Re: can't we kill DIE_GPF ? (Was: x86/traps: cleanup DO_ERROR*() to prepare for uprobes fixes)

[Oleg Nesterov <oleg@redhat.com>]

On 05/13, Masami Hiramatsu wrote:
>
> (2014/05/09 23:07), Oleg Nesterov wrote:
> > On 05/08, Oleg Nesterov wrote:
> >>
> >> For example, after this series
> >> we can convert math_error() into the "normal" DO_ERROR() user, and most probably
> >> we can do the same with do_general_protection().
> >
> > As for do_general_protection(), the problem is DIE_GPF.
> >
> > Masami, could you explain why it is needed ? kprobe_exceptions_notify()
> > is the only user, can't it use DIE_TRAP and check trapnr = X86_TRAP_GP ?
>
> Actually, this may be only for something which will happen on
> single-stepping out-of-line. And yes, I can move it onto the DIE_TRAP :)
>
> >
> > And if it can, probably we can do notify_die() at the start like other
> > DO_ERROR() functions do ?
>
> Agreed, it seems OK to me. (and seems better, since we can handle GPF
> before changing task->thread struct)
>
> Acked-by: Masami Hiramatsu <masami.hiramatsu.pt@hitachi.com>

Great, thanks!

I'll resend this patch with the changelog and other changes.

Oleg.