[Bug optimization/12630] New: [3.4 regression] Various unrecognizable insns and ICEs at -O3

[Bug optimization/12630] New: [3.4 regression] Various unrecognizable insns and ICEs at -O3

[danglin at gcc dot gnu dot org]

PLEASE REPLY TO gcc-bugzilla@gcc.gnu.org ONLY, *NOT* gcc-bugs@gcc.gnu.org.

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=12630

           Summary: [3.4 regression] Various unrecognizable insns and ICEs
                    at -O3
           Product: gcc
           Version: 3.4
            Status: UNCONFIRMED
          Severity: critical
          Priority: P2
         Component: optimization
        AssignedTo: unassigned at gcc dot gnu dot org
        ReportedBy: danglin at gcc dot gnu dot org
                CC: gcc-bugs at gcc dot gnu dot org,jh at suse dot cz
 GCC build triplet: hppa-unknown-linux-gnu, hppa2.0w-hp-hpux11.11
  GCC host triplet: hppa-unknown-linux-gnu, hppa2.0w-hp-hpux11.11
GCC target triplet: hppa-unknown-linux-gnu, hppa2.0w-hp-hpux11.11

The following patch introduced numerous regressions on the PA at -O3:

2003-10-11  Jan Hubicka  <jh@suse.cz>

        * Makefile.in (web.o): New.
        * web.c: New file.
        * rtl.h (web_main): Declare.
        * timervar.def (TV_WEB): New.
        * toplev.c (dump_file_index, dump_file_info): Add DFI_web.
        (rest_of_hanle_web): New.
        (flag_web): New static variable.
        (lang_independent_options): Add "web".
        (rest_of_compilation): Call rest_of_handle_web.
        * invoke.texi (-fweb): Document.
        * common.opt (fweb): New.
        * flags.h (flag_web): New.
        * opts.c (decode_options): Set flag_web at -O3.

        * passes.texi (web construction):  Document.
        * invoke.texi (-O3): Document that -fweb is enabled.

        * regrename.c (regrename_optimize): Deal better with situation when
        replacement failed.

        * sched-ebb.c: Include params.h and profile.h
        (schedule_ebbs):  Use tracer parameters to discover superblocks
        * Makefile.in (sched-ebb.o):  Add dependencies.

See for example the difference in testsuite results between
<http://gcc.gnu.org/ml/gcc-testresults/2003-10/msg00653.html>
and
<http://gcc.gnu.org/ml/gcc-testresults/2003-10/msg00612.html>.

Here is an example of one of the common failures:

dave@hiauly6:~/gnu/gcc-3.4/objdir/gcc/testsuite$
/home/dave/gnu/gcc-3.4/objdir/gcc/stage1/xgcc
-B/home/dave/gnu/gcc-3.4/objdir/gcc/stage1/   -O3  -w -c  -o 20011109-1.o
/home/dave/gnu/gcc-3.4/gcc/gcc/testsuite/gcc.c-torture/compile/20011109-1.c
/home/dave/gnu/gcc-3.4/gcc/gcc/testsuite/gcc.c-torture/compile/20011109-1.c: In
function `die':

/home/dave/gnu/gcc-3.4/gcc/gcc/testsuite/gcc.c-torture/compile/20011109-1.c:51:
error: unrecognizable insn:
(insn 188 187 189 17 (parallel [
            (set (mem/s:BLK (reg/f:SI 169 [ u ]) [12 S8 A16])
                (mem/s:BLK (reg/f:SI 170) [12 S8 A16]))
            (clobber (reg/f:SI 206 [ u ]))
            (clobber (reg/f:SI 205))
            (clobber (reg:SI 171))
            (clobber (reg:SI 172))
            (clobber (reg:SI 173))
            (use (const_int 8 [0x8]))
            (use (const_int 2 [0x2]))
        ]) -1 (insn_list 186 (insn_list 187 (nil)))
    (expr_list:REG_DEAD (reg/f:SI 170)
        (expr_list:REG_DEAD (reg/f:SI 169 [ u ])
            (expr_list:REG_UNUSED (reg/f:SI 206 [ u ])
                (expr_list:REG_UNUSED (reg/f:SI 205)
                    (expr_list:REG_UNUSED (reg:SI 171)
                        (expr_list:REG_UNUSED (reg:SI 172)
                            (expr_list:REG_UNUSED (reg:SI 173)
                                (expr_list:REG_EQUAL (mem/s:BLK (reg/f:SI 167)
[12 S8 A16])
                                    (nil))))))))))
/home/dave/gnu/gcc-3.4/gcc/gcc/testsuite/gcc.c-torture/compile/20011109-1.c:51:
internal compiler error: in extract_insn, at recog.c:2064
Please submit a full bug report,
with preprocessed source if appropriate.
See <URL:http://gcc.gnu.org/bugs.html> for instructions.

Another is

dave@hiauly6:~/gnu/gcc-3.4/objdir/gcc/testsuite$
/home/dave/gnu/gcc-3.4/objdir/gcc/stage1/xgcc
-B/home/dave/gnu/gcc-3.4/objdir/gcc/stage1/   -O3 -fomit-frame-pointer  -w -c 
-o structs.o
/home/dave/gnu/gcc-3.4/gcc/gcc/testsuite/gcc.c-torture/compile/structs.c
/home/dave/gnu/gcc-3.4/gcc/gcc/testsuite/gcc.c-torture/compile/structs.c: In
function `main':

/home/dave/gnu/gcc-3.4/gcc/gcc/testsuite/gcc.c-torture/compile/structs.c:263:
error: unrecognizable insn:
(insn 735 93 705 0 (set (plus:SI (reg/f:SI 30 %r30)
            (const_int -296 [0xfffffed8]))
        (reg:SI 23 %r23)) -1 (nil)
    (nil))
/home/dave/gnu/gcc-3.4/gcc/gcc/testsuite/gcc.c-torture/compile/structs.c:263:
internal compiler error: in extract_insn, at recog.c:2064

[Bug optimization/12630] [3.4 regression] Various unrecognizable insns and ICEs at -O3

[jh at suse dot cz]

PLEASE REPLY TO gcc-bugzilla@gcc.gnu.org ONLY, *NOT* gcc-bugs@gcc.gnu.org.

