Re: [PATCH 2/3 v3] x86: entry_64.S: always allocate complete "struct pt_regs"

[Andy Lutomirski]

On Feb 26, 2015 1:55 AM, "Denys Vlasenko" <vda.linux@xxxxxxxxxxxxxx> wrote:
>
> On Wed, Feb 25, 2015 at 10:59 PM, Andy Lutomirski <luto@xxxxxxxxxxxxxx> wrote:
> > On Wed, Feb 25, 2015 at 1:28 PM, Denys Vlasenko <dvlasenk@xxxxxxxxxx> wrote:
> >> On 02/25/2015 09:10 PM, Andy Lutomirski wrote:
> >> This part?
> >>
> >>         .macro FORK_LIKE func
> >>  ENTRY(stub_\func)
> >>         CFI_STARTPROC
> >> -       popq    %r11                    /* save return address */
> >> -       PARTIAL_FRAME 0
> >> -       SAVE_REST
> >> -       pushq   %r11                    /* put it back on stack */
> >> +       DEFAULT_FRAME 0, 8              /* offset 8: return address */
> >> +       SAVE_EXTRA_REGS 8
> >>         FIXUP_TOP_OF_STACK %r11, 8
> >> -       DEFAULT_FRAME 0 8               /* offset 8: return address */
> >>         call sys_\func
> >>         RESTORE_TOP_OF_STACK %r11, 8
> >> -       ret $REST_SKIP          /* pop extended registers */
> >> +       ret
> >>         CFI_ENDPROC
> >>  END(stub_\func)
> >>         .endm
> >>
> >>         FORK_LIKE  clone
> >>         FORK_LIKE  fork
> >>         FORK_LIKE  vfork
> >>
> >> But the old code (SAVE_REST thing) was also saving registers here.
> >> It had to jump through hoops (pop return address, SAVE_REST,
> >> push return address) to do that.
> >> After the patch, "SAVE_EXTRA_REGS 8" does the same, just without
> >> pop/push pair.
> >>
> >> I just don't see what's wrong with it. Can you elaborate?
> >
> > SAVE_REST pushed the regs onto the stack, whereas SAVE_EXTRA_REGS just
> > writes them in place.  It's possible for this to be called when the
> > regs have already been saved.
>
> If that would be the case - that is, if SAVE_REST was saving extra copy
> of registers on stack, then FIXUP_TOP_OF_STACK %r11, 8 would be working
> on wrong locations. The "8" there says "we have full pt_regs on stack,
> plus extra 8 bytes (the return address)". Your conjecture would mean
> that in fact there would be more bytes on stack, and FIXUP_TOP_OF_STACK
> would corrupt iret stack. Evidently, since old code was not crashing,
> this wasn't happening. SAVE_REST was really creating the "tail" of pt_regs

Ugh, you're right.

The FIXUP_TOP_OF_STACK indeed looks duplicated, bit t that's less
harmful and was already the case.

--Andy
.
>
> In addition to my previous tests, I ran my home machine with
> patched kernel. Unfortunately, it works for me :(
>
> Will try on yet another machine.
>
> --
> vda
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/