[Bug 420526] New: libata *ATA device mappings used should be consistent & non-duplicative
https://bugzilla.novell.com/show_bug.cgi?id=420526 Summary: libata *ATA device mappings used should be consistent & non-duplicative Product: openSUSE 11.1 Version: Factory Platform: PC OS/Version: Other Status: NEW Severity: Normal Priority: P5 - None Component: Basesystem AssignedTo: bnc-team-screening@forge.provo.novell.com ReportedBy: mrmazda@ij.net QAContact: qa@suse.de Found By: --- On a ata_piix single HD system with partitions >15 using mapper so that all partitions can be accessed, the result of 'mount | sort' shows that mappings are inconsistent. IOW, sometimes the output is: /dev/dm-18 on / type ext3 (rw,noatime,nodiratime,acl) /dev/dm-22 on /home type ext3 (rw,noatime,nodiratime,acl) /dev/dm-24 on /pub type ext3 (rw,noatime,nodiratime,acl) /dev/dm-26 on /srv type ext3 (rw,noatime,nodiratime,noacl) /dev/dm-28 on /usr/local type ext3 (rw,noatime,nodiratime,noacl) /dev/dm-8 on /disks/hda/boot type ext2 (rw,noatime,nodiratime,noacl) while other times the output is: /dev/dm-19 on / type ext3 (rw,noatime,nodiratime,acl) /dev/dm-23 on /home type ext3 (rw,noatime,nodiratime,acl) /dev/dm-25 on /pub type ext3 (rw,noatime,nodiratime,acl) /dev/dm-27 on /srv type ext3 (rw,noatime,nodiratime,noacl) /dev/dm-29 on /usr/local type ext3 (rw,noatime,nodiratime,noacl) /dev/dm-9 on /disks/hda/boot type ext2 (rw,noatime,nodiratime,noacl) and yet other times: /dev/dm-12 on /disks/hda/boot type ext2 (rw,noatime,nodiratime,noacl) /dev/dm-19 on / type ext3 (rw,noatime,nodiratime,acl) /dev/dm-23 on /home type ext3 (rw,noatime,nodiratime,acl) /dev/dm-25 on /pub type ext3 (rw,noatime,nodiratime,acl) /dev/dm-27 on /srv type ext3 (rw,noatime,nodiratime,noacl) /dev/dm-29 on /usr/local type ext3 (rw,noatime,nodiratime,noacl) I think there are other variations, and I don't see any pattern of why the variation exists. All the above were created without any changes to fstab, and all are from the same system on successive boots. This behavior makes it a problem to determine what command to use to mount an unmounted partition without first doing a fstab lookup, and then only if its content is not by-UUID. /dev/sda7 is sometimes dm-13, other times dm-16, while dm-13 is sometimes /dev/sda7, and other times /dev/sda9. It's sometimes possible to mount /dev/sda7 as both dm-12 and dm-9 at the same time: /dev/dm-12 on /disks/hda/boot type ext2 (rw,noatime,nodiratime,noacl) /dev/dm-18 on /disks/hda/cooker type ext3 (rw,noatime,nodiratime,noacl) /dev/dm-20 on / type ext3 (rw,noatime,nodiratime,acl) /dev/dm-24 on /home type ext3 (rw,noatime,nodiratime,acl) /dev/dm-26 on /pub type ext3 (rw,noatime,nodiratime,acl) /dev/dm-28 on /srv type ext3 (rw,noatime,nodiratime,noacl) /dev/dm-30 on /usr/local type ext3 (rw,noatime,nodiratime,noacl) /dev/dm-9 on /disks/hda/boot type ext2 (rw,noatime,nodiratime,noacl) Lately I've often tried to mount a partition, and as output got an error message that the (unmounted) device is "busy" or that there is an error in the fstab entry or the fstab entry does not exist. I had to boot DOS and remove spaces from the volume labels, then change FSTAB to mount everything LABEL= to be able to get everything to mount, which then works only if I use those labels instead of attempting to guess which is the appropriate dm- name to use. On all these boots I've seen kernel (2.6.26-14-pae) dumps that include "dmsetup Tainted". Also, when Grub loads itself the PC speaker beeps a high pitch. Neither happens booting same system to Mandriva Cooker with 2.6.26.3-desktop-1mnb. The dumps don't happen when I boot an older (2.6.26-2-pae) kernel. Lately init 0/6 randomly fails at "Shutting down D-Bus daemon", a hard reset or power off is required to actually get a reboot or powerdown, and most partitions are dirty at next init. Whether this might be related I only suspect, but since I've sometimes seen mount output showing / twice, I believe it likely. -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.
