ath9k_hif_usb_probe() requests firmware asynchronically and
there is some initialization postponed till firmware is ready.
In particular, ath9k_hif_usb_firmware_cb() callback initializes
hif_dev->tx.tx_buf and hif_dev->tx.tx_pending lists.
At the same time, ath9k_hif_usb_suspend() iterates that lists through
ath9k_hif_usb_dealloc_urbs(). If suspend happens before request_firmware_nowait()
callback is called, it can lead to oops.
Similar issue could be in ath9k_hif_usb_disconnect(), but it is prevented
using hif_dev->fw_done completion and HIF_USB_READY flag. The patch extends
this approach to suspend() as well.
Found by Linux Driver Verification project (linuxtesting.org).
Signed-off-by: Alexey Khoroshilov <khoroshilov@ispras.ru>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
struct device *dev = &hif_dev->udev->dev;
struct device *parent = dev->parent;
- complete(&hif_dev->fw_done);
+ complete_all(&hif_dev->fw_done);
if (parent)
device_lock(parent);
release_firmware(fw);
hif_dev->flags |= HIF_USB_READY;
- complete(&hif_dev->fw_done);
+ complete_all(&hif_dev->fw_done);
return;
if (!(hif_dev->flags & HIF_USB_START))
ath9k_htc_suspend(hif_dev->htc_handle);
- ath9k_hif_usb_dealloc_urbs(hif_dev);
+ wait_for_completion(&hif_dev->fw_done);
+
+ if (hif_dev->flags & HIF_USB_READY)
+ ath9k_hif_usb_dealloc_urbs(hif_dev);
return 0;
}