Hi Dave,
Could you please pick this patch for v3.10 ?
On Tuesday 12 March 2013 15:38:43 Laurent Pinchart wrote:
> The page flip handler stores the page flip event pointer and then calls
> drm_vblank_get() to enable the vblank interrupt. Due to the vblank off
> delay, the vblank interrupt can be enabled in the hardware at that
> point, even if the vblank reference count is equal to 0. If a vblank
> interrupt is triggered between storing the event pointer and calling
> drm_vblank_get(), the page flip completion handler will process the
> event and call drm_vblank_put() with a reference count equal to 0. This
> will result in a BUG_ON.
>
> Fix the race condition by calling drm_vblank_get() before storing the
> event pointer.
>
> Signed-off-by: Laurent Pinchart <laurent.pinchart+renesas at ideasonboard.com>
> ---
> drivers/gpu/drm/shmobile/shmob_drm_crtc.c | 2 +-
> 1 file changed, 1 insertion(+), 1 deletion(-)
>
> diff --git a/drivers/gpu/drm/shmobile/shmob_drm_crtc.c
> b/drivers/gpu/drm/shmobile/shmob_drm_crtc.c index d917a41..7dff49e 100644
> --- a/drivers/gpu/drm/shmobile/shmob_drm_crtc.c
> +++ b/drivers/gpu/drm/shmobile/shmob_drm_crtc.c
> @@ -494,10 +494,10 @@ static int shmob_drm_crtc_page_flip(struct drm_crtc
> *crtc,
>
> if (event) {
> event->pipe = 0;
> + drm_vblank_get(dev, 0);
> spin_lock_irqsave(&dev->event_lock, flags);
> scrtc->event = event;
> spin_unlock_irqrestore(&dev->event_lock, flags);
> - drm_vblank_get(dev, 0);
> }
>
> return 0;
--
Regards,
Laurent Pinchart
