Re: [PATCH 4/8] x86/vm86: Use the normal pt_regs area for vm86

[Brian Gerst]

On Wed, Jul 29, 2015 at 1:16 PM, Linus Torvalds
<torvalds@xxxxxxxxxxxxxxxxxxxx> wrote:
> On Wed, Jul 29, 2015 at 10:14 AM, Brian Gerst <brgerst@xxxxxxxxx> wrote:
>>
>> I think it was causing signal handling to fail, but I can't remember
>> exactly.
>
> Ugh.
>
> If that hunk made a difference, then there is something wrong with
> your patch-series. So please double-check.
>
>                 Linus

I think I remember now what the issue was.  Since entering vm86 mode
uses force_iret(), the work_pending code path was being taken.  I had
to move the call to save_v86_state out of here to handle_signal(),
otherwise it would just restore the 32-bit regs and exit the syscall
without ever entering vm86 mode.  But that meant that the test for
kernel CS was seeing the vm86 regs instead of 32-bit regs, and was
failing because it didn't account for the VM bit (if the real-mode CS
looked like RPL 0).  A fault would get stuck in a loop because it
couldn't exit to the signal handling code.

--
Brian Gerst
--
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/