- Some boots it finds the root hub but nothing beyond it
Is khubd alive when this happens? All the hcds suffer when it goes away, and it seems like sometimes it will exit. (Seems like it receives a signal, which it treats as a reason to exit.)
- If you reload it then it does this Oct 29 11:49:18 irongate kernel: drivers/usb/host/ohci-pci.c: 00:07.4: AMD756 erratum 4 workaround Oct 29 11:49:18 irongate kernel: drivers/usb/core/hcd.c: usb_register_root_hub_Rsmp_121543ac - usb_new_device failed with value -17 Oct 29 11:49:18 irongate kernel: drivers/usb/host/ohci-pci.c: can't start 00:07.4 Oct 29 11:49:18 irongate kernel: drivers/usb/core/hcd-pci.c: remove: 00:07.4, state 0 Oct 29 11:49:18 irongate kernel: drivers/usb/core/hcd.c: USB bus 1 deregistered
Haven't seen this one, but it looks to be because of some driver model state that didn't get cleaned up (which I certainly have seen). The failure from root hub registration, -EEXIST, would seem to be reported by device_register() from usb_new_device().
- Other times it does this
Oct 29 11:56:13 irongate kernel: drivers/usb/host/ohci-q.c: 00:07.4 bad entry f6f6341
At which point touching the USB mouse results in an oops in ohci_irq called from usb_hcd_irq and dies with no useful looking info except that the eax
register holds the pattern A7A7A7A7
There's a patch I sent around last Friday that's gotten some initial positive results, message title "OHCI problems in latest 2.5 kernels", that simplifies some of the logic that could cause this, and which one person reported made things work better ... I'll send it separately. That's the only one of these instability problem I'd lay at the hands of ohci. - Dave ------------------------------------------------------- This sf.net email is sponsored by:ThinkGeek Welcome to geek heaven. http://thinkgeek.com/sf _______________________________________________ [EMAIL PROTECTED] To unsubscribe, use the last form field at: https://lists.sourceforge.net/lists/listinfo/linux-usb-devel
