Re: [PATCH 5/6] x86/ptrace: down with test_thread_flag(TIF_IA32)

From: Oleg Nesterov
Date: Mon Jun 13 2016 - 09:50:35 EST


To avoid the confusion, let me first say that I am not going to argue
with these changes, I simply do not understand the problem space enough.

On 06/10, Andy Lutomirski wrote:
>
> On Fri, Jun 10, 2016 at 1:07 PM, Oleg Nesterov <oleg@xxxxxxxxxx> wrote:
> >
> > IIRC, CRIU can't c/r the 32-bit applications, or this is no longer true?
> >
>
> CRIU has a horrible, nasty, brilliant idea: it will start restoring
> 32-bit processes by treating them mostly like 64-bit processes. The
> restorer will start out 64-bit, set everything up, and long
> jump/return/sigreturn/whatever back to 32-bit mode.

OK, I see,

> My proposal was
> that, rather than coming up with nasty hacks to switch the kernel's
> idea of the task bitness,

Well, I can't resist but to me SA_IA32_ABI/SA_X32_ABI looks like a hack
too. We actually shift TIF_*32 into k_sigaction->flags, and the fact
that we do this per-signal looks, well, interesting ;)

And at first glance it would be very simple to change the task bitness,
CRIU can simply exec a dummy 32-bit application before anything else.
In this case (I think) we also do not need do_map_vdso/ARCH_MAP_VDSO_*
at least right now.

Yes, I guess this will complicate CRIU significantly.

> we instead teach the kernel to respect that
> actual bitness as indicated by CS and the syscalls used to the extent
> possible.

I am still not sure the idea to remove TIF_IA32/TIF_X32 is really good.
But again, I won't argue, I do not feel I understand pro/cons enough.

> So, yes, a restored 32-bit process that crashes should dump core as
> though it's 32-bit even though it was 64-bit when execve was last
> called :)

OK, thanks for you explanation Andy.

Oleg.