Hello
Since linux-next-20170517 at least I got the following DEADLOCK warning:
[ 4.311614] ============================================
[ 4.316919] WARNING: possible recursive locking detected
[ 4.322227] 4.12.0-rc1-next-20170517+ #273 Not tainted
[ 4.327360] --------------------------------------------
[ 4.332665] swapper/0/1 is trying to acquire lock:
[ 4.337451] (cpu_hotplug_lock.rw_sem){++++++}, at: [<c01cc998>] stop_machine+0x1c/0x3c
[ 4.345468]
but task is already holding lock:
[ 4.351294] (cpu_hotplug_lock.rw_sem){++++++}, at: [<c01f4534>] static_key_slow_inc+0x14/0x24
[ 4.359911]
other info that might help us debug this:
[ 4.366431] Possible unsafe locking scenario:
[ 4.372344] CPU0
[ 4.374789] ----
[ 4.377233] lock(cpu_hotplug_lock.rw_sem);
[ 4.381504] lock(cpu_hotplug_lock.rw_sem);
[ 4.385775]
*** DEADLOCK ***
[ 4.391691] May be due to missing lock nesting notation
[ 4.398472] 5 locks held by swapper/0/1:
[ 4.402390] #0: (net_mutex){+.+.+.}, at: [<c05d4800>] register_pernet_subsys+0x28/0x48
[ 4.410491] #1: (register_ipv4_hooks){+.+.+.}, at: [<c06a409c>] ipv4_hooks_register+0xdc/0x1e0
[ 4.419285] #2: (defrag4_mutex){+.+.+.}, at: [<c06a4d90>] nf_defrag_ipv4_enable+0x48/0x8c
[ 4.427643] #3: (cpu_hotplug_lock.rw_sem){++++++}, at: [<c01f4534>] static_key_slow_inc+0x14/0x24
[ 4.436694] #4: (jump_label_mutex){+.+...}, at: [<c01f44ac>] __static_key_slow_inc+0x78/0xec
[ 4.445312]
stack backtrace:
[ 4.449669] CPU: 1 PID: 1 Comm: swapper/0 Not tainted 4.12.0-rc1-next-20170517+ #273
[ 4.457402] Hardware name: Allwinner sun8i Family
[ 4.462100] Backtrace:
[ 4.464557] [<c010c6d8>] (dump_backtrace) from [<c010c9b0>] (show_stack+0x18/0x1c)
[ 4.472121] r7:c0c2dbd0 r6:00000000 r5:20000093 r4:c0c2dbd0
[ 4.477780] [<c010c998>] (show_stack) from [<c03e863c>] (dump_stack+0xac/0xd8)
[ 4.485002] [<c03e8590>] (dump_stack) from [<c017640c>] (__lock_acquire+0xbc0/0x19f0)
[ 4.492829] r10:ef05b200 r9:c0d8779c r8:00000000 r7:c0c2dcc0 r6:00000000 r5:c1480500
[ 4.500649] r4:c0d8779c r3:60000093
[ 4.504228] [<c017584c>] (__lock_acquire) from [<c0177a00>] (lock_acquire+0x74/0x90)
[ 4.511967] r10:c0b37858 r9:c0c55348 r8:00000001 r7:00000001 r6:60000013 r5:00000000
[ 4.519786] r4:ffffe000
[ 4.522324] [<c017798c>] (lock_acquire) from [<c0125500>] (get_online_cpus+0x58/0xe0)
[ 4.530140] r8:c0959570 r7:c01cc998 r6:c0c18ff4 r5:00000000 r4:c0c19644
[ 4.536838] [<c01254a8>] (get_online_cpus) from [<c01cc998>] (stop_machine+0x1c/0x3c)
[ 4.544662] r7:c09bd668 r6:00000000 r5:ef04dce8 r4:c010f964
[ 4.550320] [<c01cc97c>] (stop_machine) from [<c010f9b0>] (patch_text+0x2c/0x34)
[ 4.557709] r7:c09bd668 r6:c1498a14 r5:c0c58b64 r4:c06ad878
[ 4.563367] [<c010f984>] (patch_text) from [<c010f79c>] (arch_jump_label_transform+0x28/0x44)
[ 4.571886] [<c010f774>] (arch_jump_label_transform) from [<c01f3a68>] (__jump_label_update+0x94/0x9c)
[ 4.581181] r5:c0c58b64 r4:c0c58a68
[ 4.584758] [<c01f39d4>] (__jump_label_update) from [<c01f415c>] (jump_label_update+0x94/0x130)
[ 4.593448] r7:c09bd668 r6:eea252c0 r5:c1498a14 r4:c0c58b64
[ 4.599105] [<c01f40c8>] (jump_label_update) from [<c01f450c>] (__static_key_slow_inc+0xd8/0xec)
[ 4.607881] r7:c09bd668 r6:eea252c0 r5:c0c57964 r4:c1498a14
[ 4.613538] [<c01f4434>] (__static_key_slow_inc) from [<c01f453c>] (static_key_slow_inc+0x1c/0x24)
[ 4.622486] r5:c0c57964 r4:c1498a14
[ 4.626067] [<c01f4520>] (static_key_slow_inc) from [<c0617284>] (nf_register_net_hook+0x148/0x1a8)
