On Tue, 22 Jul 2003, Johannes Erdfelt wrote:
> On Tue, Jul 22, 2003, Alan Stern <[EMAIL PROTECTED]> wrote:
>
> > It sets the SPD flag for all queued IN packets that don't also
> > have IOC set. We need to know as soon as possible if a non-iso.
> > transfer gets a short read.
>
> I don't remember discussing this and I don't think it's a good idea.
>
> What problem are you trying to fix here?
>
> Why should we set SPD when the caller specifically tells us not to?
We did talk about this previously, but only very briefly and it was
about a month ago.
There are several items this change addresses:
Somewhat surprisingly, the UHCI Design Guide doesn't say anywhere
what the controller will do when it encounters a short IN packet, at least
not in the general case. All it does say is in section 3.2.2, under the
entry for bit 29 (Short Packet Detect). There it specifies what will
happen when the SPD bit is set and a short packet arrives -- but it says
nothing about what will happen if SPD _isn't_ set. Rather than rely on
manufacturers to Do The Right Thing, it seems safer simply to set the
bit.
The Guide also mentions that the behavior is _undefined_ when SPD
is set for OUT packets. The current code will do that if the user passes
in the wrong flag setting.
The specified behavior when SPD is set is exactly what we want to
happen: the TD is marked inactive, the QH is not updated, and the
interrupt pending bit gets set. Maybe if SPD isn't set the first two
would still occur; almost certainly there wouldn't be an interrupt
request. But we _want_ that interrupt, because a short read terminates a
transfer and we have to give back the URB.
The setting of SPD is not directly related to the URB_SHORT_NOT_OK
flag. All that flag means is that a short read transfer should return
with status -EREMOTEIO rather than status 0. This is already tested for
and handled properly in uhci_result_control() and uhci_result_common().
The flag doesn't mean that SPD should or should not be set.
Taken together, these arguments show that SPD should always be set for
every queued (i.e., not isochronous) input TD that's not the last one in
its URB.
Alan Stern
-------------------------------------------------------
This SF.net email is sponsored by: VM Ware
With VMware you can run multiple operating systems on a single machine.
WITHOUT REBOOTING! Mix Linux / Windows / Novell virtual machines at the
same time. Free trial click here: http://www.vmware.com/wl/offer/345/0
_______________________________________________
[EMAIL PROTECTED]
To unsubscribe, use the last form field at:
https://lists.sourceforge.net/lists/listinfo/linux-usb-devel
