Fixed
Created: May 14, 2015
Updated: Dec 3, 2018
Resolved Date: May 28, 2015
Found In Version: 6.0.0.18
Severity: Standard
Applicable for: Wind River Linux 6
Component/s: Userspace
Whilst trying to reproduce a snapshot issue observed by a customer I hit a segfault in "vgchange".
It could be that the root cause of both the customer issue and mine is the same, though I was using RCPL 18 and they were using RCPL 17. Other differences were that the customer has the rootfs on the LV, whereas I didn't get as far as setting up a rootfs on the LV, so the LV was just mounted manually.
Configure an intel-x86-64 platform with standard kernel and glibc_std rootfs.
Build and deploy to target using pxeboot - tftp kernel and NFS rootfs.
The target in my tests was a Matxm with two hard disks attached.
After booting via pxe, follow these steps:
1. Create 2 partitions on sda - one for /boot and one for / (root)
2. Created 1 partition on sdb - for extending /
3. Format sda1 with ext2 ready for use as /boot
4. Create physical volumes:
root@localhost:~# pvcreate /dev/sda2
Writing physical volume data to disk "/dev/sda2"
Physical volume "/dev/sda2" successfully created
root@localhost:~# pvcreate /dev/sdb1
Writing physical volume data to disk "/dev/sdb1"
Physical volume "/dev/sdb1" successfully created
5. Create vol group:
root@localhost:~# vgcreate rootvg /dev/sda2 /dev/sdb1
Volume group "rootvg" successfully created
6. Create logical volume:
root@localhost:~# lvcreate -L200G -n rootlv rootvg
bio: create slab <bio-2> at 2
Logical volume "rootlv" created
7. Format & Mount
root@localhost:~# mkfs -t ext4 /dev/rootvg/rootlv
...
root@localhost:~# mount /dev/rootvg/rootlv /mnt
EXT4-fs (dm-0): mounted filesystem with ordered data mode. Opts: (null)
8. Populate with some data, I just used a dist.tar.bz2 file
9. Create snapshot
root@localhost:/mnt# lvcreate -L10G -s -n backup /dev/rootvg/rootlv
bio: create slab <bio-3> at 3
Logical volume "backup" created
10. Modify the LV with some new data
11. Start restore from backup
root@localhost:/mnt# lvconvert --merge -i 1 /dev/rootvg/backup
Can't merge over open origin volume
Merging of snapshot backup will start next activation.
12. Reboot
13. Remount LV
root@localhost:~# vgscan
Reading all physical volumes. This may take a while...
Found volume group "rootvg" using metadata type lvm2
root@localhost:~# vgchange -ay
bio: create slab <bio-2> at 2
1 logical volume(s) in volume group "rootvg" now active
root@localhost:~# mount /dev/rootvg/rootlv /mnt
EXT4-fs (dm-0): mounted filesystem with ordered data mode. Opts: (null)
14. Check restore progress
root@localhost:~# lvs /dev/rootvg/backup
LV VG Attr LSize Pool Origin Data% Move Log Copy% Convert
[backup] rootvg Swi-a-s- 10.00g rootlv 0.77
at this point, and even if I don't do the "lvs" command, I see this after a few seconds:
vgchange[651]: segfault at 28 ip 000000000047ba59 sp 00007fffa3e1a7d0 error 4 in lvm[400000+d5000]
This seems to be reproducible every time.
