I've tried the upgrade to 068, but it didn't work for me. I have a second IDE controller on a PCI card (using the siimage kernel module which is loaded in the autoload). Attached to it are two HD's, giving me /dev/hde and /dev/hdf. Their partitions are mounted at boot-up (by having entries in /etc/fstab). All was working fine on udev-058. After the upgrade to udev-068 these partitions aren't accessible by fsck during startup (automounting of /etc/fstab entries), complaining about bad superblocks. Removing them from /etc/fstab allowed me to boot again. Once in X (thus after a while) I was able to mount them without any problem. It seems that the /dev/hde* and /dev/hdf* nodes are created "too late". The /dev/hda* up to /dev/hdd (including cdroms) are detected correctly however. Might this have to do with the new udev.rules? Reproducible: Always Steps to Reproduce: 1.Use a siimage kernel module enabled IDE Controller PCI card 2.emerge =udev-068 3.have and entry /dev/hde in /etc/fstab 4.reboot Actual Results: fsck checking during booting (just before it should mount /dev/hde) fails. Complains about bad superblock. Asking for root password to correct Expected Results: Boot (correctly find /dev/hde)
It's a probably a race between loading the module with the modprobe loading the driver, and the mount. Can you just build your siimage kernel driver into the kernel? Or can you use the genkernel initrd to load the driver before udev is loaded?
I've got a similar problem with an external firewire HDD: after upgrade to udev-068 I wasn't able to access the drive. Got timeouts from the device and unable to mount or fsck. After downgrade to udev-58 the problem disappeared. This is the output using udev-58: Sep 6 16:12:09 [kernel] ieee1394: Error parsing configrom for node 0-01:1023 Sep 6 16:12:09 [kernel] ieee1394: The root node is not cycle master capable; se lecting a new root node and resetting... Sep 6 16:12:09 [kernel] ieee1394: Error parsing configrom for node 0-00:1023 Sep 6 16:12:09 [kernel] ieee1394: Node changed: 0-00:1023 -> 0-01:1023 Sep 6 16:12:14 [kernel] ieee1394: Node added: ID:BUS[0-00:1023] GUID[20075901e 0454647] Sep 6 16:12:14 [kernel] scsi0 : SCSI emulation for IEEE-1394 SBP-2 Devices Sep 6 16:12:16 [kernel] ieee1394: sbp2: Logged into SBP-2 device Sep 6 16:12:16 [kernel] ieee1394: Node 0-00:1023: Max speed [S400] - Max payloa d [2048] Sep 6 16:12:16 [kernel] Vendor: Maxtor 6 Model: Y160P0 Rev: Sep 6 16:12:16 [kernel] SCSI device sda: 320173056 512-byte hdwr sectors (16392 9 MB) Sep 6 16:12:16 [kernel] sda: asking for cache data failed Sep 6 16:12:16 [kernel] SCSI device sda: 320173056 512-byte hdwr sectors (16392 9 MB) Sep 6 16:12:16 [kernel] sda: asking for cache data failed And this is the output with udev-068 Sep 5 17:06:36 [kernel] ieee1394: Error parsing configrom for node 0-01:1023 Sep 5 17:06:36 [kernel] ieee1394: The root node is not cycle master capable; se lecting a new root node and resetting... Sep 5 17:06:36 [kernel] ieee1394: Error parsing configrom for node 0-00:1023 Sep 5 17:06:36 [kernel] ieee1394: Node changed: 0-00:1023 -> 0-01:1023 Sep 5 17:06:42 [kernel] ieee1394: Node added: ID:BUS[0-00:1023] GUID[20075901e 0454647] Sep 5 17:06:42 [kernel] sbp2: $Rev: 1219 $ Ben Collins <bcollins@debian.org> Sep 5 17:06:42 [kernel] scsi0 : SCSI emulation for IEEE-1394 SBP-2 Devices Sep 5 17:06:43 [kernel] ieee1394: sbp2: Logged into SBP-2 device Sep 5 17:06:43 [kernel] ieee1394: Node 0-00:1023: Max speed [S400] - Max payloa d [2048] Sep 5 17:06:43 [kernel] Vendor: Maxtor 6 Model: Y160P0 Rev: Sep 5 17:06:43 [kernel] SCSI device sda: 320173056 512-byte hdwr sectors (16392 9 MB) Sep 5 17:06:43 [kernel] sda: asking for cache data failed Sep 5 17:06:43 [kernel] SCSI device sda: 320173056 512-byte hdwr sectors (16392 9 MB) Sep 5 17:06:50 [kernel] ieee1394: Node changed: 0-01:1023 -> 0-00:1023 Sep 5 17:06:50 [kernel] ieee1394: Node suspended: ID:BUS[0-00:1023] GUID[20075 901e0454647] Sep 5 17:07:14 [kernel] ieee1394: sbp2: aborting sbp2 command Sep 5 17:07:24 [kernel] ieee1394: sbp2: Bus reset in progress - rejecting comma nd - Last output repeated twice - Tried with 2.6.13-ck1, scsi and firewire as modules and in the kernel with same results.
about comment #2: the problem went away after removing hotplug (and udev-070)
Does 070 fix the problem?
No it doesn't. I still have to revert to 058 for correct functionality. However 070-r1 allows me to boot without fsck complaining. It does state during boot that the /dev/hde* and /dev/hdf* partitions don't exist. We're a bit closer to the solution.
I've tried udev-079, but this one even kills my sound. /dev/sound entries are not created when starting alsasound. However that's another issue. Anyway, just mentioning that udev-079 doesn't fix the problem for me yet. Meanwhile I've indeed noticed that it is a race problem. In very rare occassions the partions do get mounted.
