[linux-yocto] [kernel-cache][yocto-4.12][PATCH] features/i915/i915.cfg: compile i915 as a module

[Liwei Song]

On 12/06/2017 09:03 PM, Bruce Ashfield wrote:
> On 12/05/2017 09:10 PM, Liwei Song wrote:
>>
>>
>> On 12/06/2017 06:26 AM, Bruce Ashfield wrote:
>>> On 12/04/2017 08:16 PM, Liwei Song wrote:
>>>> Set i915 as a moudle to trigger the firmware load
>>>
>>> typo: module
>>>
>>>> at the same time the module is loaded.
>>>>
>>>> This can aviod i915 driver load rootfs's firmware before
>>>> rootfs was mounted error.
>>>
>>> I'm not following this problem description, can you rephrase ?
>>> Is it a timing problem between the driving starting (as the
>>> built in) and it triggering a firmware load (that fails due
>>> to the rootfs not being ready) ?
>>
>> Yes, it is a timing problem, how about rephrase it as following:
>>
>> This can aviod a timing problem between the driver starting
> 
> s/aviod/avoid/
> 
>> and it triggering a firmware load, after compile it as module
>> i915 driver will start after the rootfs is ready, then the
>> firmware store in /lib/firmware/ can be load successful by
>> i915 driver.
>>
> 
> But yes, that log is good for a v2.
> 
>>  
>>>
>>> presumably the i915 was built in for a reason, i.e. so it
>>> is available in early boot without the need to load a module.
>>> And this change would break that functionality. Also, are
>>> there any other build settings that need to arrange for the
>>> module autoload ?
>>
>> I suppose no during my test on board.
> 
> So the device is likely triggering the module auto-load
> via the helper ? It would be good to fully understand this part,
> since it also means that any initramfs needs the module.
> 
> What type of boot are you doing for the test ? Is there an
> initramfs/initrd involved ?

I have test the hdd image before you ask, but did not test initramfs/initrd.
But now I also test it with initramfs, the i915 module also can be auto-load
by device trigger it. If we do not have i915 device, the module won't be loaded either.


Thanks,
Liwei.

> 
> Bruce
> 
>>
>> Thanks,
>> Liwei.
>>
>>
>>>
>>> Since there is that possibility, restating the problem you are
>>> trying to fix would be good.
>>>
>>> Bruce
>>>
>>>>
>>>> Signed-off-by: Liwei Song <liwei.song at windriver.com>
>>>> ---
>>>>    features/i915/i915.cfg | 2 +-
>>>>    1 file changed, 1 insertion(+), 1 deletion(-)
>>>>
>>>> diff --git a/features/i915/i915.cfg b/features/i915/i915.cfg
>>>> index 6dcc60f2dfc7..60cb963e014b 100644
>>>> --- a/features/i915/i915.cfg
>>>> +++ b/features/i915/i915.cfg
>>>> @@ -8,6 +8,6 @@ CONFIG_AGP=y
>>>>    CONFIG_AGP_INTEL=y
>>>>    CONFIG_DRM=y
>>>>    CONFIG_DRM_KMS_HELPER=y
>>>> -CONFIG_DRM_I915=y
>>>> +CONFIG_DRM_I915=m
>>>>    CONFIG_DRM_KMS_FB_HELPER=y
>>>>    CONFIG_DRM_I915_ALPHA_SUPPORT=y
>>>>
>>>
>>>
>>>
> 
> 
>