Re: [PATCH 1/3] drm/mediatek: Dynamically allocate CMDQ and use helper functions

[AngeloGioacchino Del Regno <angelogioacchino.delregno@collabora.com>]

Il 03/08/23 10:37, CK Hu (胡俊光) ha scritto:
> Hi, Angelo:
> 
> On Thu, 2023-08-03 at 10:25 +0200, AngeloGioacchino Del Regno wrote:
>> Il 03/08/23 08:28, CK Hu (胡俊光) ha scritto:
>>> Hi, Angelo:
>>>
>>> On Wed, 2023-08-02 at 12:41 +0200, AngeloGioacchino Del Regno
>>> wrote:
>>>> Il 02/08/23 08:24, CK Hu (胡俊光) ha scritto:
>>>>> Hi, Angelo:
>>>>>
>>>>> On Fri, 2023-06-23 at 11:49 +0200, AngeloGioacchino Del Regno
>>>>> wrote:
>>>>>>     	
>>>>>> External email : Please do not click links or open
>>>>>> attachments
>>>>>> until
>>>>>> you have verified the sender or the content.
>>>>>>     Instead of stack allocating the cmdq_client and
>>>>>> cmdq_handle
>>>>>> structures
>>>>>> switch them to pointers, allowing us to migrate this driver
>>>>>> to
>>>>>> use
>>>>>> the
>>>>>> common functions provided by mtk-cmdq-helper.
>>>>>> In order to do this, it was also necessary to add a `priv`
>>>>>> pointer to
>>>>>> struct cmdq_client, as that's used to pass (in this case) a
>>>>>> mtk_crtc
>>>>>> handle to the ddp_cmdq_cb() mailbox RX callback function.
>>>>>>
>>>>>> Signed-off-by: AngeloGioacchino Del Regno <
>>>>>> angelogioacchino.delregno@collabora.com>
>>>>>> ---
>>>>>>     drivers/gpu/drm/mediatek/mtk_drm_crtc.c | 107 +++++++-----
>>>>>> ---
>>>>>> -------
>>>>>> --
>>>>>>     include/linux/soc/mediatek/mtk-cmdq.h   |   1 +
>>>>>>     2 files changed, 32 insertions(+), 76 deletions(-)
>>>>>>
>>>>>> diff --git a/drivers/gpu/drm/mediatek/mtk_drm_crtc.c
>>>>>> b/drivers/gpu/drm/mediatek/mtk_drm_crtc.c
>>>>>> index 0df62b076f49..b63289ab6787 100644
>>>>>> --- a/drivers/gpu/drm/mediatek/mtk_drm_crtc.c
>>>>>> +++ b/drivers/gpu/drm/mediatek/mtk_drm_crtc.c
>>>>>> @@ -50,8 +50,8 @@ struct mtk_drm_crtc {
>>>>>>     	bool				pending_async_planes;
>>>>>>     
>>>>>>     #if IS_REACHABLE(CONFIG_MTK_CMDQ)
>>>>>> -	struct cmdq_client		cmdq_client;
>>>>>> -	struct cmdq_pkt			cmdq_handle;
>>>>>> +	struct cmdq_client		*cmdq_client;
>>>>>> +	struct cmdq_pkt			*cmdq_handle;
>>>>>>     	u32				cmdq_event;
>>>>>>     	u32				cmdq_vblank_cnt;
>>>>>>     	wait_queue_head_t		cb_blocking_queue;
>>>>>> @@ -108,47 +108,6 @@ static void
>>>>>> mtk_drm_finish_page_flip(struct
>>>>>> mtk_drm_crtc *mtk_crtc)
>>>>>>     	}
>>>>>>     }
>>>>>>     
>>>>>> -#if IS_REACHABLE(CONFIG_MTK_CMDQ)
>>>>>> -static int mtk_drm_cmdq_pkt_create(struct cmdq_client
>>>>>> *client,
>>>>>> struct cmdq_pkt *pkt,
>>>>>> -				   size_t size)
>>>>>> -{
>>>>>> -	struct device *dev;
>>>>>> -	dma_addr_t dma_addr;
>>>>>> -
>>>>>> -	pkt->va_base = kzalloc(size, GFP_KERNEL);
>>>>>> -	if (!pkt->va_base) {
>>>>>> -		kfree(pkt);
>>>>>> -		return -ENOMEM;
>>>>>> -	}
>>>>>> -	pkt->buf_size = size;
>>>>>> -	pkt->cl = (void *)client;
>>>>>
>>>>> I have a plan to remove cl in struct cmdq_pkt. But this
>>>>> modification
>>>>> would make this plan more difficult. So I would pending this
>>>>> patch
>>>>> until cl is removed from struct cmdq_pkt.
>>>>>
>>>>
>>>> I think that this ifdef cleanup is more urgent than the removal
>>>> of
>>>> `cl` from
>>>> struct cmdq_pkt, as those ifdefs shouldn't have reached upstream
>>>> in
>>>> the first
>>>> place, don't you agree?
>>>
>>> I think removing ifdefs and using helper function are different
>>> things.
>>> You could remove ifdefs and keep mtk_drm_cmdq_pkt_create().
>>>
>>
>> I chose to do it like that because this function would otherwise be a
>> 100% duplicate of the related cmdq helper :-)
> 
> Removing cl would change the interface of cmdq_pkt_create(). And this
> is related to different maintainer's tree. So it would be a long time
> to process. For you, only removing ifdes is urgent, so use
> cmdq_pkt_create() is not urgent. So let's keep
> mtk_drm_cmdq_pkt_create() and you could remove ifdefs.
> 

Hello CK,

my CMDQ cleanup has been stuck on your intention to remove `cl` from the CMDQ
helpers for ** six months ** now.

Are you performing that removal, or can we just get this cleanup finally done?

Regards,
Angelo