Re: Call for testers for pch-branch

[Geoff Keating <geoffk at geoffk dot org>]

Andreas Schwab <schwab@suse.de> writes:

> I'm getting many PCH testsuite failures, see
> <http://gcc.gnu.org/ml/gcc-testresults/2002-12/msg00096.html>.  These
> "test for excess errors" failures are all segfaults in the compiler.
> Unfortunately I haven't been able to reproduce them by running exactly the
> same command outside the testsuite.

Hmmm.  I should really add a bit more logging to the testsuite, there
are a few file renames that don't get logged.

Basically, what the testsuite does is

cp foo.h foo.hp
gcc foo.c -I. -S -o foo.c.nopch
rm foo.hp
gcc foo.h -o foo.hp.pch
gcc foo.c -I. -S -o foo.c.pch
diff foo.c.nopch foo.c.pch

... but it's also quite possible that you've hit a bug that depends on
the stack position or something.

I do know that all the testruns that don't use -g pass on both Darwin
native and Darwin X powerpc-eabisim, and in no case is there a crash.
But, if there was some subtle random-memory-access bug, this could
just be a coincidence.


-- 
- Geoffrey Keating <geoffk@geoffk.org>