Hi Chris,
I think there are hardware problems with your Cyrix MediaGX
machine. As you wrote no one of the devices work as expected
on this machine with 2.2 and 2.3 kernels, which make me think
so. To be sure you can test 2.3 kernel on your workstation
(which appears to behave fine) or on third machine (if you have
one).
pegasus_rw_srom_word() is called only once during driver
operation and this happens when you try to ifconfig the interface.
This procedure gets HW MAC address of the device, which is
written on the device's SROM and is pretty simple. The problem
is definitely not in it.
best,
Petkan
Chris Worley wrote:
>
> Petko Manolov wrote:
> > Please, try it with some of the latest development
> > kernels.
>
> Before receiving this, I had another one of these Linksys USB100TX's
> still in the box, so I tried it on another computer (my workstation,
> UHCI USB, SuSE 6.3 w/ 2.2.14 un-backported kernel, Becker's drivers,
> since I didn't want to backport my workstation kernel).
>
> This worked. Two warnings mixed in the driver load:
>
> Found recognized USB device 'Linksys USB100Tx'.
> eth1: Linksys USB100Tx at 2 (+0), 00:e0:98:7d:2e:e1.
> eth1: GPIO pin values are 27 00.
> usbcore: ADMtek Pegasus USB-Ethernet driver claimed interface c69411c0
> usb-uhci: Root-Hub stat port1: 95 port2: 80
> eth1: Failed submitting receive URB #1, status -6, pipe 0xc0008280.
> eth1: This USB stack cannot handle multiple receive buffers.
>
> I received Petko's response, and tried a pristine linux-2.3.99-pre6
> kernel. I tried this on the original target machine (ohci USB)... same
> errors as before: "pegasus_rw_srom_word() failed".
>
> So, I'm thinking it's a hardware problem. I swapped the one in my
> workstation for the one in the machine running the 2.3 kernel. Now it
> works on the target (Except, I get three errors at ifconfig from
> pegasus.c: "party [not parity] FAIL ffff" followed by three "eth1: set
> Rx mode"). I swap them back again: it doesn't work. Swap again: it
> works. I pull eth0 cables just to make sure... I really am talking thru
> USB ethernet.
>
> Aha! It must be a hardware problem.
>
> So, I take the one that doesn't work, and put that in my workstation,
> load the 2.2 drivers, now it works!
>
> I check for the phase of the moon... damn! it's cloudy, no clues there!
>
> Either USB100TX device works on my workstation, repeatedly; both at
> about 500K bps (on a 100Mbit network). Only one will work on the target
> machine. In performance testing, the target machine (with the USB100TX
> that works), starts out at about 500K bps, then starts spewing error
> messages:
>
> eth1: RX status -110
>
> (and the driver needs to be reloaded on the target machine before I can
> get it to communicate again).
>
> I switch the target machine back to a 2.2.14 un-backported kernel (still
> SuSE 6.3 distribution) and try Becker's drivers. These drivers work on
> both USB100TX devices! In performance testing, both devices die at
> around 10MBytes just as with the 2.3 kernel, but no errors are reported.
>
> Whether the device works or not, the LED's don't behave as the linksys
> manual says. The link light never comes on, and the tx/rx light
> occasionally blinks green (on a 100BaseT lan).
>
> The markings on the USB100TX devices both say "Rev 3".
>
> The target motherboards are Cyrix MediaGX based OHCI USB's.
>
> In summary:
>
> On the Intel UHCI based hardware running a stock 2.2.14 kernel, Becker's
> drivers work with both USB100TX devices just fine. I have no data point
> for this hardware running a 2.3 kernel.
>
> On the Cyrix MediaGX OHCI based hardware, the stock 2.2.14 kernel works
> somewhat with both USB100TX devices (but dies under stress), but only
> works with one of these (exactly the same) devices when running either
> the backported 2.2.14 or 2.3.99 kernels. The one that detects correctly
> dies under stress. The one that doesn't work dies during ifconfig with
> a "pegasus_rw_srom_word() failed".
>
> Go figure!
>
> Any ideas?
>
> Thanks,
>
> Chris
---------------------------------------------------------------------
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]