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=12630



------- Additional Comments From jh at suse dot cz  2003-10-15 23:31 -------
Subject: Re:  New: [3.4 regression] Various unrecognizable insns and ICEs at -O3

> PLEASE REPLY TO gcc-bugzilla@gcc.gnu.org ONLY, *NOT* gcc-bugs@gcc.gnu.org.
> 
> http://gcc.gnu.org/bugzilla/show_bug.cgi?id=12630
> 
>            Summary: [3.4 regression] Various unrecognizable insns and ICEs
>                     at -O3
>            Product: gcc
>            Version: 3.4
>             Status: UNCONFIRMED
>           Severity: critical
>           Priority: P2
>          Component: optimization
>         AssignedTo: unassigned at gcc dot gnu dot org
>         ReportedBy: danglin at gcc dot gnu dot org
>                 CC: gcc-bugs at gcc dot gnu dot org,jh at suse dot cz
>  GCC build triplet: hppa-unknown-linux-gnu, hppa2.0w-hp-hpux11.11
>   GCC host triplet: hppa-unknown-linux-gnu, hppa2.0w-hp-hpux11.11
> GCC target triplet: hppa-unknown-linux-gnu, hppa2.0w-hp-hpux11.11
> 
> The following patch introduced numerous regressions on the PA at -O3:
> 
> 2003-10-11  Jan Hubicka  <jh@suse.cz>
> 
>         * Makefile.in (web.o): New.
>         * web.c: New file.
>         * rtl.h (web_main): Declare.
>         * timervar.def (TV_WEB): New.
>         * toplev.c (dump_file_index, dump_file_info): Add DFI_web.
>         (rest_of_hanle_web): New.
>         (flag_web): New static variable.
>         (lang_independent_options): Add "web".
>         (rest_of_compilation): Call rest_of_handle_web.
>         * invoke.texi (-fweb): Document.
>         * common.opt (fweb): New.
>         * flags.h (flag_web): New.
>         * opts.c (decode_options): Set flag_web at -O3.
> 
>         * passes.texi (web construction):  Document.
>         * invoke.texi (-O3): Document that -fweb is enabled.
> 
>         * regrename.c (regrename_optimize): Deal better with situation when
>         replacement failed.
> 
>         * sched-ebb.c: Include params.h and profile.h
>         (schedule_ebbs):  Use tracer parameters to discover superblocks
>         * Makefile.in (sched-ebb.o):  Add dependencies.
> 
> See for example the difference in testsuite results between
> <http://gcc.gnu.org/ml/gcc-testresults/2003-10/msg00653.html>
> and
> <http://gcc.gnu.org/ml/gcc-testresults/2003-10/msg00612.html>.
> 
> Here is an example of one of the common failures:
> 
> dave@hiauly6:~/gnu/gcc-3.4/objdir/gcc/testsuite$
> /home/dave/gnu/gcc-3.4/objdir/gcc/stage1/xgcc
> -B/home/dave/gnu/gcc-3.4/objdir/gcc/stage1/   -O3  -w -c  -o 20011109-1.o
> /home/dave/gnu/gcc-3.4/gcc/gcc/testsuite/gcc.c-torture/compile/20011109-1.c
> /home/dave/gnu/gcc-3.4/gcc/gcc/testsuite/gcc.c-torture/compile/20011109-1.c: In
> function `die':
> 
> /home/dave/gnu/gcc-3.4/gcc/gcc/testsuite/gcc.c-torture/compile/20011109-1.c:51:
> error: unrecognizable insn:
> (insn 188 187 189 17 (parallel [
>             (set (mem/s:BLK (reg/f:SI 169 [ u ]) [12 S8 A16])
>                 (mem/s:BLK (reg/f:SI 170) [12 S8 A16]))
>             (clobber (reg/f:SI 206 [ u ]))
>             (clobber (reg/f:SI 205))

This is pa.md that is wrong.  
Does the attached patch help?

Index: pa.md
===================================================================
RCS file: /cvs/gcc/gcc/gcc/config/pa/pa.md,v
retrieving revision 1.131
diff -c -3 -p -r1.131 pa.md
*** pa.md	24 Sep 2003 01:38:32 -0000	1.131
--- pa.md	15 Oct 2003 23:30:14 -0000
***************
*** 3016,3025 ****
  ;; therefore it is forced to operand 2.  If the count is compile-time
  ;; determined, we need two scratch registers for the unrolled code.
  (define_insn "movstrsi_internal"
!   [(set (mem:BLK (match_operand:SI 0 "register_operand" "+r,r"))
! 	(mem:BLK (match_operand:SI 1 "register_operand" "+r,r")))
!    (clobber (match_dup 0))
!    (clobber (match_dup 1))
     (clobber (match_operand:SI 2 "register_operand" "=r,r"))	;loop cnt/tmp
     (clobber (match_operand:SI 3 "register_operand" "=&r,&r"))	;item tmp
     (clobber (match_operand:SI 6 "register_operand" "=&r,&r"))	;item tmp2
--- 3016,3025 ----
  ;; therefore it is forced to operand 2.  If the count is compile-time
  ;; determined, we need two scratch registers for the unrolled code.
  (define_insn "movstrsi_internal"
!   [(set (mem:BLK (match_operand:SI 0 "register_operand" "r,r"))
! 	(mem:BLK (match_operand:SI 1 "register_operand" "r,r")))
!    (clobber (match_operand:SI 7 "register_operand" "=0,0")))
!    (clobber (match_operand:SI 8 "register_operand" "=1,1"))
     (clobber (match_operand:SI 2 "register_operand" "=r,r"))	;loop cnt/tmp
     (clobber (match_operand:SI 3 "register_operand" "=&r,&r"))	;item tmp
     (clobber (match_operand:SI 6 "register_operand" "=&r,&r"))	;item tmp2

[Bug optimization/12630] [3.4 regression] Various unrecognizable insns and ICEs at -O3

[dave at hiauly1 dot hia dot nrc dot ca]

PLEASE REPLY TO gcc-bugzilla@gcc.gnu.org ONLY, *NOT* gcc-bugs@gcc.gnu.org.

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=12630



------- Additional Comments From dave at hiauly1 dot hia dot nrc dot ca  2003-10-16 01:00 -------
Subject: Re:  [3.4 regression] Various unrecognizable

> This is pa.md that is wrong.  
> Does the attached patch help?

Looks promising!  It would have taken me a long time to figure this
one out.  movstrsi_internal has been this way for a long time.

The patch appears to fix the second failure with structs.c.

Thanks,
Dave

[Bug optimization/12630] [3.4 regression] Various unrecognizable insns and ICEs at -O3

[jh at suse dot cz]

PLEASE REPLY TO gcc-bugzilla@gcc.gnu.org ONLY, *NOT* gcc-bugs@gcc.gnu.org.

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=12630



------- Additional Comments From jh at suse dot cz  2003-10-16 14:19 -------
Subject: Re:  New: [3.4 regression] Various unrecognizable insns and ICEs at -O3

