Hi Johan,
On Thu, Mar 7, 2013 at 4:15 AM, Johan Hedberg <johan.hedberg@xxxxxxxxx> wrote:
> Hi Andre,
>
> On Wed, Mar 06, 2013, Andre Guedes wrote:
>> @@ -1042,6 +1042,9 @@ int hci_unregister_cb(struct hci_cb *hcb);
>> struct hci_request {
>> struct hci_dev *hdev;
>> struct sk_buff_head cmd_q;
>> +
>> + /* This flag is set if something goes wrong during request creation */
>> + bool error;
>> };
>
> I'd prefer if you'd use "int err". That'd allow you to record the exact
> error from when it occurred and have hci_req_run return it.
I'll change this in v2.
>> int hci_req_run(struct hci_request *req, hci_req_complete_t complete)
>> @@ -2460,6 +2461,15 @@ int hci_req_run(struct hci_request *req, hci_req_complete_t complete)
>> if (skb_queue_empty(&req->cmd_q))
>> return -ENODATA;
>>
>> + /*
>> + * If error flag is set, remove all HCI commands queued on the HCI
>> + * request queue.
>> + */
>> + if (req->error) {
>> + skb_queue_purge(&req->cmd_q);
>> + return -EIO;
>> + }
>
> I think this check belongs before the queue_empty check. If the first
> command failed to be added we should properly have hci_req_run fail too.
> If you do it your way e.g. hci_req_sync would just succeed even though
> there was a memory allocation error.
Fair enough. I'll fix it in v2.
Regards,
Andre
--
To unsubscribe from this list: send the line "unsubscribe linux-bluetooth" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
