[Bug target/109541] New: ICE in extract_constrain_insn on sparc64-unknown-linux-gnu when building rhash-1.4.3

[Bug target/109541] New: ICE in extract_constrain_insn on sparc64-unknown-linux-gnu when building rhash-1.4.3

[sjames at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=109541

            Bug ID: 109541
           Summary: ICE in extract_constrain_insn on
                    sparc64-unknown-linux-gnu when building rhash-1.4.3
           Product: gcc
           Version: 13.0
            Status: UNCONFIRMED
          Severity: normal
          Priority: P3
         Component: target
          Assignee: unassigned at gcc dot gnu.org
          Reporter: sjames at gcc dot gnu.org
  Target Milestone: ---
              Host: sparc64-unknown-linux-gnu
            Target: sparc64-unknown-linux-gnu
             Build: sparc64-unknown-linux-gnu

Created attachment 54880
  --> https://gcc.gnu.org/bugzilla/attachment.cgi?id=54880&action=edit
a-sha512.i

Hit this when building rhash-1.4.3
```
$ sparc64-unknown-linux-gnu-gcc -DUSE_OPENSSL -DNDEBUG -pipe
-ffunction-sections -fdata-sections -fomit-frame-pointer -Wall -W
-Wstrict-prototypes -Wnested-externs -Winline -Wpointer-arith
-Wbad-function-cast -Wmissing-prototypes -Wmissing-declarations
-Wdeclaration-after-statement -O2 -mcpu=niagara4 -pipe -fpic
-DRHASH_XVERSION=0x01040200 librhash/sha512.c
librhash/sha512.c: In function ‘rhash_sha512_process_block’:
librhash/sha512.c:179:1: error: insn does not satisfy its constraints:
  179 | }
      | ^
(insn 5 2048 675 2 (set (reg/f:DI 72 %f40 [2613])
        (mem/u/c:DI (plus:DI (reg:DI 3 %g3 [109])
                (symbol_ref:DI ("rhash_k512") [flags 0x2] <var_decl
0xfff800010eb1eea0 rhash_k512>)) [0  S8 A64])) "librhash/sha512.c":158:17 125
{*movdi_insn_sp64}
     (expr_list:REG_EQUIV (symbol_ref:DI ("rhash_k512") [flags 0x2] <var_decl
0xfff800010eb1eea0 rhash_k512>)
        (nil)))
during RTL pass: reload
librhash/sha512.c:179:1: internal compiler error: in extract_constrain_insn, at
recog.cc:2692
0x18521f7 internal_error(char const*, ...)
        ???:0
0x2cbd17 fancy_abort(char const*, int, char const*)
        ???:0
0x2b2b27 _fatal_insn(char const*, rtx_def const*, char const*, int, char
const*)
        ???:0
0x2b2bdf _fatal_insn_not_found(rtx_def const*, char const*, int, char const*)
        ???:0
0xa3b7b7 extract_constrain_insn(rtx_insn*)
        ???:0
0x8c3687 lra(_IO_FILE*)
        ???:0
Please submit a full bug report, with preprocessed source (by using
-freport-bug).
Please include the complete backtrace with any bug report.
See <https://bugs.gentoo.org/> for instructions.
```

