* Alan Stern <[EMAIL PROTECTED]>:
I mentioned it on LKML, you probably didn't get it, sorry: nforce2 with ehci, some reported he doesn't have this problem with ohci (but I really don't want to do backups over USB 1.1). I'm not sure if you also got the error forwarded: "host_reset requested but not implemented". My original report is archived here: <http://marc.theaimsgroup.com/?l=linux-usb-users&m=105886063115712&w=2>
That report wasn't very helpful. I recommend reviewing the "How do I report a bug?" FAQ at www.linux-usb.org next time, and for now forward:
- /proc/bus/usb/devices content showing your host controllers and the "problem" device.
- /proc/interrupts showing those host drivers. sometimes this is useful when acpi irq setup didn't break things.
- lspci -vv output for your usb controllers.
- dmesg output including all EHCI messages, with the usb code compiled with CONFIG_USB_DEBUG. The most important bits will be any problems it reports, near when the scsi layer gets unhappy.
- Results of using the same drivers on 2.6.0-test3, which has much better code in the layers over the ehci driver.
If the problem reproduces there, then include the "async", "periodic", and "registers" files for that hc in sysfs (/sys/class/usb_host/something/*) when this problem is appearing. (Likely just snapshot the files every few seconds until the fault happens.)
Does this device work with the OHCI driver? Just rmmod ehci-hcd and the OHCI should take over (on your NForce2).
You might try changing the EHCI module's "log2_irq_thresh" setting. Maybe try 3 (2^3 eights of a millisecond).
There were changes to the EHCI driver between 2.4.21 and 2.4.22;
Called bugfixes ... they resolved every confirmed bug in the EHCI control and bulk code, including several nasty ones!
Of course, there are enough things interacting that fixing one thing can sometimes break another, even in another layer. That may be what you're seeing. I've seen ehci fixes turn up bugs in usbcore, usb-storage, the scsi stack, and even once in sd_mod. I think most of those components changed between those two releases too.
maybe they are responsible for what you're seeing. I don't know if it's possible to use the 2.4.21 version of the EHCI driver in 2.4.22, or vice versa -- but it would make a good test. Does anyone else know the answer?
It works, just copied 2.4.21/drivers/usb/host/* to 2.4.22/drivers/usb/host/ and recompiled the kernel.
All 2.4.22 errors disappeared, so my problem is definitly related to the new host-code in 2.4.22.
Could be ... but I've certainly seen cases where looking at only part of the picture makes a relationship seem to exist, which may not exist or may be caused by a different problem..
Heck, the White House did that several times already this month! :(
Would be great if we could figure the problem out before 2.4.22 goes stable.
Would also have been great to get bug report before we reached 2.4.22-rc1 of course.
- Dave
------------------------------------------------------- This SF.Net email sponsored by: Free pre-built ASP.NET sites including Data Reports, E-commerce, Portals, and Forums are available now. Download today and enter to win an XBOX or Visual Studio .NET. http://aspnet.click-url.com/go/psa00100003ave/direct;at.aspnet_072303_01/01 _______________________________________________ [EMAIL PROTECTED] To unsubscribe, use the last form field at: https://lists.sourceforge.net/lists/listinfo/linux-usb-devel
