alphaev6-unknown-linux-gnu bootstrap fails RTL checking

alphaev6-unknown-linux-gnu bootstrap fails RTL checking

[Toon Moene]

And quite quickly too (I had settled for an overnight build, but within
a minute I got the following):

gcc  -DIN_GCC    -g  -W -Wall -Wtraditional -Wwrite-strings
-Wstrict-prototypes -Wmissing-prototypes -pedantic -Wno-long-long 
-DHAVE_CONFIG_H  -o genattr \
 genattr.o rtl.o bitmap.o ggc-none.o gensupport.o print-rtl.o errors.o `
case "obstack.o" in ?*) echo obstack.o ;; esac ` ` case "" in ?*) echo 
;; esac ` ` case "" in ?*) echo  ;; esac ` ` case "" in ?*) echo  ;;
esac ` ` case "" in ?*) echo  ;; esac ` 
./genattr ../../egcs/gcc/config/alpha/alpha.md > tmp-attr.h
genattr: RTL check: access of elt 7 of `define_insn' with last elt 4
genattr: Internal compiler error in `process_rtx', at gensupport.c:212
Please submit a full bug report.
See <URL: http://www.gnu.org/software/gcc/bugs.html > for instructions.

If noone with an Alpha can reproduce this, I'm willing to dig into it
further ....

-- 
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)

Re: alphaev6-unknown-linux-gnu bootstrap fails RTL checking

[Clinton Popetz]

On Tue, Jul 11, 2000 at 07:53:08PM +0200, Toon Moene wrote:
> And quite quickly too (I had settled for an overnight build, but within
> a minute I got the following):
> 
> gcc  -DIN_GCC    -g  -W -Wall -Wtraditional -Wwrite-strings
> -Wstrict-prototypes -Wmissing-prototypes -pedantic -Wno-long-long 
> -DHAVE_CONFIG_H  -o genattr \
>  genattr.o rtl.o bitmap.o ggc-none.o gensupport.o print-rtl.o errors.o `
> case "obstack.o" in ?*) echo obstack.o ;; esac ` ` case "" in ?*) echo 
> ;; esac ` ` case "" in ?*) echo  ;; esac ` ` case "" in ?*) echo  ;;
> esac ` ` case "" in ?*) echo  ;; esac ` 
> ./genattr ../../egcs/gcc/config/alpha/alpha.md > tmp-attr.h
> genattr: RTL check: access of elt 7 of `define_insn' with last elt 4
> genattr: Internal compiler error in `process_rtx', at gensupport.c:212
> Please submit a full bug report.
> See <URL: http://www.gnu.org/software/gcc/bugs.html > for instructions.

Hmm, evidently rtl checking has never been enabled when someone did a
define_insn_and_split.  The insn in question is ze_and_ne, which has
been in the tree for over a month.

			-Clint

Tue Jul 11 15:28:21 CDT 2000  Clinton Popetz  <cpopetz@cygnus.com>

	* gensupport.c (process_rtx): Make rtl checking stop
	complaining about the define_insn while it is being
	converted from a define_insn_and_split.

Index: gensupport.c
===================================================================
RCS file: /cvs/gcc/egcs/gcc/gensupport.c,v
retrieving revision 1.9
diff -c -2 -p -r1.9 gensupport.c
*** gensupport.c	2000/05/27 20:03:31	1.9
--- gensupport.c	2000/07/11 20:25:29
*************** process_rtx (desc, lineno)
*** 174,178 ****
        {
  	const char *split_cond;
! 	rtx split;
  	int i;
  
--- 174,178 ----
        {
  	const char *split_cond;
! 	rtx split, attr;
  	int i;
  
*************** process_rtx (desc, lineno)
*** 209,214 ****
  
  	/* Fix up the DEFINE_INSN.  */
  	PUT_CODE (desc, DEFINE_INSN);
! 	XVEC (desc, 4) = XVEC (desc, 7);
  
  	/* Queue them.  */
--- 209,215 ----
  
  	/* Fix up the DEFINE_INSN.  */
+ 	attr = XVEC (desc, 7);
  	PUT_CODE (desc, DEFINE_INSN);
! 	XVEC (desc, 4) = attr;
  
  	/* Queue them.  */

Re: alphaev6-unknown-linux-gnu bootstrap fails RTL checking

[Richard Henderson]

On Tue, Jul 11, 2000 at 03:33:30PM -0500, Clinton Popetz wrote:
> 	* gensupport.c (process_rtx): Make rtl checking stop
> 	complaining about the define_insn while it is being
> 	converted from a define_insn_and_split.

Ok.


r~

Solved, was: Re: alphaev6-unknown-linux-gnu bootstrap fails RTL checking

[Toon Moene]

Clinton Popetz wrote:

> On Tue, Jul 11, 2000 at 07:53:08PM +0200, I wrote:

> > And quite quickly too (I had settled for an overnight build, but within
> > a minute I got the following):

[ ... ./genattr dies because of RTL check .... ]

> Hmm, evidently rtl checking has never been enabled when someone did a
> define_insn_and_split.  The insn in question is ze_and_ne, which has
> been in the tree for over a month.
> 
>                         -Clint
> 
> Tue Jul 11 15:28:21 CDT 2000  Clinton Popetz  <cpopetz@cygnus.com>
> 
>         * gensupport.c (process_rtx): Make rtl checking stop
>         complaining about the define_insn while it is being
>         converted from a define_insn_and_split.

Thanks, that did it - I just completed a full bootstrap with
--enable-checking=misc,tree,rtl,gc in 3h10m for a CVS download at
2000071210UTC on alphaev6-unknown-linux-gnu.

Make -k check results will be at gcc-testresults@gcc.gnu.org shortly.

-- 
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)