> PLEASE REPLY TO gcc-bugzilla@gcc.gnu.org ONLY, *NOT* gcc-bugs@gcc.gnu.org.
> 
> http://gcc.gnu.org/bugzilla/show_bug.cgi?id=12630
> 
>            Summary: [3.4 regression] Various unrecognizable insns and ICEs
>                     at -O3
>            Product: gcc
>            Version: 3.4
>             Status: UNCONFIRMED
>           Severity: critical
>           Priority: P2
>          Component: optimization
>         AssignedTo: unassigned at gcc dot gnu dot org
>         ReportedBy: danglin at gcc dot gnu dot org
>                 CC: gcc-bugs at gcc dot gnu dot org,jh at suse dot cz
>  GCC build triplet: hppa-unknown-linux-gnu, hppa2.0w-hp-hpux11.11
>   GCC host triplet: hppa-unknown-linux-gnu, hppa2.0w-hp-hpux11.11
> GCC target triplet: hppa-unknown-linux-gnu, hppa2.0w-hp-hpux11.11
> 
> The following patch introduced numerous regressions on the PA at -O3:
> 
> 2003-10-11  Jan Hubicka  <jh@suse.cz>
> 
>         * Makefile.in (web.o): New.
>         * web.c: New file.
>         * rtl.h (web_main): Declare.
>         * timervar.def (TV_WEB): New.
>         * toplev.c (dump_file_index, dump_file_info): Add DFI_web.
>         (rest_of_hanle_web): New.
>         (flag_web): New static variable.
>         (lang_independent_options): Add "web".
>         (rest_of_compilation): Call rest_of_handle_web.
>         * invoke.texi (-fweb): Document.
>         * common.opt (fweb): New.
>         * flags.h (flag_web): New.
>         * opts.c (decode_options): Set flag_web at -O3.
> 
>         * passes.texi (web construction):  Document.
>         * invoke.texi (-O3): Document that -fweb is enabled.
> 
>         * regrename.c (regrename_optimize): Deal better with situation when
>         replacement failed.
> 
>         * sched-ebb.c: Include params.h and profile.h
>         (schedule_ebbs):  Use tracer parameters to discover superblocks
>         * Makefile.in (sched-ebb.o):  Add dependencies.
> 
> See for example the difference in testsuite results between
> <http://gcc.gnu.org/ml/gcc-testresults/2003-10/msg00653.html>
> and
> <http://gcc.gnu.org/ml/gcc-testresults/2003-10/msg00612.html>.
> 
> Here is an example of one of the common failures:
> 
> dave@hiauly6:~/gnu/gcc-3.4/objdir/gcc/testsuite$
> /home/dave/gnu/gcc-3.4/objdir/gcc/stage1/xgcc
> -B/home/dave/gnu/gcc-3.4/objdir/gcc/stage1/   -O3  -w -c  -o 20011109-1.o
> /home/dave/gnu/gcc-3.4/gcc/gcc/testsuite/gcc.c-torture/compile/20011109-1.c
> /home/dave/gnu/gcc-3.4/gcc/gcc/testsuite/gcc.c-torture/compile/20011109-1.c: In
> function `die':
> 
> /home/dave/gnu/gcc-3.4/gcc/gcc/testsuite/gcc.c-torture/compile/20011109-1.c:51:
> error: unrecognizable insn:
> (insn 188 187 189 17 (parallel [
>             (set (mem/s:BLK (reg/f:SI 169 [ u ]) [12 S8 A16])
>                 (mem/s:BLK (reg/f:SI 170) [12 S8 A16]))
>             (clobber (reg/f:SI 206 [ u ]))
>             (clobber (reg/f:SI 205))
>             (clobber (reg:SI 171))
>             (clobber (reg:SI 172))
>             (clobber (reg:SI 173))
>             (use (const_int 8 [0x8]))
>             (use (const_int 2 [0x2]))
>         ]) -1 (insn_list 186 (insn_list 187 (nil)))
>     (expr_list:REG_DEAD (reg/f:SI 170)
>         (expr_list:REG_DEAD (reg/f:SI 169 [ u ])
>             (expr_list:REG_UNUSED (reg/f:SI 206 [ u ])
>                 (expr_list:REG_UNUSED (reg/f:SI 205)
>                     (expr_list:REG_UNUSED (reg:SI 171)
>                         (expr_list:REG_UNUSED (reg:SI 172)
>                             (expr_list:REG_UNUSED (reg:SI 173)
>                                 (expr_list:REG_EQUAL (mem/s:BLK (reg/f:SI 167)
> [12 S8 A16])
>                                     (nil))))))))))
> /home/dave/gnu/gcc-3.4/gcc/gcc/testsuite/gcc.c-torture/compile/20011109-1.c:51:
> internal compiler error: in extract_insn, at recog.c:2064
> Please submit a full bug report,
> with preprocessed source if appropriate.
> See <URL:http://gcc.gnu.org/bugs.html> for instructions.
BTW did the patch passed testing?  If so, I will install it as obvious
so we can close this bug...

Honza

[Bug optimization/12630] [3.4 regression] Various unrecognizable insns and ICEs at -O3

[dave at hiauly1 dot hia dot nrc dot ca]

PLEASE REPLY TO gcc-bugzilla@gcc.gnu.org ONLY, *NOT* gcc-bugs@gcc.gnu.org.

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=12630



------- Additional Comments From dave at hiauly1 dot hia dot nrc dot ca  2003-10-16 16:38 -------
Subject: Re:  [3.4 regression] Various unrecognizable

> BTW did the patch passed testing?  If so, I will install it as obvious
> so we can close this bug...

Sorry, I spoke too soon.  An ICE occured in stage2:

stage1/xgcc -Bstage1/ -B/home/dave/opt/gnu/hppa-linux/bin/ -c   -g -O2 -DIN_GCC -W -Wall -Wwrite-strings -Wstrict-prototypes -Wmissing-prototypes -pedantic -Wno-long-long -Werror -fno-common   -DHAVE_CONFIG_H    -I. -I. -I../../gcc/gcc -I../../gcc/gcc/. -I../../gcc/gcc/../include  ../../gcc/gcc/varasm.c -o varasm.o
../../gcc/gcc/varasm.c: In function `assemble_real':

../../gcc/gcc/varasm.c:1933: internal compiler error: in schedule_insns, at sched-rgn.c:2744

Haven't had time to investigate further.

Dave

[Bug optimization/12630] [3.4 regression] Various unrecognizable insns and ICEs at -O3

[dave at hiauly1 dot hia dot nrc dot ca]

PLEASE REPLY TO gcc-bugzilla@gcc.gnu.org ONLY, *NOT* gcc-bugs@gcc.gnu.org.

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=12630



------- Additional Comments From dave at hiauly1 dot hia dot nrc dot ca  2003-10-16 19:59 -------
Subject: Re:  [3.4 regression] Various unrecognizable

> Sorry, I spoke too soon.  An ICE occured in stage2:
> 
> stage1/xgcc -Bstage1/ -B/home/dave/opt/gnu/hppa-linux/bin/ -c   -g -O2
> -DIN_GCC -W -Wall -Wwrite-strings -Wstrict-prototypes -Wmissing-prototypes
> -pedantic -Wno-long-long -Werror -fno-common   -DHAVE_CONFIG_H    -I. -I.
> -I../../gcc/gcc -I../../gcc/gcc/. -I../../gcc/gcc/../include
> ../../gcc/gcc/varasm.c -o varasm.o
> ../../gcc/gcc/varasm.c: In function `assemble_real':
> 
> ../../gcc/gcc/varasm.c:1933: internal compiler error: in schedule_insns, at
> sched-rgn.c:2744

This may be a gcse bug.  We have this rtl after addressof:

(insn 3 161 8 0 ../../gcc/gcc/varasm.c:1902 (set (reg/v/f:SI 94 [ d ])
        (reg:SI 26 %r26 [ d ])) 37 {*pa.md:2265} (nil)
    (nil))

...

