[Bug bootstrap/105551] New: [nvptx] ICE in final_scan_insn_1, at final.cc:2629 when building libgcc2.c since r13-259-g76db543db88727789a6c117608a23edc2eace713

[Bug bootstrap/105551] New: [nvptx] ICE in final_scan_insn_1, at final.cc:2629 when building libgcc2.c since r13-259-g76db543db88727789a6c117608a23edc2eace713

[burnus at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=105551

            Bug ID: 105551
           Summary: [nvptx] ICE in final_scan_insn_1, at final.cc:2629
                    when building libgcc2.c since
                    r13-259-g76db543db88727789a6c117608a23edc2eace713
           Product: gcc
           Version: 13.0
            Status: UNCONFIRMED
          Keywords: build, ice-on-valid-code
          Severity: normal
          Priority: P3
         Component: bootstrap
          Assignee: unassigned at gcc dot gnu.org
          Reporter: burnus at gcc dot gnu.org
                CC: rguenth at gcc dot gnu.org, tschwinge at gcc dot gnu.org,
                    vries at gcc dot gnu.org
  Target Milestone: ---
            Target: nvptx-none

..../gcc/xgcc \
-B.../gcc/ -B.../nvptx-none/bin/ -B.../nvptx-none/lib/ -isystem
.../nvptx-none/include -isystem .../nvptx-none/sys-include
--sysroot=.../nvptx-none   -g -O2 -mgomp -O2  -g -O2 -DIN_GCC 
-DCROSS_DIRECTORY_STRUCTURE  -W -Wall -Wno-narrowing -Wwrite-strings
-Wcast-qual -Wno-format -Wstrict-prototypes -Wmissing-prototypes
-Wold-style-definition  -isystem ./include   -g -DIN_LIBGCC2 -fbuilding-libgcc
-fno-stack-protector -Dinhibit_libc  -I. -I. -I../../.././gcc
-I.../src/gcc-mainline/libgcc -I.../src/gcc-mainline/libgcc/.
-I.../src/gcc-mainline/libgcc/../gcc -I.../src/gcc-mainline/libgcc/../include 
-DHAVE_CC_TLS -DUSE_EMUTLS  -o _muldi3.o -MT _muldi3.o -MD -MP -MF _muldi3.dep
-DL_muldi3 -c .../src/gcc-mainline/libgcc/libgcc2.c

during RTL pass: final
.../src/gcc-mainline/libgcc/libgcc2.c: In function '__multi3':
.../src/gcc-mainline/libgcc/libgcc2.c:538:1: internal compiler error: in
final_scan_insn_1, at final.cc:2629
  538 | }
      | ^
0x9a6288 final_scan_insn_1
        .../src/gcc-mainline/gcc/final.cc:2629
0x9a642b final_scan_insn(rtx_insn*, _IO_FILE*, int, int, int*)
        .../src/gcc-mainline/gcc/final.cc:2940
0x9a6545 final_1
        .../src/gcc-mainline/gcc/final.cc:1997
0x9a7564 rest_of_handle_final
        .../src/gcc-mainline/gcc/final.cc:4285
0x9a7564 execute
        .../src/gcc-mainline/gcc/final.cc:4363
Please submit a full bug report, with preprocessed source (by using
-freport-bug).



Bisecting points to r13-259-g76db543db88727789a6c117608a23edc2eace713

commit 76db543db88727789a6c117608a23edc2eace713
Author: Richard Biener <rguenther@suse.de>
Date:   Tue May 10 11:44:40 2022 +0200

    middle-end/105537 - debug processing

    The following makes sure to have a consistent state of
    flag_var_tracking_assignments with the distributed handling
    in process_options and finish_options by moving everything to
    finish_options which also restores diagnostics for
    -g0 -fvar-tracking which was lost with previous changes.

    2022-05-10  Richard Biener  <rguenther@suse.de>

            PR middle-end/105537
            * toplev.cc (process_options): Move flag_var_tracking
            handling ...
            * opts.cc (finish_options): ... here.

            * gcc.dg/torture/pr105537.c: New testcase.

