[Intel-gfx] Possible regression in 3.17-rc2 in i915 driver

[Jani Nikula]

+intel-gfx

Ville, Daniel, any thoughts before we queue a revert?

BR,
Jani.


On Sun, 31 Aug 2014, Tibor Billes <tbilles at gmx.com> wrote:
> Hi!
>
> I tried to upgrade my kernel from 3.16 to 3.17-rc2 and I found that my laptop
> was unable to boot. The boot process hangs after 2-3 seconds (according to
> timestamps of log messages). The last kernel log line I usually see is the
> discovery of my touchpad: "input: SynPS/2 Synaptics TouchPad as
> /devices/platform/i8042/serio2/input/input11".
>
> I did a git bisect and it pointed me to the following commit:
> commit 208bf9fdcd3575aa4a5d48b3e0295f7cdaf6fc44
> Author: Ville Syrjälä <ville.syrjala at linux.intel.com>
> Date:   Mon Aug 11 13:15:35 2014 +0300
>  
>     drm/i915: Fix locking for intel_enable_pipe_a()
>  
>     intel_enable_pipe_a() gets called with all the modeset locks already
>     held (by drm_modeset_lock_all()), so trying to grab the same
>     locks using another drm_modeset_acquire_ctx is going to fail miserably.
>      
>     Move most of the drm_modeset_acquire_ctx handling (init/drop/fini)
>     out from intel_{get,release}_load_detect_pipe() into the callers
>     (intel_{crt,tv}_detect()). Only the actual locking and backoff
>     handling is left in intel_get_load_detect_pipe(). And in
>     intel_enable_pipe_a() we just share the mode_config.acquire_ctx from
>     drm_modeset_lock_all() which is already holding all the relevant locks.
>      
>     It's perfectly legal to lock the same ww_mutex multiple times using the
>     same ww_acquire_ctx. drm_modeset_lock() will convert the returned
>     -EALREADY into 0, so the caller doesn't need to do antyhing special.
>      
>     Fixes a hang on resume on my 830.
>      
>     Signed-off-by: Ville Syrjälä <ville.syrjala at linux.intel.com>
>     Reviewed-by: Daniel Vetter <daniel.vetter at ffwll.ch>
>     Cc: stable at vger.kernel.org
>     Signed-off-by: Jani Nikula <jani.nikula at intel.com>
>  
> I tried booting with the above commit reverted on top of 3.17-rc2 and it
> booted successfully.
>  
> I also did a MagicSyrq-W (Display list of blocked (D state) tasks) after the
> boot process stopped (using plain 3.17-rc2, without reverting the above commit)
> and it showed that plymouthd was blocked with the following call trace (hand
> copied):
> irq_exit
> common_interrupt
> schedule_preemt_disabled
> __mutex_lock_slowpath
> mutex_lock
> drm_modeset_lock
> drm_getconnector
> drm_mode_getcrttc
> drm_ioctl
> ...
>  
> This may have something to do with the hang or it may not, I don't know but is
> related to drm locking so I thought it was a good idea to mention it.
>  
> My laptop is an old Fujitsu-Siemens Amilo M1450g, running Linux Mint 16.
>  
> Let me know if I can help further debug this issue.
>  
> Tibor Billes

-- 
Jani Nikula, Intel Open Source Technology Center