Commit ab949d519601 ("ALSA: hda - Fix deadlock of controller device
lock at unbinding") added a temporary device_unlock()/device_lock()
pair around probe-work cancellation to avoid a deadlock between
controller unbind and codec probe.

That deadlock depended on the driver core taking both a device lock and
its parent lock during bind and unbind. Since commit 8c97a46af04b
("driver core: hold dev's parent lock when needed") and follow-up
fixes, the parent lock is only taken when bus->need_parent_lock is set.
The HDA bus does not set that flag, so codec binding no longer locks
the controller device as the codec's parent.

Keep cancel_delayed_work_sync(), since the async probe/remove race
still needs to be serialized, but drop the stale unlock/relock
workaround and its outdated FIXME comment. Keeping it around only
opens an unnecessary unlocked window inside azx_remove().

Signed-off-by: Cássio Gabriel <cassiogabrielcontato@gmail.com>
---
 sound/hda/controllers/intel.c | 13 -------------
 1 file changed, 13 deletions(-)

diff --git a/sound/hda/controllers/intel.c b/sound/hda/controllers/intel.c
index 8a7bd49e411f..1b22dbf7a719 100644
--- a/sound/hda/controllers/intel.c
+++ b/sound/hda/controllers/intel.c
@@ -2433,20 +2433,7 @@ static void azx_remove(struct pci_dev *pci)
 		/* cancel the pending probing work */
 		chip = card->private_data;
 		hda = container_of(chip, struct hda_intel, chip);
-		/* FIXME: below is an ugly workaround.
-		 * Both device_release_driver() and driver_probe_device()
-		 * take *both* the device's and its parent's lock before
-		 * calling the remove() and probe() callbacks.  The codec
-		 * probe takes the locks of both the codec itself and its
-		 * parent, i.e. the PCI controller dev.  Meanwhile, when
-		 * the PCI controller is unbound, it takes its lock, too
-		 * ==> ouch, a deadlock!
-		 * As a workaround, we unlock temporarily here the controller
-		 * device during cancel_work_sync() call.
-		 */
-		device_unlock(&pci->dev);
 		cancel_delayed_work_sync(&hda->probe_work);
-		device_lock(&pci->dev);
 
 		clear_bit(chip->dev_index, probed_devs);
 		pci_set_drvdata(pci, NULL);

---
base-commit: 39b68f430a1b299a041ce44a695ae9ddb2576ca1
change-id: 20260328-hda-intel-drop-obsolete-probe-workaround-d7d2a5528caf

Best regards,
--  
Cássio Gabriel <cassiogabrielcontato@gmail.com>

On Sun, 29 Mar 2026 00:42:01 +0100,
Cássio Gabriel wrote:
> 
> Commit ab949d519601 ("ALSA: hda - Fix deadlock of controller device
> lock at unbinding") added a temporary device_unlock()/device_lock()
> pair around probe-work cancellation to avoid a deadlock between
> controller unbind and codec probe.
> 
> That deadlock depended on the driver core taking both a device lock and
> its parent lock during bind and unbind. Since commit 8c97a46af04b
> ("driver core: hold dev's parent lock when needed") and follow-up
> fixes, the parent lock is only taken when bus->need_parent_lock is set.
> The HDA bus does not set that flag, so codec binding no longer locks
> the controller device as the codec's parent.
> 
> Keep cancel_delayed_work_sync(), since the async probe/remove race
> still needs to be serialized, but drop the stale unlock/relock
> workaround and its outdated FIXME comment. Keeping it around only
> opens an unnecessary unlocked window inside azx_remove().
> 
> Signed-off-by: Cássio Gabriel <cassiogabrielcontato@gmail.com>

A good catch.  Applied to for-next branch now.


thanks,

Takashi