Bluetooth: hci_qca: Fix BT not getting powered-off on rmmod

The BT core skips calling the hci_dev's shutdown method when the HCI
is unregistered. This means that qca_power_off() was not getting called
leaving BT powered on.

This causes regulators / pwrseq providers to not get disabled which also
causes problem when re-loading the module because regulators and pwrseq
providers have an enablecount which now has never dropped to 0, causing
the BT to not get properly reset between rmmod and re-load which causes
initialization failure on the re-load.

Fix this by calling qca_power_off() from qca_close() when BT has not
already been powered off through a qca_hci_shutdown() call.

hci_ldisc.c will call qca_close() after freeing the hdev, so this
means that qca_power_off() can now no longer deref hu->hdev, change
the logging in qca_power_off() to no longer use hu->hdev.

Signed-off-by: Hans de Goede <johannes.goede@oss.qualcomm.com>
Reviewed-by: Bartosz Golaszewski <bartosz.golaszewski@oss.qualcomm.com>
Signed-off-by: Luiz Augusto von Dentz <luiz.von.dentz@intel.com>
This commit is contained in:
Hans de Goede
2026-02-12 15:17:21 +01:00
committed by Luiz Augusto von Dentz
parent 7e2e1e5859
commit 9ff5ff0b91

View File

@@ -722,6 +722,10 @@ static int qca_close(struct hci_uart *hu)
BT_DBG("hu %p qca close", hu);
/* BT core skips qca_hci_shutdown() which calls qca_power_off() on rmmod */
if (!test_bit(QCA_BT_OFF, &qca->flags))
qca_power_off(hu);
serial_clock_vote(HCI_IBS_VOTE_STATS_UPDATE, hu);
skb_queue_purge(&qca->tx_wait_q);
@@ -2260,7 +2264,7 @@ static void qca_power_off(struct hci_uart *hu)
qca_regulator_disable(qcadev);
if (qcadev->sw_ctrl) {
sw_ctrl_state = gpiod_get_value_cansleep(qcadev->sw_ctrl);
bt_dev_dbg(hu->hdev, "SW_CTRL is %d", sw_ctrl_state);
BT_DBG("SW_CTRL is %d", sw_ctrl_state);
}
break;