[Bug target/107786] New: [13 Regression] ICE in extract_insn, at recog.cc:2791 since r13-4151-gacbb5ef06ee978

[Bug target/107786] New: [13 Regression] ICE in extract_insn, at recog.cc:2791 since r13-4151-gacbb5ef06ee978

[marxin at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=107786

            Bug ID: 107786
           Summary: [13 Regression] ICE in extract_insn, at recog.cc:2791
                    since r13-4151-gacbb5ef06ee978
           Product: gcc
           Version: 13.0
            Status: UNCONFIRMED
          Severity: normal
          Priority: P3
         Component: target
          Assignee: unassigned at gcc dot gnu.org
          Reporter: marxin at gcc dot gnu.org
                CC: ptomsich at gcc dot gnu.org
  Target Milestone: ---
              Host: x86_64-linux-gnu
            Target: riscv64-unknown-linux-gnu

The following crashes now:

$ cat riscv.c
int c;

int main() {
  for (;;) {
    char h = c * 100;
    if (h)
      break;
  }
}

$ /dev/shm/objdir2/gcc/xgcc -B /dev/shm/objdir2/gcc/ riscv.c -O2 -S
riscv.c: In function ‘main’:
riscv.c:9:1: error: unrecognizable insn:
    9 | }
      | ^
(jump_insn 37 36 15 2 (set (pc)
        (if_then_else (ne (reg:SI 14 a4 [147])
                (const_int 0 [0]))
            (label_ref 18)
            (pc))) "riscv.c":6:8 -1
     (int_list:REG_BR_PROB 355222868 (nil))
 -> 18)
during RTL pass: split2
riscv.c:9:1: internal compiler error: in extract_insn, at recog.cc:2791
0x13ef4be _fatal_insn(char const*, rtx_def const*, char const*, int, char
const*)
        /home/marxin/Programming/gcc2/gcc/rtl-error.cc:108
0x13ef4ff _fatal_insn_not_found(rtx_def const*, char const*, int, char const*)
        /home/marxin/Programming/gcc2/gcc/rtl-error.cc:116
0x139e8a7 extract_insn(rtx_insn*)
        /home/marxin/Programming/gcc2/gcc/recog.cc:2791
0x139e563 extract_insn_cached(rtx_insn*)
        /home/marxin/Programming/gcc2/gcc/recog.cc:2680
0xf3dd44 cleanup_subreg_operands(rtx_insn*)
        /home/marxin/Programming/gcc2/gcc/final.cc:3054
0x139ff38 split_insn
        /home/marxin/Programming/gcc2/gcc/recog.cc:3420
0x13a00b8 split_all_insns()
        /home/marxin/Programming/gcc2/gcc/recog.cc:3488
0x13a2022 execute
        /home/marxin/Programming/gcc2/gcc/recog.cc:4457
Please submit a full bug report, with preprocessed source (by using
-freport-bug).
Please include the complete backtrace with any bug report.
See <https://gcc.gnu.org/bugs/> for instructions.

[Bug target/107786] [13 Regression] ICE in extract_insn, at recog.cc:2791 since r13-4151-gacbb5ef06ee978

[marxin at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=107786

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

           What    |Removed                     |Added
----------------------------------------------------------------------------
   Last reconfirmed|                            |2022-11-21
     Ever confirmed|0                           |1
   Target Milestone|---                         |13.0
             Status|UNCONFIRMED                 |NEW
           Keywords|                            |ice-on-valid-code

[Bug target/107786] [13 Regression] ICE in extract_insn, at recog.cc:2791 since r13-4151-gacbb5ef06ee978

[ptomsich at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=107786

ptomsich at gcc dot gnu.org changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|NEW                         |ASSIGNED
           Assignee|unassigned at gcc dot gnu.org      |ptomsich at gcc dot gnu.org

--- Comment #1 from ptomsich at gcc dot gnu.org ---
Looks like a dependency between a recent change of ours and another change that
is still in discussion.

I'll take this one.

[Bug target/107786] [13 Regression] ICE in extract_insn, at recog.cc:2791 since r13-4151-gacbb5ef06ee978

[cvs-commit at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=107786

--- Comment #2 from CVS Commits <cvs-commit at gcc dot gnu.org> ---
The master branch has been updated by Philipp Tomsich <ptomsich@gcc.gnu.org>:

https://gcc.gnu.org/g:4c7d336b673df2f3bf23bc5e7a69c445a2320c04

commit r13-4201-g4c7d336b673df2f3bf23bc5e7a69c445a2320c04
Author: Philipp Tomsich <philipp.tomsich@vrull.eu>
Date:   Mon Nov 21 12:44:23 2022 +0100

    RISC-V: Fix ICE in branch<ANYI:mode>_shiftedarith_equals_zero

    With the recent improvements to the splitting of special cases of
    branch patterns on RISC-V, a dependency on an unmerged/in-discussion
    change for branch-equals-zero slipped in: this allowed a non-X mode to
    be presented to branch-equals-zero (where only X mode is permissible).

    This addresses the issue by wrapping the ANYI operand in a paradoxical
    SUBREG:X (the high bits can be safely ignored, as we we perform an
    and-immediate before the branch in the pattern).

    Tested against the GCC testsuite and committed as obvious.

    gcc/ChangeLog:

            PR target/107786
            * config/riscv/riscv.md
            (*branch<ANYI:mode>_shiftedarith_equals_zero): Wrap ANYI
            in a subreg, as our branch instructions only supports X.

    gcc/testsuite/ChangeLog:

            * gcc.target/riscv/pr107786-2.c: New test.
            * gcc.target/riscv/pr107786.c: New test.

[Bug target/107786] [13 Regression] ICE in extract_insn, at recog.cc:2791 since r13-4151-gacbb5ef06ee978

[marxin at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=107786

--- Comment #3 from Martin Liška <marxin at gcc dot gnu.org> ---
Thanks for the quick fix. Can I close it as fixed?

[Bug target/107786] [13 Regression] ICE in extract_insn, at recog.cc:2791 since r13-4151-gacbb5ef06ee978

[ptomsich at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=107786

ptomsich at gcc dot gnu.org changed:

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

--- Comment #4 from ptomsich at gcc dot gnu.org ---
I kicked off some additional testing that will run overnight.
Let's close it for now and I will reopen in case that there is further fallout.