/usr/bin/x86_64-pc-linux-gnu-as --disable-libstdcxx-pch
--prefix=/repo/gcc-trunk//binary-trunk-r12-3878-20210924101619-g710c6ab4ad5-checking-yes-rtl-df-extra-nobootstrap-amd64
Thread model: posix
Supported LTO compression algorithms: zlib zstd
gcc version 12.0.0 20210924 (experimental) (GCC)

[Bug target/102477] ICE: in emit_move_insn, at expr.c:4026 at -Og with __builtin_shufflevector()

[rguenth at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=102477

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

           What    |Removed                     |Added
----------------------------------------------------------------------------
   Last reconfirmed|                            |2021-09-24
             Status|UNCONFIRMED                 |NEW
     Ever confirmed|0                           |1

--- Comment #1 from Richard Biener <rguenth at gcc dot gnu.org> ---
Confirmed.  We have

_2 = (vector(16) signed int) v.0_1;

where v.0_1 is vector(16) unsigned int.  So appearantly we have an optab for
that but it doesn't work here.

[Bug target/102477] ICE: in emit_move_insn, at expr.c:4026 at -Og with __builtin_shufflevector()

[jakub at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=102477

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

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

--- Comment #2 from Jakub Jelinek <jakub at gcc dot gnu.org> ---
Such conversion is just reinterpretation of the bits in the same mode, nop at
RTL level.  So it should just work.

[Bug target/102477] ICE: in emit_move_insn, at expr.c:4026 at -Og with __builtin_shufflevector()

[zsojka at seznam dot cz]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=102477

--- Comment #3 from Zdenek Sojka <zsojka at seznam dot cz> ---
Created attachment 51517
  --> https://gcc.gnu.org/bugzilla/attachment.cgi?id=51517&action=edit
different testcase failing at -O without __builtin_shufflevector()

$ x86_64-pc-linux-gnu-gcc -O testcase.c 
during RTL pass: expand
testcase.c: In function 'foo':
testcase.c:7:9: internal compiler error: in emit_move_insn, at expr.c:4026
    7 |   V v = -(g >> sizeof (g) - 1);
      |         ^~~~~~~~~~~~~~~~~~~~~~
0x6b8206 emit_move_insn(rtx_def*, rtx_def*)
        /repo/gcc-trunk/gcc/expr.c:4026
0xe36893 expand_gimple_stmt_1
        /repo/gcc-trunk/gcc/cfgexpand.c:4006
0xe36893 expand_gimple_stmt
        /repo/gcc-trunk/gcc/cfgexpand.c:4040
0xe3c98f expand_gimple_basic_block
        /repo/gcc-trunk/gcc/cfgexpand.c:6082
0xe3e55e execute
        /repo/gcc-trunk/gcc/cfgexpand.c:6808
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 target/102477] [12 Regression] ICE: in emit_move_insn, at expr.c:4026 at -Og and -O

[zsojka at seznam dot cz]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=102477

--- Comment #4 from Zdenek Sojka <zsojka at seznam dot cz> ---
Created attachment 51557
  --> https://gcc.gnu.org/bugzilla/attachment.cgi?id=51557&action=edit
another testcase, failing at -O

$ x86_64-pc-linux-gnu-gcc -O testcase.c
during RTL pass: expand
testcase.c: In function 'foo':
testcase.c:7:8: internal compiler error: in emit_move_insn, at expr.c:4026
    7 |   v *= -(v >> 4 * 8 - 1);
      |        ^~~~~~~~~~~~~~~~~
0x6b91fa emit_move_insn(rtx_def*, rtx_def*)
        /repo/gcc-trunk/gcc/expr.c:4026
0xe38a03 expand_gimple_stmt_1
        /repo/gcc-trunk/gcc/cfgexpand.c:4006
0xe38a03 expand_gimple_stmt
        /repo/gcc-trunk/gcc/cfgexpand.c:4040
0xe3eaff expand_gimple_basic_block
        /repo/gcc-trunk/gcc/cfgexpand.c:6082
0xe406ce execute
        /repo/gcc-trunk/gcc/cfgexpand.c:6808
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 target/102477] [12 Regression] ICE: in emit_move_insn, at expr.c:4026 at -Og and -O

[pinskia at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=102477

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

           What    |Removed                     |Added
----------------------------------------------------------------------------
   Target Milestone|---                         |12.0

[Bug target/102477] [12 Regression] ICE: in emit_move_insn, at expr.c:4026 at -Og and -O

[rguenth at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=102477

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

           What    |Removed                     |Added
----------------------------------------------------------------------------
         Resolution|---                         |FIXED
             Blocks|82738                       |
             Status|NEW                         |RESOLVED

--- Comment #5 from Richard Biener <rguenth at gcc dot gnu.org> ---
All of the testcases work now.


Referenced Bugs:

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=82738
[Bug 82738] [meta-bug] issues with the -Og optimization level