Re: drm/debugfs: Drop conditionals around of_node pointers

From: Sui Jingfeng
Date: Wed May 01 2024 - 00:36:41 EST


Hi,


On 2024/5/1 05:33, Doug Anderson wrote:
Hi,

On Mon, Apr 29, 2024 at 6:16 PM 隋景峰 <suijingfeng@xxxxxxxxxx> wrote:
Hi,


-----原始邮件-----
发件人: "Maxime Ripard" <mripard@xxxxxxxxxx>
发送时间: 2024-04-29 19:30:24 (星期一)
收件人: "Sui Jingfeng" <suijingfeng@xxxxxxxxxx>
抄送: "Sui Jingfeng" <sui.jingfeng@xxxxxxxxx>, "Maarten Lankhorst" <maarten.lankhorst@xxxxxxxxxxxxxxx>, "Thomas Zimmermann" <tzimmermann@xxxxxxx>, "David Airlie" <airlied@xxxxxxxxx>, "Daniel Vetter" <daniel@xxxxxxxx>, "Douglas Anderson" <dianders@xxxxxxxxxxxx>, "Laurent Pinchart" <laurent.pinchart+renesas@xxxxxxxxxxxxxxxx>, "Biju Das" <biju.das.jz@xxxxxxxxxxxxxx>, dri-devel@xxxxxxxxxxxxxxxxxxxxx, linux-kernel@xxxxxxxxxxxxxxx
主题: Re: drm/debugfs: Drop conditionals around of_node pointers

On Sun, Apr 28, 2024 at 04:52:13PM +0800, Sui Jingfeng wrote:
ping

在 2024/3/22 06:22, Sui Jingfeng 写道:
Having conditional around the of_node pointer of the drm_bridge structure
turns out to make driver code use ugly #ifdef blocks.
The code being ugly is an opinion, what problem is it causing exactly?

Drop the conditionals to simplify debugfs.
What does it simplifies?

Fixes: d8dfccde2709 ("drm/bridge: Drop conditionals around of_node pointers")
Signed-off-by: Sui Jingfeng <sui.jingfeng@xxxxxxxxx>
Why do we want to backport that patch to stable?
Technically it's not CCing stable and so it's not really incorrect.
...but I agree that this is a bit of a stretch to call it a "Fix".
Maybe drop the "Fixes" line?


OK, good idea, acceptable.

Originally, I add the fix tag to hint that my modification belongs to
the commit d8dfccde2709 ("drm/bridge: Drop conditionals around of_node pointers") too.
But get missed.



My commit message is written based on commit of d8dfccde2709

$ git show c9e358dfc4a8

This patch is based on commit c9e358dfc4a8 ("driver-core: remove
conditionals around devicetree pointers").

Having conditional around the of_node pointer of the drm_bridge
structure turns out to make driver code use ugly #ifdef blocks. Drop the
conditionals to simplify drivers. While this slightly increases the size
of struct drm_bridge on non-OF system, the number of bridges used today
and foreseen tomorrow on those systems is very low, so this shouldn't be
an issue.

So drop #if conditionals by adding struct device_node forward declaration.

Maxime
I'm just start to contribute by mimic other people's tone, there seems no need
to over read.
I think the fact that you skipped the reference to commit c9e358dfc4a8
("driver-core: remove conditionals around devicetree pointers") was
relevant here. Referencing that commit makes it easy for the reader to
see that you are following convention used throughout the kernel and
not just asserting your own opinion about style.

If you add that reference into your commit message and send a v2, I'm
happy to apply it.


OK, thanks a lot.


-Doug

--
Best regards,
Sui