[Bug bootstrap/105551] [13 Regression] [nvptx] ICE in final_scan_insn_1, at final.cc:2629 when building libgcc2.c since r13-259-g76db543db88727789a6c117608a23edc2eace713

[burnus at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=105551

Tobias Burnus <burnus at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
   Target Milestone|---                         |13.0
            Summary|[nvptx] ICE in              |[13 Regression] [nvptx] ICE
                   |final_scan_insn_1, at       |in final_scan_insn_1, at
                   |final.cc:2629 when building |final.cc:2629 when building
                   |libgcc2.c since             |libgcc2.c since
                   |r13-259-g76db543db88727789a |r13-259-g76db543db88727789a
                   |6c117608a23edc2eace713      |6c117608a23edc2eace713

--- Comment #1 from Tobias Burnus <burnus at gcc dot gnu.org> ---
Glancing at the current patch, it seems as if the

      || debug_hooks->var_location == do_nothing_debug_hooks.var_location)

check got lost.

[Bug bootstrap/105551] [13 Regression] [nvptx] ICE in final_scan_insn_1, at final.cc:2629 when building libgcc2.c since r13-259-g76db543db88727789a6c117608a23edc2eace713

[rguenther at suse dot de]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=105551

--- Comment #2 from rguenther at suse dot de <rguenther at suse dot de> ---
On Tue, 10 May 2022, burnus at gcc dot gnu.org wrote:

> https://gcc.gnu.org/bugzilla/show_bug.cgi?id=105551
> 
> Tobias Burnus <burnus at gcc dot gnu.org> changed:
> 
>            What    |Removed                     |Added
> ----------------------------------------------------------------------------
>    Target Milestone|---                         |13.0
>             Summary|[nvptx] ICE in              |[13 Regression] [nvptx] ICE
>                    |final_scan_insn_1, at       |in final_scan_insn_1, at
>                    |final.cc:2629 when building |final.cc:2629 when building
>                    |libgcc2.c since             |libgcc2.c since
>                    |r13-259-g76db543db88727789a |r13-259-g76db543db88727789a
>                    |6c117608a23edc2eace713      |6c117608a23edc2eace713
> 
> --- Comment #1 from Tobias Burnus <burnus at gcc dot gnu.org> ---
> Glancing at the current patch, it seems as if the
> 
>       || debug_hooks->var_location == do_nothing_debug_hooks.var_location)
> 
> check got lost.

Yes, there's no way to check it at this point since the debug hooks
are not yet set up.  I wonder whether this would be better done in
the var-tracking gate then?  Or where exactly does the failure
materialize?

I suppose since we check for dwarf_debuginfo_p () we could also
simply look at dwarf2_debug_hooks.var_location but then that
_always_ is dwarf2out_var_location, so maybe I'm missing something...

[Bug bootstrap/105551] [13 Regression] [nvptx] ICE in final_scan_insn_1, at final.cc:2629 when building libgcc2.c since r13-259-g76db543db88727789a6c117608a23edc2eace713

[rguenth at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=105551

--- Comment #3 from Richard Biener <rguenth at gcc dot gnu.org> ---
Ah, so I see nvptx_option_override sets flag_var_tracking to 0, but that should
be in effect at finish_option time already.

But as said, since we have

  if (opts->x_debug_info_level < DINFO_LEVEL_NORMAL
      || !dwarf_debuginfo_p (opts))

the debug_hooks->var_location == do_nothing_debug_hooks.var_location case
should be covered already - the debug hooks will be dwarf2_debug_hooks
(hopefully).

I will try to reproduce and understand.

[Bug bootstrap/105551] [13 Regression] [nvptx] ICE in final_scan_insn_1, at final.cc:2629 when building libgcc2.c since r13-259-g76db543db88727789a6c117608a23edc2eace713

[rguenth at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=105551

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

           What    |Removed                     |Added
----------------------------------------------------------------------------
   Last reconfirmed|                            |2022-05-11
             Status|UNCONFIRMED                 |ASSIGNED
     Ever confirmed|0                           |1
           Assignee|unassigned at gcc dot gnu.org      |rguenth at gcc dot gnu.org

--- Comment #4 from Richard Biener <rguenth at gcc dot gnu.org> ---
diff --git a/gcc/opts.cc b/gcc/opts.cc
index c9badd241a0..23de3c579a6 100644
--- a/gcc/opts.cc
+++ b/gcc/opts.cc
@@ -1334,11 +1334,15 @@ finish_options (struct gcc_options *opts, struct
gcc_options *opts_set,
              || opts->x_flag_selective_scheduling2));

