[PATCH v5 3/9] drm/hisilicon/hibmc: Add support for frame buffer

[Rongrong Zou]

在 2016/10/26 20:59, Daniel Vetter 写道:
> On Wed, Oct 26, 2016 at 05:19:31PM +0800, Rongrong Zou wrote:
>> Hi Daniel,
>>
>> Thansk for reviewing!
>>
>> 在 2016/10/26 13:56, Daniel Vetter 写道:
>>> On Wed, Oct 26, 2016 at 10:37:00AM +0800, Rongrong Zou wrote:
>>>> Add support for fbdev and kms fb management.
>>>>
>>>> Signed-off-by: Rongrong Zou <zourongrong at gmail.com>
>>>
>>> Small drive-by comment below.
>>>
>>>> diff --git a/drivers/gpu/drm/hisilicon/hibmc/hibmc_drm_drv.h b/drivers/gpu/drm/hisilicon/hibmc/hibmc_drm_drv.h
>>>> index db8d80e..d41138a 100644
>>>> --- a/drivers/gpu/drm/hisilicon/hibmc/hibmc_drm_drv.h
>>>> +++ b/drivers/gpu/drm/hisilicon/hibmc/hibmc_drm_drv.h
>>>> @@ -20,9 +20,23 @@
>>>>    #define HIBMC_DRM_DRV_H
>>>>
>>>>    #include <drm/drmP.h>
>>>> +#include <drm/drm_fb_helper.h>
>>>>    #include <drm/ttm/ttm_bo_driver.h>
>>>>    #include <drm/drm_gem.h>
>>>>
>>>> +struct hibmc_framebuffer {
>>>> +	struct drm_framebuffer fb;
>>>> +	struct drm_gem_object *obj;
>>>> +	bool is_fbdev_fb;
>>>> +};
>>>> +
>>>> +struct hibmc_fbdev {
>>>> +	struct drm_fb_helper helper;
>>>> +	struct hibmc_framebuffer fb;
>>>
>>> I wouldn't embed the single framebuffer here, but instead have a pointer
>>> and just refcount it. This here is a pattern that predates framebuffer
>>> refcounting, and it leads to plenty of surprises.
>>
>> will allocate fbdev in next patch version, thanks.
>
> Not the fbdev, the hibmc_framebuffer.

Understood, thanks.

>
>>> Maybe we should update the documentation of
>>> drm_framebuffer_unregister_private() to mention that it is deprecated? The
>>> overview doc in drm_framebuffer.c already explains that, but I guess
>>> that's not obvious enough.
>>
>> Just replace drm_framebuffer_unregister_private() with
>> drm_framebuffer_remove()?
>>
>> I found many other drivers use the following two functions in their
>> own xxx_fbdev_destroy():
>> 	drm_framebuffer_unregister_private(fbdev->fb);
>> 	drm_framebuffer_remove(fbdev->fb);
>> so the former is redundant and can be removed directly?
>
> They all embed the fb instead of having a pointer, because those drivers
> are all older than the fb refcounting support. In general good practice is
> to look at the most recently merged driver, not at all of them. Only the
> most recently one has a good chance to be up-to-date with latest best
> practices. The function to call would be drm_framebuffer_unreference(),
> and your struct above should be
>
> struct hibmc_fbdev {
> 	struct drm_fb_helper helper;
> 	struct hibmc_framebuffer *fb;
> 	...
> };
>
> Note how fb is a pointer here, not an embedded struct.

And in hibmc_user_framebuffer_destroy(), it should call
kfree(hibmc_framebuffer *hibmc_fb) not kfree(drm_framebuffer *fb), thanks.

regards,
Rongrong

> -Daniel
>
>>
>>>
>>> Can you pls do that patch? And pls make sure it all looks pretty when
>>> building the docs with
>>
>> No problem, i'll send another patch later.
>>
>> Regards,
>> Rongrong
>>
>>>
>>> $ make htmldocs
>>>
>>> Thanks, Daniel
>>>
>>
>>
>>
>