Hi David,

> > ***Mar  6 11:59:27 Webdesktop kernel: PCI: 00:10.3 PCI cache line size
> > set incorrectly (32 bytes) by BIOS/FW.
> > ***Mar  6 11:59:27 Webdesktop kernel: PCI: 00:10.3 cache line size too
> > large - expecting 16.
> 
> Odd -- what kind of CPU does that have?  

An AthlonXP 1600+ (Palomino). Would this really impact on the stability of
the USB 2.0? If that's a possibility, I have an Athlon 1GHz CPU that I
used to run very well with a KT133 board and an NEC USB 2.0 card that I
can test on the same motherboard.

>That seems likely to be some
> kind of misconfiguration.  Should be harmless, since it's not going to
> enable the feature that would then break.

I wouldn't be surprised either, but considering this message was being
displayed almost immediately after ehci-hcd had been loaded, I thought it
was worth mentioning.

> > 00:10.3 is the VT8235's USB 2.0 controller. There were no options in
> > the CMOS to configure the PCI cache line size, so I haven't been able
> > to test other values. Could it be that this issue is causing problems
> > with the EHCI (particularly reading data from the storage device)? 
> 
> I'd more suspect that the errata on your VT8235 aren't the same as
> on mine ... 

I'm not sure if it would help, but here's the exact numbering on my
VT8235:

VT8235
0244CD TAIWAN
23A3486641

>if you haven't been seeing lost IAA statistics when you
> "rmmod ehci-hcd", like "(lost 223)", that'd be my guess as to why
> our results seem different.

After unloading ehci-hcd, there were some messages. After copying 530Mb of
files to the USB 2.0 HDD:
ehci-hcd 00:0a.2 (NEC USB 2.0 Unused): irq normal 0 err 0 reclaim 0 (lost
0)
ehci-hcd 00:10.3 (VIA USB 2.0 Used): irq normal 158604 err 1 reclaim 11330
(lost 102)

I then cold booted the computer and copied 70Mb from the USB 2.0 HDD to
the IDE HDD (more than that usually causes the system to hang):
ehci-hcd 00:0a.2: irq normal 0 err 0 reclaim 0 (lost 0)
ehci-hcd 00:10.3: irq normal 20339 err 1 reclaim 2517 (lost 11)

Does this mean anything?

Thanks,
Jonathan


-------------------------------------------------------
This SF.net email is sponsored by: Etnus, makers of TotalView, The debugger 
for complex code. Debugging C/C++ programs can leave you feeling lost and 
disoriented. TotalView can help you find your way. Available on major UNIX 
and Linux platforms. Try it free. www.etnus.com
_______________________________________________
[EMAIL PROTECTED]
To unsubscribe, use the last form field at:
https://lists.sourceforge.net/lists/listinfo/linux-usb-devel

Reply via email to