Is this still an issue with 0.5.7.1 or later and recent udev versions? Or can I
close this as FIXED?

The Ubuntu user reports it as fixed. Closing the bug.

This bug just reappeared for USB DVD drives, and seems to happen fairly
consistently: See https://launchpad.net/bugs/56484.

11:01:20.396 [I] osspec.c:232: SEQNUM=2636, ACTION=add, SUBSYSTEM=scsi_device,
DEVPATH=/sys/class/scsi_device/11:0:0:0, DEVNAME=, IFINDEX=0
11:01:20.397 [I] hotplug.c:205: /sys/class/scsi_device/11:0:0:0 is a class
device (devpath)
11:01:20.397 [I] classdev.c:1378: class_add: subsys=scsi_device
sysfs_path=/sys/class/scsi_device/11:0:0:0 dev= physdev=0x100c86c8
11:01:20.617 [I] osspec.c:232: SEQNUM=2637, ACTION=add, SUBSYSTEM=scsi_generic,
DEVPATH=/sys/class/scsi_generic/sg0, DEVNAME=/dev/sg0, IFINDEX=0
11:01:20.619 [I] hotplug.c:205: /sys/class/scsi_generic/sg0 is a class device
(devpath)
11:01:20.619 [I] classdev.c:1378: class_add: subsys=scsi_generic
sysfs_path=/sys/class/scsi_generic/sg0 dev=/dev/sg0 physdev=0x100c86c8
11:01:20.633 [I] classdev.c:1241: Add callouts completed
udi=/org/freedesktop/Hal/devices/usb_device_409_56_000000000001_if0_scsi_host_scsi_device_lun0_scsi_generic
11:01:20.633 [I] hald.c:82: Added device to GDL;
udi=/org/freedesktop/Hal/devices/usb_device_409_56_000000000001_if0_scsi_host_scsi_device_lun0_scsi_generic

Message from syslogd@localhost at Wed Aug 23 11:01:20 2006 ...
localhost kernel: [33867.200668] sr0: scsi3-mmc drive: 8x/48x writer cd/rw
xa/form2 cdda tray
11:01:27.329 [I] osspec.c:232: SEQNUM=2635, ACTION=add, SUBSYSTEM=block,
DEVPATH=/sys/block/sr0, DEVNAME=/dev/scd0, IFINDEX=0
11:01:27.330 [I] hotplug.c:208: /sys/block/sr0 is a block device (devpath)
11:01:27.331 [I] blockdev.c:589: block_add: sysfs_path=/sys/block/sr0
dev=/dev/scd0 is_part=1, parent=0x00000000
11:01:27.332 [I] blockdev.c:499: get_luks_uuid: device_file=/dev/scd0
11:01:27.332 [I] blockdev.c:625: Ignoring hotplug event - no parent
11:01:27.332 [W] blockdev.c:990: Not adding device object

Thus hal has no idea whatsoever about /dev/sr0, although the device is present
in the system and works quite well.

OK, I think I have an idea about what goes wrong now:

* The kernel creates /sys/block/sr0, which is the block device for the USB CD-ROM.

* ./hald/linux2/hotplug.c, around line 285:

   is_partition = isdigit(hotplug_event->sysfs.sysfs_path[len - 1]) ||
      strstr (hotplug_event->sysfs.sysfs_path, "/fakevolume") ;

  hal assumes that devices ending with a number are partitions on a block
device, which is not the case for SCSI CD-ROMs (i.e. it's not
/sys/block/sra/sra0, SCSI CD-ROMs are enumerated with numbers, not with letters
as ATAPI devices are). Thus is_partition becomes true.

    if (is_partition) {
       gchar *parent_path;
       parent_path = hal_util_get_parent_path (hotplug_event->sysfs.sysfs_path);

sysfs.sysfs_path is /sys/block/sr0, thus parent_path is /sys/block, which is of
course not a device.

       parent = hal_device_store_match_key_value_string (hald_get_gdl (),      
                                                                   
"linux.sysfs_path_device", parent_path);

parent becomes NULL here, which is then directly passed to
hotplug_event_begin_add_blockdev(), which throws the event away.

Created attachment 6658 [details] [review]
hack to fix it

I created a small patch that fixes it for now; it's pretty ugly, i. e.
eventually the original is_partition condition should be reworked properly, but
the patch should point out what the actual problem is.

   * Add 16-nonpartitions-ending-in-nums.patch:
     - Hal currently assumes that device names ending in numbers are
       partitions. This patch adds a safe and easy, but slightly ugly
       workaround. (Real fix will be discussed with upstream at
       https://bugs.freedesktop.org/show_bug.cgi?id=5558)
     - This breaks SCSI, USB, and Firewire CD drives, which are usually called
       'scd0', 'sr1', etc.

This is IIRC fixed in actual HAL versions by check for the range file in sysfs. Close as Fixed