Re: [PATCH] drm/i915: Init power domains early in driver load

[Jani Nikula <jani.nikula@intel.com>]

On Thu, 07 Jan 2016, Daniel Vetter <daniel.vetter@ffwll.ch> wrote:
> On Thu, Jan 7, 2016 at 1:52 PM, Ville Syrjälä
> <ville.syrjala@linux.intel.com> wrote:
>> On Thu, Jan 07, 2016 at 12:44:21PM +0100, Daniel Vetter wrote:
>>> Since
>>>
>>> commit ac9b8236551d1177fd07b56aef9b565d1864420d
>>> Author: Ville Syrjälä <ville.syrjala@linux.intel.com>
>>> Date:   Fri Nov 27 18:55:26 2015 +0200
>>>
>>>     drm/i915: Introduce a gmbus power domain
>>>
>>> gmbus also needs the power domain infrastructure right from the start,
>>> since as soon as we register the i2c controllers someone can use them.
>>>
>>> v2: Adjust cleanup paths too (Chris).
>>>
>>> Cc: Ville Syrjälä <ville.syrjala@linux.intel.com>
>>> Cc: Patrik Jakobsson <patrik.jakobsson@linux.intel.com>
>>> Cc: Imre Deak <imre.deak@intel.com>
>>> Cc: Jani Nikula <jani.nikula@intel.com>
>>> Cc: Meelis Roos <mroos@linux.ee>
>>> Cc: Chris Wilson <chris@chris-wilson.co.uk>
>>> Fixes: ac9b8236551d ("drm/i915: Introduce a gmbus power domain")
>>> Cc: stable@vger.kernel.org
>>> References: http://www.spinics.net/lists/intel-gfx/msg83075.html
>>> Tested-by: Meelis Roos <mroos@linux.ee>
>>> Signed-off-by: Daniel Vetter <daniel.vetter@intel.com>
>>> ---
>>>  drivers/gpu/drm/i915/i915_dma.c | 11 +++++------
>>>  1 file changed, 5 insertions(+), 6 deletions(-)
>>>
>>> diff --git a/drivers/gpu/drm/i915/i915_dma.c b/drivers/gpu/drm/i915/i915_dma.c
>>> index 988a3806512a..490d8b0d931e 100644
>>> --- a/drivers/gpu/drm/i915/i915_dma.c
>>> +++ b/drivers/gpu/drm/i915/i915_dma.c
>>> @@ -398,7 +398,6 @@ static int i915_load_modeset_init(struct drm_device *dev)
>>>       if (ret)
>>>               goto cleanup_vga_switcheroo;
>>>
>>> -     intel_power_domains_init_hw(dev_priv, false);
>>>
>>>       intel_csr_ucode_init(dev_priv);
>>>
>>> @@ -1025,6 +1024,8 @@ int i915_driver_load(struct drm_device *dev, unsigned long flags)
>>>
>>>       intel_irq_init(dev_priv);
>>>       intel_uncore_sanitize(dev);
>>> +     intel_power_domains_init(dev_priv);
>>> +     intel_power_domains_init_hw(dev_priv);
>>
>> I think intel_init_dpio() needs to be moved too. We need to know the
>> DPIO IOSF ports before attempting to talk to the PHY (which can happen
>> from intel_power_domains_init_hw()).
>
> Ugh, will change.
>
>> I'm also wondering why we're doing gmbus init this early. We shouldn't
>> need it until modeset init.
>
> Anyone can access the gmbus controller once we register it. Userspace
> can (like what seems to happen on Meelis' box), but also the i2c core
> has some auto-probed stuff in some configs afaik.

Ville's question was why we register the i2c adapters this early.

As I explained in [1], the auto-probing happens when there's an i2c
driver with matching class (I2C_CLASS_DDC). That's what happens on
Meelis' box. But yes, we need to take userspace into account as well.

In short, when we call intel_gmbus_setup(), we need to be ready for i2c
communication.

BR,
Jani.




[1] http://mid.gmane.org/87vb75znpz.fsf@intel.com


-- 
Jani Nikula, Intel Open Source Technology Center