bcache_gc: BUG: soft lockup - CPU#4 stuck for 22s!

Hi,

while trying to use bcache on 3.17-rc4 i got those messages and a load of 1000.

Is this a known problem?

14-09-12 02:32:22 BUG: soft lockup - CPU#4 stuck for 23s! [bcache_gc:1585]

2014-09-12 02:31:54     INFO: rcu_sched self-detected stall on CPU { 4} (t=150009 jiffies g=1762124 c=1762123 q=235323)
2014-09-12 02:31:42     BUG: soft lockup - CPU#4 stuck for 22s! [bcache_gc:1585]
2014-09-12 02:31:14     BUG: soft lockup - CPU#4 stuck for 22s! [bcache_gc:1585]
2014-09-12 02:30:46     BUG: soft lockup - CPU#4 stuck for 22s! [bcache_gc:1585]
2014-09-12 02:30:18     BUG: soft lockup - CPU#4 stuck for 22s! [bcache_gc:1585]
2014-09-12 02:29:50     BUG: soft lockup - CPU#4 stuck for 22s! [bcache_gc:1585]
2014-09-12 02:29:22     BUG: soft lockup - CPU#4 stuck for 23s! [bcache_gc:1585]
2014-09-12 02:28:54     INFO: rcu_sched self-detected stall on CPU { 4} (t=105006 jiffies g=1762124 c=1762123 q=209365)
2014-09-12 02:28:42     BUG: soft lockup - CPU#4 stuck for 23s! [bcache_gc:1585]
2014-09-12 02:28:14     BUG: soft lockup - CPU#4 stuck for 22s! [bcache_gc:1585]
2014-09-12 02:27:46     BUG: soft lockup - CPU#4 stuck for 22s! [bcache_gc:1585]
2014-09-12 02:27:18     BUG: soft lockup - CPU#4 stuck for 22s! [bcache_gc:1585]
2014-09-12 02:26:50     BUG: soft lockup - CPU#4 stuck for 22s! [bcache_gc:1585]
2014-09-12 02:26:22     BUG: soft lockup - CPU#4 stuck for 22s! [bcache_gc:1585]
2014-09-12 02:25:54     INFO: rcu_sched self-detected stall on CPU { 4} (t=60003 jiffies g=1762124 c=1762123 q=136221)
2014-09-12 02:25:42     BUG: soft lockup - CPU#4 stuck for 23s! [bcache_gc:1585]
2014-09-12 02:25:14     BUG: soft lockup - CPU#4 stuck for 23s! [bcache_gc:1585]
2014-09-12 02:24:46     BUG: soft lockup - CPU#4 stuck for 23s! [bcache_gc:1585]

Stefan

Greetings,

This was supposed to be corrected in the 3.17 push. I haven't seen it on my systems over the past few weeks of testing. Can you provide more details what was running when this occurred? What FS, hardware etc.

Ross Anderson

Hi Ross,
Am 12.09.2014 19:51, schrieb Ross Anderson:

I'm sorry i was using a 3.16.2 kernel but with all bcache patches from 3.17-rc4 added - my fault.

Which one should fix this?

List:
commit 4f6ce97baa7cf98afbd1962a2a184b9c05775e61

Author: Kent Overstreet <[email protected]>
Date:   Mon Jul 7 13:03:36 2014 -0700

bcache: Drop unneeded blk_sync_queue() calls

this is needed for the queue/block device we created (it's done by blk_cleanup_queue() which we do call) - but calling it for the
block devices we

    only opened is pointless.

