[Bug c++/115038] New: ICE in seh_cfa_offset since 14.1.0

[Bug c++/115038] New: ICE in seh_cfa_offset since 14.1.0

[reiter.christoph at gmail dot com]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=115038

            Bug ID: 115038
           Summary: ICE in seh_cfa_offset since 14.1.0
           Product: gcc
           Version: 14.1.0
            Status: UNCONFIRMED
          Severity: normal
          Priority: P3
         Component: c++
          Assignee: unassigned at gcc dot gnu.org
          Reporter: reiter.christoph at gmail dot com
  Target Milestone: ---

Downstream issue: https://github.com/msys2/MINGW-packages/issues/20861

The following code using GCC 14.1.0 on Windows targeting x86_64-w64-mingw32:

```
// gcc -c -fno-omit-frame-pointer -O2 repro.cpp
template <typename a> void b(a, a);
template <typename a> void c(a, a, float);
float d;
void e() {
  float f;
  c(f, f, d);
  b(0.0f, f);
}
```

results in

```
repro.cpp: In function 'void e()':
repro.cpp:8:1: internal compiler error: in seh_cfa_offset, at
config/i386/winnt.cc:1137
    8 | }
      | ^
```

* Replacing "-O2" with "-O1" fixes the issue
* Removing "-fno-omit-frame-pointer" fixes the issue
* Downgrading to GCC 13.2.0 fixes the issue

If there is any other information I can provide, let me know.

[Bug target/115038] ICE in seh_cfa_offset since 14.1.0

[reiter.christoph at gmail dot com]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=115038

--- Comment #1 from Christoph Reiter <reiter.christoph at gmail dot com> ---
Another downstream report with the same error:
https://github.com/msys2/MINGW-packages/issues/20864

I've c-reduced that as well to:

```
// gcc -c -fno-omit-frame-pointer -O2 repro.cpp
struct d {
  d();
};
struct e {
  e() : c(1.0) {}
  float c;
};
struct j {
  using ad = d;
};
class k : j {
  d g;
  e h;
};
template <typename, typename, typename, typename> using i = k;
class {
  i<int, int, int, int> f;
} a;
k b;
```

[Bug target/115038] [14/15 regression] internal error in seh_cfa_offset with -O2 -fno-omit-frame-pointer

[ebotcazou at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=115038

Eric Botcazou <ebotcazou at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |ebotcazou at gcc dot gnu.org
   Target Milestone|---                         |14.2
             Status|UNCONFIRMED                 |NEW
     Ever confirmed|0                           |1
   Last reconfirmed|                            |2024-05-12
            Summary|ICE in seh_cfa_offset since |[14/15 regression] internal
                   |14.1.0                      |error in seh_cfa_offset
                   |                            |with -O2
                   |                            |-fno-omit-frame-pointer

--- Comment #2 from Eric Botcazou <ebotcazou at gcc dot gnu.org> ---
I can reproduce.

[Bug target/115038] [14/15 regression] internal error in seh_cfa_offset with -O2 -fno-omit-frame-pointer

[ebotcazou at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=115038

--- Comment #3 from Eric Botcazou <ebotcazou at gcc dot gnu.org> ---
It's the -ffold-mem-offsets pass messing up with the prologue and the
frame-related instructions (so -fno-fold-mem-offsets is a workaround).

[Bug target/115038] [14/15 regression] internal error in seh_cfa_offset with -O2 -fno-omit-frame-pointer

[ebotcazou at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=115038

--- Comment #4 from Eric Botcazou <ebotcazou at gcc dot gnu.org> ---
Tentative fixlet:

diff --git a/gcc/fold-mem-offsets.cc b/gcc/fold-mem-offsets.cc
index 2e15b05529e..84b9623058b 100644
--- a/gcc/fold-mem-offsets.cc
+++ b/gcc/fold-mem-offsets.cc
@@ -491,7 +491,7 @@ fold_offsets (rtx_insn *insn, rtx reg, bool analyze, bitmap
foldable_insns)
 {
   rtx_insn *def = get_single_def_in_bb (insn, reg);

-  if (!def || GET_CODE (PATTERN (def)) != SET)
+  if (!def || RTX_FRAME_RELATED_P (def) || GET_CODE (PATTERN (def)) != SET)
     return 0;

   rtx dest = SET_DEST (PATTERN (def));

[Bug target/115038] [14/15 regression] internal error in seh_cfa_offset with -O2 -fno-omit-frame-pointer

[reiter.christoph at gmail dot com]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=115038

--- Comment #5 from Christoph Reiter <reiter.christoph at gmail dot com> ---
Thanks! I can confirm that "-fno-fold-mem-offsets" works around the issue. I'll
test the proposed patch tomorrow.

In the meantime we've reduced another instance of this when building emacs, but
this time with "simpler" C code and without -fno-omit-frame-pointer:

```
// gcc -c -O2 repro.c -> internal compiler error: in seh_cfa_offset
// gcc -c -g3 -O2 -gdwarf-2 repro.c -> internal compiler error: in
dwarf2out_frame_debug_expr
void a() {
    __builtin_unwind_init();
}
```

Adding this here for completeness.

