[Bug target/90689] [10 Regression] ICE in extract_insn, at recog.c:2310 on ppc64le

[amodra at gmail dot com]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=90689

Alan Modra  changed:

   What|Removed |Added

 Status|NEW |RESOLVED
 CC|amodra at gcc dot gnu.org, |
   |amodra at gmail dot com|
 Resolution|--- |FIXED

--- Comment #7 from Alan Modra  ---
Fixed

[Bug target/90689] [10 Regression] ICE in extract_insn, at recog.c:2310 on ppc64le

[amodra at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=90689

--- Comment #6 from Alan Modra  ---
Author: amodra
Date: Tue Jun  4 00:13:07 2019
New Revision: 271895

URL: https://gcc.gnu.org/viewcvs?rev=271895=gcc=rev
Log:
PR90689, ICE in extract_insn on ppc64le

PR target/90689
* config/rs6000/rs6000.c (rs6000_call_aix): Correct r271753 merge
error.

Modified:
trunk/gcc/ChangeLog
trunk/gcc/config/rs6000/rs6000.c

[Bug target/90689] [10 Regression] ICE in extract_insn, at recog.c:2310 on ppc64le

[amodra at gmail dot com]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=90689

Alan Modra  changed:

   What|Removed |Added

 CC||amodra at gmail dot com
   Assignee|unassigned at gcc dot gnu.org  |amodra at gmail dot com

--- Comment #5 from Alan Modra  ---
It's a merge error.  The correct condition for the block shown in comment #4 is

  if (is_pltseq_longcall)

I'll give that a quick test and commit as obvious.

[Bug target/90689] [10 Regression] ICE in extract_insn, at recog.c:2310 on ppc64le

[segher at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=90689

--- Comment #4 from Segher Boessenkool  ---
rs6000.c does

  if (HAVE_AS_PLTSEQ
  && DEFAULT_ABI == ABI_ELFv2
  && GET_CODE (func_desc) == SYMBOL_REF)
{
  rtvec v = gen_rtvec (3, toc_reg, func_desc, tlsarg);
  rtx mark_toc_reg = gen_rtx_UNSPEC (Pmode, v, UNSPEC_PLTSEQ);
  emit_insn (gen_rtx_SET (stack_toc_mem, mark_toc_reg));
}
  else
emit_move_insn (stack_toc_mem, toc_reg);

but the insn condition for *pltseq_tocsave_ is

  "TARGET_PLTSEQ
   && DEFAULT_ABI == ABI_ELFv2"

I think that HAVE_AS_PLTSEQ there should be TARGET_PLTSEQ?

[Bug target/90689] [10 Regression] ICE in extract_insn, at recog.c:2310 on ppc64le

[segher at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=90689

Segher Boessenkool  changed:

   What|Removed |Added

 Status|UNCONFIRMED |NEW
   Last reconfirmed|2019-05-31 00:00:00 |2019-06-03
 Ever confirmed|0   |1

--- Comment #3 from Segher Boessenkool  ---
I have reproduced it now (on a native build).  It needs all of
  -mno-pltseq -fno-inline -fno-plt -Os -mabi=elfv2
to fail.

[Bug target/90689] [10 Regression] ICE in extract_insn, at recog.c:2310 on ppc64le

[marxin at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=90689

Martin Liška  changed:

   What|Removed |Added

 CC||amodra at gcc dot gnu.org,
   ||wschmidt at gcc dot gnu.org

--- Comment #2 from Martin Liška  ---
Started with r271753. I use normal cross compiler:

Configured with: ../configure --enable-languages=c,c++,fortran
--disable-multilib --prefix=/home/marxin/bin/gcc2 --disable-bootstrap
--disable-libsanitizer --target=ppc64le-linux-gnu
--with-as=/usr/bin/powerpc64le-suse-linux-as : (reconfigured)

[Bug target/90689] [10 Regression] ICE in extract_insn, at recog.c:2310 on ppc64le

[segher at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=90689

--- Comment #1 from Segher Boessenkool  ---
I cannot reproduce this (on a native compiler).  Do you use some non-default
defaults, maybe?  And, what SVN revision did you test?