On Wed, 18 Jan 2006, Clay wrote: > Howdy, > > I am a bit of a newbie in regards to Linux, not computers in general. I am > running Debian Sarge with a 2.6.15.1 kernel. My kernel has the same config > as the Debian Sarge 2.6.8-2-386 kernel except for "CONFIG_EFI_PARTITION" > support disabled. > > Here's my issue: I am attempting to transfer my music collection to my iPod > Video via USB 2.0. After a period of time, the device is reset with the > "reset high speed USB device using ehci_hcd" hitting kern.log. This reset > triggers a lot of other failures. The result is that my transfer fails when > the device becomes inaccessible.
Could you provide more details, such as a kernel log? Even better, compile your kernel with CONFIG_USB_DEBUG set and provide the debugging log. Resets like that are a normal part of error recovery. However they aren't supposed to trigger lots of other failures. On the other hand, other people have also noted problems with iPods; they seem to be a little strange (not fully conformant with the USB specifications). > I have been searching for three days for real fix for this and now know that > this is not issue specific to iPod. It seems to be a problem with USB 2.0 > and possibly 1.1 devices. It's a general phenomenon, often caused by noise or timing errors in the USB data signals. Since high-speed (USB 2.0) transmissions are much more sensitive to noise and have closer timing tolerances than full-speed (USB 1.1), they tend to show the problem more. It can be caused by other things too, like buggy firmware in the USB device. > Although it also seems to not be restricted to > USB storage devices, the problem occurs much more often with this type of > device. It would show up with any device that transfers as much data as a storage device. Keep in mind too that storage devices are among the most popular kinds of USB devices. > It also seems to also be experienced by many people on a variety of > hardware running a range of kernel versions. There a lot of postings and > such related to this with some suggesting possible fixes. One such was to > disable EFI support. No, EFI support is completely unconnected with this problem. The EFI issue has to do with the fact that many USB storage devices claim to have one more sector than they actually do, and they crash when the computer tries to read that non-existent "last" sector. Combine that with the fact that EFI stores important information in the last sector of a disk and you have a recipe for trouble. > So, my questions for you all finally are: Has this issue been officially > recognized by the team responsible for USB support in Linux? What team is that? Linux development is carried out almost entirely by volunteers. Nobody is responsible for anything and there are no "teams". Having said this, of course the people who work on USB support for Linux are aware of the problem. We could hardly be unaware of it, given how often it crops up. > Has there been > a real fix released for this issue? If so, where would I find it? The real fix is to improve the hardware. Frequently these problems are caused by bad combinations of USB controller, USB cable, and USB device. For instance, I have a USB disk drive and cable that won't work on one of my computers. Changing any one of the three ingredients fixes things. In short, this does not appear to be a software issue. It would be nice, though, if we could find a software fix. If you have any suggestions, I'll certainly listen to them. Alan Stern ------------------------------------------------------- This SF.net email is sponsored by: Splunk Inc. Do you grep through log files for problems? Stop! Download the new AJAX search engine that makes searching your log files as easy as surfing the web. DOWNLOAD SPLUNK! http://sel.as-us.falkag.net/sel?cmd=lnk&kid=103432&bid=230486&dat=121642 _______________________________________________ Linux-usb-users@lists.sourceforge.net To unsubscribe, use the last form field at: https://lists.sourceforge.net/lists/listinfo/linux-usb-users
