Re: async pf: INFO: SOFTIRQ-safe -> SOFTIRQ-unsafe lock order detected

[Paul E. McKenney]

On Thu, Apr 26, 2012 at 04:50:35PM +0200, Sasha Levin wrote:
> Hi all.
>
> I got a "INFO: SOFTIRQ-safe -> SOFTIRQ-unsafe lock order detected" warning
> while running LTP inside a KVM guest using the recent -next kernel.
>
> It seems that it was initially originated from rcu_torture_rea(), but I
> don't think that it's a problem with RCU itself (I'll cc. Paul just in
> case). RCU torture was indeed configured to run during the testing.

I might be missing something, but I don't see how RCU is involved in
this one.

Thanx, Paul

> The output is attached since it's a bit long.
>
> Thanks.

> [ 4248.579019] ======================================================
> [ 4248.579019] [ INFO: SOFTIRQ-safe -> SOFTIRQ-unsafe lock order detected ]
> [ 4248.579019] 3.4.0-rc4-next-20120424-sasha #94 Tainted: G W
> [ 4248.579019] ------------------------------------------------------
> [ 4248.579019] rcu_torture_rea/1615 [HC0[0]:SC0[0]:HE0:SE1] is trying to acquire:
> [ 4248.579019] (pgd_lock){+.+...}, at: [<ffffffff81089476>] pgd_free+0x26/0xa0
> [ 4248.579019]
> [ 4248.579019] and this task is already holding:
> [ 4248.579019] (&(&async_pf_sleepers[i].lock)->rlock){..-...}, at: [<ffffffff8107cdc8>] kvm_async_pf_task_wake+0x48/0x130
> [ 4248.579019] which would create a new lock dependency:
> [ 4248.579019] (&(&async_pf_sleepers[i].lock)->rlock){..-...} -> (pgd_lock){+.+...}
> [ 4248.579019]
> [ 4248.579019] but this new dependency connects a SOFTIRQ-irq-safe lock:
> [ 4248.579019] (&(&async_pf_sleepers[i].lock)->rlock){..-...}
> [ 4248.579019] ... which became SOFTIRQ-irq-safe at:
> [ 4248.579019] [<ffffffff8111ab85>] mark_irqflags+0x95/0x170
> [ 4248.579019] [<ffffffff8111c56b>] __lock_acquire+0x2bb/0x4c0
> [ 4248.579019] [<ffffffff8111c84c>] lock_acquire+0xdc/0x120
> [ 4248.579019] [<ffffffff8289823b>] _raw_spin_lock+0x3b/0x70
> [ 4248.579019] [<ffffffff8107cdc8>] kvm_async_pf_task_wake+0x48/0x130
> [ 4248.579019] [<ffffffff8107d1f6>] do_async_page_fault+0x96/0xb0
> [ 4248.579019] [<ffffffff82899695>] async_page_fault+0x25/0x30
> [ 4248.579019] [<ffffffff8289b1ac>] call_softirq+0x1c/0x30
> [ 4248.579019] [<ffffffff81050235>] do_softirq+0x75/0x120
> [ 4248.579019] [<ffffffff810be1cb>] irq_exit+0x5b/0xf0
> [ 4248.579019] [<ffffffff8107361a>] smp_apic_timer_interrupt+0x8a/0xa0
> [ 4248.579019] [<ffffffff8289a7af>] apic_timer_interrupt+0x6f/0x80
> [ 4248.579019] [<ffffffff8107d1be>] do_async_page_fault+0x5e/0xb0
> [ 4248.579019] [<ffffffff82899695>] async_page_fault+0x25/0x30
> [ 4248.579019] [<ffffffff811d7de8>] do_huge_pmd_wp_page+0x308/0x520
