Am Samstag, 9. August 2003 02:04 schrieb David Brownell:
> The code that manges the synchronous control/bulk calls has
> been a mess for ages. This patch rewrites it using:
>
> - "struct completion" instead of a usb-internal clone therof,
> - prepare_to_wait()/finish_wait() instead of the tangled
> mess it now uses (or a new wait_event_timeout call, as in
> previous versions of this patch).
wait_event_timeout still seems ideally suited to the task.
> It's a net code shrink and simplification.
>
> Please review and merge.
>
> - Dave
[..]
// Starts urb and waits for completion or timeout
static int usb_start_wait_urb(struct urb *urb, int timeout, int* actual_length)
{
- DECLARE_WAITQUEUE(wait, current);
- struct usb_api_data awd;
+ DEFINE_WAIT(wait);
+ DECLARE_COMPLETION(done);
int status;
- init_waitqueue_head(&awd.wqh);
- awd.done = 0;
-
- set_current_state(TASK_UNINTERRUPTIBLE);
- add_wait_queue(&awd.wqh, &wait);
+ urb->transfer_flags |= URB_ASYNC_UNLINK;
Why async?
+ urb->context = &done;
- urb->context = &awd;
+ might_sleep();
+ prepare_to_wait(&done.wait, &wait, TASK_UNINTERRUPTIBLE);
status = usb_submit_urb(urb, GFP_ATOMIC);
If you are doing it that way, why do you use a struct completion?
A simple wait queue is sufficient. And doesn't export details of how
struct completion looks internally.
- if (status) {
- // something went wrong
- usb_free_urb(urb);
- set_current_state(TASK_RUNNING);
- remove_wait_queue(&awd.wqh, &wait);
- return status;
- }
-
- while (timeout && !awd.done)
- {
- timeout = schedule_timeout(timeout);
- set_current_state(TASK_UNINTERRUPTIBLE);
- rmb();
- }
+ if (status != 0) {
+ urb->actual_length = 0;
- set_current_state(TASK_RUNNING);
- remove_wait_queue(&awd.wqh, &wait);
+ /* normal case: we woke without a timeout */
+ } else if (schedule_timeout((timeout == 0)
+ ? MAX_SCHEDULE_TIMEOUT
+ : timeout) != 0) {
+ status = urb->status;
- if (!timeout && !awd.done) {
- if (urb->status != -EINPROGRESS) { /* No callback?!! */
- printk(KERN_ERR "usb: raced timeout, "
- "pipe 0x%x status %d time left %d\n",
- urb->pipe, urb->status, timeout);
- status = urb->status;
- } else {
- warn("usb_control/bulk_msg: timeout");
- usb_unlink_urb(urb); // remove urb safely
- status = -ETIMEDOUT;
+ /* abnormal: timed out, so force completion via unlink */
+ } else {
+ status = usb_unlink_urb(urb);
+ switch (status) {
+ case -EINPROGRESS: /* normal */
+ case -EBUSY: /* already completing */
+ dev_err(&urb->dev->dev, "control/bulk timeout\n");
+ break;
+ default:
+ /* shouldn't happen */
+ dev_err(&urb->dev->dev, "c/b unlink err %d\n", status);
+ break;
}
- } else
- status = urb->status;
+ wait_for_completion(&done);
This is a nop unless you timed out. Why not do the unlink synchronously?
+ /* NOTE: HCDs return this status too */
+ status = -ETIMEDOUT;
+ }
+ finish_wait(&done.wait, &wait);
if (actual_length)
*actual_length = urb->actual_length;
-
usb_free_urb(urb);
return status;
}
-------------------------------------------------------
This SF.Net email sponsored by: Free pre-built ASP.NET sites including
Data Reports, E-commerce, Portals, and Forums are available now.
Download today and enter to win an XBOX or Visual Studio .NET.
http://aspnet.click-url.com/go/psa00100003ave/direct;at.aspnet_072303_01/01
_______________________________________________
[EMAIL PROTECTED]
To unsubscribe, use the last form field at:
https://lists.sourceforge.net/lists/listinfo/linux-usb-devel
