Bugzilla – Bug 139317
USB device not accepted ("device not accepting address") until USB module is reloaded
Last modified: 2006-04-12 16:29:31 UTC
(This problem had been observed on a different hardware with the DVD-release kernel as well): An USB device (DVD burner via USB 2.0 in this case, a USB memory stick via USB 1 on the other case) is not working after being connected, even though the same device was working with that hard and software before. Kernel messages say: "device not accepting address". Unplugging and replugging the device doesn't help, but unloading the USB driver ("rmmod ehci_hcd" in this case) does help. The device is accepted then. See syslog messages that will be attached in a moment.
Created attachment 60936 [details] Extract of /var/log/messages (host and user names changed for privacy/security) The USB DVD burner was not accepted until the USB module was unloaded (the module did reload automatically).
Greg, any idea?
If you unload the ehci-hcd driver, it reconnects to the device using the uhci-hcd driver, which will run at 1.1 speeds (not very fast.) Odd that your device isn't accepting the connection. Does this device work properly in a different machine? Does this machine accept any USB 2.0 devices connected to it? Also, can you attach the output of the 'hwinfo' program?
OK, some hardware info: Computer is a HP Vectra VL with on-board USB 1. The add-on card for USB 2 is a "Digitus Connect USB 2.0 PCI Add-On Card" using a popular VIA chip (VIA 6212L I think).
Created attachment 61740 [details] Output of "hwinfo" (after rmmod ehci_hcd, ...; modprobe ehci_hcd)
Ugh, not a VIA chip :( Those are known to have issues with Linux, and as VIA will not work with the Linux developers to workaround their bugs, there's not much we can do at all here :( One last thing, can you attach 'hwinfo' first thing after booting and without messing with the 'rmmod ehci-hcd' stuff? That will give me a clearer picture.
Created attachment 62521 [details] Output of hwinfo after a reboot with the DVD writer attached Boot was yesterday, device attached today. I hope that's OK.
Created attachment 62522 [details] Output of lspci -vv Just in case you want to have the "lspci -vv" output as well.
Still happen with Beta3 of the 10.1 release?
I got notified about http://bugzilla.kernel.org/show_bug.cgi?id=5777 which seems to deal with the same bug. I had no chance to test another release on that hardware, because i'm quite tight on diskspace here.
Can you just try the latest kernel of the day instead? That should be all that is needed to verify this.
Is there a "Kernel of the day HOWTO"? I need more details, please.
Not really, just download the rpm from the ftp site. You can try the Beta6 rpm which is now public if you want to.
Am marking closed due to inactivity. If you do test this on the latest kernel and still have the problem, please reopen the bug.