[ 4.635102] r5:c0c57964 r4:c0c501c0
[ 4.638682] [<c061713c>] (nf_register_net_hook) from [<c0617bd8>] (nf_register_net_hooks+0x40/0x78)
[ 4.647721] r9:c0c55348 r8:00000002 r7:c0c4fd00 r6:c0c57b15 r5:00000000 r4:c0c55348
[ 4.655461] [<c0617b98>] (nf_register_net_hooks) from [<c06a4dbc>] (nf_defrag_ipv4_enable+0x74/0x8c)
[ 4.664587] r9:c0c578ae r8:00000009 r7:00000009 r6:c0c57b15 r5:c0c4fd00 r4:00000000
[ 4.672319] [<c06a4d48>] (nf_defrag_ipv4_enable) from [<c06a4164>] (ipv4_hooks_register+0x1a4/0x1e0)
[ 4.681441] r5:c0c4fd00 r4:eea25180
[ 4.685022] [<c06a3fc0>] (ipv4_hooks_register) from [<c0623e10>] (nf_ct_l3proto_pernet_register+0x30/0x3c)
[ 4.694666] r7:ef0dc800 r6:c0c4fd00 r5:c0c4fd00 r4:00000000
[ 4.700325] [<c0623de0>] (nf_ct_l3proto_pernet_register) from [<c06a45f0>] (ipv4_net_init+0x30/0x68)
[ 4.709452] [<c06a45c0>] (ipv4_net_init) from [<c05d3e04>] (ops_init+0x104/0x16c)
[ 4.716926] r5:eea25180 r4:c0c552a0
[ 4.720506] [<c05d3d00>] (ops_init) from [<c05d4734>] (register_pernet_operations+0x108/0x1ac)
[ 4.729111] r9:c0c43cb4 r8:ef04de70 r7:c0c4fcd0 r6:c0c552a0 r5:00000000 r4:c0c4fd00
[ 4.736852] [<c05d462c>] (register_pernet_operations) from [<c05d480c>] (register_pernet_subsys+0x34/0x48)
[ 4.746497] r9:00000000 r8:c0c607c0 r7:c0b37850 r6:c0c552a0 r5:c0c4fc44 r4:c0c4fc40
[ 4.754241] [<c05d47d8>] (register_pernet_subsys) from [<c0b2cf5c>] (nf_conntrack_l3proto_ipv4_init+0x38/0xb4)
[ 4.764232] r7:c0b37850 r6:ffffe000 r5:c0b2cf24 r4:00000000
[ 4.769893] [<c0b2cf24>] (nf_conntrack_l3proto_ipv4_init) from [<c0101954>] (do_one_initcall+0x5c/0x198)
[ 4.779361] r5:c0b2cf24 r4:c0c0f4cc
[ 4.782942] [<c01018f8>] (do_one_initcall) from [<c0b00ff0>] (kernel_init_freeable+0x254/0x2e8)
[ 4.791635] r9:00000007 r8:c0c607c0 r7:c0b37850 r6:c0c607c0 r5:c0b48060 r4:c09ff118
[ 4.799377] [<c0b00d9c>] (kernel_init_freeable) from [<c070c1d8>] (kernel_init+0x10/0x118)
[ 4.807635] r10:00000000 r9:00000000 r8:00000000 r7:00000000 r6:00000000 r5:c070c1c8
[ 4.815454] r4:00000000
[ 4.817995] [<c070c1c8>] (kernel_init) from [<c0107f70>] (ret_from_fork+0x14/0x24)
[ 4.825555] r5:c070c1c8 r4:00000000
I bisected the issue to commit d215aab82d81974f438bfbc80aa437132f3c37c3 "cpu/hotplug: Convert hotplug locking to percpu rwsem"
i try to revert the patch but get:
CC kernel/jump_label.o
/linux-next/kernel/jump_label.c:124:6: warning: no previous prototype for '__static_key_slow_inc' [-Wmissing-prototypes]
void __static_key_slow_inc(struct static_key *key)
^~~~~~~~~~~~~~~~~~~~~
/linux-next/kernel/jump_label.c: In function '__static_key_slow_dec':
/linux-next/kernel/jump_label.c:194:2: error: implicit declaration of function 'lockdep_assert_hotplug_held' [-Werror=implicit-function-declaration]
didnt try to go further.
Regards
Corentin Labbe
On Sun, 21 May 2017, Corentin Labbe wrote:
> Since linux-next-20170517 at least I got the following DEADLOCK warning:
> [ 4.311614] ============================================
> [ 4.316919] WARNING: possible recursive locking detected
> [ 4.322227] 4.12.0-rc1-next-20170517+ #273 Not tainted
> [ 4.327360] --------------------------------------------
> [ 4.332665] swapper/0/1 is trying to acquire lock:
> [ 4.337451] (cpu_hotplug_lock.rw_sem){++++++}, at: [<c01cc998>] stop_machine+0x1c/0x3c
The offending commit has been removed from -next. We are working on a replacement.
Thanks,
tglx