https://bugzilla.novell.com/show_bug.cgi?id=420526
User mrmazda@ij.net added comment
https://bugzilla.novell.com/show_bug.cgi?id=420526#c1
--- Comment #1 from Felix Miata
https://bugzilla.novell.com/show_bug.cgi?id=420526
User mrmazda@ij.net added comment
https://bugzilla.novell.com/show_bug.cgi?id=420526#c2
--- Comment #2 from Felix Miata
https://bugzilla.novell.com/show_bug.cgi?id=420526
User mrmazda@ij.net added comment
https://bugzilla.novell.com/show_bug.cgi?id=420526#c3
--- Comment #3 from Felix Miata
https://bugzilla.novell.com/show_bug.cgi?id=420526
Robert Vojcik
https://bugzilla.novell.com/show_bug.cgi?id=420526
User gregkh@novell.com added comment
https://bugzilla.novell.com/show_bug.cgi?id=420526#c4
Greg Kroah-Hartman
https://bugzilla.novell.com/show_bug.cgi?id=420526
User kasievers@novell.com added comment
https://bugzilla.novell.com/show_bug.cgi?id=420526#c5
Kay Sievers
https://bugzilla.novell.com/show_bug.cgi?id=420526
User mrmazda@ij.net added comment
https://bugzilla.novell.com/show_bug.cgi?id=420526#c6
--- Comment #6 from Felix Miata
What's the output of: tree /dev/disk/
This attachment.
Do you miss any device from this list of links? All fs-label, fs-uuid and "physical path" links should be there, which you can use in fstab, or from the commandline.
by-id: ata*: all accounted for dm-name-ata: all accounted for dm-name-scsi: all acounted for dm-uuid*linear-ata: big mess makes me cross-eyed trying to tell dm-uuid*linear-scsi: big mess makes me cross-eyed trying to tell edd*: all >15 missing scsi-SATA: all accounted for by-path: all >15 missing by-label: HPFS (#1) missing IBM BM (#2) missing extended (#4) missing created but unformatted ext3 (#22, #23) missing by-uuid: extended, HPFS & IBM BM partitions missing (NAICT) -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.
https://bugzilla.novell.com/show_bug.cgi?id=420526
User kasievers@novell.com added comment
https://bugzilla.novell.com/show_bug.cgi?id=420526#c7
--- Comment #7 from Kay Sievers
by-id: ata*: all accounted for dm-name-ata: all accounted for dm-name-scsi: all acounted for dm-uuid*linear-ata: big mess makes me cross-eyed trying to tell dm-uuid*linear-scsi: big mess makes me cross-eyed trying to tell
It's the device mapper supplied uuid, the link just prepends dm-uuid. :)
edd*: all >15 missing
That's fine, It's BIOS signature stuff, which works only for some disks/setups.
scsi-SATA: all accounted for by-path: all >15 missing
Yeah, dm devices are virtual, and have no real "path to the device".
by-label: HPFS (#1) missing
It's probably not recognized by libvol_id. What does: /lib/udev/vol_id /dev/<device with hpfs> print?
IBM BM (#2) missing
What's IBM BM? How can it be detected?
extended (#4) missing created but unformatted ext3 (#22, #23) missing
What's an "unformatted ext3"? Does it contain a fs, or not? -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.
https://bugzilla.novell.com/show_bug.cgi?id=420526
User mrmazda@ij.net added comment
https://bugzilla.novell.com/show_bug.cgi?id=420526#c8
Felix Miata
(In reply to comment #6 from Felix Miata)
by-label: HPFS (#1) missing
It's probably not recognized by libvol_id. What does: /lib/udev/vol_id /dev/<device with hpfs> print?
ID_FS_USAGE=filesystem ID_FS_TYPE=hpfs ID_FS_VERSION=197 ID_FS_UUID= ID_FS_UUID_ENC= ID_FS_LABEL= ID_FS_LABEL_ENC= ID_FS_LABEL_SAFE= BTW, it does have a volume label, beginning at offset 0x2B: "P01 S16A". I can attach that first sector if you want.
IBM BM (#2) missing
What's IBM BM?
Answered in attachment 235623 (IBM Boot Manager)
How can it be detected?
Can't answer that, but I can attach a copy of its first sector if you want.
extended (#4) missing created but unformatted ext3 (#22, #23) missing
What's an "unformatted ext3"? Does it contain a fs, or not?
