public inbox for gcc@gcc.gnu.org
 help / color / mirror / Atom feed
* Bootstrap fails with recently CVS'd sources on alphaev6-unknown-linux-gnu.
@ 2000-05-29 13:53 Toon Moene
  2000-05-29 14:14 ` Richard Henderson
  0 siblings, 1 reply; 3+ messages in thread
From: Toon Moene @ 2000-05-29 13:53 UTC (permalink / raw)
  To: gcc

L.S.,

Sources CVS'd at 200005291830 UTC.  This is the failure on
alphaev6-unknown-linux-gnu:

stage1/xgcc -Bstage1/ -B/usr/snp/alphaev6-unknown-linux-gnu/bin/ -c 
-DIN_GCC    -W -Wall -Wtraditional -Wwrite-strings -Wstrict-prototypes
-Wmissing-prototypes -pedantic -Wno-long-long -O2 -g -O2 
-DHAVE_CONFIG_H    -I. -I../../egcs/gcc -I../../egcs/gcc/config
-I../../egcs/gcc/../include ../../egcs/gcc/optabs.c
xgcc: Internal compiler error: program cc1 got fatal signal 11
make[2]: *** [optabs.o] Error 1
make[2]: Leaving directory `/home/toon/compilers/snapshots/obj/gcc'
make[1]: *** [bootstrap] Error 2
make[1]: Leaving directory `/home/toon/compilers/snapshots/obj/gcc'
make: *** [bootstrap] Error 2

Using gdb, i get:

(gdb) where
#0  0x12021af60 in recog_3 (x0=0x20000bd69e0, insn=0x200004ec480, 
    pnum_clobbers=0x11fffdc9c) at insn-recog.c:5723
#1  0x12021ec54 in recog (x0=0x20000bd69e0, insn=0x200004ec480, 
    pnum_clobbers=0x11fffdc9c) at insn-recog.c:8491
#2  0x12018ed68 in recog_for_combine (pnewpat=0x11fffdd48,
insn=0x200004ec480, 
    pnotes=0x11fffddb8) at ../../egcs/gcc/combine.c:9575
#3  0x120176284 in try_combine (i3=0x200004ec4c0, i2=0x200004ec480, 
    i1=0x200004f1bc0, new_direct_jump_p=0x11fffdeac)
    at ../../egcs/gcc/combine.c:2232
#4  0x120172430 in combine_instructions (f=0x200004f55c0, nregs=88)
    at ../../egcs/gcc/combine.c:731
#5  0x120009760 in rest_of_compilation (decl=0x20000989700)
    at ../../egcs/gcc/toplev.c:3183
#6  0x1202a45a8 in finish_function (nested=0) at
../../egcs/gcc/c-decl.c:6549
#7  0x12028a03c in yyparse () at c-parse.y:313
#8  0x1200076f0 in compile_file (name=0x2000003f0d0 "\030")
    at ../../egcs/gcc/toplev.c:2288
#9  0x12000d14c in main (argc=18, argv=0x11fffe948)
    at ../../egcs/gcc/toplev.c:4702
#10 0x2000014ffb0 in __libc_start_main (main=0x12000c740 <main>,
argc=18, 
    argv=0x11fffe948, init=0x120001d00 <_init>, fini=0x1202c5460
<_fini>, 
    rtld_fini=0x20000bce220, stack_end=0x11fffe930)
    at ../sysdeps/generic/libc-start.c:78
(gdb) list
5718      x1 = XEXP (x0, 0);
5719      goto L1672;
5720    
5721     L834: ATTRIBUTE_UNUSED_LABEL
5722      x2 = XEXP (x1, 1);
5723      if (GET_CODE (x2) == CONST_INT
5724          && XWINT (x2, 0) == 0
5725          && (GET_MODE_CLASS (GET_MODE (operands[0])) == MODE_INT
5726       && GET_MODE_SIZE (GET_MODE (operands[0])) <= 8
5727       && GET_MODE (operands[0]) == GET_MODE (SET_SRC (PATTERN
(insn)))))
(gdb) p *x2
$1 = {code = CONST_INT, mode = VOIDmode, jump = 0, call = 0, unchanging
= 0, 
  volatil = 0, in_struct = 0, used = 0, integrated = 0, frame_related =
0, 
  fld = {{rtwint = 0, rtint = 0, rtuint = 0, rtstr = 0x0, rtx = 0x0, 
      rtvec = 0x0, rttype = VOIDmode, rt_addr_diff_vec_flags =
{min_align = 0, 
        base_after_vec = 0, min_after_vec = 0, max_after_vec = 0, 
        min_after_base = 0, max_after_base = 0, offset_unsigned = 0,  =
0, 
        scale = 0}, rt_cselib = 0x0, rtbit = 0x0, rttree = 0x0, bb =
0x0}}}
(gdb) p *x1
$2 = {code = NE, mode = VOIDmode, jump = 32, call = 32, unchanging = 32, 
  volatil = 32, in_struct = 32, used = 32, integrated = 32, 
  frame_related = 32, fld = {{rtwint = 2199035634208, rtint = 12378656, 
      rtuint = 12378656, rtstr = 0x20000bce220 "9", rtx = 0x20000bce220, 
      rtvec = 0x20000bce220, rttype = 12378656, rt_addr_diff_vec_flags =
{
        min_align = 12378656, base_after_vec = 48354, min_after_vec =
48354, 
        max_after_vec = 48354, min_after_base = 48354, max_after_base =
48354, 
        offset_unsigned = 48354,  = 48354, scale = 33554620}, 
      rt_cselib = 0x20000bce220, rtbit = 0x20000bce220, 
      rttree = 0x20000bce220, bb = 0x20000bce220}}}
(gdb) p x2
$3 = 0x2000001e480
(gdb) p x1
$4 = 0x20000bcfcc0
(gdb) p *(0x20000bce220)
$5 = 524345

