[LU-7965] mkfs.lustre sometimes fails with a segmentation fault

• Type: Bug
• Resolution: Fixed
• Priority: Minor
• Fix Version/s: Lustre 2.9.0
• Affects Version/s: Lustre 2.7.0
• Labels: patch
• Severity: 3
• Assignee: Gregoire Pichon
• Reporter: Gregoire Pichon

Description

During format of a Lustre target, the mkfs.lustre command sometimes crash with a segmentation fault.

It occurs at the end of the process so the target is well formatted anyway.

# mkfs.lustre -v --reformat --quiet "--fsname=fs1" --ost --index=0 "--mgsnode=mexico9@o2ib5" "--mkfsoptions=-m 0 -E lazy_itable_init=0,lazy_journal_init=0" --device-size=4194304 /var/loop/ost0

   Permanent disk data:
Target: fs1:OST0000
Index: 0
Lustre FS: fs1
Mount type: ldiskfs
Flags: 0x62
              (OST first_time update )
Persistent mount opts: errors=remount-ro
Parameters: mgsnode=10.128.0.13@o2ib5

formatting backing filesystem ldiskfs on /dev/loop0
        target name fs1:OST0000
        4k blocks 1048576
        options -m 0 -J size=163 -I 256 -q -O extents,uninit_bg,dir_nlink,quota,huge_file,flex_bg -G 256 -E lazy_itable_init=0,lazy_journal_init=0,resize=4290772992 -F
mkfs_cmd = mke2fs -j -b 4096 -L fs1:OST0000 -m 0 -J size=163 -I 256 -q -O extents,uninit_bg,dir_nlink,quota,huge_file,flex_bg -G 256 -E lazy_itable_init=0,lazy_journal_init=0,resize=4290772992 -F /dev/loop0 1048576
Writing CONFIGS/mountdata
Segmentation fault

After analysis of the problem, it appears the stack is corrupted.
Issue comes from enable_default_ext4_features() routine.

I am going to push a patch that fixes the problem.

The bug probably affects other Lustre versions.

Issue Links

• is related to: LU-7919 Buffer overflow in mount_lustre: parse_ldd(),append_option()