    Change-Id: I53dfded14ed15b9581d10ca8399d5e1b3abbf9f2
    (cherry picked from commit 0781c8748cf1ea2b0dcd966571103909528c4efa)

commit 486b2c0d5254fa541634116cf9427089aca92105

Author: Jianjian Huo <[email protected]>
Date:   Sun Jul 13 09:08:59 2014 -0700

bcache: add mutex lock for bch_is_open

Since bch_is_open will iterate linked list bch_cache_sets and uncached_devices, it needs bch_register_lock.

commit 3b126259ee5ace5d3df27e7af1f5b623f091e9aa

Author: Surbhi Palande <[email protected]>
Date:   Thu Apr 17 12:07:04 2014 -0700

bcache: Correct printing of btree_gc_max_duration_ms

time_stats::btree_gc_max_duration_mc is not bit shifted by 8

Fixes BUG #138

commit e2c7fe1094ec597b5290f7b7030368ad303b66a5

Author: Slava Pestov <[email protected]>
Date:   Sat Jul 12 00:22:53 2014 -0700

    bcache: try to set b->parent properly

    bcache_flash_dev.ktest would reliably crash with 8k and 16k bucket size
    before; now it passes.

    Change-Id: Ib542232235e39298c3a7548fe52b645cabb823d1
    (cherry picked from commit 2452cc89063a2a6890368f185c4b6d7d8802179e)

commit 827381306e94ba3e2d18b8bf5eabb07cd99bbeb6

Author: Slava Pestov <[email protected]>
Date:   Thu Jun 19 15:05:59 2014 -0700

    bcache: fix memory corruption in init error path

    If register_cache_set() failed, we would touch ca->set after
    it had already been freed. Also, fix an assertion to catch
    this.

    Change-Id: I748e5f5b223e2d9b2602075dec2f997cced2394d
    (cherry picked from commit c9a78332b42cbdcdd386a95192a716b67d1711a4)

commit e53833bb678f9a02888c2b51789ed3c679bb72c7

Author: Slava Pestov <[email protected]>
Date:   Fri Jul 11 12:17:41 2014 -0700

    bcache: fix crash with incomplete cache set

    Change-Id: I6abde52afe917633480caaf4e2518f42a816d886
    (cherry picked from commit bf0c55c986540483c34ca640f2eef4c3314388b1)

commit c917ccd4c371082117ef09b6e1dd95b98db34359

Author: Kent Overstreet <[email protected]>
Date:   Wed Jun 11 19:44:49 2014 -0700

bcache: Fix more early shutdown bugs

commit 5a95fa33c0652c4ec8e354284e432a8f5f89b2ff

Author: Slava Pestov <[email protected]>
Date:   Sat Jul 12 21:53:11 2014 -0700

bcache: fix use-after-free in btree_gc_coalesce()

If we goto out_nocoalesce after we free new_nodes[0], we end up freeing new_nodes[0] again. This was generating a lockdep warning. The fix is to set new_nodes[0] to NULL, since the out_nocoalesce path safely ignores NULL entries in the new_nodes array.

This regression was introduced in 2d7f9531.

Change-Id: I76564d7257800583214376b4bacf236cda90c89c (cherry picked from commit 400ffaa2acd72274e2c7293a9724382383bebf3e)

commit 21690f2df19df170a8ebdb8bc53123529e74bbba

Author: Kent Overstreet <[email protected]>
Date:   Mon Jun 2 15:39:44 2014 -0700

bcache: Fix an infinite loop in journal replay

When running with multiple cache devices, if one of the devices has
a completely
empty journal but we'd already found some journal entries on a
previosu device
we'd go into an infinite loop.

commit 7559a9aa48ca73972f8f68f61735828cc6407e43

Author: Slava Pestov <[email protected]>
Date:   Fri May 23 11:18:35 2014 -0700

bcache: fix crash in bcache_btree_node_alloc_fail tracepoint

'b' was NULL.

Change-Id: Icac0fd04afa2d23f213d96d51afd53374e6dd0c0 (cherry picked from commit 913dc33fb2720fb5f979011664294137ddd8b13b)

commit cc6b3ec3da3fb190d10f8310f182200e1cf29efc

Author: Slava Pestov <[email protected]>
Date:   Thu May 22 12:14:24 2014 -0700

bcache: bcache_write tracepoint was crashing

commit bdcf832c86e3833d85a021129eefc9e2f4780cea

Author: Slava Pestov <[email protected]>
Date:   Mon Jun 30 22:31:20 2014 -0700

bcache: fix typo in bch_bkey_equal_header

commit 8265043558c66f53c3032f30f5c764b716504daa

Author: Kent Overstreet <[email protected]>
Date:   Mon May 19 08:55:40 2014 -0700

bcache: Allocate bounce buffers with GFP_NOWAIT

There's no point in blocking on these allocations, since our
fallback paths will

    probably go faster than blocking.