---

With the attached preprocessed source, this is enough:
```
$ sparc64-unknown-linux-gnu-gcc -O1 -mcpu=niagara4 -fpic -c a-sha512.i
librhash/sha512.c: In function ‘rhash_sha512_process_block’:
librhash/sha512.c:179:1: error: insn does not satisfy its constraints:
(insn 5 2048 675 2 (set (reg/f:DI 72 %f40 [2613])
        (mem/u/c:DI (plus:DI (reg:DI 3 %g3 [109])
                (symbol_ref:DI ("rhash_k512") [flags 0x2] <var_decl
0xfff800010ed1e900 rhash_k512>)) [0  S8 A64])) "librhash/sha512.c":158:17 125
{*movdi_insn_sp64}
     (expr_list:REG_EQUIV (symbol_ref:DI ("rhash_k512") [flags 0x2] <var_decl
0xfff800010ed1e900 rhash_k512>)
        (nil)))
during RTL pass: reload
librhash/sha512.c:179:1: internal compiler error: in extract_constrain_insn, at
recog.cc:2692
0x18521f7 internal_error(char const*, ...)
        ???:0
0x2cbd17 fancy_abort(char const*, int, char const*)
        ???:0
0x2b2b27 _fatal_insn(char const*, rtx_def const*, char const*, int, char
const*)
        ???:0
0x2b2bdf _fatal_insn_not_found(rtx_def const*, char const*, int, char const*)
        ???:0
0xa3b7b7 extract_constrain_insn(rtx_insn*)
        ???:0
0x8c3687 lra(_IO_FILE*)
        ???:0
Please submit a full bug report, with preprocessed source (by using
-freport-bug).
Please include the complete backtrace with any bug report.
See <https://bugs.gentoo.org/> for instructions.
```

[Bug target/109541] ICE in extract_constrain_insn on sparc64-unknown-linux-gnu when building rhash-1.4.3

[sjames at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=109541

--- Comment #1 from Sam James <sjames at gcc dot gnu.org> ---
Maybe a dupe of PR108703? Need to rebuild with debugging symbols as well now.

[Bug target/109541] [12 regression] ICE in extract_constrain_insn on sparc64-unknown-linux-gnu when building rhash-1.4.3

[sjames at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=109541

Sam James <sjames at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
            Summary|ICE in                      |[12 regression] ICE in
                   |extract_constrain_insn on   |extract_constrain_insn on
                   |sparc64-unknown-linux-gnu   |sparc64-unknown-linux-gnu
                   |when building rhash-1.4.3   |when building rhash-1.4.3

--- Comment #2 from Sam James <sjames at gcc dot gnu.org> ---
With < 12, need -ftree-vectorize because the default changed. I can repro w/
10/11/12/13.

It's not very easy for me to bisect older versions because they don't e.g.
build with newer glibc and I always lose track of the patch(es) for those kind
of problems. But I'm pretty suspicious of this being a dupe still, so I guess
it's not so important.

[Bug target/109541] [12 regression] ICE in extract_constrain_insn on sparc64-unknown-linux-gnu when building rhash-1.4.3

[sjames at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=109541

--- Comment #3 from Sam James <sjames at gcc dot gnu.org> ---
(In reply to Sam James from comment #2)
> With < 12, need -ftree-vectorize because the default changed. I can repro w/
> 10/11/12/13.

Calling it a 12 regression because it now fails by default while the other PR
does not, but it's fine if you disagree.

[Bug target/109541] [12 regression] ICE in extract_constrain_insn on sparc64-unknown-linux-gnu when building rhash-1.4.3

[sjames at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=109541

--- Comment #4 from Sam James <sjames at gcc dot gnu.org> ---
(other PR fails with 10 etc even without -ftree-vectorize)

[Bug target/109541] [12/13/14 regression] ICE in extract_constrain_insn on sparc64-unknown-linux-gnu when building rhash-1.4.3

[rguenth at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=109541

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

           What    |Removed                     |Added
----------------------------------------------------------------------------
   Target Milestone|---                         |12.3
            Summary|[12 regression] ICE in      |[12/13/14 regression] ICE
                   |extract_constrain_insn on   |in extract_constrain_insn
                   |sparc64-unknown-linux-gnu   |on
                   |when building rhash-1.4.3   |sparc64-unknown-linux-gnu
                   |                            |when building rhash-1.4.3

[Bug target/109541] [12/13/14 regression] ICE in extract_constrain_insn on sparc64-unknown-linux-gnu when building rhash-1.4.3

[rguenth at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=109541

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

           What    |Removed                     |Added
----------------------------------------------------------------------------
   Target Milestone|12.3                        |12.4

--- Comment #5 from Richard Biener <rguenth at gcc dot gnu.org> ---
GCC 12.3 is being released, retargeting bugs to GCC 12.4.

