[PATCH] drm/i915/audio: 'scheduling while atomic' during system resume

Jani Nikula jani.nikula at linux.intel.com
Tue Apr 9 10:30:20 UTC 2024


On Tue, 09 Apr 2024, Brent Lu <brent.lu at intel.com> wrote:
> The i915_audio_component_codec_wake_override() function will be called
> during resume process so replace usleep_range() function calls with
> udelay() to avoid the 'scheduling while atomic' error.

That explanation does not hold water.

For example, what about the mutexes on the same paths?


BR,
Jani.


>
> [22138.069052] BUG: scheduling while atomic: cras/2682/0x00000000
> [22138.069359] Call Trace:
> [22138.069369]  <TASK>
> [22138.069380]  dump_stack_lvl+0x69/0x97
> [22138.069407]  __schedule_bug+0x87/0x9a
> [22138.069422]  __schedule+0x4f9/0xf88
> [22138.069450]  schedule+0x4e/0xd0
> [22138.069462]  schedule_hrtimeout_range_clock+0xab/0x11b
> [22138.069488]  usleep_range_state+0x63/0x8c
> [22138.069502]  i915_audio_component_codec_wake_override+0x8e/0xf2
> [22138.069518]  snd_hdac_set_codec_wakeup+0x39/0x44 [snd_hda_core (HASH:381b 97)]
> [22138.069541]  hda_dsp_ctrl_init_chip+0x45/0x414 [snd_sof_intel_hda_common (HASH:ff4d 98)]
> [22138.069578]  hda_resume+0x4c/0xf2 [snd_sof_intel_hda_common (HASH:ff4d 98)]
> [22138.069598]  hda_dsp_runtime_resume+0x32/0x9f [snd_sof_intel_hda_common (HASH:ff4d 98)]
> [22138.069618]  sof_resume+0x67/0x234 [snd_sof (HASH:ad31 99)]
> [22138.069655]  __rpm_callback+0x87/0x130
> [22138.069680]  rpm_callback+0x22/0x77
> [22138.069691]  rpm_resume+0x3bb/0x508
>
> Signed-off-by: Brent Lu <brent.lu at intel.com>
> ---
>  drivers/gpu/drm/i915/display/intel_audio.c | 4 ++--
>  1 file changed, 2 insertions(+), 2 deletions(-)
>
> diff --git a/drivers/gpu/drm/i915/display/intel_audio.c b/drivers/gpu/drm/i915/display/intel_audio.c
> index 07e0c73204f3..9a85e6da63bf 100644
> --- a/drivers/gpu/drm/i915/display/intel_audio.c
> +++ b/drivers/gpu/drm/i915/display/intel_audio.c
> @@ -1120,12 +1120,12 @@ static void i915_audio_component_codec_wake_override(struct device *kdev,
>  	 */
>  	intel_de_rmw(i915, HSW_AUD_CHICKENBIT,
>  		     SKL_AUD_CODEC_WAKE_SIGNAL, 0);
> -	usleep_range(1000, 1500);
> +	udelay(1000);
>  
>  	if (enable) {
>  		intel_de_rmw(i915, HSW_AUD_CHICKENBIT,
>  			     0, SKL_AUD_CODEC_WAKE_SIGNAL);
> -		usleep_range(1000, 1500);
> +		udelay(1000);
>  	}
>  
>  	i915_audio_component_put_power(kdev, cookie);

-- 
Jani Nikula, Intel


More information about the Intel-gfx mailing list