(insn 5 4 6 0 ../../gcc/gcc/varasm.c:1902 (set (reg/f:SI 96 [ d ])
        (reg/v/f:SI 94 [ d ])) 37 {*pa.md:2265} (nil)
    (nil))

(insn 6 5 7 0 ../../gcc/gcc/varasm.c:1902 (parallel [
            (set (mem/s:BLK (reg/f:SI 95) [102 d+0 S24 A32])
		 (mem/s:BLK (reg/f:SI 96 [ d ]) [102 d+0 S24 A32]))
	    (clobber (reg/f:SI 95))
	    (clobber (reg/f:SI 96 [ d ]))
	    (clobber (reg:SI 97))
	    (clobber (reg:SI 98))
	    (clobber (reg:SI 99))
	    (use (const_int 24 [0x18]))
	    (use (const_int 4 [0x4]))
	]) 72 {movstrsi_internal} (nil)
    (nil))

After gcse, we have the following insn which doesn't satisfy the constraints:

(insn 6 4 7 0 ../../gcc/gcc/varasm.c:1902 (parallel [
            (set (mem/s:BLK (reg/f:SI 95) [102 d+0 S24 A32])
		(mem/s:BLK (reg/v/f:SI 94 [ d ]) [102 d+0 S24 A32]))
	    (clobber (reg/f:SI 95))
	    (clobber (reg/f:SI 96 [ d ]))
	    (clobber (reg:SI 97))
	    (clobber (reg:SI 98))
	    (clobber (reg:SI 99))
	    (use (const_int 24 [0x18]))
	    (use (const_int 4 [0x4]))
	]) 72 {movstrsi_internal} (nil)
    (nil))

Insn 5 has been deleted.

I don't see any way to have an insn with an operand that's both an input
and clobbered.

Dave

[Bug optimization/12630] [3.4 regression] Various unrecognizable insns and ICEs at -O3

[jh at suse dot cz]

PLEASE REPLY TO gcc-bugzilla@gcc.gnu.org ONLY, *NOT* gcc-bugs@gcc.gnu.org.

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=12630



------- Additional Comments From jh at suse dot cz  2003-10-16 20:33 -------
Subject: Re:  [3.4 regression] Various unrecognizable insns and ICEs at -O3

> PLEASE REPLY TO gcc-bugzilla@gcc.gnu.org ONLY, *NOT* gcc-bugs@gcc.gnu.org.
> 
> http://gcc.gnu.org/bugzilla/show_bug.cgi?id=12630
> 
> 
> 
> ------- Additional Comments From dave at hiauly1 dot hia dot nrc dot ca  2003-10-16 19:59 -------
> Subject: Re:  [3.4 regression] Various unrecognizable
> 
> > Sorry, I spoke too soon.  An ICE occured in stage2:
> > 
> > stage1/xgcc -Bstage1/ -B/home/dave/opt/gnu/hppa-linux/bin/ -c   -g -O2
> > -DIN_GCC -W -Wall -Wwrite-strings -Wstrict-prototypes -Wmissing-prototypes
> > -pedantic -Wno-long-long -Werror -fno-common   -DHAVE_CONFIG_H    -I. -I.
> > -I../../gcc/gcc -I../../gcc/gcc/. -I../../gcc/gcc/../include
> > ../../gcc/gcc/varasm.c -o varasm.o
> > ../../gcc/gcc/varasm.c: In function `assemble_real':
> > 
> > ../../gcc/gcc/varasm.c:1933: internal compiler error: in schedule_insns, at
> > sched-rgn.c:2744
> 
> This may be a gcse bug.  We have this rtl after addressof:
> 
> (insn 3 161 8 0 ../../gcc/gcc/varasm.c:1902 (set (reg/v/f:SI 94 [ d ])
>         (reg:SI 26 %r26 [ d ])) 37 {*pa.md:2265} (nil)
>     (nil))
> 
> ...
> 
> (insn 5 4 6 0 ../../gcc/gcc/varasm.c:1902 (set (reg/f:SI 96 [ d ])
>         (reg/v/f:SI 94 [ d ])) 37 {*pa.md:2265} (nil)
>     (nil))
> 
> (insn 6 5 7 0 ../../gcc/gcc/varasm.c:1902 (parallel [
>             (set (mem/s:BLK (reg/f:SI 95) [102 d+0 S24 A32])
> 		 (mem/s:BLK (reg/f:SI 96 [ d ]) [102 d+0 S24 A32]))
> 	    (clobber (reg/f:SI 95))
> 	    (clobber (reg/f:SI 96 [ d ]))
> 	    (clobber (reg:SI 97))
> 	    (clobber (reg:SI 98))
> 	    (clobber (reg:SI 99))
> 	    (use (const_int 24 [0x18]))
> 	    (use (const_int 4 [0x4]))
> 	]) 72 {movstrsi_internal} (nil)
>     (nil))
> 
> After gcse, we have the following insn which doesn't satisfy the constraints:
> 
> (insn 6 4 7 0 ../../gcc/gcc/varasm.c:1902 (parallel [
>             (set (mem/s:BLK (reg/f:SI 95) [102 d+0 S24 A32])
> 		(mem/s:BLK (reg/v/f:SI 94 [ d ]) [102 d+0 S24 A32]))
> 	    (clobber (reg/f:SI 95))
> 	    (clobber (reg/f:SI 96 [ d ]))
> 	    (clobber (reg:SI 97))
> 	    (clobber (reg:SI 98))
> 	    (clobber (reg:SI 99))
> 	    (use (const_int 24 [0x18]))
> 	    (use (const_int 4 [0x4]))
> 	]) 72 {movstrsi_internal} (nil)
>     (nil))
> 
> Insn 5 has been deleted.
> 
> I don't see any way to have an insn with an operand that's both an input
> and clobbered.

This is not problem - you simply load the value into input operant and
then don't expect it to stay there after the instruction.
The transfromation above looks valid to me.  Reload ought to be able to
manage the set operand match clobbers.
(in fact we will likely produce better code if we did clobber
match_scratch with matching operand, like i386 umuldi3_highpart_rex64
does)

Honza
> 
> Dave
> 
> 
> 
> 
> ------- You are receiving this mail because: -------
> You are on the CC list for the bug, or are watching someone who is.

[Bug optimization/12630] [3.4 regression] Various unrecognizable insns and ICEs at -O3

[dave at hiauly1 dot hia dot nrc dot ca]

PLEASE REPLY TO gcc-bugzilla@gcc.gnu.org ONLY, *NOT* gcc-bugs@gcc.gnu.org.

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=12630



------- Additional Comments From dave at hiauly1 dot hia dot nrc dot ca  2003-10-16 20:46 -------
Subject: Re:  [3.4 regression] Various unrecognizable

> > (insn 6 4 7 0 ../../gcc/gcc/varasm.c:1902 (parallel [
> >             (set (mem/s:BLK (reg/f:SI 95) [102 d+0 S24 A32])
> > 		(mem/s:BLK (reg/v/f:SI 94 [ d ]) [102 d+0 S24 A32]))
> > 	    (clobber (reg/f:SI 95))
> > 	    (clobber (reg/f:SI 96 [ d ]))
> > 	    (clobber (reg:SI 97))
> > 	    (clobber (reg:SI 98))
> > 	    (clobber (reg:SI 99))
> > 	    (use (const_int 24 [0x18]))
> > 	    (use (const_int 4 [0x4]))
> > 	]) 72 {movstrsi_internal} (nil)
> >     (nil))
> > 
> > Insn 5 has been deleted.
> > 
> > I don't see any way to have an insn with an operand that's both an input
> > and clobbered.
> 
> This is not problem - you simply load the value into input operant and
> then don't expect it to stay there after the instruction.
> The transfromation above looks valid to me.  Reload ought to be able to
> manage the set operand match clobbers.

