Re: regression after "drm/i915: fix page flip finish vs. prepare on plane B"

[Jesse Barnes <jbarnes-Y1mF5jBUw70BENJcbMCuUQ@public.gmane.org>]

On Sun, 04 Jul 2010 20:08:13 +0200
Alexey Fisher <bug-track-M18mAb7Tlt0yCq4wW13eYl6hYfS7NtTn@public.gmane.org> wrote:

> Hallo all,
> i have suspend problem after this patch:
> 
> commit 70565d00db6ef5735819db973fa8da95bd34a6ab
> Author: Jesse Barnes <jbarnes-i+zFwBhNuU+v7KeE/7ZovXzaFgp7qdfK@public.gmane.org>
> Date:   Thu Jul 1 04:45:43 2010 -0700
> 
>     drm/i915: fix page flip finish vs. prepare on plane B
>     
>     The refreshed patch had a copy & paste bug.
>     
>     Reported-by: Simon Farnsworth <simon.farnsworth-ByZzAK2WNt/QT0dZR+AlfA@public.gmane.org>
>     Signed-off-by: Jesse Barnes <jbarnes-Y1mF5jBUw70BENJcbMCuUQ@public.gmane.org>
>     Signed-off-by: Linus Torvalds <torvalds-de/tnXTf+JLsfHDXvbKv3WD2FQJk+8+b@public.gmane.org>
> 
> diff --git a/drivers/gpu/drm/i915/i915_irq.c
> b/drivers/gpu/drm/i915/i915_irq.c
> index d676e55..dba53d4 100644
> --- a/drivers/gpu/drm/i915/i915_irq.c
> +++ b/drivers/gpu/drm/i915/i915_irq.c
> @@ -947,9 +947,9 @@ irqreturn_t i915_driver_irq_handler(DRM_IRQ_ARGS)
>                 }
>  
>                 if (iir & I915_DISPLAY_PLANE_B_FLIP_PENDING_INTERRUPT) {
> +                       intel_prepare_page_flip(dev, 1);
>                         if (dev_priv->flip_pending_is_done)
>                                 intel_finish_page_flip_plane(dev, 1);
> -                       intel_prepare_page_flip(dev, 1);
>                 }
>  
>                 if (pipea_stats & vblank_status) {
> 
> =============================================
> 
> for testing i use plain "echo mem > /sys/power/state"
> after this, pc will initiate suspending and freeze with black screen and
> cursor on the top left corner.
> after reversing this patch suspend working again.
> My conf: Intel DG45ID board (G45 + ICH10) and 64bit ubuntu 10.04

Sounds like your kernel may have panic'd with the change.  The previous
code was definitely wrong though, so it may be a more general problem
with the flip done vs. pending bit.

Can you find the "vt/console: try harder to print output when panicing"
patch and apply that?  You may get some better output and we can figure
out what's going on.

Thanks,
-- 
Jesse Barnes, Intel Open Source Technology Center