[Intel-gfx] [PATCH] drm/i915: Fix up the forcewake timer initialization

[Daniel Vetter]

On Wed, Mar 19, 2014 at 8:26 AM, Chris Wilson <chris at chris-wilson.co.uk> wrote:
> On Tue, Mar 18, 2014 at 05:41:54PM -0700, Ben Widawsky wrote:
>> On Tue, Mar 18, 2014 at 04:31:03PM +0100, Daniel Vetter wrote:
>> > This is a regression introduced in
>> >
>> > commit 0294ae7b44bba7ab0d4cef9a8736287f38bdb4fd
>> > Author: Chris Wilson <chris at chris-wilson.co.uk>
>> > Date:   Thu Mar 13 12:00:29 2014 +0000
>> >
>> >     drm/i915: Consolidate forcewake resetting to a single function
>> >
>> > The reordered setup sequence ended up calling del_timer_sync before
>> > the timer was set up correctly, resulting in endless hilarity when
>> > loading the driver.
>> >
>> > Compared to Ben's patch (which moved around the setup_timer call to
>> > sanitize_early) this moves the sanitize_early call around in the
>> > driver load call. This way we avoid calling setup_timer again in the
>> > resume code (where we also call sanitize_early).
>> >
>> > Cc: Chris Wilson <chris at chris-wilson.co.uk>
>> > Cc: Mika Kuoppala <mika.kuoppala at intel.com>
>> > Cc: Ben Widawsky <benjamin.widawsky at intel.com>
>> > Tested-by: Rodrigo Vivi <rodrigo.vivi at gmail.com>
>> > Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=76242
>> > Signed-off-by: Daniel Vetter <daniel.vetter at ffwll.ch>
>> > ---
>> >  drivers/gpu/drm/i915/i915_dma.c     | 2 --
>> >  drivers/gpu/drm/i915/intel_uncore.c | 2 ++
>> >  2 files changed, 2 insertions(+), 2 deletions(-)
>> >
>> > diff --git a/drivers/gpu/drm/i915/i915_dma.c b/drivers/gpu/drm/i915/i915_dma.c
>> > index e4d2b9f15ae2..9faee49f210d 100644
>> > --- a/drivers/gpu/drm/i915/i915_dma.c
>> > +++ b/drivers/gpu/drm/i915/i915_dma.c
>> > @@ -1608,8 +1608,6 @@ int i915_driver_load(struct drm_device *dev, unsigned long flags)
>> >             goto put_bridge;
>> >     }
>> >
>> > -   intel_uncore_early_sanitize(dev);
>> > -
>> >     /* This must be called before any calls to HAS_PCH_* */
>> >     intel_detect_pch(dev);
>> >
>> > diff --git a/drivers/gpu/drm/i915/intel_uncore.c b/drivers/gpu/drm/i915/intel_uncore.c
>> > index e2e328d86aff..c3832d9270a6 100644
>> > --- a/drivers/gpu/drm/i915/intel_uncore.c
>> > +++ b/drivers/gpu/drm/i915/intel_uncore.c
>> > @@ -736,6 +736,8 @@ void intel_uncore_init(struct drm_device *dev)
>> >     setup_timer(&dev_priv->uncore.force_wake_timer,
>> >                 gen6_force_wake_timer, (unsigned long)dev_priv);
>> >
>> > +   intel_uncore_early_sanitize(dev);
>> > +
>> >     if (IS_VALLEYVIEW(dev)) {
>> >             dev_priv->uncore.funcs.force_wake_get = __vlv_force_wake_get;
>> >             dev_priv->uncore.funcs.force_wake_put = __vlv_force_wake_put;
>>
>> If you only want to setup_timer once, the setup_timer call should be in
>> intel_uncore_init() which is the only one called only at load time. And
>> of course, this is where the bug is. Otherwise, thaw calls
>> uncore_early_sanitize, which will setup_timer again (which I thought was
>> your complaint with my original patch).
>
> I don't think we want to stop checking for garbage in the forcewake
> registers upon resume though?

We still have a call to early_sanitize in i915_thaw, which should take
care of this. My intention here was to only touch codepaths
exclusively run at driver load time, and both i915_driver_load and
uncore_init aren't used anywhere else. Or do I miss something and
should wait for the morning coffee to kick in?

Cheers, Daniel
-- 
Daniel Vetter
Software Engineer, Intel Corporation
+41 (0) 79 365 57 48 - http://blog.ffwll.ch