Possibly reload would fix this up but we didn't get that far.  The problem
as I see it is that there is no indication that SI 94 is clobbered and
has to die.  I think this is why there is an inconsistency in the death
notes and the ICE.  If SI 94 were used after this instruction, the
transformation wouldn't be valid.

Dave

[Bug optimization/12630] [3.4 regression] Various unrecognizable insns and ICEs at -O3

[jh at suse dot cz]

PLEASE REPLY TO gcc-bugzilla@gcc.gnu.org ONLY, *NOT* gcc-bugs@gcc.gnu.org.

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=12630



------- Additional Comments From jh at suse dot cz  2003-10-16 21:00 -------
Subject: Re:  [3.4 regression] Various unrecognizable insns and ICEs at -O3

> PLEASE REPLY TO gcc-bugzilla@gcc.gnu.org ONLY, *NOT* gcc-bugs@gcc.gnu.org.
> 
> http://gcc.gnu.org/bugzilla/show_bug.cgi?id=12630
> 
> 
> 
> ------- Additional Comments From dave at hiauly1 dot hia dot nrc dot ca  2003-10-16 20:46 -------
> Subject: Re:  [3.4 regression] Various unrecognizable
> 
> > > (insn 6 4 7 0 ../../gcc/gcc/varasm.c:1902 (parallel [
> > >             (set (mem/s:BLK (reg/f:SI 95) [102 d+0 S24 A32])
> > > 		(mem/s:BLK (reg/v/f:SI 94 [ d ]) [102 d+0 S24 A32]))
> > > 	    (clobber (reg/f:SI 95))
> > > 	    (clobber (reg/f:SI 96 [ d ]))
> > > 	    (clobber (reg:SI 97))
> > > 	    (clobber (reg:SI 98))
> > > 	    (clobber (reg:SI 99))
> > > 	    (use (const_int 24 [0x18]))
> > > 	    (use (const_int 4 [0x4]))
> > > 	]) 72 {movstrsi_internal} (nil)
> > >     (nil))
> > > 
> > > Insn 5 has been deleted.
> > > 
> > > I don't see any way to have an insn with an operand that's both an input
> > > and clobbered.
> > 
> > This is not problem - you simply load the value into input operant and
> > then don't expect it to stay there after the instruction.
> > The transfromation above looks valid to me.  Reload ought to be able to
> > manage the set operand match clobbers.
> 
> Possibly reload would fix this up but we didn't get that far.  The problem
> as I see it is that there is no indication that SI 94 is clobbered and
> has to die.  I think this is why there is an inconsistency in the death

Before reload this does not matter.  As long as clobber mentions other
register, 94 does not die.
Reload later should realize that 94 and 95 needs to sit in the same
register and copy 94 around...
Perhaps that is 95/96 that is just clobbered for no obvious reason
bringing GCC down, but that ought to work as well on the other hand I
can't come up with sane use for this construct so it is most probably
not tested.
I am attaching the patch to use match_scratch instead so we don't end up
with insane clobber of the pseudo, but I don't even have time to compile it
right now.  I am not sure whether the use of match_scratch in expander
is valid.  If not, it will be necessary to expand it by hand...

Honza

Index: pa.md
===================================================================
RCS file: /cvs/gcc/gcc/gcc/config/pa/pa.md,v
retrieving revision 1.131
diff -c -3 -p -r1.131 pa.md
*** pa.md	24 Sep 2003 01:38:32 -0000	1.131
--- pa.md	16 Oct 2003 20:58:12 -0000
***************
*** 2945,2952 ****
  (define_expand "movstrsi"
    [(parallel [(set (match_operand:BLK 0 "" "")
  		   (match_operand:BLK 1 "" ""))
! 	      (clobber (match_dup 7))
! 	      (clobber (match_dup 8))
  	      (clobber (match_dup 4))
  	      (clobber (match_dup 5))
  	      (clobber (match_dup 6))
--- 2945,2952 ----
  (define_expand "movstrsi"
    [(parallel [(set (match_operand:BLK 0 "" "")
  		   (match_operand:BLK 1 "" ""))
! 	      (clobber (match_scratch:SI 7 ""))
! 	      (clobber (match_scratch:SI 8 ""))
  	      (clobber (match_dup 4))
  	      (clobber (match_dup 5))
  	      (clobber (match_dup 6))
***************
*** 3016,3025 ****
  ;; therefore it is forced to operand 2.  If the count is compile-time
  ;; determined, we need two scratch registers for the unrolled code.
  (define_insn "movstrsi_internal"
!   [(set (mem:BLK (match_operand:SI 0 "register_operand" "+r,r"))
! 	(mem:BLK (match_operand:SI 1 "register_operand" "+r,r")))
!    (clobber (match_dup 0))
!    (clobber (match_dup 1))
     (clobber (match_operand:SI 2 "register_operand" "=r,r"))	;loop cnt/tmp
     (clobber (match_operand:SI 3 "register_operand" "=&r,&r"))	;item tmp
     (clobber (match_operand:SI 6 "register_operand" "=&r,&r"))	;item tmp2
--- 3016,3025 ----
  ;; therefore it is forced to operand 2.  If the count is compile-time
  ;; determined, we need two scratch registers for the unrolled code.
  (define_insn "movstrsi_internal"
!   [(set (mem:BLK (match_operand:SI 0 "register_operand" "r,r"))
! 	(mem:BLK (match_operand:SI 1 "register_operand" "r,r")))
!    (clobber (match_scratch:SI 7 "=0,0")))
!    (clobber (match_scratch:SI 8 "=1,1"))
     (clobber (match_operand:SI 2 "register_operand" "=r,r"))	;loop cnt/tmp
     (clobber (match_operand:SI 3 "register_operand" "=&r,&r"))	;item tmp
     (clobber (match_operand:SI 6 "register_operand" "=&r,&r"))	;item tmp2

[Bug optimization/12630] [3.4 regression] Various unrecognizable insns and ICEs at -O3

[dave at hiauly1 dot hia dot nrc dot ca]

PLEASE REPLY TO gcc-bugzilla@gcc.gnu.org ONLY, *NOT* gcc-bugs@gcc.gnu.org.

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=12630



------- Additional Comments From dave at hiauly1 dot hia dot nrc dot ca  2003-10-16 22:29 -------
Subject: Re:  [3.4 regression] Various unrecognizable

> Before reload this does not matter.  As long as clobber mentions other
> register, 94 does not die.

Ah, I think I understand the subtle reason to use match_scratch.

> I am attaching the patch to use match_scratch instead so we don't end up
> with insane clobber of the pseudo, but I don't even have time to compile it
> right now.  I am not sure whether the use of match_scratch in expander
> is valid.  If not, it will be necessary to expand it by hand...

This patch seems to work.  I will install if the check completes
successfully with no regressions.

There needs to be a documentation update.  rtl.texi states

  There is one other known use for clobbering a pseudo register in a
  @code{parallel}: when one of the input operands of the insn is also
  clobbered by the insn.  In this case, using the same pseudo register in
  the clobber and elsewhere in the insn produces the expected results.

Thanks for all your help,
Dave

[Bug optimization/12630] [3.4 regression] Various unrecognizable insns and ICEs at -O3

[cvs-commit at gcc dot gnu dot org]

PLEASE REPLY TO gcc-bugzilla@gcc.gnu.org ONLY, *NOT* gcc-bugs@gcc.gnu.org.

