linux-kernel - Re: [PATCH v2 10/17] drm/vram-helper: make drm_vram_mm_debugfs

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Message-ID: <CAKMK7uGbg5Lax+eXJda4k9LNd7JBb+LRtRw4S+bZ4GbNGT--ZA@mail.gmail.com>
Date:   Wed, 18 Mar 2020 20:10:43 +0100
From:   Daniel Vetter <daniel@...ll.ch>
To:     Greg KH <gregkh@...uxfoundation.org>
Cc:     Wambui Karuga <wambui.karugax@...il.com>,
        Dave Airlie <airlied@...ux.ie>,
        Maarten Lankhorst <maarten.lankhorst@...ux.intel.com>,
        Maxime Ripard <mripard@...nel.org>,
        Thomas Zimmermann <tzimmermann@...e.de>,
        dri-devel <dri-devel@...ts.freedesktop.org>,
        Linux Kernel Mailing List <linux-kernel@...r.kernel.org>
Subject: Re: [PATCH v2 10/17] drm/vram-helper: make drm_vram_mm_debugfs_init()
 return 0

On Wed, Mar 18, 2020 at 5:58 PM Greg KH <gregkh@...uxfoundation.org> wrote:
>
> On Wed, Mar 18, 2020 at 05:31:47PM +0100, Daniel Vetter wrote:
> > On Wed, Mar 18, 2020 at 5:03 PM Wambui Karuga <wambui.karugax@...il.com> wrote:
> > >
> > >
> > >
> > > On Wed, 18 Mar 2020, Daniel Vetter wrote:
> > >
> > > > On Tue, Mar 10, 2020 at 04:31:14PM +0300, Wambui Karuga wrote:
> > > >> Since 987d65d01356 (drm: debugfs: make
> > > >> drm_debugfs_create_files() never fail), drm_debugfs_create_files() never
> > > >> fails and should return void. Therefore, remove its use as the
> > > >> return value of drm_vram_mm_debugfs_init(), and have the function
> > > >> return 0 directly.
> > > >>
> > > >> v2: have drm_vram_mm_debugfs_init() return 0 instead of void to avoid
> > > >> introducing build issues and build breakage.
> > > >>
> > > >> References: https://lists.freedesktop.org/archives/dri-devel/2020-February/257183.html
> > > >> Signed-off-by: Wambui Karuga <wambui.karugax@...il.com>
> > > >> Acked-by: Thomas Zimmermann <tzimmermann@...e.de>
> > > >> ---
> > > >>  drivers/gpu/drm/drm_gem_vram_helper.c | 10 ++++------
> > > >>  1 file changed, 4 insertions(+), 6 deletions(-)
> > > >>
> > > >> diff --git a/drivers/gpu/drm/drm_gem_vram_helper.c b/drivers/gpu/drm/drm_gem_vram_helper.c
> > > >> index 92a11bb42365..c8bcc8609650 100644
> > > >> --- a/drivers/gpu/drm/drm_gem_vram_helper.c
> > > >> +++ b/drivers/gpu/drm/drm_gem_vram_helper.c
> > > >> @@ -1048,14 +1048,12 @@ static const struct drm_info_list drm_vram_mm_debugfs_list[] = {
> > > >>   */
> > > >>  int drm_vram_mm_debugfs_init(struct drm_minor *minor)
> > > >>  {
> > > >> -    int ret = 0;
> > > >> -
> > > >>  #if defined(CONFIG_DEBUG_FS)
> > > >
> > > > Just noticed that this #if here is not needed, we already have a dummy
> > > > function for that case. Care to write a quick patch to remove it? On top
> > > > of this patch series here ofc, I'm in the processing of merging the entire
> > > > pile.
> > > >
> > > > Thanks, Daniel
> > > Hi Daniel,
> > > Without this check here, and compiling without CONFIG_DEBUG_FS, this
> > > function is run and the drm_debugfs_create_files() does not have access to
> > > the parameters also protected by an #if above this function. So the change
> > > throws an error for me. Is that correct?
> >
> > Hm right. Other drivers don't #ifdef out their debugfs file functions
> > ... kinda a bit disappointing that we can't do this in the neatest way
> > possible.
> >
> > Greg, has anyone ever suggested to convert the debugfs_create_file
> > function (and similar things) to macros that don't use any of the
> > arguments, and then also annotating all the static functions/tables as
> > __maybe_unused and let the compiler garbage collect everything?
> > Instead of explicit #ifdef in all the drivers ...
>
> No, no one has suggested that, having the functions be static inline
> should make it all "just work" properly if debugfs is not enabled.  The
> variables will not be used, so the compiler should just optimize them
> away properly.
>
> No checks for CONFIG_DEBUG_FS should be needed anywhere in .c code.

So the trouble with this one is that the static inline functions for
the debugfs file are wrapped in a #if too, and hence if we drop the
#if around the function call stuff won't compile. Should we drop all
the #if in the .c file and assume the compiler will remove all the
dead code and dead functions?
-Daniel (who has no idea how this all works really)
-- 
Daniel Vetter
Software Engineer, Intel Corporation
+41 (0) 79 365 57 48 - http://blog.ffwll.ch