linux-kernel - Re: [PATCH 14/15] drm/amd: cleanup: drm_modeset_lock_all() --> DRM_MODESET_LOCK_ALL

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]

Date:   Sat, 18 Sep 2021 11:42:40 +0200
From:   Fernando Ramos <greenfoo@....eu>
To:     Sean Paul <sean@...rly.run>
Cc:     dri-devel@...ts.freedesktop.org, linux-kernel@...r.kernel.org,
        linux-doc@...r.kernel.org, amd-gfx@...ts.freedesktop.org,
        intel-gfx@...ts.freedesktop.org, linux-arm-msm@...r.kernel.org,
        freedreno@...ts.freedesktop.org, nouveau@...ts.freedesktop.org,
        linux-renesas-soc@...r.kernel.org, linux-tegra@...r.kernel.org
Subject: Re: [PATCH 14/15] drm/amd: cleanup: drm_modeset_lock_all() -->
 DRM_MODESET_LOCK_ALL_BEGIN()

On 21/09/18 01:17AM, Fernando Ramos wrote:

> > > +#include <drm/drm_drv.h>
> > 
> > Top-level headers generally come above the driver headers. Also, now that I think
> > about this a bit more, all of the new includes in this set should probably be
> > for 'drm_modeset_lock.h' instead of 'drm_drv.h'.
> 
> Ok. Let me try that.

Turns out that the DRM_MODESET_LOCK_ALL_*() macros expansion includes a call
to drm_drv_uses_atomic_modeset() which is defined in "drm_drv.h".

Thus, #include'ing <drm/drm_drv.h> cannot be avoided.

This makes me wonder...

  1. "drm_drv.h" includes "drm_device.h", which includes "drm_mode_config.h",
     which includes "drm_modeset_lock.h"

  2. "drm_modeset_lock.h" defines DRM_MODESET_LOCK_ALL_*() which expands into
     drm_drv_uses_atomic_modeset()

  3. drm_drv_uses_atomic_modeset() is declared in "drm_drv.h"

There seems to be a circular dependency here.

We can try to fix this, but I suggest to do it in a different patch series.