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=12630



------- Additional Comments From cvs-commit at gcc dot gnu dot org  2003-10-17 03:32 -------
Subject: Bug 12630

CVSROOT:	/cvs/gcc
Module name:	gcc
Changes by:	danglin@gcc.gnu.org	2003-10-17 03:32:21

Modified files:
	gcc            : ChangeLog 
	gcc/config/pa  : pa.md 

Log message:
	PR optimization/12630
	* pa.md (movstrsi, movstrsi_internal): Use match_scratch in clobbers
	for operands 7 and 8.

Patches:
http://gcc.gnu.org/cgi-bin/cvsweb.cgi/gcc/gcc/ChangeLog.diff?cvsroot=gcc&r1=2.1438&r2=2.1439
http://gcc.gnu.org/cgi-bin/cvsweb.cgi/gcc/gcc/config/pa/pa.md.diff?cvsroot=gcc&r1=1.131&r2=1.132

[Bug optimization/12630] [3.4 regression] Various unrecognizable insns and ICEs at -O3

[danglin at gcc dot gnu dot org]

PLEASE REPLY TO gcc-bugzilla@gcc.gnu.org ONLY, *NOT* gcc-bugs@gcc.gnu.org.

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=12630


danglin at gcc dot gnu dot org changed:

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


------- Additional Comments From danglin at gcc dot gnu dot org  2003-10-17 03:50 -------
Fixed.

[Bug optimization/12630] [3.4 regression] Various unrecognizable insns and ICEs at -O3

[pinskia at gcc dot gnu dot org]

PLEASE REPLY TO gcc-bugzilla@gcc.gnu.org ONLY, *NOT* gcc-bugs@gcc.gnu.org.

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=12630


pinskia at gcc dot gnu dot org changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
   Target Milestone|---                         |3.4

[Bug optimization/12630] [3.4 regression] Various unrecognizable insns and ICEs at -O3

[danglin at gcc dot gnu dot org]

PLEASE REPLY TO gcc-bugzilla@gcc.gnu.org ONLY, *NOT* gcc-bugs@gcc.gnu.org.

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=12630


danglin at gcc dot gnu dot org changed:

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


------- Additional Comments From danglin at gcc dot gnu dot org  2003-10-20 17:08 -------
There's still a problem.  I haven't normally being testing the bootstrap
with ada.  With ada, the following error occurs:

stage1/xgcc -Bstage1/ -B/home/dave/opt/gnu/hppa-linux/bin/ -c -g -O2      -gnatp
g -gnata -I- -I. -Iada -I../../gcc/gcc/ada ../../gcc/gcc/ada/sem_prag.adb -o ada
/sem_prag.o
../../gcc/gcc/ada/sem_prag.adb: In function `sem_prag.analyze_pragma':

../../gcc/gcc/ada/sem_prag.adb:3559: error: unable to find a register to spill i
n class `R1_REGS'
../../gcc/gcc/ada/sem_prag.adb:3559: error: this is the insn:
(insn 10661 16193 10665 1947 ../../gcc/gcc/ada/sem_prag.adb:7511 (parallel [
            (set (mem/s:BLK (reg/f:SI 2767) [101 sem__scope_suppress+0 S11 A8])
                (mem/s/u:BLK (reg/f:SI 2769) [1 S11 A8]))
            (clobber (scratch:SI))
            (clobber (scratch:SI))
            (clobber (reg:SI %r21 [2773]))
            (clobber (reg:SI %r20 [2774]))
            (clobber (reg:SI %r19 [2775]))
            (use (const_int 11 [0xb]))
            (use (const_int 1 [0x1]))
        ]) 72 {movstrsi_internal} (nil)
    (expr_list:REG_UNUSED (scratch:SI)
        (expr_list:REG_UNUSED (scratch:SI)
            (expr_list:REG_UNUSED (reg:SI %r21 [2773])
                (expr_list:REG_UNUSED (reg:SI %r20 [2774])
                    (expr_list:REG_UNUSED (reg:SI %r19 [2775])
                        (nil)))))))
+===========================GNAT BUG DETECTED==============================+
| 3.4 20031020 (experimental) (hppa-unknown-linux-gnu) GCC error:          |
| in spill_failure, at reload1.c:1862                                      |
| Error detected at ../../gcc/gcc/ada/sem_prag.adb:3559:7                  |
| Please submit a bug report; see http://gcc.gnu.org/bugs.html.            |
| Include the entire contents of this bug box in the report.               |
| Include the exact gcc or gnatmake command that you entered.              |
| Also include sources listed below in gnatchop format                     |
| (concatenated together with no headers between files).                   |
+==========================================================================+

Switching from match_dup to match_scratch introduced a need for two additional
temporaries.  We only have one register in the class R1_REGS.

Re: [Bug optimization/12630] [3.4 regression] Various unrecognizable insns and ICEs at -O3

[Jan Hubicka]

> PLEASE REPLY TO gcc-bugzilla@gcc.gnu.org ONLY, *NOT* gcc-bugs@gcc.gnu.org.
> 
> http://gcc.gnu.org/bugzilla/show_bug.cgi?id=12630
> 
> 
> danglin at gcc dot gnu dot org changed:
> 
>            What    |Removed                     |Added
> ----------------------------------------------------------------------------
>              Status|RESOLVED                    |UNCONFIRMED
>          Resolution|FIXED                       |
> 
> 
> ------- Additional Comments From danglin at gcc dot gnu dot org  2003-10-20 17:08 -------
> There's still a problem.  I haven't normally being testing the bootstrap
> with ada.  With ada, the following error occurs:
> 
> stage1/xgcc -Bstage1/ -B/home/dave/opt/gnu/hppa-linux/bin/ -c -g -O2      -gnatp
> g -gnata -I- -I. -Iada -I../../gcc/gcc/ada ../../gcc/gcc/ada/sem_prag.adb -o ada
> /sem_prag.o
> ../../gcc/gcc/ada/sem_prag.adb: In function `sem_prag.analyze_pragma':
> 
> ../../gcc/gcc/ada/sem_prag.adb:3559: error: unable to find a register to spill i
> n class `R1_REGS'
> ../../gcc/gcc/ada/sem_prag.adb:3559: error: this is the insn:
> (insn 10661 16193 10665 1947 ../../gcc/gcc/ada/sem_prag.adb:7511 (parallel [
>             (set (mem/s:BLK (reg/f:SI 2767) [101 sem__scope_suppress+0 S11 A8])
>                 (mem/s/u:BLK (reg/f:SI 2769) [1 S11 A8]))
>             (clobber (scratch:SI))
>             (clobber (scratch:SI))
>             (clobber (reg:SI %r21 [2773]))
>             (clobber (reg:SI %r20 [2774]))
>             (clobber (reg:SI %r19 [2775]))
>             (use (const_int 11 [0xb]))
>             (use (const_int 1 [0x1]))
>         ]) 72 {movstrsi_internal} (nil)
>     (expr_list:REG_UNUSED (scratch:SI)
>         (expr_list:REG_UNUSED (scratch:SI)
>             (expr_list:REG_UNUSED (reg:SI %r21 [2773])
>                 (expr_list:REG_UNUSED (reg:SI %r20 [2774])
>                     (expr_list:REG_UNUSED (reg:SI %r19 [2775])
>                         (nil)))))))
> +===========================GNAT BUG DETECTED==============================+
> | 3.4 20031020 (experimental) (hppa-unknown-linux-gnu) GCC error:          |
> | in spill_failure, at reload1.c:1862                                      |
> | Error detected at ../../gcc/gcc/ada/sem_prag.adb:3559:7                  |
> | Please submit a bug report; see http://gcc.gnu.org/bugs.html.            |
> | Include the entire contents of this bug box in the report.               |
> | Include the exact gcc or gnatmake command that you entered.              |
> | Also include sources listed below in gnatchop format                     |
> | (concatenated together with no headers between files).                   |
> +==========================================================================+
> 
> Switching from match_dup to match_scratch introduced a need for two additional
> temporaries.  We only have one register in the class R1_REGS.