> [ 4248.579019] [<ffffffff811a4e0c>] handle_mm_fault+0x25c/0x350
> [ 4248.579019] [<ffffffff81085031>] do_page_fault+0x421/0x450
> [ 4248.579019] [<ffffffff8107d191>] do_async_page_fault+0x31/0xb0
> [ 4248.579019] [<ffffffff82899695>] async_page_fault+0x25/0x30
> [ 4248.579019]
> [ 4248.579019] to a SOFTIRQ-irq-unsafe lock:
> [ 4248.579019] (pgd_lock){+.+...}
> [ 4248.579019] ... which became SOFTIRQ-irq-unsafe at:
> [ 4248.579019] ... [<ffffffff8111ac00>] mark_irqflags+0x110/0x170
> [ 4248.579019] [<ffffffff8111c56b>] __lock_acquire+0x2bb/0x4c0
> [ 4248.579019] [<ffffffff8111c84c>] lock_acquire+0xdc/0x120
> [ 4248.579019] [<ffffffff8289823b>] _raw_spin_lock+0x3b/0x70
> [ 4248.579019] [<ffffffff81083061>] sync_global_pgds+0x61/0x1b0
> [ 4248.579019] [<ffffffff81083d5d>] vmalloc_sync_all+0x1d/0x20
> [ 4248.579019] [<ffffffff810e1a71>] register_die_notifier+0x11/0x30
> [ 4248.579019] [<ffffffff83b93537>] init_uprobes+0x75/0x79
> [ 4248.579019] [<ffffffff83b71629>] do_one_initcall+0x7a/0x155
> [ 4248.579019] [<ffffffff83b717cb>] do_basic_setup+0x9c/0xba
> [ 4248.579019] [<ffffffff83b71a42>] kernel_init+0x1fe/0x280
> [ 4248.579019] [<ffffffff8289b0b4>] kernel_thread_helper+0x4/0x10
> [ 4248.579019]
> [ 4248.579019] other info that might help us debug this:
> [ 4248.579019]
> [ 4248.579019] Possible interrupt unsafe locking scenario:
> [ 4248.579019]
> [ 4248.579019] CPU0 CPU1
> [ 4248.579019] ---- ----
> [ 4248.579019] lock(pgd_lock);
> [ 4248.579019] local_irq_disable();
> [ 4248.579019] lock(&(&async_pf_sleepers[i].lock)->rlock);
> [ 4248.579019] lock(pgd_lock);
> [ 4248.579019] <Interrupt>
> [ 4248.579019] lock(&(&async_pf_sleepers[i].lock)->rlock);
> [ 4248.579019]
> [ 4248.579019] *** DEADLOCK ***
> [ 4248.579019]
> [ 4248.579019] 1 lock held by rcu_torture_rea/1615:
> [ 4248.579019] #0: (&(&async_pf_sleepers[i].lock)->rlock){..-...}, at: [<ffffffff8107cdc8>] kvm_async_pf_task_wake+0x48/0x130
> [ 4248.579019]
> [ 4248.579019] the dependencies between SOFTIRQ-irq-safe lock and the holding lock:
> [ 4248.579019] -> (&(&async_pf_sleepers[i].lock)->rlock){..-...} ops: 1556 {
> [ 4248.579019] IN-SOFTIRQ-W at:
> [ 4248.579019] [<ffffffff8111ab85>] mark_irqflags+0x95/0x170
> [ 4248.579019] [<ffffffff8111c56b>] __lock_acquire+0x2bb/0x4c0
> [ 4248.579019] [<ffffffff8111c84c>] lock_acquire+0xdc/0x120
> [ 4248.579019] [<ffffffff8289823b>] _raw_spin_lock+0x3b/0x70
> [ 4248.579019] [<ffffffff8107cdc8>] kvm_async_pf_task_wake+0x48/0x130
> [ 4248.579019] [<ffffffff8107d1f6>] do_async_page_fault+0x96/0xb0
> [ 4248.579019] [<ffffffff82899695>] async_page_fault+0x25/0x30
> [ 4248.579019] [<ffffffff8289b1ac>] call_softirq+0x1c/0x30
