Re: some signal tests on amd64

[Antti Kantee <pooka%cs.hut.fi@localhost>]

On Mon Feb 21 2011 at 15:53:01 +0000, Christos Zoulas wrote:
> In article <20110221072753.GB2111%cs.hut.fi@localhost>,
> Antti Kantee  <pooka%cs.hut.fi@localhost> wrote:
> >Between 2011-02-15 19:20:02 and 2011-02-18 23:50:02 two signal-related
> >tests on amd64 started failing.  Unfortunately it's not possible to give
> >a more accurate interval due to the build being broken and the kernel
> >being panicky in that interval.  Seems like it's not an issue on i386
> >or macppc, so it might be only and amd64 issue (though sparc64 results
> >are not in yet).
> >
> >Can someone who has an idea have a quick look?
> >thanks
> >
> >http://www.whooppee.com/~paul/amd64-results/1133_atf.html#lib_libc_gen_t_siginfo_sigchild_dump
> >http://www.whooppee.com/~paul/amd64-results/1133_atf.html#atf_atf-c_detail_process_test_status_coredump
> 
> Well, the test is fine, nothing can coredump anymore since coredumping became
> a module.
> 
> Try 
> 
> $ ulimit -c unlimited
> $ sleep 10 &
> $ kill -QUIT $!

Oh, heh, that's a bit ridiculous.  Apparently i386 works because exec
formats are modules which depend on the coredump module, but on amd64
they're builtins, and coredumps are not loaded as a dependency.

I put coredumps back into GENERIC as a re-regression until someone can
figure out a solution.

-- 
älä karot toivorikkauttas, kyl rätei ja lumpui piisaa