Why these two temporaries are needed?
R1_REGS has constraint 'a' that is not used in the pattern at all so why
reload chooses to use it?

Honza

[Bug optimization/12630] [3.4 regression] Various unrecognizable insns and ICEs at -O3

[hubicka at ucw dot cz]

PLEASE REPLY TO gcc-bugzilla@gcc.gnu.org ONLY, *NOT* gcc-bugs@gcc.gnu.org.

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=12630



------- Additional Comments From hubicka at ucw dot cz  2003-10-20 20:31 -------
Subject: Re:  [3.4 regression] Various unrecognizable insns and ICEs at -O3

> PLEASE REPLY TO gcc-bugzilla@gcc.gnu.org ONLY, *NOT* gcc-bugs@gcc.gnu.org.
> 
> http://gcc.gnu.org/bugzilla/show_bug.cgi?id=12630
> 
> 
> danglin at gcc dot gnu dot org changed:
> 
>            What    |Removed                     |Added
> ----------------------------------------------------------------------------
>              Status|RESOLVED                    |UNCONFIRMED
>          Resolution|FIXED                       |
> 
> 
> ------- Additional Comments From danglin at gcc dot gnu dot org  2003-10-20 17:08 -------
> There's still a problem.  I haven't normally being testing the bootstrap
> with ada.  With ada, the following error occurs:
> 
> stage1/xgcc -Bstage1/ -B/home/dave/opt/gnu/hppa-linux/bin/ -c -g -O2      -gnatp
> g -gnata -I- -I. -Iada -I../../gcc/gcc/ada ../../gcc/gcc/ada/sem_prag.adb -o ada
> /sem_prag.o
> ../../gcc/gcc/ada/sem_prag.adb: In function `sem_prag.analyze_pragma':
> 
> ../../gcc/gcc/ada/sem_prag.adb:3559: error: unable to find a register to spill i
> n class `R1_REGS'
> ../../gcc/gcc/ada/sem_prag.adb:3559: error: this is the insn:
> (insn 10661 16193 10665 1947 ../../gcc/gcc/ada/sem_prag.adb:7511 (parallel [
>             (set (mem/s:BLK (reg/f:SI 2767) [101 sem__scope_suppress+0 S11 A8])
>                 (mem/s/u:BLK (reg/f:SI 2769) [1 S11 A8]))
>             (clobber (scratch:SI))
>             (clobber (scratch:SI))
>             (clobber (reg:SI %r21 [2773]))
>             (clobber (reg:SI %r20 [2774]))
>             (clobber (reg:SI %r19 [2775]))
>             (use (const_int 11 [0xb]))
>             (use (const_int 1 [0x1]))
>         ]) 72 {movstrsi_internal} (nil)
>     (expr_list:REG_UNUSED (scratch:SI)
>         (expr_list:REG_UNUSED (scratch:SI)
>             (expr_list:REG_UNUSED (reg:SI %r21 [2773])
>                 (expr_list:REG_UNUSED (reg:SI %r20 [2774])
>                     (expr_list:REG_UNUSED (reg:SI %r19 [2775])
>                         (nil)))))))
> +===========================GNAT BUG DETECTED==============================+
> | 3.4 20031020 (experimental) (hppa-unknown-linux-gnu) GCC error:          |
> | in spill_failure, at reload1.c:1862                                      |
> | Error detected at ../../gcc/gcc/ada/sem_prag.adb:3559:7                  |
> | Please submit a bug report; see http://gcc.gnu.org/bugs.html.            |
> | Include the entire contents of this bug box in the report.               |
> | Include the exact gcc or gnatmake command that you entered.              |
> | Also include sources listed below in gnatchop format                     |
> | (concatenated together with no headers between files).                   |
> +==========================================================================+
> 
> Switching from match_dup to match_scratch introduced a need for two additional
> temporaries.  We only have one register in the class R1_REGS.

Why these two temporaries are needed?
R1_REGS has constraint 'a' that is not used in the pattern at all so why
reload chooses to use it?

Honza

[Bug optimization/12630] [3.4 regression] Various unrecognizable insns and ICEs at -O3

[dave at hiauly1 dot hia dot nrc dot ca]

PLEASE REPLY TO gcc-bugzilla@gcc.gnu.org ONLY, *NOT* gcc-bugs@gcc.gnu.org.

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=12630



------- Additional Comments From dave at hiauly1 dot hia dot nrc dot ca  2003-10-20 22:06 -------
Subject: Re:  [3.4 regression] Various unrecognizable

> > Switching from match_dup to match_scratch introduced a need for two
> additional
> > temporaries.  We only have one register in the class R1_REGS.
> 
> Why these two temporaries are needed?
> R1_REGS has constraint 'a' that is not used in the pattern at all so why
> reload chooses to use it?

I don't know.  The PA backend only uses class R1_REGS for two very specific
situations and this isn't one of them.

Dave

[Bug optimization/12630] [3.4 regression] Various unrecognizable insns and ICEs at -O3

[jh at suse dot cz]

PLEASE REPLY TO gcc-bugzilla@gcc.gnu.org ONLY, *NOT* gcc-bugs@gcc.gnu.org.

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=12630



------- Additional Comments From jh at suse dot cz  2003-10-20 22:29 -------
Subject: Re:  [3.4 regression] Various unrecognizable insns and ICEs at -O3

> PLEASE REPLY TO gcc-bugzilla@gcc.gnu.org ONLY, *NOT* gcc-bugs@gcc.gnu.org.
> 
> http://gcc.gnu.org/bugzilla/show_bug.cgi?id=12630
> 
> 
> 
> ------- Additional Comments From dave at hiauly1 dot hia dot nrc dot ca  2003-10-20 22:06 -------
> Subject: Re:  [3.4 regression] Various unrecognizable
> 
> > > Switching from match_dup to match_scratch introduced a need for two
> > additional
> > > temporaries.  We only have one register in the class R1_REGS.
> > 
> > Why these two temporaries are needed?
> > R1_REGS has constraint 'a' that is not used in the pattern at all so why
> > reload chooses to use it?
> 
> I don't know.  The PA backend only uses class R1_REGS for two very specific
> situations and this isn't one of them.

Do you have any idea how can I reproduce it w/o PA machine?
(don't know much about Ada...)

