[LU-1727] lprocfs_counter_add() could be called in interrupt context

• Status: Resolved
• Project: Lustre
• Affects Version/s: Lustre 2.3.0, Lustre 2.1.5
• Fix Version/s: Lustre 2.3.0, Lustre 2.1.5
• Type: Bug
• Priority: Blocker
• Reporter: Maloo
• Assignee: Zhenyu Xu
• Resolution: Fixed
• Severity: 3

Description

This issue was created by maloo for bobijam <[e-mail]>

This issue relates to the following test suite run: https://maloo.whamcloud.com/test_sets/c2763a2e-de5b-11e1-b7ed-52540035b04c.

The sub-test test_compilebench failed with the following error:

test failed to respond and timed out

Info required for matching: parallel-scale-nfsv4 compilebench

the MDT test_log shows

08:12:04:LustreError: 30719:0:(lvfs_lib.c:94:lprocfs_counter_add()) ASSERTION(!cfs_in_interrupt()) failed

essential call trace

08:12:09: [<ffffffffa0440d79>] libcfs_assertion_failed+0x59/0x70 [libcfs]
08:12:09: [<ffffffffa04a36b5>] lprocfs_counter_add+0x165/0x170 [lvfs]
08:12:09: [<ffffffffa0623367>] ldlm_pool_shrink+0x57/0xf0 [ptlrpc]
08:12:09: [<ffffffff81096a8f>] ? up+0x2f/0x50
08:12:09: [<ffffffffa062414b>] ldlm_pools_shrink+0x27b/0x330 [ptlrpc]
08:12:09: [<ffffffffa0624233>] ldlm_pools_srv_shrink+0x13/0x20 [ptlrpc]
08:12:09: [<ffffffff81129bfa>] shrink_slab+0x12a/0x1a0
08:12:09: [<ffffffff8112ba5f>] do_try_to_free_pages+0x2ff/0x520
08:12:09: [<ffffffff81114def>] ? zone_watermark_ok+0x1f/0x30
08:12:09: [<ffffffff8112be6d>] try_to_free_pages+0x9d/0x130
08:12:09: [<ffffffff81132746>] ? next_online_pgdat+0x26/0x50
08:12:09: [<ffffffff8112cfc0>] ? isolate_pages_global+0x0/0x350
08:12:09: [<ffffffff81123ded>] __alloc_pages_nodemask+0x40d/0x940
08:12:09: [<ffffffff8115e102>] kmem_getpages+0x62/0x170
08:12:10: [<ffffffff8115ed1a>] fallback_alloc+0x1ba/0x270
08:12:10: [<ffffffff8115e76f>] ? cache_grow+0x2cf/0x320
08:12:10: [<ffffffff8115ea99>] ____cache_alloc_node+0x99/0x160
08:12:10: [<ffffffffa0436a13>] ? cfs_alloc+0x63/0x90 [libcfs]
08:12:10: [<ffffffff8115f6c9>] __kmalloc+0x189/0x220
08:12:10: [<ffffffffa0436a13>] cfs_alloc+0x63/0x90 [libcfs]
08:12:10: [<ffffffffa062fc9a>] ptlrpc_prep_bulk_imp+0x7a/0x350 [ptlrpc]

Comments

[Jodi Levi (Inactive)]

We think this should be assigned to BobiJam.

[Peter Jones]

Bobijam

Could you please look into this one?

Thanks

Peter

[Zhenyu Xu]

patch tracking at http://review.whamcloud.com/3597

patch description

lprocfs: fix an old assertion

lprocfs_counter_add() can be called in interrupt context, as a memory
allocation can trigger memory shrinker call ldlm_pool_shrink(), which
calls lprocfs_counter_add().

[Peter Jones]

Landed for 2.3

[Wojciech Turek (Inactive)]

I have seen this on Lustre-2.1.2, has the patch landed for it as well?

[Zhenyu Xu]

patch port for b2_1 is tracking at http://review.whamcloud.com/3729

[Andreas Dilger]

Patch still needs to land for b2_1.

[Peter Jones]

Landed for 2.1.5