> [ 4248.579019] [<ffffffff81050235>] do_softirq+0x75/0x120
> [ 4248.579019] [<ffffffff810be1cb>] irq_exit+0x5b/0xf0
> [ 4248.579019] [<ffffffff8107361a>] smp_apic_timer_interrupt+0x8a/0xa0
> [ 4248.579019] [<ffffffff8289a7af>] apic_timer_interrupt+0x6f/0x80
> [ 4248.579019] [<ffffffff8107d1be>] do_async_page_fault+0x5e/0xb0
> [ 4248.579019] [<ffffffff82899695>] async_page_fault+0x25/0x30
> [ 4248.579019] [<ffffffff811d7de8>] do_huge_pmd_wp_page+0x308/0x520
> [ 4248.579019] [<ffffffff811a4e0c>] handle_mm_fault+0x25c/0x350
> [ 4248.579019] [<ffffffff81085031>] do_page_fault+0x421/0x450
> [ 4248.579019] [<ffffffff8107d191>] do_async_page_fault+0x31/0xb0
> [ 4248.579019] [<ffffffff82899695>] async_page_fault+0x25/0x30
> [ 4248.579019] INITIAL USE at:
> [ 4248.579019] [<ffffffff8111c595>] __lock_acquire+0x2e5/0x4c0
> [ 4248.579019] [<ffffffff8111c84c>] lock_acquire+0xdc/0x120
> [ 4248.579019] [<ffffffff8289823b>] _raw_spin_lock+0x3b/0x70
> [ 4248.579019] [<ffffffff8107cb30>] apf_task_wake_all+0x20/0x80
> [ 4248.579019] [<ffffffff8107cdb2>] kvm_async_pf_task_wake+0x32/0x130
> [ 4248.579019] [<ffffffff8107d1f6>] do_async_page_fault+0x96/0xb0
> [ 4248.579019] [<ffffffff82899695>] async_page_fault+0x25/0x30
> [ 4248.579019] [<ffffffff83bb1f7a>] con_init+0x1b/0x23f
> [ 4248.579019] [<ffffffff83bb168d>] console_init+0x19/0x2a
> [ 4248.579019] [<ffffffff83b71fa2>] start_kernel+0x30c/0x3f8
> [ 4248.579019] [<ffffffff83b71314>] x86_64_start_reservations+0xff/0x104
> [ 4248.579019] [<ffffffff83b7146e>] x86_64_start_kernel+0x155/0x164
> [ 4248.579019] }
> [ 4248.579019] ... key at: [<ffffffff83d3fc84>] __key.35787+0x0/0x8
> [ 4248.579019] ... acquired at:
> [ 4248.579019] [<ffffffff8111a43c>] check_usage+0x1ac/0x1d0
> [ 4248.579019] [<ffffffff8111a4d1>] check_irq_usage+0x71/0xf0
> [ 4248.579019] [<ffffffff8111b7bb>] check_prev_add+0x16b/0x4d0
> [ 4248.579019] [<ffffffff8111c1be>] validate_chain+0x69e/0x790
> [ 4248.579019] [<ffffffff8111c6d3>] __lock_acquire+0x423/0x4c0
> [ 4248.579019] [<ffffffff8111c84c>] lock_acquire+0xdc/0x120
> [ 4248.579019] [<ffffffff8289823b>] _raw_spin_lock+0x3b/0x70
> [ 4248.579019] [<ffffffff81089476>] pgd_free+0x26/0xa0
> [ 4248.579019] [<ffffffff810b4d81>] __mmdrop+0x31/0xf0
> [ 4248.579019] [<ffffffff8107cacc>] apf_task_wake_one+0x4c/0x90
> [ 4248.579019] [<ffffffff8107cdf5>] kvm_async_pf_task_wake+0x75/0x130
> [ 4248.579019] [<ffffffff8107d1f6>] do_async_page_fault+0x96/0xb0
> [ 4248.579019] [<ffffffff82899695>] async_page_fault+0x25/0x30
> [ 4248.579019] [<ffffffff8289695e>] __schedule+0xce/0x880
> [ 4248.579019] [<ffffffff82897275>] schedule+0x55/0x60
> [ 4248.579019] [<ffffffff8115a3e2>] rcu_torture_reader+0x2a2/0x380