Honza
> 
> Dave
> 
> 
> 
> 
> ------- You are receiving this mail because: -------
> You are on the CC list for the bug, or are watching someone who is.

[Bug optimization/12630] [3.4 regression] Various unrecognizable insns and ICEs at -O3

[dave at hiauly1 dot hia dot nrc dot ca]

PLEASE REPLY TO gcc-bugzilla@gcc.gnu.org ONLY, *NOT* gcc-bugs@gcc.gnu.org.

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=12630



------- Additional Comments From dave at hiauly1 dot hia dot nrc dot ca  2003-10-20 22:38 -------
Subject: Re:  [3.4 regression] Various unrecognizable

> Do you have any idea how can I reproduce it w/o PA machine?
> (don't know much about Ada...)

Neither do I.  I don't how to reproduce this w/o a PA machine.
However, I should be able to arrange a PA account.  If you send
your ssh key to randolph@tausq.org <Randolph Chung>, he should
be able to arrange an account for you on gsyprf11.external.hp.com.
He recently added an account for Alan Modra.

I will send a note to Randolph.

Dave

[Bug optimization/12630] [3.4 regression] Various unrecognizable insns and ICEs at -O3

[dave at hiauly1 dot hia dot nrc dot ca]

PLEASE REPLY TO gcc-bugzilla@gcc.gnu.org ONLY, *NOT* gcc-bugs@gcc.gnu.org.

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=12630



------- Additional Comments From dave at hiauly1 dot hia dot nrc dot ca  2003-11-03 01:32 -------
Subject: Re:  [3.4 regression] Various unrecognizable

> > Switching from match_dup to match_scratch introduced a need for two
> additional
> > temporaries.  We only have one register in the class R1_REGS.
> 
> Why these two temporaries are needed?
> R1_REGS has constraint 'a' that is not used in the pattern at all so why
> reload chooses to use it?

I'm not sure that I understand this yet but I have the sense that
the problem is related to loop optimization.  The principal place
that %r1 is needed is to load the address of a symbol that isn't
a read only operand.  For this, we use a pair of high and losum 
insns.  The high insn requires %r1.  In the case at hand, I noticed
that the movstr insn is in a loop where the high insn has been
pulled out of the loop.  Possibly, by doing this, we have introduced
a situation that reload can't handle.  Both the source and destination
registers for the movstr operation are loaded in the loop using
high-losum sequences.

Dave

[Bug optimization/12630] [3.4 regression] Various unrecognizable insns and ICEs at -O3

[danglin at gcc dot gnu dot org]

PLEASE REPLY TO gcc-bugzilla@gcc.gnu.org ONLY, *NOT* gcc-bugs@gcc.gnu.org.

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=12630



------- Additional Comments From danglin at gcc dot gnu dot org  2003-11-05 17:39 -------
I have looked at the problem some more.  Using match_scratch fails because
it requires a secondary reload.  The secondary reload uses the equivalent
MEM for X when it is a pseudo register:

  /* If X is a pseudo-register that has an equivalent MEM (actually, if it
     is still a pseudo-register by now, it *must* have an equivalent MEM
     but we don't want to assume that), use that equivalent when seeing if
     a secondary reload is needed since whether or not a reload is needed
     might be sensitive to the form of the MEM.  */

  if (GET_CODE (x) == REG && REGNO (x) >= FIRST_PSEUDO_REGISTER
      && reg_equiv_mem[REGNO (x)] != 0)
    x = reg_equiv_mem[REGNO (x)];

When the MEM is symbolic, we need to use %r1 for the reload (this is
absolutely required in pic code and desireable in non-pic code).  As there
are two inputs for the movstr pattern, we fail when both memory references
are symbolic.  See the rtl shown in comment #12.

The possible fixes are:

1) Delete the PA movstr pattern :(
2) Use two additional registers for the movstr operation and do a split
   after reload to allow unnecessary copies to be deleted.  I'm not
   particularly in favor of doing this as it seems like a hack.
3) Revert and/or fix the regrename.c patch so that the clobber match_dup
   semantic works again.
4) Somehow fix reload to not generate secondary reloads when we have
   numeric constraints.  I don't understand reload well enough to know
   whether this is a viable option.

[Bug optimization/12630] [3.4 regression] Various unrecognizable insns and ICEs at -O3

[pinskia at gcc dot gnu dot org]

PLEASE REPLY TO gcc-bugzilla@gcc.gnu.org ONLY, *NOT* gcc-bugs@gcc.gnu.org.

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=12630


pinskia at gcc dot gnu dot org changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|UNCONFIRMED                 |NEW
     Ever Confirmed|                            |1
           Keywords|                            |ice-on-valid-code
   Last reconfirmed|0000-00-00 00:00:00         |2003-11-07 15:56:00
               date|                            |

[Bug optimization/12630] [3.4 regression] Various unrecognizable insns and ICEs at -O3

[cvs-commit at gcc dot gnu dot org]

------- Additional Comments From cvs-commit at gcc dot gnu dot org  2003-11-08 20:36 -------
Subject: Bug 12630

CVSROOT:	/cvs/gcc
Module name:	gcc
Changes by:	danglin@gcc.gnu.org	2003-11-08 20:36:30

Modified files:
	gcc            : ChangeLog 
	gcc/config/pa  : pa.c pa.md pa-protos.h 

Log message:
	PR optimization/12630
	* pa.c (compute_movstrsi_length): Rename to compute_movstr_length.
	Handle length computation 64-bit moves.
	(compute_clrstr_length, output_block_clear): Implement block clear.
	(output_block_move): Handle 64-bit moves.
	(pa_adjust_insn_length): Use compute_movstr_length and
	compute_clrstr_length.
	* pa.md (movstrsi): Revise operand order and comments.  Don't use
	match_scratch.
	(movstrsi_internal): Delete.
	(movstrsi_prereload, movstrsi_postreload): New insns.  Define splitter
	and peephole2 patterns to transform prereload to postreload form.
	(movstrdi, movstrdi_prereload, movstrdi_postreload, clrstrsi,
	clrstrsi_prereload, clrstrsi_postreload, clrstrdi, clrstrdi_prereload,
	clrstrdi_postreload): New patterns for 64-bit block move, and block
	clear.
	* pa-protos.h (output_block_clear): New prototype.

Patches:
http://gcc.gnu.org/cgi-bin/cvsweb.cgi/gcc/gcc/ChangeLog.diff?cvsroot=gcc&r1=2.1704&r2=2.1705
http://gcc.gnu.org/cgi-bin/cvsweb.cgi/gcc/gcc/config/pa/pa.c.diff?cvsroot=gcc&r1=1.231&r2=1.232
http://gcc.gnu.org/cgi-bin/cvsweb.cgi/gcc/gcc/config/pa/pa.md.diff?cvsroot=gcc&r1=1.133&r2=1.134
http://gcc.gnu.org/cgi-bin/cvsweb.cgi/gcc/gcc/config/pa/pa-protos.h.diff?cvsroot=gcc&r1=1.27&r2=1.28



-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=12630

[Bug optimization/12630] [3.4 regression] Various unrecognizable insns and ICEs at -O3

[danglin at gcc dot gnu dot org]

------- Additional Comments From danglin at gcc dot gnu dot org  2003-11-09 00:41 -------
Workaround applied.


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


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=12630