Re: [PATCH, RFC] Bluetooth: btusb, hci_intel: Fix wait_on_bit_timeout() return value checks

On 08/12/2016 02:12 AM, Johan Hedberg wrote:

On Thu, Aug 11, 2016, Bart Van Assche wrote:

quoted

wait_on_bit_timeout() returns one of the following three values:
* 0 to indicate success.
* -EINTR to indicate that a signal has been received;
* -EAGAIN to indicate timeout;
Make the wait_on_bit_timeout() callers check for these values.
This patch has not yet been tested.

Signed-off-by: Bart Van Assche <redacted>
Cc: Marcel Holtmann <marcel@holtmann.org>
Cc: Gustavo Padovan <redacted>
Cc: Johan Hedberg <redacted>
---
 drivers/bluetooth/btusb.c     | 5 ++---
 drivers/bluetooth/hci_intel.c | 6 +++---
 2 files changed, 5 insertions(+), 6 deletions(-)

Good catch. It seems the API was changed not to return 1 by this commit
(which failed to update any users of of it):

  commit 68985633bccb6066bf1803e316fbc6c1f5b796d6
  Author: Peter Zijlstra [off-list ref]
  Date:   Tue Dec 1 14:04:04 2015 +0100

      sched/wait: Fix signal handling in bit wait helpers

Do you have some doubts about your fix? You put "RFC" in the subject
which usually means the author isn't completely sure about it.

Hello Johan,

The reason I put "RFC" in the subject is because I wouldn't like this patch to be accepted upstream without having been tested first. Mentioning "RFC" makes people think twice before applying a patch. That's the only reason.

Bart.