[Bug target/109541] [12/13/14 regression] ICE in extract_constrain_insn on when building rhash-1.4.3

[ebotcazou at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=109541

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

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |ebotcazou at gcc dot gnu.org
            Summary|[12/13/14 regression] ICE   |[12/13/14 regression] ICE
                   |in extract_constrain_insn   |in extract_constrain_insn
                   |on                          |on when building
                   |sparc64-unknown-linux-gnu   |rhash-1.4.3
                   |when building rhash-1.4.3   |
             Status|UNCONFIRMED                 |NEW
   Last reconfirmed|                            |2023-05-12
     Ever confirmed|0                           |1

--- Comment #6 from Eric Botcazou <ebotcazou at gcc dot gnu.org> ---
I can reproduce.

[Bug target/109541] [12/13/14 regression] ICE in extract_constrain_insn on when building rhash-1.4.3

[ebotcazou at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=109541

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

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

--- Comment #7 from Eric Botcazou <ebotcazou at gcc dot gnu.org> ---
The problem is that LRA assigns a floating-point register to the PIC
pseudo-register (pic_offset_table_rtx) and the SPARC back-end is not prepared
for it.

Vladimir, would it be feasible to prevent this from happening?

[Bug target/109541] [12/13/14 regression] ICE in extract_constrain_insn on when building rhash-1.4.3

[ebotcazou at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=109541

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

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

[Bug target/109541] [12/13/14 regression] ICE in extract_constrain_insn on when building rhash-1.4.3

[vmakarov at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=109541

--- Comment #8 from Vladimir Makarov <vmakarov at gcc dot gnu.org> ---
(In reply to Eric Botcazou from comment #7)
> The problem is that LRA assigns a floating-point register to the PIC
> pseudo-register (pic_offset_table_rtx) and the SPARC back-end is not
> prepared for it.
> 
> Vladimir, would it be feasible to prevent this from happening?

Sure.  I'll work on this after my vacation (in one week).

[Bug target/109541] [12/13/14 regression] ICE in extract_constrain_insn on when building rhash-1.4.3

[vmakarov at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=109541

--- Comment #9 from Vladimir Makarov <vmakarov at gcc dot gnu.org> ---
(In reply to Eric Botcazou from comment #7)
> The problem is that LRA assigns a floating-point register to the PIC
> pseudo-register (pic_offset_table_rtx) and the SPARC back-end is not
> prepared for it.
> 
> Vladimir, would it be feasible to prevent this from happening?

Sorry, I can not reproduce it on gcc-11, gcc-12, and master using -O1
-mcpu=niagara4 -fpic -c a-sha512.i.

Fortunately, I can reproduce
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=108703.  So I'll start on PR108703
first.

[Bug target/109541] [12/13/14 regression] ICE in extract_constrain_insn on when building rhash-1.4.3

[ebotcazou at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=109541

--- Comment #10 from Eric Botcazou <ebotcazou at gcc dot gnu.org> ---
> Sorry, I can not reproduce it on gcc-11, gcc-12, and master using -O1
> -mcpu=niagara4 -fpic -c a-sha512.i.

You need a properly configured cross-compiler, which means that you need to
build and install binutils first.  Then the most reduced testcase is the one
I'm going to attach:

eric@fomalhaut:~/build/gcc/sparc64-linux> gcc/xgcc -Bgcc -S a-sha512.c -O
-mcpu=niagara4 -fpic
a-sha512.c: In function 'rhash_sha512_process_block':
a-sha512.c:15:9: warning: cast from pointer to integer of different size
[-Wpointer-to-int-cast]
   15 |     F = (unsigned) &G;
      |         ^
a-sha512.c:16:10: warning: cast from pointer to integer of different size
[-Wpointer-to-int-cast]
   16 |     T1 = (unsigned) (&T1 + (W_3 += rhash_sha512_process_block_block > 9
> W_4));
      |          ^
a-sha512.c:17:9: warning: cast from pointer to integer of different size
[-Wpointer-to-int-cast]
   17 |     C = (unsigned) (T1 + &E);
      |         ^
a-sha512.c:20:1: error: insn does not satisfy its constraints:
   20 | }
      | ^
(insn 8 107 9 2 (set (reg/f:DI 3 %g3 [165])
        (mem/u/c:DI (plus:DI (reg:DI 3 %g3 [109])
                (symbol_ref:DI ("rhash_sha512_process_block_i") <var_decl
0x7f9e0ed6d2d0 rhash_sha512_process_block_i>)) [0  S8 A64])) "a-sha512.c":10:10
discrim 1 125 {*movdi_insn_sp64}
     (expr_list:REG_EQUIV (symbol_ref:DI ("rhash_sha512_process_block_i")
<var_decl 0x7f9e0ed6d2d0 rhash_sha512_process_block_i>)
        (nil)))
during RTL pass: reload
a-sha512.c:20:1: internal compiler error: in extract_constrain_insn, at
recog.cc:2692
0x620dd7 _fatal_insn(char const*, rtx_def const*, char const*, int, char
const*)
        /home/eric/cvs/gcc/gcc/rtl-error.cc:108
0x620e63 _fatal_insn_not_found(rtx_def const*, char const*, int, char const*)
        /home/eric/cvs/gcc/gcc/rtl-error.cc:118
0xd3dadd extract_constrain_insn(rtx_insn*)
        /home/eric/cvs/gcc/gcc/recog.cc:2692
0xbb9c37 check_rtl
        /home/eric/cvs/gcc/gcc/lra.cc:2126
0xbc08db lra(_IO_FILE*)
        /home/eric/cvs/gcc/gcc/lra.cc:2544
0xb5d889 do_reload
        /home/eric/cvs/gcc/gcc/ira.cc:5967
0xb5d889 execute
        /home/eric/cvs/gcc/gcc/ira.cc:6153

> Fortunately, I can reproduce
> https://gcc.gnu.org/bugzilla/show_bug.cgi?id=108703.  So I'll start on
> PR108703 first.

This one is pathological though, so of a lower priority IMO.

[Bug target/109541] [12/13/14 regression] ICE in extract_constrain_insn on when building rhash-1.4.3

[ebotcazou at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=109541

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

           What    |Removed                     |Added
----------------------------------------------------------------------------
  Attachment #54880|0                           |1
        is obsolete|                            |

--- Comment #11 from Eric Botcazou <ebotcazou at gcc dot gnu.org> ---
Created attachment 55227
  --> https://gcc.gnu.org/bugzilla/attachment.cgi?id=55227&action=edit
Reduced testcase

[Bug target/109541] [12/13/14 regression] ICE in extract_constrain_insn on when building rhash-1.4.3

[sjames at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=109541

--- Comment #12 from Sam James <sjames at gcc dot gnu.org> ---
(Vladimir (and anyone else interested): For bugs like this, you're welcome to
use gentoo's testing hardware if desired. Just email me an SSH key.)

[Bug target/109541] [12/13/14 regression] ICE in extract_constrain_insn on when building rhash-1.4.3

[sjames at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=109541

--- Comment #13 from Sam James <sjames at gcc dot gnu.org> ---
By the way, I think this needs --enable-checking=rtl, which is maybe why you
couldn't hit it before.

[Bug target/109541] [12/13/14 regression] ICE in extract_constrain_insn on when building rhash-1.4.3

[ebotcazou at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=109541

--- Comment #14 from Eric Botcazou <ebotcazou at gcc dot gnu.org> ---
> By the way, I think this needs --enable-checking=rtl, which is maybe why you
> couldn't hit it before.

No, that's incorrect.  You need a properly configured compiler instead.

[Bug target/109541] [12/13/14 regression] ICE in extract_constrain_insn on when building rhash-1.4.3

[sjames at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=109541

--- Comment #15 from Sam James <sjames at gcc dot gnu.org> ---
Okay. I couldn't hit it on some native machines until I rebuilt with more
checking, but maybe that was a fluke.

[Bug target/109541] [12/13/14 regression] ICE in extract_constrain_insn on when building rhash-1.4.3

[vmakarov at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=109541

--- Comment #16 from Vladimir Makarov <vmakarov at gcc dot gnu.org> ---
Sam, thank you for your help.  I've reproduced the problem on your machine.

The fix most probably will be ready this week.

