Re: [PATCH v2 2/3] drm: Plumb debugfs_init through to panels

From: Doug Anderson <dianders@chromium.org>
To: Javier Martinez Canillas <javierm@redhat.com>
Cc: Jani Nikula <jani.nikula@linux.intel.com>,
	Andrzej Hajda <andrzej.hajda@intel.com>,
	Neil Armstrong <narmstrong@baylibre.com>,
	David Airlie <airlied@linux.ie>,
	Robert Foss <robert.foss@linaro.org>,
	dri-devel <dri-devel@lists.freedesktop.org>,
	Jonas Karlman <jonas@kwiboo.se>,
	LKML <linux-kernel@vger.kernel.org>,
	Jernej Skrabec <jernej.skrabec@gmail.com>,
	Thierry Reding <thierry.reding@gmail.com>,
	Laurent Pinchart <Laurent.pinchart@ideasonboard.com>,
	Thomas Zimmermann <tzimmermann@suse.de>,
	lschyi@chromium.org, Sam Ravnborg <sam@ravnborg.org>,
	jjsu@chromium.org
Subject: Re: [PATCH v2 2/3] drm: Plumb debugfs_init through to panels
Date: Tue, 22 Feb 2022 15:47:26 -0800	[thread overview]
Message-ID: <CAD=FV=VRa5DKUpGru9_0sOUXB1y_BZ2rd6uwToRYnx6a_TNRcw@mail.gmail.com> (raw)
In-Reply-To: <ce339fc7-b6bf-b19a-be10-54ee86487b96@redhat.com>

Hi,

On Wed, Feb 16, 2022 at 1:36 AM Javier Martinez Canillas
<javierm@redhat.com> wrote:
>
> On 2/16/22 10:25, Jani Nikula wrote:
>
> [snip]
>
> >>
> >> I actually wrote said follow-up patches and they were ready to go, but
> >> when I was trying to come up with the right "Fixes" tag I found commit
> >> b792e64021ec ("drm: no need to check return value of debugfs_create
> >> functions"). So what's being requested is nearly the opposite of what
> >> Greg did there.
> >>
> >> I thought about perhaps only checking for directories but even that
> >> type of check was removed by Greg's patch. Further checking shows that
> >> start_creating() actually has:
> >>
> >> if (IS_ERR(parent))
> >>   return parent;
> >>
> >> ...so I guess that explains why it's fine to skip the check even for parents?
> >>
> >> Sure enough I confirmed that if I pass `ERR_PTR(-EINVAL)` as the root
> >> for `panel->funcs->debugfs_init()` that nothing bad seems to happen...
> >
> > Yeah, the idea is that you don't need to check for debugfs function
> > return values and you can safely pass error pointers to debugfs
> > functions. The worst that can happen is you don't get the debugfs, but
> > hey, it's debugfs so you shouldn't fail anything else because of that
> > anyway.
> >
>
> Thanks a lot Doug and Jani for the explanations. That makes sense and it
> explains why most code I looked was not checking for the return value.
>
> I guess we should write something about this in the debugfs functions
> kernel doc so it's mentioned explicitly and people don't have to guess.

For anyone interested, I've taken Javier's suggestion and tried to
update the docs:

https://lore.kernel.org/r/20220222154555.1.I26d364db7a007f8995e8f0dac978673bc8e9f5e2@changeid