testcase for validate_replace_rtx_1 fix.

[Jan Hubicka]

> 
> I have committed these two testcases.  They don't fail for me on
> powerpc-linux, but I expect they do fail on some other platform (Jan,
> which platform?).  I added the following ChangeLog entry:
i386 fails for me:
u-pl1:/tmp/moje/egcs-20001002.64/gcc$ ./xgcc -O2 -B ./ g.c
u-pl1:/tmp/moje/egcs-20001002.64/gcc$ ./a.out
Aborted (core dumped)
u-pl1:/tmp/moje/egcs-20001002.64/gcc$ ./xgcc -O2 -B ./ t.c
u-pl1:/tmp/moje/egcs-20001002.64/gcc$ ./a.out
Aborted (core dumped)
u-pl1:/tmp/moje/egcs-20001002.64/gcc$ ./xgcc  -B ./ t.c
u-pl1:/tmp/moje/egcs-20001002.64/gcc$ ./a.out
u-pl1:/tmp/moje/egcs-20001002.64/gcc$ ./xgcc  -B ./ g.c
u-pl1:/tmp/moje/egcs-20001002.64/gcc$ ./a.out
I am using the snapshot, but no steps to fix the problems has been done so
far and at (modified) current CVS version it fails as well.
> 
> 2000-10-09  Geoff Keating  <geoffk@cygnus.com>
> 
> 	* gcc.c-torture/compile/20001009-1.c: New testcase.
> 	* gcc.c-torture/execute/20001009-1.c: New testcase.
Both testcases fails at runtime only... Will you move the first to execute,
or may I do it myself?
Can you also please send me the -da files for the testcase?  Fact that it
don't fail for PPC is interesting, since this is generic bug.

Honza
> 	Testcases provided by Jan Hubicka <jh@suse.cz>.
> 
> -- 
> - Geoffrey Keating <geoffk@cygnus.com>