[Bug target/109541] [12/13/14 regression] ICE in extract_constrain_insn on when building rhash-1.4.3

[sjames at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=109541

--- Comment #17 from Sam James <sjames at gcc dot gnu.org> ---
Thank you Vladimir! I appreciate the time both of you are spending on
alternative platforms for bugs like these.

Also, Vladimir, if you want to continue to use the machine after this, if you
ever have an occasion to do something on sparc, please feel free. Your account
will remain. The machine is made available for sourceware CI and for developers
to test for general stuff like this. Just email me if any questions or if you
want something installed, etc.

[Bug target/109541] [12/13/14 regression] ICE in extract_constrain_insn on when building rhash-1.4.3

[cvs-commit at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=109541

--- Comment #18 from CVS Commits <cvs-commit at gcc dot gnu.org> ---
The master branch has been updated by Vladimir Makarov <vmakarov@gcc.gnu.org>:

https://gcc.gnu.org/g:8cc8707446b77f9413654b31704f5a639673c916

commit r14-1610-g8cc8707446b77f9413654b31704f5a639673c916
Author: Vladimir N. Makarov <vmakarov@redhat.com>
Date:   Wed Jun 7 09:51:54 2023 -0400

    RA: Constrain class of pic offset table pseudo to general regs

    On some targets an integer pseudo can be assigned to a FP reg.  For
    pic offset table pseudo it means we will reload the pseudo in this
    case and, as a consequence, memory containing the pseudo might be
    recognized as wrong one.  The patch fix this problem.

            PR target/109541

    gcc/ChangeLog:

            * ira-costs.cc: (find_costs_and_classes): Constrain classes of pic
            offset table pseudo to a general reg subset.

    gcc/testsuite/ChangeLog:

            * gcc.target/sparc/pr109541.c: New.

[Bug target/109541] [12/13/14 regression] ICE in extract_constrain_insn on when building rhash-1.4.3

[ebotcazou at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=109541

--- Comment #19 from Eric Botcazou <ebotcazou at gcc dot gnu.org> ---
Thanks Vladimir!  Would you be OK with a backport to the 13 branch?

[Bug target/109541] [12/13 regression] ICE in extract_constrain_insn on when building rhash-1.4.3

[cvs-commit at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=109541

--- Comment #20 from CVS Commits <cvs-commit at gcc dot gnu.org> ---
The releases/gcc-13 branch has been updated by Vladimir Makarov
<vmakarov@gcc.gnu.org>:

https://gcc.gnu.org/g:5568d224f2e179b8e91a90fa4fba10bd8a7f963d

commit r13-7451-g5568d224f2e179b8e91a90fa4fba10bd8a7f963d
Author: Vladimir N. Makarov <vmakarov@redhat.com>
Date:   Wed Jun 7 09:51:54 2023 -0400

    RA: Constrain class of pic offset table pseudo to general regs

    On some targets an integer pseudo can be assigned to a FP reg.  For
    pic offset table pseudo it means we will reload the pseudo in this
    case and, as a consequence, memory containing the pseudo might be
    recognized as wrong one.  The patch fix this problem.

            PR target/109541

    gcc/ChangeLog:

            * ira-costs.cc: (find_costs_and_classes): Constrain classes of pic
            offset table pseudo to a general reg subset.

    gcc/testsuite/ChangeLog:

            * gcc.target/sparc/pr109541.c: New.

[Bug target/109541] [12 regression] ICE in extract_constrain_insn on when building rhash-1.4.3

[ebotcazou at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=109541

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

           What    |Removed                     |Added
----------------------------------------------------------------------------
            Summary|[12/13 regression] ICE in   |[12 regression] ICE in
                   |extract_constrain_insn on   |extract_constrain_insn on
                   |when building rhash-1.4.3   |when building rhash-1.4.3
         Resolution|---                         |FIXED
   Target Milestone|12.4                        |13.2
             Status|ASSIGNED                    |RESOLVED

--- Comment #21 from Eric Botcazou <ebotcazou at gcc dot gnu.org> ---
Fixed on 13 branch and mainline.