Type 0x83 partitions 22 & 23 were created, but not yet given filesystems. -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.
https://bugzilla.novell.com/show_bug.cgi?id=420526
User kasievers@novell.com added comment
https://bugzilla.novell.com/show_bug.cgi?id=420526#c9
Kay Sievers
(In reply to comment #7 from Kay Sievers)
(In reply to comment #6 from Felix Miata)
by-label: HPFS (#1) missing
It's probably not recognized by libvol_id. What does: /lib/udev/vol_id /dev/<device with hpfs> print?
ID_FS_USAGE=filesystem ID_FS_TYPE=hpfs ID_FS_VERSION=197 ID_FS_UUID= ID_FS_UUID_ENC= ID_FS_LABEL= ID_FS_LABEL_ENC= ID_FS_LABEL_SAFE=
BTW, it does have a volume label, beginning at offset 0x2B: "P01 S16A". I can attach that first sector if you want.
Yeah, a copy of the sector would be nice, I'll go add the label reading then.
IBM BM (#2) missing
What's IBM BM?
Answered in attachment 235623 [details] (IBM Boot Manager)
How can it be detected?
Can't answer that, but I can attach a copy of its first sector if you want.
Seems, it can not be reliably detected from volume metadata, only the type number in the partition table signifies this.
extended (#4) missing created but unformatted ext3 (#22, #23) missing
What's an "unformatted ext3"? Does it contain a fs, or not?
Type 0x83 partitions 22 & 23 were created, but not yet given filesystems.
Ah, ok. If we get the label link for the hpfs partition, and you would use /dev/disk/by-*/* links in fstab, which information you would still miss to identify your volumes? Thanks! -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.
https://bugzilla.novell.com/show_bug.cgi?id=420526
User mrmazda@ij.net added comment
https://bugzilla.novell.com/show_bug.cgi?id=420526#c10
--- Comment #10 from Felix Miata
(In reply to comment #8 from Felix Miata)
BTW, it does have a volume label, beginning at offset 0x2B: "P01 S16A". I can attach that first sector if you want.
Yeah, a copy of the sector would be nice, I'll go add the label reading then.
If we get the label link for the hpfs partition, and you would use /dev/disk/by-*/* links in fstab, which information you would still miss to identify your volumes? Thanks!
I would not use /dev/disk/by-*/* links in fstab. Like UUID labeling, they are much too long for my feeble brain to deal with. I use LABEL= in Grub and fstab, and generally the mountpoint from fstab for mounting and unmounting. The main problem is for volumes that are not in fstab, such as those newly created or those subjected to a LABEL or UUID change. Really it seems I should be able to use blkid to get the needed information, but as is usual for man pages, there are no examples. Without examples, I am typically unable to understand, and this is such a case. Without parameters, blkid produces output overflowing the screen. With parameters, I either get a usage message, or too little output. Combined with tune2fs -l and grep'd blkid, I am able to fumble well enough to find the following duplicative result that makes no sense to me except to point to the root cause of this bug: # blkid | grep 12home | grep dm- /dev/dm-23: LABEL="S16A-12home" UUID="af9a1d43-5154-4cb0-91d9-1a62d3050ef2" TYPE="ext3" SEC_TYPE="ext2" /dev/dm-26: LABEL="S16A-12home" UUID="af9a1d43-5154-4cb0-91d9-1a62d3050ef2" TYPE="ext3" -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.
https://bugzilla.novell.com/show_bug.cgi?id=420526
User kasievers@novell.com added comment
https://bugzilla.novell.com/show_bug.cgi?id=420526#c11
Kay Sievers
https://bugzilla.novell.com/show_bug.cgi?id=420526
User mrmazda@ij.net added comment
https://bugzilla.novell.com/show_bug.cgi?id=420526#c12
--- Comment #12 from Felix Miata
https://bugzilla.novell.com/show_bug.cgi?id=420526
User kasievers@novell.com added comment
https://bugzilla.novell.com/show_bug.cgi?id=420526#c13
Kay Sievers
https://bugzilla.novell.com/show_bug.cgi?id=420526
User kasievers@novell.com added comment
https://bugzilla.novell.com/show_bug.cgi?id=420526#c14
--- Comment #14 from Kay Sievers
https://bugzilla.novell.com/show_bug.cgi?id=420526
User kasievers@novell.com added comment
https://bugzilla.novell.com/show_bug.cgi?id=420526#c15
Kay Sievers
participants (1)
-
bugzilla_noreply＠novell.com