Re: [RFC/PATCH 3/3] x86/signal/64: Add explicit controls for sigcontext SS handling

[Cyrill Gorcunov <gorcunov@gmail.com>]

On Thu, Aug 13, 2015 at 01:18:50PM -0700, Andy Lutomirski wrote:
> This adds two new uc_flags flags.  UC_SAVED_SS will be set for all
> 64-bit signals (including x32).  It indicates that the saved SS field
> is valid and that the kernel understands UC_RESTORE_SS.
> 
> The kernel will *not* set UC_RESTORE_SS.  User signal handlers can
> set UC_RESTORE_SS themselves to indicate that sigreturn should
> restore SS from the sigcontext.
> 
> 64-bit programs that use segmentation are encouraged to check
> UC_SAVED_SS and set UC_RESTORE_SS in their signal handlers.  This is
> the only straightforward way to cause sigreturn to restore SS.  (The
> only non-test program that I know of that uses segmentation in a
> 64-bit binary is DOSEMU, and DOSEMU currently uses a nasty
> trampoline to work around the lack of this mechanism in old kernels.
> It could detect UC_RESTORE_SS and use it to avoid needing a
> trampoline.
> 
> Cc: Stas Sergeev <stsp@list.ru>
> Cc: Linus Torvalds <torvalds@linux-foundation.org>
> Cc: Cyrill Gorcunov <gorcunov@gmail.com>
> Cc: Pavel Emelyanov <xemul@parallels.com>
> Signed-off-by: Andy Lutomirski <luto@kernel.org>

Looks reasonable to me. Andy, Linus, what the final conclusion --
are we about to introduce this flag or simply continue with
revert? Should I test this one? (from the code I don't excpect it
break criu anyhow but still).