After upgrading to udev 031, i have lost all my ide cdroms. They are still present as normal in /proc/ide/ They appear to be found as normal under /sys under /dev, there is no cdroms directory, although the hdc and hdd device nodes are still present. there is only one disk available under the /dev/ide tree which is /dev/ide/host0/bus0/target0/lun0/disc which is a symlink to /dev/hdd I'm not certain, but if I had to hazard a guess, I'd say its a problem with ide-devfs.sh not creating the devices properly, since it doesn't seem to have stepped through, but rather overwritten each ide device in the /dev/ide tree with the symlink of the next ide device, rather than creating the rest of that file tree. Reproducible: Always Steps to Reproduce: 1. emerge udev 031 2. Reboot 3. Actual Results: No cdrom devices found. Expected Results: created /dev/cdroms/cdrom0 and cdrom1, and other devices under the /dev/ide tree
*** Bug 63675 has been marked as a duplicate of this bug. ***
I've done a bit more testing, and apparently it isn't the scripts themselves, as udev 030 doesnt show the same problem even with the scripts from 031. I'm guessing that somewhere along the line, some part of udev is passing bad options to the device maker rules and scripts.
I can confirm similar problems with scsi-devices. I have a home-built /etc/udev/rules.d/10-local.rules with BUS="scsi", KERNEL="sd*", PROGRAM="/etc/udev/scripts/scsidisk.sh %b %n", SYMLINK="sd/%c" And /etc/udev/scripts/scsidisk.sh with id=$1 part=$2 # Extract SCSI logical address components host=`echo $id | cut -f 1 -d:` bus=`echo $id | cut -f 2 -d:` target=`echo $id | cut -f 3 -d:` lun=`echo $id | cut -f 4 -d:` if [ "$part" ]; then echo "c${host}b${bus}t${target}u${lun}p${part}" else echo "c${host}b${bus}t${target}u${lun}" fi All I get is /dev/sd/c0b0t0u0 -> ../sde3 Since sde3 is the last detected scsi-device my guess is that udev don't pass The "kernel number" of the device. (%n)
My cdrom still worked, but hdparm puked for me, there was only one /dev/disc entry, none of the others. Anyways, reverting to 030 solved my problems, and I think the probelm is the new {scsi|ide}-devfs.sh but I am by no means a udev expert.
The problem has now been noticed in the linux-hoptplug-devel mailing-list. And http://marc.theaimsgroup.com/?l=linux-hotplug-devel&m=109503879112491&w=2 contains a patch.
Created attachment 39485 [details] Updated udev-031 ebuild with namedev patch ebuild
Created attachment 39486 [details, diff] udev patch for namedev This is the patch for the above ebuild.
fixed in release 032
oops, forgot to mark fixed...