I want to add a few comments here that should help the HCD developers....
On Fri, 31 Mar 2000, Geoffrey Hausheer wrote:
> I've been working with Matt Dharm on what I initially assumed was a
> usb-storage problem.� However, I am seeing different results with
> usb-uhci and uhci, which indicates there is likely an HCD problem.�
To be clear on the point: There _may_ be a problem in usb-storage, but at
this point I'm not seeing one. I am seeing behavior which seems
attributable only to an HCD problem.
> The problem is that I cannot re-load the usb-storage module.� The
> first time it (usually) works fine, but after unloading it, I cannot
> re-load it. This is using the usb-uhci module.
Geoff's initial report indicated that the unload/reload problem only
occured about 20% of the time. Geoff, are you now saying it occurs all
the time?
As the logs Geoff send show, it would appear that the device is NAKing. I
doubt that this is really what's happening -- it looks like another place
where the data-toggle isn't being reset properly, or perhaps is being
reset when it shouldn't be. At least, this has been the cause of
"continuous NAKing" behavior like this in the past. Just a thought as to
a good place to start looking.
>�When I try the uhci module, the
> usb-storage driver won't load successfully at all.� The drive does seem
> to get detected successfully by the usb-uhci/uhci modules (as indicated
> by /proc/bus/usb/devices)
Note that is problem appears to be slightly different. At one point, we
get a "Bulk logical error" from usb-storage, which happens when the magic
number in the status stage (also known as the signature) is incorrect. If
you read through his logs, you'll notice that on successful commands, the
signature is 0x53425355. However, at one point when using uhci.o, we get
a signature back of 0xc773493c -- and the rest of the data in the status
looks pretty bogus too. As an interesting note, the command itself seems
to have functioned just fine. The first command where this happens is an
INQUIRY, and it successfully (or so it seems) transferred 96 bytes of
data, just like the times when INQUIRY worked.
> When usb-storage fails to load (with either module) I get messages like:
> scsi0: usb scsi0
> scsi: 1 host.
> usb_control/bulk_msg: timeout
> usb_storage: abort() requested but not implemented
> usb_storage: bus_request() requested but not implemented
When he says 'fails to load', he really means 'fails to work'. The
module is loaded, but it's not communicating properly with the device.
That should be evident from the logs.
Matt Dharm
--
Matthew Dharm Home: [EMAIL PROTECTED]
Engineer, Qualcomm, Inc. Work: [EMAIL PROTECTED]
What, are you one of those Microsoft-bashing Linux freaks?
-- Customer to Greg
User Friendly, 2/10/1999
---------------------------------------------------------------------
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]
