2003-03-29 00:22:34

by Kallol Biswas

[permalink] [raw]
Subject: netchip's net2280 usb 2.0 device

Hi,
We have been using the net2280 chip (usb 2.0) as a usb target
printer device. We have been seeing data corruption problems
during a bulk out transfer when data is taken out the device
quite slow. The corruption size is only 4 bytes suggesting
a device problem. Is anyone using this chip and has anyone
encountered similar problem?

Is there any other 2.0 usb chip that can be used as a target mode
device?


Kallol


2003-03-29 16:09:07

by David Brownell

[permalink] [raw]
Subject: Re: netchip's net2280 usb 2.0 device

Hi,

> We have been using the net2280 chip (usb 2.0) as a usb target
> printer device. We have been seeing data corruption problems
> during a bulk out transfer when data is taken out the device
> quite slow. The corruption size is only 4 bytes suggesting
> a device problem. Is anyone using this chip and has anyone
> encountered similar problem?

I've certainly been using it ... there's a Linux driver for it,
part of a "USB Gadget" framework that could support other such
"target mode" hardware, at http://kernel.bkbits.net/~david-b
for general use. (Download from BK, the 2.5.64 patch doesn't
include the net2280 driver and there's no 2.4 patch yet. I'll
send out an announcement soon, when I update the patches.)

I haven't run into that particular problem, but it sounds
like it might be a known erratum ... have you asked NetChip, or
checked the 14-March errata at their website? They've been
pretty responsive to my questions.


> Is there any other 2.0 usb chip that can be used as a target mode
> device?

The net2280 is the only such chip I know of that talks PCI
directly. So it's particularly Linux-friendly: it doesn't
need special bus adapter hardware.

- Dave


2003-03-31 19:32:33

by David Brownell

[permalink] [raw]
Subject: Re: netchip's net2280 usb 2.0 device

David Brownell wrote:
> Hi,
>
>> We have been using the net2280 chip (usb 2.0) as a usb target
>> printer device. We have been seeing data corruption problems
>> during a bulk out transfer when data is taken out the device
>> quite slow. The corruption size is only 4 bytes suggesting
>> a device problem. Is anyone using this chip and has anyone
>> encountered similar problem?
>
>
> I've certainly been using it ... there's a Linux driver for it,
> part of a "USB Gadget" framework that could support other such
> "target mode" hardware, at http://kernel.bkbits.net/~david-b
> for general use. (Download from BK, the 2.5.64 patch doesn't
> include the net2280 driver and there's no 2.4 patch yet. I'll
> send out an announcement soon, when I update the patches.)

Just done. There are now patches for 2.4 and 2.5 kernels.


> I haven't run into that particular problem, but it sounds
> like it might be a known erratum ... have you asked NetChip, or
> checked the 14-March errata at their website? They've been
> pretty responsive to my questions.

I have in mind particularly erratum 0106 ... which might have
been what was making my TTCP testing from working. Well, at
least the latest code for that is now behaving.

- Dave



>> Is there any other 2.0 usb chip that can be used as a target mode
>> device?
>
>
> The net2280 is the only such chip I know of that talks PCI
> directly. So it's particularly Linux-friendly: it doesn't
> need special bus adapter hardware.
>
> - Dave
>
>