-- 
Toon Moene - mailto:toon@moene.indiv.nluug.nl - phoneto: +31 346 214290
Saturnushof 14, 3738 XG  Maartensdijk, The Netherlands
GNU Fortran 77: http://gcc.gnu.org/onlinedocs/g77_news.html
GNU Fortran 95: http://g95.sourceforge.net/ (under construction)

^ permalink raw reply	[flat|nested] 3+ messages in thread

* Re: Bootstrap fails with recently CVS'd sources on alphaev6-unknown-linux-gnu.
  2000-05-29 13:53 Bootstrap fails with recently CVS'd sources on alphaev6-unknown-linux-gnu Toon Moene
@ 2000-05-29 14:14 ` Richard Henderson
  2000-05-30  8:54   ` Richard Henderson
  0 siblings, 1 reply; 3+ messages in thread
From: Richard Henderson @ 2000-05-29 14:14 UTC (permalink / raw)
  To: Toon Moene; +Cc: gcc

On Mon, May 29, 2000 at 10:49:08PM +0200, Toon Moene wrote:
> Sources CVS'd at 200005291830 UTC.  This is the failure on
> alphaev6-unknown-linux-gnu:
> 
> stage1/xgcc -Bstage1/ -B/usr/snp/alphaev6-unknown-linux-gnu/bin/ -c 
> -DIN_GCC    -W -Wall -Wtraditional -Wwrite-strings -Wstrict-prototypes
> -Wmissing-prototypes -pedantic -Wno-long-long -O2 -g -O2 
> -DHAVE_CONFIG_H    -I. -I../../egcs/gcc -I../../egcs/gcc/config
> -I../../egcs/gcc/../include ../../egcs/gcc/optabs.c
> xgcc: Internal compiler error: program cc1 got fatal signal 11

I know.  I believe it to be a bug in genrecog, but have yet to find it.
In the meantime I'm hacking around the problem like so.



r~



Index: alpha.md
===================================================================
RCS file: /cvs/gcc/egcs/gcc/config/alpha/alpha.md,v
retrieving revision 1.123
diff -c -p -d -r1.123 alpha.md
*** alpha.md	2000/05/29 07:52:26	1.123
--- alpha.md	2000/05/29 21:13:38
***************
*** 2526,2531 ****
--- 2526,2532 ----
  	    (const_int 0)))]
    "GET_MODE_CLASS (GET_MODE (operands[0])) == MODE_INT
     && GET_MODE_SIZE (GET_MODE (operands[0])) <= 8
+    && GET_CODE (PATTERN (insn)) == SET
     && GET_MODE (operands[0]) == GET_MODE (SET_SRC (PATTERN (insn)))"
    "cmpult $31,%1,%0"
    [(set_attr "type" "icmp")])

