Hi Andrea,
Do you have chance to have a try on the attached patch?
Thanks,
Gavin
On Wed, Oct 17, 2012 at 11:53:38PM +0200, Andrea Righi wrote:
>Just got this on 3.7.0-rc1 (last git commit 1867353):
>
>[49048.262912] =================================
>[49048.262913] [ INFO: inconsistent lock state ]
>[49048.262916] 3.7.0-rc1+ #518 Not tainted
>[49048.262918] ---------------------------------
>[49048.262919] inconsistent {RECLAIM_FS-ON-W} -> {IN-RECLAIM_FS-W} usage.
>[49048.262922] kswapd0/35 [HC0[0]:SC0[0]:HE1:SE1] takes:
>[49048.262924] (&mapping->i_mmap_mutex){+.+.?.}, at: [<ffffffff81192fbc>] page_referenced+0x9c/0x2e0
>[49048.262933] {RECLAIM_FS-ON-W} state was registered at:
>[49048.262935] [<ffffffff810ed5d6>] mark_held_locks+0x86/0x150
>[49048.262938] [<ffffffff810edce7>] lockdep_trace_alloc+0x67/0xc0
>[49048.262942] [<ffffffff811a9323>] kmem_cache_alloc_trace+0x33/0x230
>[49048.262945] [<ffffffff811a1a27>] do_mmu_notifier_register+0x87/0x180
>[49048.262948] [<ffffffff811a1b53>] mmu_notifier_register+0x13/0x20
>[49048.262951] [<ffffffff81006738>] kvm_dev_ioctl+0x428/0x510
>[49048.262955] [<ffffffff811c7ce8>] do_vfs_ioctl+0x98/0x570
>[49048.262959] [<ffffffff811c8251>] sys_ioctl+0x91/0xb0
>[49048.262962] [<ffffffff815df302>] system_call_fastpath+0x16/0x1b
>[49048.262966] irq event stamp: 825
>[49048.262968] hardirqs last enabled at (825): [<ffffffff815d6fa0>] _raw_spin_unlock_irq+0x30/0x60
>[49048.262971] hardirqs last disabled at (824): [<ffffffff815d6659>] _raw_spin_lock_irq+0x19/0x80
>[49048.262975] softirqs last enabled at (0): [<ffffffff81082170>] copy_process+0x630/0x17c0
>[49048.262979] softirqs last disabled at (0): [< (null)>] (null)
>[49048.262981]
>[49048.262981] other info that might help us debug this:
>[49048.262983] Possible unsafe locking scenario:
>[49048.262983]
>[49048.262984] CPU0
>[49048.262986] ----
>[49048.262987] lock(&mapping->i_mmap_mutex);
>[49048.262989] <Interrupt>
>[49048.262991] lock(&mapping->i_mmap_mutex);
>[49048.262993]
>[49048.262993] *** DEADLOCK ***
>[49048.262993]
>[49048.262995] no locks held by kswapd0/35.
>[49048.262996]
>[49048.262996] stack backtrace:
>[49048.262999] Pid: 35, comm: kswapd0 Not tainted 3.7.0-rc1+ #518
>[49048.263000] Call Trace:
>[49048.263005] [<ffffffff815cd988>] print_usage_bug+0x1f5/0x206
>[49048.263008] [<ffffffff8105a21f>] ? save_stack_trace+0x2f/0x50
>[49048.263011] [<ffffffff810ea865>] mark_lock+0x295/0x2f0
>[49048.263014] [<ffffffff810e9c70>] ? print_irq_inversion_bug.part.42+0x1f0/0x1f0
>[49048.263017] [<ffffffff810eae5d>] __lock_acquire+0x59d/0x1c20
>[49048.263020] [<ffffffff815cf163>] ? put_cpu_partial+0x65/0xbd
>[49048.263024] [<ffffffff81052d06>] ? native_sched_clock+0x26/0x90
>[49048.263028] [<ffffffff810c5555>] ? sched_clock_cpu+0xc5/0x120
>[49048.263031] [<ffffffff810ecbe0>] lock_acquire+0x90/0x210
>[49048.263034] [<ffffffff81192fbc>] ? page_referenced+0x9c/0x2e0
>[49048.263038] [<ffffffff815d2ea3>] mutex_lock_nested+0x73/0x3d0
>[49048.263041] [<ffffffff81192fbc>] ? page_referenced+0x9c/0x2e0
>[49048.263044] [<ffffffff81192fbc>] ? page_referenced+0x9c/0x2e0
>[49048.263047] [<ffffffff810e764e>] ? put_lock_stats.isra.26+0xe/0x40
>[49048.263051] [<ffffffff810e7a84>] ? lock_release_holdtime.part.27+0xd4/0x150
>[49048.263055] [<ffffffff8116edab>] ? __remove_mapping+0xab/0x120
>[49048.263058] [<ffffffff81192fbc>] page_referenced+0x9c/0x2e0
>[49048.263061] [<ffffffff81171b94>] shrink_page_list+0x3e4/0xa20
>[49048.263064] [<ffffffff81052d06>] ? native_sched_clock+0x26/0x90
>[49048.263068] [<ffffffff811726f5>] ? shrink_inactive_list+0x165/0x4b0
>[49048.263071] [<ffffffff815d6fa0>] ? _raw_spin_unlock_irq+0x30/0x60
>[49048.263075] [<ffffffff81172787>] shrink_inactive_list+0x1f7/0x4b0
>[49048.263079] [<ffffffff81172e8d>] shrink_lruvec+0x44d/0x550
>[49048.263082] [<ffffffff81173693>] kswapd+0x703/0xdf0
>[49048.263086] [<ffffffff810af470>] ? __init_waitqueue_head+0x60/0x60
>[49048.263090] [<ffffffff81172f90>] ? shrink_lruvec+0x550/0x550
>[49048.263093] [<ffffffff810ae98d>] kthread+0xed/0x100
>[49048.263097] [<ffffffff810ae8a0>] ? flush_kthread_worker+0x190/0x190
>[49048.263100] [<ffffffff815df25c>] ret_from_fork+0x7c/0xb0
>[49048.263103] [<ffffffff810ae8a0>] ? flush_kthread_worker+0x190/0x190
>
>Should we use a GFP_NOFS allocation in mmu_notifier_register() or is
>there a better way to fix/avoid this?
