* [PATCH] xtensa: don't clear cpenable unconditionally on release
@ 2018-11-27 0:47 Max Filippov
2018-11-27 1:27 ` Max Filippov
0 siblings, 1 reply; 2+ messages in thread
From: Max Filippov @ 2018-11-27 0:47 UTC (permalink / raw)
To: linux-xtensa; +Cc: Chris Zankel, Max Filippov, stable
fast_coprocessor exception handler expects that the coprocessor owner
task never gets the "coprocessor disabled" exception for a coprocessor
that is marked as enabled in the thread_info::cpenable. If that happens
it will reload potentially outdated context from the thread_info
structure into the coprocessor registers.
Only call coprocessor_clear_cpenable from the coprocessor_release_all
if the latter is called for the current task.
Cc: stable@vger.kernel.org
Signed-off-by: Max Filippov <jcmvbkbc@gmail.com>
---
arch/xtensa/kernel/process.c | 3 ++-
1 file changed, 2 insertions(+), 1 deletion(-)
diff --git a/arch/xtensa/kernel/process.c b/arch/xtensa/kernel/process.c
index 3fa0c440f664..6c7c80106f60 100644
--- a/arch/xtensa/kernel/process.c
+++ b/arch/xtensa/kernel/process.c
@@ -87,7 +87,8 @@ void coprocessor_release_all(struct thread_info *ti)
}
ti->cpenable = cpenable;
- coprocessor_clear_cpenable();
+ if (ti == current_thread_info())
+ coprocessor_clear_cpenable();
preempt_enable();
}
--
2.11.0
^ permalink raw reply related [flat|nested] 2+ messages in thread
* Re: [PATCH] xtensa: don't clear cpenable unconditionally on release
2018-11-27 0:47 [PATCH] xtensa: don't clear cpenable unconditionally on release Max Filippov
@ 2018-11-27 1:27 ` Max Filippov
0 siblings, 0 replies; 2+ messages in thread
From: Max Filippov @ 2018-11-27 1:27 UTC (permalink / raw)
To: linux-xtensa; +Cc: Chris Zankel, stable
On Mon, Nov 26, 2018 at 4:47 PM Max Filippov <jcmvbkbc@gmail.com> wrote:
>
> fast_coprocessor exception handler expects that the coprocessor owner
> task never gets the "coprocessor disabled" exception for a coprocessor
> that is marked as enabled in the thread_info::cpenable. If that happens
> it will reload potentially outdated context from the thread_info
> structure into the coprocessor registers.
After looking at it some more I see that the above is not true. It will
flush current coprocessor context to the thread_info and then load it
back from there. No correctness issue, but some extra work that
can be avoided.
--
Thanks.
-- Max
^ permalink raw reply [flat|nested] 2+ messages in thread
end of thread, other threads:[~2018-11-27 12:23 UTC | newest]
Thread overview: 2+ messages (download: mbox.gz follow: Atom feed
-- links below jump to the message on this page --
2018-11-27 0:47 [PATCH] xtensa: don't clear cpenable unconditionally on release Max Filippov
2018-11-27 1:27 ` Max Filippov
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox