[LU-2546] test: sanity FAIL: test_140 Invalid symlink depth

• Status: Resolved
• Project: Lustre
• Affects Version/s: Lustre 2.3.0
• Type: Bug
• Priority: Minor
• Reporter: Jay Lan (Inactive)
• Assignee: Nathaniel Clark
• Resolution: Duplicate
• Labels: sles11

Environment

server: 2.1.3-1nasS, centos 6.3, 2.6.32_279.2.1.el6
client: 2.3.0-1nasC, sles11sp2, 3.0.42_0.7.3
1 mds, 2 oss, 2 clients.

• Severity: 3

Description

== sanity test 140: Check reasonable stack depth (shouldn't LBUG) ==================================== 17:37:58 (1356658678)
The symlink depth = 40
sanity test_140: @@@@@@ FAIL: Invalid symlink depth

The test_logs are uploaded as 'sanity.test_140.tgz'.
The test was run on service331.

Issue Links

• duplicates: LU-2020 sanity 140 should allow 40 consecutiv...

Comments

[Peter Jones]

Nathaniel

I think that you have worked in this area recently so could you please comment on this one?

Thanks

Peter

[Bob Glossman (Inactive)]

I think this is due to the 3.6 kernel in sles11 sp2 having the commit b356379a020bb7197603118bb1cbc903963aa198. This restructures the handling of trailing symlinks in pathnames. They are now recognized and handled as iterative, not as nested and recursive. Iterative sequences of symlinks have the much higher limit of 40, not MAX_NESTED_LINKS (8) in vfs code.
Either the test can be changed to accept 40 as a valid loop count, or the test can be changed to create symlink sequences that will be recognized as recursive in even the newest kernel versions of vfs code.

[Bob Glossman (Inactive)]

sorry, got a bit confused about kernel versions. It's 3.0 kernel in sles11 sp2, not 3.6. My point about the commit changing symlink handling is valid anyway. It's there in 3.0 too.

[Jay Lan (Inactive)]

Do you have a link to the commit, Bob? Thanks!

[Bob Glossman (Inactive)]

This is the link I found: http://git.kernel.org/?p=linux/kernel/git/stable/linux-stable.git;a=commit;h=b356379a020bb7197603118bb1cbc903963aa198

It appears to be one of a series of related commits revising symlink handling in vfs path walk code from ~3/2011

[Nathaniel Clark]

This was fixed in master by bug LU-2020

The following change can be cleanly cherry-picked onto both 2.1 and 2.3
http://review.whamcloud.com/4079

[Jay Lan (Inactive)]

Thanks, Nathaniel. Test_140 passed with this patch.