On Sun, 2 Jan 2011, vichy wrote:
> after studying the source again, I guess when timeout happen, the
> ehci_urb_done is called by end_unlink_async to screen out the previous
> commands.
By way of qh_completions. That's right.
> Why I ask this question is I found something abnormal
> I purposely dump the the token of the qtd in timeout urb under
> qh_completions and it shows:
> (I set a bit when driver fail on wait_for_completion_timeout, and
> check this bit in qh_completions)
>
> urb 84718200 toke 0x80000e00
> ehci -ehci.0: ehci_urb_done 1 urb 84718200 ep0out status -150 len 0/0
> usb 2-1: ifconfig timed out on ep0out len=0/0
Do you know which qtd this is? A control transfer uses 3 qtds.
> The active bit is 0, and if I remember correctly this bit should be
> already set in qh_urb_transaction.
Yes. But it isn't set in the dummy qtd.
> Meanwhile, I also add message as below:
> wmb ();
> dummy->hw_token = token;
> if(token != dummy->hw_token)
> printk("Not identy token=0x%x, dummy->hw_token=0x%x\n",token,
> dummy->hw_token);
> but i didn't see it in the log.
>
> I also grep the source and I didn't find any place to clear this bit.
No, the hardware clears it automatically.
> Did I miss anything or I need to do more experiments?
Maybe. Have you seen this email?
http://marc.info/?l=linux-mips&m=129397158114595&w=2
People have been using USB on MIPS for a long time. Why do you think
you're seeing problems now?
Alan Stern
--
To unsubscribe from this list: send the line "unsubscribe linux-usb" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
