On Thu, 25 Mar 2004, David Brownell wrote: > Guennadi Liakhovetski wrote: > > Hello > > > > I am trying to get CardBus running on a PXA-based system. And as a > > test-card I am using a USB 2.0 CardBus card. Sometimes it works. Or, at > > least, inserting a USB device produces an expeced result. But sometimes I > > am getting: > > > > ... > > err klogd: host/usb-ohci.c: OHCI Unrecoverable Error, controller usb-05:00.0 > > disabled > > ... > > > > Any idea where to look? Unfortunately, I have no other CardBus cards at > > the moment, but 16-bit PCMCIA cards work. > > As a rule the UE exceptions imply some kind of DMA error, > like a PCI master or target abort. Once the controller is > disabled, all the other faults you showed are normal. > > That could be caused by hardware interfacing issues, since > it sounds like you may not have other DMA masters working > stably on your system.
On the PCI bus, you mean. Hm, I have a PCI-controller (IT8152) internal OHCI USB 1.1, 2 RealTek chips, tried other network cards, all seem to work normal. "Seem" means, that although I cannot 100% guarantee, network cards are likely to use bus-mastering. But, what indeed can be a problem - DMA on the CardBus bus. That is a good hint. I remember there were some registers there:-) I'll double-check that tomorrow, thanks a lot. > I understand PXA can be sensitive > about how address lines are managed, and that'd be a good > match for that kind of intermittent failure. > > > In terms of software problems, I'd guess you're using some > code on the order of 2.4.25-vrs2, which doesn't have any > particular _known_ ohci issues ... beyond the various races > and quirks that are fixed in the 2.6 "ohci-hcd". Sorry, forgot to say - 2.4.21. Anything particularly bad there? But, as already said, the internal OHCI controller works ok. > If you can't try a 2.6 kernel yet, there's a slightly elderly > backport of "ohci-hcd" in bk://usb-gadget.bkbits.net/gadget-2.4 > (along with pxa2xx_udc for 2.4, possibly interesting) that > might be interesting. If that helps at all, it'd be worth > your making that code match 2.6.latest more closely. Thanks, I'll see how bad it goes. Guennadi --- Guennadi Liakhovetski ------------------------------------------------------- This SF.Net email is sponsored by: IBM Linux Tutorials Free Linux tutorial presented by Daniel Robbins, President and CEO of GenToo technologies. Learn everything from fundamentals to system administration.http://ads.osdn.com/?ad_id=1470&alloc_id=3638&op=click _______________________________________________ [EMAIL PROTECTED] To unsubscribe, use the last form field at: https://lists.sourceforge.net/lists/listinfo/linux-usb-devel