Damn, sorry for noise. On 10/06, Oleg Nesterov wrote: > > Forgot to mention... > > I didn't test this patch and it is in fact buggy, we should > set sig = context->siginfo->si_signo when !valid_signal(sig).
Heh. And I just realized SEND_SIG_NOINFO == NULL, so this patch can't work. But I think it is clear what it should do. off-topic, but can't resist. Personally I don't like the fact SEND_SIG_NOINFO == NULL. Regardless of this patch, what do you think if we change these SEND_XXX defines so that no one can be confused with NULL ? Oleg.