[PATCH] drm/rockchip: vop: power off until vop standby take effect

[Daniel Kurtz]

On Feb 4, 2015 11:38 AM, "Mark yao" <mark.yao at rock-chips.com> wrote:
>
> On 2015年02月02日 15:53, Daniel Vetter wrote:
>>
>> On Mon, Feb 02, 2015 at 10:30:09AM +0800, Mark yao wrote:
>>>
>>> On 2015年02月02日 10:07, Daniel Kurtz wrote:
>>>>
>>>> Hi Mark, Heiko,
>>>>
>>>> On Sat, Jan 31, 2015 at 4:41 PM, Mark Yao <mark.yao at rock-chips.com> wrote:
>>>>>
>>>>> Vop standby will take effect end of current frame,
>>>>> if dsp_hold_valid_irq happen, it means vop standby complete.
>>>>>
>>>>> we must wait standby complete when we want to disable aclk,
>>>>> if not, memory bus maybe dead.
>>>>>
>>>>> Signed-off-by: Mark Yao <mark.yao at rock-chips.com>
>>>>> ---
>>>>>   drivers/gpu/drm/rockchip/rockchip_drm_vop.c |   76 ++++++++++++++++++++++-----
>>>>>   1 file changed, 63 insertions(+), 13 deletions(-)
>>>>>
>>>>> diff --git a/drivers/gpu/drm/rockchip/rockchip_drm_vop.c b/drivers/gpu/drm/rockchip/rockchip_drm_vop.c
>>>>> index fb25836..47ea61f 100644
>>>>> --- a/drivers/gpu/drm/rockchip/rockchip_drm_vop.c
>>>>> +++ b/drivers/gpu/drm/rockchip/rockchip_drm_vop.c
>>>>> @@ -89,6 +89,7 @@ struct vop {
>>>>>          /* mutex vsync_ work */
>>>>>          struct mutex vsync_mutex;
>>>>>          bool vsync_work_pending;
>>>>> +       struct completion dsp_hold_completion;
>>>>>
>>>>>          const struct vop_data *data;
>>>>>
>>>>> @@ -382,6 +383,34 @@ static bool is_alpha_support(uint32_t format)
>>>>>          }
>>>>>   }
>>>>>
>>>>> +static void vop_dsp_hold_valid_irq_enable(struct vop *vop)
>>>>> +{
>>>>> +       unsigned long flags;
>>>>> +
>>>>> +       BUG_ON(!vop->is_enabled);
>>>>
>>>> Re: Heiko "use a WARN_ON":
>>>>
>>>> If the VOP clock is off, then the system will just hang when trying to
>>>> write the VOP register so in this case, BUG_ON gives a more reliable
>>>> crash dump than the hang.
>>>
>>>    In this way, you are right, if vop clocks is disabled, write vop register
>>> will hang system, and the WARN_ON
>>
>> if (WARN_ON(cond))
>>         return;
>>
>> is what we commonly use in i915. Because it's really not any good to use
>> BUG_ON in drm drivers, at least if you expect anyone to use fbcon on top
>> of it (i.e. anything even remotely resembling a normal distro): Then the
>> intial modeset is all done under console_lock, which means that _none_ of
>> your BUG_ON will ever get out over serial console.
>>
>> The problem is fbcon locking, but no fool yet signed up to fix it (it's
>> horrible, I looked ...).
>> -Daniel
>
> I know this problem, if fbcon locking, and BUG_ON, we can't get any log from serial console, Trouble
> me many times. So know, if no drm log output, the first thing I do is remove the fbcon lock, and then
> we can see the crash dump.
>
> But here, I expect to use BUG_ON, because if BUG happen, vop register will hang system, and WARN_ON
> may be have no chance to output.

Yeah, but danvet is suggesting we just return immediately on the
WARN_ON condition, to avoid accessing the vop registers and hanging.

Thus, the WARN_ON will have a chance to put out (unless the early
return path goes off and finds some other unprotected path goes to
access unlocked registers).

-Dan

>
> Mark
>