   /* We know which debug output will be used so we can set flag_var_tracking
-     and flag_var_tracking_uninit if the user has not specified them.  Note
-     we have not yet initialized debug_hooks so we might uselessly run
-     var-tracking on targets without var_location debug hook support.  */
+     and flag_var_tracking_uninit if the user has not specified them.  */
   if (opts->x_debug_info_level < DINFO_LEVEL_NORMAL
-      || !dwarf_debuginfo_p (opts))
+      || !dwarf_debuginfo_p (opts)
+      /* We have not yet initialized debug hooks so match that to check
+        whether we're only doing DWARF2_LINENO_DEBUGGING_INFO.  */
+#ifndef DWARF2_DEBUGGING_INFO
+      || 1
+#endif
+     )
     {
       if ((opts_set->x_flag_var_tracking && opts->x_flag_var_tracking == 1)
          || (opts_set->x_flag_var_tracking_uninit


seems to get past the failure point - can you check a full build with this?

[Bug bootstrap/105551] [13 Regression] [nvptx] ICE in final_scan_insn_1, at final.cc:2629 when building libgcc2.c since r13-259-g76db543db88727789a6c117608a23edc2eace713

[burnus at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=105551

--- Comment #5 from Tobias Burnus <burnus at gcc dot gnu.org> ---
(In reply to Richard Biener from comment #4)
> --- a/gcc/opts.cc
> +++ b/gcc/opts.cc

> +#ifndef DWARF2_DEBUGGING_INFO
> +      || 1
> +#endif

> seems to get past the failure point - can you check a full build with this?

Yes - it builds with that fix. Thanks! Side remark: I personally find 'true'
more readable than '1'.

[Bug bootstrap/105551] [13 Regression] [nvptx] ICE in final_scan_insn_1, at final.cc:2629 when building libgcc2.c since r13-259-g76db543db88727789a6c117608a23edc2eace713

[cvs-commit at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=105551

--- Comment #6 from CVS Commits <cvs-commit at gcc dot gnu.org> ---
The master branch has been updated by Richard Biener <rguenth@gcc.gnu.org>:

https://gcc.gnu.org/g:e7d9fdf5e0ee4c34a880139254340b4165016289

commit r13-285-ge7d9fdf5e0ee4c34a880139254340b4165016289
Author: Richard Biener <rguenther@suse.de>
Date:   Wed May 11 10:47:34 2022 +0200

    bootstrap/105551 - restore nvptx build

    The following makes sure to disable var-tracking if only
    dwarf2-line debuginfo is present.

    2022-05-11  Richard Biener  <rguenther@suse.de>

            PR bootstrap/105551
            * opts.cc (finish_options): Also disable var-tracking if
            !DWARF2_DEBUGGING_INFO.

[Bug bootstrap/105551] [13 Regression] [nvptx] ICE in final_scan_insn_1, at final.cc:2629 when building libgcc2.c since r13-259-g76db543db88727789a6c117608a23edc2eace713

[rguenth at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=105551

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

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

--- Comment #7 from Richard Biener <rguenth at gcc dot gnu.org> ---
Fixed.

[Bug bootstrap/105551] [13 Regression] [nvptx] ICE in final_scan_insn_1, at final.cc:2629 when building libgcc2.c since r13-259-g76db543db88727789a6c117608a23edc2eace713

[cvs-commit at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=105551

--- Comment #8 from CVS Commits <cvs-commit at gcc dot gnu.org> ---
The releases/gcc-12 branch has been updated by Jakub Jelinek
<jakub@gcc.gnu.org>:

https://gcc.gnu.org/g:a6a0f3423f3053999c0eb6e7183319c1dca6455d

commit r12-8526-ga6a0f3423f3053999c0eb6e7183319c1dca6455d
Author: Richard Biener <rguenther@suse.de>
Date:   Wed May 11 10:47:34 2022 +0200

    bootstrap/105551 - restore nvptx build

    The following makes sure to disable var-tracking if only
    dwarf2-line debuginfo is present.

    2022-05-11  Richard Biener  <rguenther@suse.de>

            PR bootstrap/105551
            * opts.cc (finish_options): Also disable var-tracking if
            !DWARF2_DEBUGGING_INFO.

    (cherry picked from commit e7d9fdf5e0ee4c34a880139254340b4165016289)