possible logic bug in raid5.c

[Jens Laas <jens.laas@data.slu.se>]

Hello!

It appears to me that raid5 assumes the number of raid_disks (in 
conf-raid_disks for example) to directly map to the rdev list in mddev.

This is not always the case. We encountered the problem when we had one 
"ghost" device in this list. Maybe this can be triggered some other way 
too.

Example:
in run() in raid5.c

when counting working_disks:

        ITERATE_RDEV(mddev,rdev,tmp) {
                 raid_disk = rdev->raid_disk;
                 if (raid_disk >= conf->raid_disks
                     || raid_disk < 0)
                         continue;

If there are more disks in mddev's list than conf->raid_disks this loop
will ignore the tail of the list.

The test "if (raid_disk >= conf->raid_disks" makes sure that in our case 
the "ghost" disk is tested (and not added) and the following disk (which 
is a working raid disk) is ignored.

Without being familiar with the code I think it should either make sure 
that "ghost" devices cannot be added or that the above kind of tests are 
changed.

In our "ghost"-disk case this is what mdadm reports:
(some parts of output deleted)
$ mdadm --detail /dev/md0
    Raid Devices : 4
   Total Devices : 4
  Active Devices : 4
Working Devices : 4
  Failed Devices : 0
   Spare Devices : 0

     Number   Major   Minor   RaidDevice State
        0       8       33        0      active sync   /dev/sdc1
        1       8       49        1      active sync   /dev/sdd1
        2       8       65        2      active sync   /dev/sde1
        3       0        0        3      removed
        4       8       81        4      active sync   /dev/sdf1

This is how it looks like in /sys:
# ls -l /sys/block/md0/md/
total 0
-rw-r--r-- 1 root root 4096 Jun  1 10:27 array_state
--w------- 1 root root 4096 Jun  1 10:27 bitmap_set_bits
-rw-r--r-- 1 root root 4096 Jun  1 10:27 chunk_size
-rw-r--r-- 1 root root 4096 Jun  1 10:27 component_size
drwxr-xr-x 2 root root    0 Jun  1 09:47 dev-sdc1
drwxr-xr-x 2 root root    0 Jun  1 02:27 dev-sdd1
drwxr-xr-x 2 root root    0 Jun  1 02:27 dev-sde1
drwxr-xr-x 2 root root    0 Jun  1 02:27 dev-sdf1
-rw-r--r-- 1 root root 4096 Jun  1 10:27 layout
-rw-r--r-- 1 root root 4096 Jun  1 10:27 level
-rw-r--r-- 1 root root 4096 Jun  1 10:27 metadata_version
-r--r--r-- 1 root root 4096 Jun  1 10:27 mismatch_cnt
--w------- 1 root root 4096 Jun  1 10:27 new_dev
-rw-r--r-- 1 root root 4096 Jun  1 10:27 raid_disks
lrwxrwxrwx 1 root root    0 Jun  1 10:27 rd0 -> 
../../../block/md0/md/dev-sdc1
lrwxrwxrwx 1 root root    0 Jun  1 10:27 rd1 -> 
../../../block/md0/md/dev-sdd1
lrwxrwxrwx 1 root root    0 Jun  1 10:27 rd2 -> 
../../../block/md0/md/dev-sde1
lrwxrwxrwx 1 root root    0 Jun  1 10:27 rd4 -> 
../../../block/md0/md/dev-sdf1

Cheers,
Jens Låås

-----------------------------------------------------------------------
     'In theory, there is no difference between theory and practice.
      But, in practice, there is.'
-----------------------------------------------------------------------
     Jens Låås                              Email: jens.laas@data.slu.se
     Department of Computer Services, SLU   Phone: +46 18 67 35 15
     Vindbrovägen 1
     P.O. Box 7079
     S-750 07 Uppsala
     SWEDEN
-----------------------------------------------------------------------
-
To unsubscribe from this list: send the line "unsubscribe linux-raid" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html