ASoC: SOF: Fix deadlock when shutdown a frozen userspace #4072
Conversation
If system shutdown has not been completed cleanly, it is possible the DMA stream shutdown has not been done, or was not clean. If this is the case, Intel TGL/ADL HDA platforms may fail to shutdown cleanly due to pending HDA DMA transactions. To avoid this, detect this scenario in the shutdown callback, and perform an additional controller reset. This has been tested to unblock S5 entry if this condition is hit. Co-developed-by: Archana Patni <archana.patni@intel.com> Signed-off-by: Archana Patni <archana.patni@intel.com> Signed-off-by: Kai Vehmanen <kai.vehmanen@linux.intel.com>
…shutdown" The unregister machine drivers call is not safe to do when kexec is used. Kexec-lite gets blocked with following backtrace: [ 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 This reverts commit 83bfc7e. Reported-by: Ricardo Ribalda <ribalda@chromium.org> Cc: Ricardo Ribalda <ribalda@chromium.org> Signed-off-by: Kai Vehmanen <kai.vehmanen@linux.intel.com>
kv2019i
requested review from
libinyang,
bardliao,
lyakh,
plbossart,
dbaluta,
lgirdwood,
ranj063,
ujfalusi,
ajitkupandey,
kuanhsuncheng and
yaochunhung
as code owners
|
@thesofproject/nxp @thesofproject/amd @thesofproject/mediatek please check, this will impact the shutdown sequence. |
|
Alternative would be to do an explicit snd_card_disconnect() like Takashi drafted in |
|
@kv2019i is the revert also necessary with the first patch? |
|
@ranj063 The revert is needed as the machine driver unregister triggers the problem with freeze. But we need the first patch, as otherwise the revert will bring the known issue with S5 entry back. So both are needed. |
| } | ||
|
|
||
| return ret; | ||
| } |
There was a problem hiding this comment.
nitpick: I would rather add the hda_dsp_shutdown_dma_flush() under the hda_dsp_shutdown(), just like if the hda_dsp_shutdown() would been a static function.
|
@kv2019i wrote
Let's wait one more day. I'll send tomorrow upstream. You can comment on alsa-devel as well of course. |
Please go ahead, no comments from us. |
An alternative to https://mailman.alsa-project.org/pipermail/alsa-devel/2022-December/209429.html
This PR reworks the shutdown and removes the unregister calls from SOF core, leaving responsibility for any cleanup actions to DSP-specific shutdown ops.