diff mbox series

[v7,2/2] ASoC: SOF: Fix deadlock when shutdown a frozen userspace

Message ID 20221127-snd-freeze-v7-2-127c582f1ca4@chromium.org
State New
Headers
Series ASoC: SOF: Fix deadlock when shutdown a frozen userspace |

Commit Message

Ricardo Ribalda Nov. 30, 2022, 4:29 p.m. UTC 
  During kexec(), the userspace might frozen. Therefore we cannot wait
for it to complete.

During a kexec with frozen processe do not unregister the clients.

This fixes:

[   84.943749] Freezing user space processes ... (elapsed 0.111 seconds) done.
[  246.784446] INFO: task kexec-lite:5123 blocked for more than 122 seconds.
[  246.819035] Call Trace:
[  246.821782]  <TASK>
[  246.824186]  __schedule+0x5f9/0x1263
[  246.828231]  schedule+0x87/0xc5
[  246.831779]  snd_card_disconnect_sync+0xb5/0x127
...
[  246.889249]  snd_sof_device_shutdown+0xb4/0x150
[  246.899317]  pci_device_shutdown+0x37/0x61
[  246.903990]  device_shutdown+0x14c/0x1d6
[  246.908391]  kernel_kexec+0x45/0xb9

And:

[  246.893222] INFO: task kexec-lite:4891 blocked for more than 122 seconds.
[  246.927709] Call Trace:
[  246.930461]  <TASK>
[  246.932819]  __schedule+0x5f9/0x1263
[  246.936855]  ? fsnotify_grab_connector+0x5c/0x70
[  246.942045]  schedule+0x87/0xc5
[  246.945567]  schedule_timeout+0x49/0xf3
[  246.949877]  wait_for_completion+0x86/0xe8
[  246.954463]  snd_card_free+0x68/0x89
...
[  247.001080]  platform_device_unregister+0x12/0x35

Cc: stable@vger.kernel.org
Fixes: 83bfc7e793b5 ("ASoC: SOF: core: unregister clients and machine drivers in .shutdown")
Signed-off-by: Ricardo Ribalda <ribalda@chromium.org>
---
 sound/soc/sof/core.c | 4 +++-
 1 file changed, 3 insertions(+), 1 deletion(-)

Comments

Kai Vehmanen Dec. 1, 2022, 9:03 a.m. UTC | #1 
Hi,

On Wed, 30 Nov 2022, Ricardo Ribalda wrote:

> During kexec(), the userspace might frozen. Therefore we cannot wait
> for it to complete.
[...]
> --- a/sound/soc/sof/core.c
> +++ b/sound/soc/sof/core.c
> @@ -9,6 +9,7 @@
>  //
>  
>  #include <linux/firmware.h>
> +#include <linux/kexec.h>
>  #include <linux/module.h>
>  #include <sound/soc.h>
>  #include <sound/sof.h>
> @@ -484,7 +485,8 @@ int snd_sof_device_shutdown(struct device *dev)
>  	 * make sure clients and machine driver(s) are unregistered to force
>  	 * all userspace devices to be closed prior to the DSP shutdown sequence
>  	 */
> -	sof_unregister_clients(sdev);
> +	if (!kexec_with_frozen_processes())
> +		sof_unregister_clients(sdev);
>  
>  	snd_sof_machine_unregister(sdev, pdata);

I think the case you hit was specifically snd_card_disconnect_sync() that 
gets called via snd_sof_machine_unregister(), right, so you'd have to skip 
both sof_unregister_clients() and the machine_unregister().

Skipping ok might be an ok solution here. There's clearly a problem and we 
cannot just drop these calls in the general case (when we are going to 
S5), but in the specific case of kexec, this is probably safe. And I agree 
one way or another this needs to be fixed. Pierre and others what do you 
think?

Br, Kai
diff mbox series

Patch

diff --git a/sound/soc/sof/core.c b/sound/soc/sof/core.c
index 3e6141d03770..4301f347bb90 100644
--- a/sound/soc/sof/core.c
+++ b/sound/soc/sof/core.c
@@ -9,6 +9,7 @@ 
 //
 
 #include <linux/firmware.h>
+#include <linux/kexec.h>
 #include <linux/module.h>
 #include <sound/soc.h>
 #include <sound/sof.h>
@@ -484,7 +485,8 @@  int snd_sof_device_shutdown(struct device *dev)
 	 * make sure clients and machine driver(s) are unregistered to force
 	 * all userspace devices to be closed prior to the DSP shutdown sequence
 	 */
-	sof_unregister_clients(sdev);
+	if (!kexec_with_frozen_processes())
+		sof_unregister_clients(sdev);
 
 	snd_sof_machine_unregister(sdev, pdata);