[Bug 567652] New: parted corrupts reiserfs superblock during resize
http://bugzilla.novell.com/show_bug.cgi?id=567652 http://bugzilla.novell.com/show_bug.cgi?id=567652#c0 Summary: parted corrupts reiserfs superblock during resize Classification: openSUSE Product: openSUSE 11.2 Version: Final Platform: Other OS/Version: Other Status: NEW Severity: Major Priority: P5 - None Component: Basesystem AssignedTo: bnc-team-screening@forge.provo.novell.com ReportedBy: aschnell@novell.com QAContact: qa@suse.de Found By: Development Blocker: --- parted corrupts the superblock of a reiser filesystem during resize (on a partitioned software RAID): Steps to reproduce: - create partitioned software RAID md1 with partition md1p2 - mkfs.reiserfs /dev/md1p2 - parted /dev/md1 print gives: 2 20964887s 54034226s 33069340s primary reiserfs type=83 resizing partition 2 gives this info: Information: The reiserfs file system passed a basic check. For a more comprehensive check, run reiserfsck --check. print now gives: 2 20964887s 64034226s 43069340s primary reiserfs type=83 - reiserfsck --check /dev/md1p2 prints the error: reiserfs_open: Your partition is not big enough to contain the filesystem of (5383666) blocks as was specified in the found super block. Failed to open the filesystem. -- Configure bugmail: http://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.
http://bugzilla.novell.com/show_bug.cgi?id=567652
http://bugzilla.novell.com/show_bug.cgi?id=567652#c1
Arvin Schnell
http://bugzilla.novell.com/show_bug.cgi?id=567652
http://bugzilla.novell.com/show_bug.cgi?id=567652#c2
--- Comment #2 from Arvin Schnell
http://bugzilla.novell.com/show_bug.cgi?id=567652
http://bugzilla.novell.com/show_bug.cgi?id=567652#c
Petr Uzel
http://bugzilla.novell.com/show_bug.cgi?id=567652
http://bugzilla.novell.com/show_bug.cgi?id=567652#c3
Petr Uzel
http://bugzilla.novell.com/show_bug.cgi?id=567652
http://bugzilla.novell.com/show_bug.cgi?id=567652#c4
Arvin Schnell
http://bugzilla.novell.com/show_bug.cgi?id=567652
http://bugzilla.novell.com/show_bug.cgi?id=567652#c5
Petr Uzel
/sys/block/mdX/range is 1 and ext_range is 256.
Parted up to 1.9.0 uses /sys/block/DEV/range to determine the highest possible partition number that might exist on DEV. Since it is 1 in your case, parted will not inform kernel about any partition except the first one. As far as I understand, /sys/block/DEV/range represents maximum number of partitions DEV can contain. Tejun: is this correct? Or should ext_range be also consulted when looking for highest possible partition number on given device? If I create the array like in comment #2, then range==64.
I suppose the manpage of mdadm is outdated with kernel 2.6.28 where the kernel simply names the devices mdXpY. I don't know, but as I wrote above, it behaves differently when created with /dev/dm_dX name wrt to range attribute. Adding mdadm maintainer to CC:
The problem isn't that device nodes or entries in /proc/partitions are missing but that the entires in /proc/partitions are not updated after resizing the partiton with parted. Rereading the partition table with blockdev updates /proc/partitions.
I get this. With device whose range==1, kernel won't be informed about any changes of partitions with no.>=2. So either 1) MD device with range==1 isn't supposed to be partitioned 2) there is a bug in MD causing range==1 3) there is a bug parted => the algorithm for determining largest partition number for a device has to be fixed Apart from that, there still might be some parted/kernel/udev/hal timing issues (the fix for bug #539521 doesn't seem to work in all cases), but without fixing the range thing, this cannot work. I'm going to investigate further. -- Configure bugmail: http://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.
http://bugzilla.novell.com/show_bug.cgi?id=567652
http://bugzilla.novell.com/show_bug.cgi?id=567652#c6
--- Comment #6 from Petr Uzel
I don't know, but as I wrote above, it behaves differently when created with /dev/dm_dX name wrt to range attribute. Adding mdadm maintainer to CC: */dev/md_dX*
-- Configure bugmail: http://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.
http://bugzilla.novell.com/show_bug.cgi?id=567652
http://bugzilla.novell.com/show_bug.cgi?id=567652#c7
Tejun Heo
As far as I understand, /sys/block/DEV/range represents maximum number of partitions DEV can contain. Tejun: is this correct? Or should ext_range be also consulted when looking for highest possible partition number on given device?
range now shows the traditional contiguous minor range while ext_range shows the full range which may be discontiguous. So, yeap, all md devices can be partitioned now and parted should tell the kernel so. Thanks. -- Configure bugmail: http://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.
http://bugzilla.novell.com/show_bug.cgi?id=567652
http://bugzilla.novell.com/show_bug.cgi?id=567652#c8
--- Comment #8 from Petr Uzel
http://bugzilla.novell.com/show_bug.cgi?id=567652
http://bugzilla.novell.com/show_bug.cgi?id=567652#c9
Petr Uzel
http://bugzilla.novell.com/show_bug.cgi?id=567652
http://bugzilla.novell.com/show_bug.cgi?id=567652#c10
Marcus Meissner
http://bugzilla.novell.com/show_bug.cgi?id=567652
http://bugzilla.novell.com/show_bug.cgi?id=567652#c11
--- Comment #11 from Petr Uzel
You dont need a seperate swamp id, right? I believe not as I'd like to release the fix together with fix for bug#539521.
-- Configure bugmail: http://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.
http://bugzilla.novell.com/show_bug.cgi?id=567652
http://bugzilla.novell.com/show_bug.cgi?id=567652#c12
Swamp Workflow Management
http://bugzilla.novell.com/show_bug.cgi?id=567652
http://bugzilla.novell.com/show_bug.cgi?id=567652#c13
Petr Uzel
http://bugzilla.novell.com/show_bug.cgi?id=567652
http://bugzilla.novell.com/show_bug.cgi?id=567652#c14
--- Comment #14 from Bernhard Wiedemann
participants (1)
-
bugzilla_noreply@novell.com