slab: fix the DEADLOCK issue on l3 alien lock

DEADLOCK will be report while running a kernel with NUMA and LOCKDEP enabled,
the process of this fake report is:

	   kmem_cache_free()	//free obj in cachep
	-> cache_free_alien()	//acquire cachep's l3 alien lock
	-> __drain_alien_cache()
	-> free_block()
	-> slab_destroy()
	-> kmem_cache_free()	//free slab in cachep->slabp_cache
	-> cache_free_alien()	//acquire cachep->slabp_cache's l3 alien lock

Since the cachep and cachep->slabp_cache's l3 alien are in the same lock class,
fake report generated.

This should not happen since we already have init_lock_keys() which will
reassign the lock class for both l3 list and l3 alien.

However, init_lock_keys() was invoked at a wrong position which is before we
invoke enable_cpucache() on each cache.

Since until set slab_state to be FULL, we won't invoke enable_cpucache()
on caches to build their l3 alien while creating them, so although we invoked
init_lock_keys(), the l3 alien lock class won't change since we don't have
them until invoked enable_cpucache() later.

This patch will invoke init_lock_keys() after we done enable_cpucache()
instead of before to avoid the fake DEADLOCK report.

Michael traced the problem back to a commit in release 3.0.0:

commit 30765b92ada267c5395fc788623cb15233276f5c
Author: Peter Zijlstra <peterz@infradead.org>
Date:   Thu Jul 28 23:22:56 2011 +0200

    slab, lockdep: Annotate the locks before using them

    Fernando found we hit the regular OFF_SLAB 'recursion' before we
    annotate the locks, cure this.

    The relevant portion of the stack-trace:

    > [    0.000000]  [<c085e24f>] rt_spin_lock+0x50/0x56
    > [    0.000000]  [<c04fb406>] __cache_free+0x43/0xc3
    > [    0.000000]  [<c04fb23f>] kmem_cache_free+0x6c/0xdc
    > [    0.000000]  [<c04fb2fe>] slab_destroy+0x4f/0x53
    > [    0.000000]  [<c04fb396>] free_block+0x94/0xc1
    > [    0.000000]  [<c04fc551>] do_tune_cpucache+0x10b/0x2bb
    > [    0.000000]  [<c04fc8dc>] enable_cpucache+0x7b/0xa7
    > [    0.000000]  [<c0bd9d3c>] kmem_cache_init_late+0x1f/0x61
    > [    0.000000]  [<c0bba687>] start_kernel+0x24c/0x363
    > [    0.000000]  [<c0bba0ba>] i386_start_kernel+0xa9/0xaf

    Reported-by: Fernando Lopez-Lezcano <nando@ccrma.Stanford.EDU>
    Acked-by: Pekka Enberg <penberg@kernel.org>
    Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
Link: http://lkml.kernel.org/r/1311888176.2617.379.camel@laptop
    Signed-off-by: Ingo Molnar <mingo@elte.hu>
The commit moved init_lock_keys() before we build up the alien, so we
failed to reclass it.

Cc: <stable@vger.kernel.org> # 3.0+
Acked-by: Christoph Lameter <cl@linux.com>
Tested-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
Signed-off-by: Michael Wang <wangyun@linux.vnet.ibm.com>
Signed-off-by: Pekka Enberg <penberg@kernel.org>

diff --git a/mm/slab.c b/mm/slab.c
index 3b4587bb7b19ed5b859a9f58c3a86a79d67cc2b9..cd5a92650303c251bf1b769feea8c807e85fa15b 100644 (file)
--- a/mm/slab.c
+++ b/mm/slab.c
@@ -1774,9 +1774,6 @@ void __init kmem_cache_init_late(void)
 
        slab_state = UP;
 
-       /* Annotate slab for lockdep -- annotate the malloc caches */
-       init_lock_keys();
-
        /* 6) resize the head arrays to their final sizes */
        mutex_lock(&slab_mutex);
        list_for_each_entry(cachep, &slab_caches, list)
@@ -1784,6 +1781,9 @@ void __init kmem_cache_init_late(void)
                        BUG();
        mutex_unlock(&slab_mutex);
 
+       /* Annotate slab for lockdep -- annotate the malloc caches */
+       init_lock_keys();
+
        /* Done! */
        slab_state = FULL;