[LU-7465] False positive memory leak in sanity tests due to iov_iter_get_pages_alloc

• Status: Resolved
• Project: Lustre
• Affects Version/s: Lustre 2.8.0
• Fix Version/s: Lustre 2.8.0
• Type: Bug
• Priority: Minor
• Reporter: Wally Wang (Inactive)
• Assignee: WC Triage
• Resolution: Fixed
• Labels: patch

Environment

3.16 and later kernels

• Severity: 3

Description

The iov_iter_get_pages_alloc function(introduced by 3.16 kernels, LU-6260) is a kernel function which does not use Lustre's hook to keep track of memory allocation and therefore the number of bytes tallied by Lustre's ll_free_user_pages() appeared to be 'extra'. This happens only to direct io read and it is not a real leak of memory. It can be reproduced by a number of sanity tests, such as 205, to result in the message similar to:

25243:0:(class_obd.c:646:cleanup_obdclass()) obd_memory max:
18446744073709549568, leaked: 18446744073709549568

Issue Links

• is related to: LU-7052 iovec created with iov_iter_get_pages...

Comments

[Gerrit Updater]

Wally Wang ([e-mail]) uploaded a new patch: http://review.whamcloud.com/17342
Subject: LU-7465 llite: Do not track memory leak for kernel function
Project: fs/lustre-release
Branch: master
Current Patch Set: 1
Commit: dc38ad8d997a7252c6f8d3e246b38dd4d85295c6

[Cory Spitz]

Frank notes that this is related to LU-7052 and http://review.whamcloud.com/#/c/16111. Note that that patch was rejected and then abandoned.

[Andreas Dilger]

I don't see any reviews on 16111, so I assume the patch rejection was something that happened internally? In any case, good to see the new patch.

[Cory Spitz]

I think that it was discussed not in Gerrit, but in LU-7052.

[Gerrit Updater]

Oleg Drokin ([e-mail]) merged in patch http://review.whamcloud.com/17342/
Subject: LU-7465 llite: Do not track memory leak for kernel function
Project: fs/lustre-release
Branch: master
Current Patch Set:
Commit: 09141c0796802e7a3471c084ea5928674b3a1862

[Joseph Gmitter (Inactive)]

Landed for 2.8