I'm using Kernel 2.6.15-1.1975_FC5 udev 084-4 It seems this kernel no longer export 'bus' symlinks in sysfs but we didn't update udev to wait for other files. Result we send the hotplug further up the stack before the files we need are ready. This means hald pukes on my USB stick 15:36:34.027 [E] util.c:463: Cannot open '/sys/devices/pci0000:00/0000:00:1d.1/usb2/2-2/2-2:1.0/host12/target12:0:0/12:0:0:0/model' 15:36:34.028 [E] util.c:463: Cannot open '/sys/devices/pci0000:00/0000:00:1d.1/usb2/2-2/2-2:1.0/host12/target12:0:0/12:0:0:0/vendor' 15:36:34.029 [E] util.c:322: Cannot open '/sys/devices/pci0000:00/0000:00:1d.1/usb2/2-2/2-2:1.0/host12/target12:0:0/12:0:0:0/type' Of course, extra points for fixing the kernel to not send out events before all sysfs files are actually ready. But I'll settle for some updated udev rules instead. I'm marking this is a FC5 Blocker as it breaks hotplugging USB sticks. I'm also marking this as high priority. Thanks.
Adding kernel folks as this is probably due to kernel ABI 'breakage'
Reverting back to udev-084-1.1 made things work for me.
Changing summary to something more appropriate.
I am seeing the same thing on old kernel FC5test2 kernel on two machines. Going back to the FC5test2 udev fixes it. It is breaking some scsi stuff too in addition to usb mouse.
> Reverting back to udev-084-1.1 made things work for me. Hmm.... * Tue Jan 24 2006 Harald Hoyer <harald> - 078-7 - put WAIT_FOR_SYSFS rules in 05-udev-early.rules
Diff from udev-084-1.1 to current udev... This means we have to wait for usb devices, too, which fails on some devices. -ACTION=="add", DEVPATH=="/devices/*", ENV{PHYSDEVBUS}=="?*", WAIT_FOR_SYSFS="b us" +ACTION=="add", DEVPATH=="/devices/*", DEVPATH!="*usb*", ENV{PHYSDEVBUS}=="?*", WAIT_FOR_SYSFS="bus"
> It is breaking some scsi stuff too in addition to usb mouse. Details please!
on the other side, there is bug #182723
and bug #182674