[Bug target/105993] [13 Regression] ICE: RTL check: expected code 'reg', have 'subreg' in rhs_regno, at rtl.h:1932 with -O -mxop

2022-06-17 Thread ubizjak at gmail dot com via Gcc-bugs 
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=105993

Uroš Bizjak  changed:

   What|Removed |Added

 Resolution|--- |FIXED
   Target Milestone|--- |13.0
 Status|UNCONFIRMED |RESOLVED

--- Comment #4 from Uroš Bizjak  ---
Fixed.

[Bug target/105993] [13 Regression] ICE: RTL check: expected code 'reg', have 'subreg' in rhs_regno, at rtl.h:1932 with -O -mxop

2022-06-17 Thread cvs-commit at gcc dot gnu.org via Gcc-bugs 
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=105993

--- Comment #3 from CVS Commits  ---
The master branch has been updated by Uros Bizjak :

https://gcc.gnu.org/g:1d6044c250e3badfa2a403fee670b295106bf4fc

commit r13-1155-g1d6044c250e3badfa2a403fee670b295106bf4fc
Author: Uros Bizjak 
Date:   Fri Jun 17 16:22:20 2022 +0200

i386: Fix VPMOV splitter [PR105993]

REGNO should not be used with register_operand before reload because
subregs of registers or even subregs of memory match the predicate.
The build with RTL checking enabled does not tolerate REGNO with
non-reg operand.
The patch splits the splitter into two related splitters and uses
(match_dup ...) RTXes instead of REGNO comparisons.

2022-06-17  Uroš Bizjak  

gcc/ChangeLog:

PR target/105993
* config/i386/sse.md (vpmov splitter): Use (match_dup ...)
instead of REGNO comparisons in combine splitter.

gcc/testsuite/ChangeLog:

PR target/105993
* gcc.target/i386/pr105993.c: New test.

[Bug target/105993] [13 Regression] ICE: RTL check: expected code 'reg', have 'subreg' in rhs_regno, at rtl.h:1932 with -O -mxop

2022-06-16 Thread ubizjak at gmail dot com via Gcc-bugs 
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=105993

--- Comment #2 from Uroš Bizjak  ---
Created attachment 53149
  --> https://gcc.gnu.org/bugzilla/attachment.cgi?id=53149=edit
Patch to fix the failure

The patch fixes this particular failure by using (match_dup X). In general,
rtx_equal_p should be used before reload (also with register_operand). REGNO on
register_operand is valid only after reload (post-reload splitters and
peephole2 patterns).

[Bug target/105993] [13 Regression] ICE: RTL check: expected code 'reg', have 'subreg' in rhs_regno, at rtl.h:1932 with -O -mxop

2022-06-15 Thread crazylht at gmail dot com via Gcc-bugs 
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=105993

Hongtao.liu  changed:

   What|Removed |Added

 CC||crazylht at gmail dot com

--- Comment #1 from Hongtao.liu  ---
It looks like there're several places we use REGNO for register_operand before
RA, they're all problematic.