    Change-Id: I733ca202c25cb36bde02607a0a60552229a4241c
    (cherry picked from commit 501d52a90cbe652b41336c206ff0e95799d5a9b5)

commit b2c9961d6120c0993c06843484ee7f8c7cf7e39a

Author: Kent Overstreet <[email protected]>
Date:   Mon May 19 08:57:55 2014 -0700

bcache: Make sure to pass GFP_WAIT to mempool_alloc()

this was very wrong - mempool_alloc() only guarantees success with
GFP_WAIT.
bcache uses GFP_NOWAIT in various other places where we have a
fallback,
circuits must've gotten crossed when writing this code or something.

commit 64786bb5585b84f41892c2df052c963b1a06ec80

Author: Slava Pestov <[email protected]>
Date:   Thu May 1 13:48:57 2014 -0700

bcache: fix uninterruptible sleep in writeback thread

There were two issues here:

- writeback thread did not start until the device first became dirty
- writeback thread used uninterruptible sleep once running

Without this patch I see kernel warnings printed and a load average of 1.52 after booting my test VM. With this patch the warnings are
gone and
the load average is near 0.00 as expected.

commit cc58857bb12c78b81f81ad838e9005d6c47b8afe

Author: Slava Pestov <[email protected]>
Date:   Mon Apr 21 18:23:12 2014 -0700

bcache: wait for buckets when allocating new btree root

Tested:
- sometimes bcache_tier test would hang on startup with a failure to allocate the btree root -- no longer seeing this

commit ed0487836568d152ef4d1f9a16de9aa5872e6c70

Author: Slava Pestov <[email protected]>
Date:   Tue May 20 12:20:28 2014 -0700

bcache: fix crash on shutdown in passthrough mode

We never started the writeback thread in this case, so don't stop it. (cherry picked from commit a664d0f05a2ec02c8f042db536d84d15d6e19e81)

commit 052aefa2ce3455c0654591e43d90fb46a2336f8c

Author: Slava Pestov <[email protected]>
Date:   Tue Apr 29 15:39:27 2014 -0700

bcache: fix lockdep warnings on shutdown
(cherry picked from commit e5112201c1285841f8b565ece5d6ae7e0d7947a2)

commit b1a3f91107bbd5e22e9f461dd70f210f15393108

Author: Slava Pestov <[email protected]>
Date:   Mon Apr 21 18:22:35 2014 -0700

bcache allocator: send discards with correct size (cherry picked from commit 8b326d3a2a76912dfed2f0ab937d59fae9512ca2)

commit 35c5161eb523784bff426f678d387545f0fa4f45

Author: Surbhi Palande <[email protected]>
Date:   Thu Apr 10 16:09:51 2014 -0700

bcache: Fix to remove the rcu_sched stalls.

while loop was executing infinitely.
This fix ends the while loop gracefully.

commit 0b119e88f5e5400018a9f5edba6c85d1431701bd

Author: Kent Overstreet <[email protected]>
Date:   Thu Apr 10 17:58:49 2014 -0700

bcache: Fix a journal replay bug

journal replay wansn't validating pointers with
bch_extent_invalid() before
derefing, fixed

commit ffccebead362a0d5f236bc7d18642b85f1fe41b1

Author: Kent Overstreet <[email protected]>
Date:   Wed Mar 19 17:49:37 2014 -0700

bcache: Fix a bug when detaching

After detaching a backing device from a cache set, a bit wasn't getting reset meaning the second detach wouldn't work correctly.

Stefan