>
>Thanks,
>-Andrea
>
>--
>To unsubscribe, send a message with 'unsubscribe linux-mm' in
>the body to majordomo@xxxxxxxxx. For more info on Linux MM,
>see: http://www.linux-mm.org/ .
>Don't email: <a href=mailto:"dont@xxxxxxxxx";> email@xxxxxxxxx </a>
>
>From 8b7dcc6afd617e8b52ed1b10221195cce0c8f442 Mon Sep 17 00:00:00 2001
From: Gavin Shan <shangw@xxxxxxxxxxxxxxxxxx>
Date: Thu, 18 Oct 2012 20:14:06 +0800
Subject: [PATCH] mm/mmu_notifier: allocate mmu_notifier in advance
While allocating mmu_notifier with parameter GFP_KERNEL, swap would
start to work in case of tight available memory. Eventually, that
would lead to dead-lock while swap deamon does swapping anonymous
pages. It was caused by commit e0f3c3f78da29b114e7c1c68019036559f715948
("mm/mmu_notifier: init notifier if necessary").
The patch simply back out the above commit.
Signed-off-by: Gavin Shan <shangw@xxxxxxxxxxxxxxxxxx>
---
mm/mmu_notifier.c | 26 +++++++++++++-------------
1 files changed, 13 insertions(+), 13 deletions(-)
diff --git a/mm/mmu_notifier.c b/mm/mmu_notifier.c
index 479a1e7..8a5ac8c 100644
--- a/mm/mmu_notifier.c
+++ b/mm/mmu_notifier.c
@@ -196,28 +196,28 @@ static int do_mmu_notifier_register(struct mmu_notifier *mn,
BUG_ON(atomic_read(&mm->mm_users) <= 0);
/*
- * Verify that mmu_notifier_init() already run and the global srcu is
- * initialized.
- */
+ * Verify that mmu_notifier_init() already run and the global srcu is
+ * initialized.
+ */
BUG_ON(!srcu.per_cpu_ref);
+ ret = -ENOMEM;
+ mmu_notifier_mm = kmalloc(sizeof(struct mmu_notifier_mm), GFP_KERNEL);
+ if (unlikely(!mmu_notifier_mm))
+ goto out;
+
if (take_mmap_sem)
down_write(&mm->mmap_sem);
ret = mm_take_all_locks(mm);
if (unlikely(ret))
- goto out;
+ goto out_clean;
if (!mm_has_notifiers(mm)) {
- mmu_notifier_mm = kmalloc(sizeof(struct mmu_notifier_mm),
- GFP_KERNEL);
- if (unlikely(!mmu_notifier_mm)) {
- ret = -ENOMEM;
- goto out_of_mem;
- }
INIT_HLIST_HEAD(&mmu_notifier_mm->list);
spin_lock_init(&mmu_notifier_mm->lock);
mm->mmu_notifier_mm = mmu_notifier_mm;
+ mmu_notifier_mm = NULL;
}
atomic_inc(&mm->mm_count);
@@ -233,12 +233,12 @@ static int do_mmu_notifier_register(struct mmu_notifier *mn,
hlist_add_head(&mn->hlist, &mm->mmu_notifier_mm->list);
spin_unlock(&mm->mmu_notifier_mm->lock);
-out_of_mem:
mm_drop_all_locks(mm);
-out:
+out_clean:
if (take_mmap_sem)
up_write(&mm->mmap_sem);
-
+ kfree(mmu_notifier_mm);
+out:
BUG_ON(atomic_read(&mm->mm_users) <= 0);
return ret;
}
--
1.7.5.4