Re: Bootstrap failure on ia64

[Roger Sayle <roger at www dot eyesopen dot com>]

Hi Andreas,
> Current mainline does not bootstrap on ia64:
>
> stage1/xgcc -Bstage1/ -B/usr/local/ia64-suse-linux/bin/ -c   -g -O2 -DIN_GCC   -W -Wall -Wwrite-strings -Wstrict-prototypes -Wmissing-prototypes -pedantic -Wno-long-long -Werror -fno-common -Wno-error  -DHAVE_CONFIG_H -DGENERATOR_FILE    -I. -I. -I../../gcc -I../../gcc/. -I../../gcc/config -I../../gcc/../include \
>   ../../gcc/gengtype-yacc.c -o gengtype-yacc.o
> /usr/share/bison/bison.simple: In function `yyparse':
> /usr/share/bison/bison.simple:778: internal compiler error: in trunc_int_for_mode, at explow.c:56
> Please submit a full bug report,
> with preprocessed source if appropriate.
> See <URL:http://gcc.gnu.org/bugs.html> for instructions.
>
> Reverting this change causes the failure to go away:
>
> 2003-05-18  Roger Sayle  <roger@eyesopen.com>
> 	    Zack Weinberg  <zack@codesourcery.com>
>
> 	PR middle-end/10472
> 	* builtins.c (expand_builtin_memcpy):  Call force_operand on
> 	expressions and use simplify_gen_binary to create the addition.


Sorry for the inconvenience.  I'm looking into it.  But any help
that you can provide would be very much appreciated.  For example,
a preprocessed source file (privately?) or what the value of mode
is in the failing trunc_int_for_mode or a stack trace.

I've compiled a cross compiler to ia64-suse-linux, which built without
a problem and all my small testcases for mempcpy and stpcpy appear to
be working fine with the cc1.  Hence a gengtype-yacc.i would really help
the process.  I'm having difficulty getting the include paths right
from a cross-compiler tree.

In the meantime, "-fno-builtin-mempcpy -fno-builtin-stpcpy" should
act as a workaround, should you need to bootstrap the compiler.


Sorry I can't provide an immediate fix.  From my understanding of it
the code looks good and is an improvement on what was there before.
I'll try my best to reproduce this on ia64-hp-hpux11.22, but the
very different handling of Pmode vs. ptr_mode on HPUX may mask the
problem.


For the record the patch was originally posted here:
http://gcc.gnu.org/ml/gcc-patches/2003-05/msg00377.html
and approved by Zack via private e-mail (with a small modification).
It was retested with a full bootstrap and full regression testing on
i686-pc-linux-gnu just yesterday with no new failures on that target.

My apologies one again.

Roger
--
Roger Sayle,                         E-mail: roger@eyesopen.com
OpenEye Scientific Software,         WWW: http://www.eyesopen.com/
Suite 1107, 3600 Cerrillos Road,     Tel: (+1) 505-473-7385
Santa Fe, New Mexico, 87507.         Fax: (+1) 505-473-0833