^ permalink raw reply	[flat|nested] 3+ messages in thread

* Re: Bootstrap fails with recently CVS'd sources on alphaev6-unknown-linux-gnu.
  2000-05-29 14:14 ` Richard Henderson
@ 2000-05-30  8:54   ` Richard Henderson
  0 siblings, 0 replies; 3+ messages in thread
From: Richard Henderson @ 2000-05-30  8:54 UTC (permalink / raw)
  To: Toon Moene; +Cc: gcc

On Mon, May 29, 2000 at 02:14:26PM -0700, Richard Henderson wrote:
> I know.  I believe it to be a bug in genrecog, but have yet to find it.

To my surprise, apparently there's no correlation to be drawn
from the pattern being matched and the insn that it belongs to.
Certainly they don't match during recog_for_combine...

So I'll try to be a little less clever next time.


r~


        * config/alpha/alpha.md (*setne_internal): Use match_operator
        instead of working down from insn.

Index: config/alpha/alpha.md
===================================================================
RCS file: /cvs/gcc/egcs/gcc/config/alpha/alpha.md,v
retrieving revision 1.123
diff -c -p -d -r1.123 alpha.md
*** alpha.md	2000/05/29 07:52:26	1.123
--- alpha.md	2000/05/30 15:50:10
***************
*** 2520,2535 ****
  ;; Next are all the integer comparisons, and conditional moves and branches
  ;; and some of the related define_expand's and define_split's.
  
- (define_insn "*setne_internal"
-   [(set (match_operand 0 "register_operand" "=r")
- 	(ne (match_operand:DI 1 "reg_or_8bit_operand" "rI")
- 	    (const_int 0)))]
-   "GET_MODE_CLASS (GET_MODE (operands[0])) == MODE_INT
-    && GET_MODE_SIZE (GET_MODE (operands[0])) <= 8
-    && GET_MODE (operands[0]) == GET_MODE (SET_SRC (PATTERN (insn)))"
-   "cmpult $31,%1,%0"
-   [(set_attr "type" "icmp")])
- 
  (define_insn "*setcc_internal"
    [(set (match_operand 0 "register_operand" "=r")
  	(match_operator 1 "alpha_comparison_operator"
--- 2520,2525 ----
***************
*** 2550,2555 ****
--- 2540,2557 ----
     && GET_MODE_SIZE (GET_MODE (operands[0])) <= 8
     && GET_MODE (operands[0]) == GET_MODE (operands[1])"
    "cmp%c1 %r3,%2,%0"
+   [(set_attr "type" "icmp")])
+ 
+ (define_insn "*setne_internal"
+   [(set (match_operand 0 "register_operand" "=r")
+ 	(match_operator 1 "signed_comparison_operator"
+ 			  [(match_operand:DI 2 "reg_or_8bit_operand" "rI")
+ 			   (const_int 0)]))]
+   "GET_MODE_CLASS (GET_MODE (operands[0])) == MODE_INT
+    && GET_MODE_SIZE (GET_MODE (operands[0])) <= 8
+    && GET_CODE (operands[1]) == NE
+    && GET_MODE (operands[0]) == GET_MODE (operands[1])"
+   "cmpult $31,%2,%0"
    [(set_attr "type" "icmp")])
  
  ;; The mode folding trick can't be used with const_int operands, since

^ permalink raw reply	[flat|nested] 3+ messages in thread

end of thread, other threads:[~2000-05-30  8:54 UTC | newest]

Thread overview: 3+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2000-05-29 13:53 Bootstrap fails with recently CVS'd sources on alphaev6-unknown-linux-gnu Toon Moene
2000-05-29 14:14 ` Richard Henderson
2000-05-30  8:54   ` Richard Henderson

This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for read-only IMAP folder(s) and NNTP newsgroup(s).