[Bug target/115038] [14/15 regression] internal error in seh_cfa_offset with -O2 -fno-omit-frame-pointer

[reiter.christoph at gmail dot com]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=115038

--- Comment #6 from Christoph Reiter <reiter.christoph at gmail dot com> ---
I can confirm that the patch from comment 4 fixes all the above cases, and also
emacs builds again.

Thanks again

[Bug target/115038] [14/15 regression] internal error in seh_cfa_offset with -O2 -fno-omit-frame-pointer

[rguenth at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=115038

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

           What    |Removed                     |Added
----------------------------------------------------------------------------
           Priority|P3                          |P2
             Target|X86_64-mingw32              |x86_64-mingw32
      Known to work|                            |13.2.0

[Bug rtl-optimization/115038] [14/15 regression] internal error in seh_cfa_offset with -O2 -fno-omit-frame-pointer

[ebotcazou at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=115038

Eric Botcazou <ebotcazou at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
          Component|target                      |rtl-optimization
           Keywords|EH                          |

--- Comment #7 from Eric Botcazou <ebotcazou at gcc dot gnu.org> ---
Recategorizing.

[Bug rtl-optimization/115038] [14/15 regression] internal error in seh_cfa_offset with -O2 -fno-omit-frame-pointer

[ebotcazou at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=115038

Eric Botcazou <ebotcazou at gcc dot gnu.org> changed:

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

[Bug rtl-optimization/115038] [14/15 regression] internal error in seh_cfa_offset with -O2 -fno-omit-frame-pointer

[law at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=115038

--- Comment #8 from Jeffrey A. Law <law at gcc dot gnu.org> ---
Yea, I would think we want to avoid anything marked as frame related. 
Otherwise we have to go back and fixup the CFI nodes and such.

Eric, do you want to handle the final bootstrap+regression test?  Or do you
want me to take it from here?

[Bug rtl-optimization/115038] [14/15 regression] internal error in seh_cfa_offset with -O2 -fno-omit-frame-pointer

[ebotcazou at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=115038

--- Comment #9 from Eric Botcazou <ebotcazou at gcc dot gnu.org> ---
> Eric, do you want to handle the final bootstrap+regression test?  Or do you
> want me to take it from here?

I can do it right now.

[Bug rtl-optimization/115038] [14/15 regression] internal error in seh_cfa_offset with -O2 -fno-omit-frame-pointer

[cvs-commit at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=115038

--- Comment #10 from GCC Commits <cvs-commit at gcc dot gnu.org> ---
The master branch has been updated by Eric Botcazou <ebotcazou@gcc.gnu.org>:

https://gcc.gnu.org/g:f14ef5cfd4c1ba1d34afda9174935e40d3c0a3ce

commit r15-776-gf14ef5cfd4c1ba1d34afda9174935e40d3c0a3ce
Author: Eric Botcazou <ebotcazou@adacore.com>
Date:   Wed May 22 18:10:39 2024 +0200

    Fix internal error in seh_cfa_offset with -O2 -fno-omit-frame-pointer

    The problem directly comes from the -ffold-mem-offsets pass messing up with
    the prologue and the frame-related instructions, which is a no-no with SEH,
    so the fix simply disconnects the pass in these circumstances.

    gcc/
            PR rtl-optimization/115038
            * fold-mem-offsets.cc (fold_offsets): Return 0 if the defining
            instruction of the register is frame related.

    gcc/testsuite/
            * g++.dg/opt/fmo1.C: New test.

[Bug rtl-optimization/115038] [14/15 regression] internal error in seh_cfa_offset with -O2 -fno-omit-frame-pointer

[cvs-commit at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=115038

--- Comment #11 from GCC Commits <cvs-commit at gcc dot gnu.org> ---
The releases/gcc-14 branch has been updated by Eric Botcazou
<ebotcazou@gcc.gnu.org>:

https://gcc.gnu.org/g:2f0e0862406a17bb8bf4ad948ae22916bae092a0

commit r14-10233-g2f0e0862406a17bb8bf4ad948ae22916bae092a0
Author: Eric Botcazou <ebotcazou@adacore.com>
Date:   Wed May 22 18:10:39 2024 +0200

    Fix internal error in seh_cfa_offset with -O2 -fno-omit-frame-pointer

    The problem directly comes from the -ffold-mem-offsets pass messing up with
    the prologue and the frame-related instructions, which is a no-no with SEH,
    so the fix simply disconnects the pass in these circumstances.

    gcc/
            PR rtl-optimization/115038
            * fold-mem-offsets.cc (fold_offsets): Return 0 if the defining
            instruction of the register is frame related.

    gcc/testsuite/
            * g++.dg/opt/fmo1.C: New test.

[Bug rtl-optimization/115038] [14/15 regression] internal error in seh_cfa_offset with -O2 -fno-omit-frame-pointer

[ebotcazou at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=115038

Eric Botcazou <ebotcazou at gcc dot gnu.org> changed:

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

--- Comment #12 from Eric Botcazou <ebotcazou at gcc dot gnu.org> ---
Thanks for reporting the problem.