Re: [PATCH] arm: Remove early stack deallocation from restore_user_regs

[Russell King - ARM Linux <linux@arm.linux.org.uk>]

On Fri, Jan 09, 2015 at 05:06:54PM +0000, Daniel Thompson wrote:
> On 09/01/15 16:46, Russell King - ARM Linux wrote:
> > On Mon, Jan 05, 2015 at 03:12:38PM +0000, Daniel Thompson wrote:
> >> Currently restore_user_regs deallocates the SVC stack early in
> >> its execution and relies on no exception being taken between
> >> the deallocation and the registers being restored. The introduction
> >> of a default FIQ handler that also uses the SVC stack breaks this
> >> assumption and can result in corrupted register state.
> >>
> >> This patch works around the problem by removing the early
> >> stack deallocation and using r2 as a temporary instead. I have
> >> not found a way to do this without introducing an extra mov
> >> instruction to the macro.
> >>
> >> Signed-off-by: Daniel Thompson <daniel.thompson@linaro.org>
> >> ---
> > 
> > Please put it in the patch system, thanks.
> 
> Will do.
> 
> 
> > I think we should queue
> > this one for stable too, as I think we need this for v3.18
> > (as a result of c0e7f7ee717e2b4c5791e7422424c96b5008c39e,
> > ARM: 8150/3: fiq: Replace default FIQ handler)?
> 
> It's a close call.

I agree.

> Before 8150/3 the system would probably crash if the default FIQ handler
> ran. After 8150/3 the system is also likely to crash since there's no
> code hooked into the handler in v3.18 that can clear the source of FIQ
> leaving us stuck re-entering the FIQ handler.
> 
> Nevertheless, this is a nasty gremlin to leave for backporters (it
> wasn't easy to find) so I'd be very happy to Cc: stable and see what
> they think.

Well, we could ask Greg now.  It's not a regression (as nothing makes
use of the previously merged changes yet), but it is a nasty latent bug
which we could easily solve.

Having thought about it some more, I'm tempted to say... leave the
stable tag off it, and we can make the decision later - after it's had
a chance to really prove itself to a much wider audience.  That'd be
the lowest risk for the 3.18 stable tree.

-- 
FTTC broadband for 0.8mile line: currently at 10.5Mbps down 400kbps up
according to speedtest.net.