> [ 4248.579019] [<ffffffff810dada2>] kthread+0xb2/0xc0
> [ 4248.579019] [<ffffffff8289b0b4>] kernel_thread_helper+0x4/0x10
> [ 4248.579019]
> [ 4248.579019]
> [ 4248.579019] the dependencies between the lock to be acquired and SOFTIRQ-irq-unsafe lock:
> [ 4248.579019] -> (pgd_lock){+.+...} ops: 2112935 {
> [ 4248.579019] HARDIRQ-ON-W at:
> [ 4248.579019] [<ffffffff8111abe0>] mark_irqflags+0xf0/0x170
> [ 4248.579019] [<ffffffff8111c56b>] __lock_acquire+0x2bb/0x4c0
> [ 4248.579019] [<ffffffff8111c84c>] lock_acquire+0xdc/0x120
> [ 4248.579019] [<ffffffff8289823b>] _raw_spin_lock+0x3b/0x70
> [ 4248.579019] [<ffffffff81083061>] sync_global_pgds+0x61/0x1b0
> [ 4248.579019] [<ffffffff81083d5d>] vmalloc_sync_all+0x1d/0x20
> [ 4248.579019] [<ffffffff810e1a71>] register_die_notifier+0x11/0x30
> [ 4248.579019] [<ffffffff83b93537>] init_uprobes+0x75/0x79
> [ 4248.579019] [<ffffffff83b71629>] do_one_initcall+0x7a/0x155
> [ 4248.579019] [<ffffffff83b717cb>] do_basic_setup+0x9c/0xba
> [ 4248.579019] [<ffffffff83b71a42>] kernel_init+0x1fe/0x280
> [ 4248.579019] [<ffffffff8289b0b4>] kernel_thread_helper+0x4/0x10
> [ 4248.579019] SOFTIRQ-ON-W at:
> [ 4248.579019] [<ffffffff8111ac00>] mark_irqflags+0x110/0x170
> [ 4248.579019] [<ffffffff8111c56b>] __lock_acquire+0x2bb/0x4c0
> [ 4248.579019] [<ffffffff8111c84c>] lock_acquire+0xdc/0x120
> [ 4248.579019] [<ffffffff8289823b>] _raw_spin_lock+0x3b/0x70
> [ 4248.579019] [<ffffffff81083061>] sync_global_pgds+0x61/0x1b0
> [ 4248.579019] [<ffffffff81083d5d>] vmalloc_sync_all+0x1d/0x20
> [ 4248.579019] [<ffffffff810e1a71>] register_die_notifier+0x11/0x30
> [ 4248.579019] [<ffffffff83b93537>] init_uprobes+0x75/0x79
> [ 4248.579019] [<ffffffff83b71629>] do_one_initcall+0x7a/0x155
> [ 4248.579019] [<ffffffff83b717cb>] do_basic_setup+0x9c/0xba
> [ 4248.579019] [<ffffffff83b71a42>] kernel_init+0x1fe/0x280
> [ 4248.579019] [<ffffffff8289b0b4>] kernel_thread_helper+0x4/0x10
> [ 4248.579019] INITIAL USE at:
> [ 4248.579019] [<ffffffff8111c595>] __lock_acquire+0x2e5/0x4c0
> [ 4248.579019] [<ffffffff8111c84c>] lock_acquire+0xdc/0x120
> [ 4248.579019] [<ffffffff8289823b>] _raw_spin_lock+0x3b/0x70
> [ 4248.579019] [<ffffffff81085c15>] update_page_count+0x25/0x50
> [ 4248.579019] [<ffffffff8289161c>] phys_pte_init+0x16d/0x180
> [ 4248.579019] [<ffffffff82891b16>] phys_pmd_init+0x250/0x337
> [ 4248.579019] [<ffffffff82891d5e>] phys_pud_init+0x161/0x3a4
> [ 4248.579019] [<ffffffff8289208e>] kernel_physical_mapping_init+0xed/0x1d9
> [ 4248.579019] [<ffffffff8284853b>] init_memory_mapping+0x38b/0x460
> [ 4248.579019] [<ffffffff83b77d66>] setup_arch+0x691/0x815
> [ 4248.579019] [<ffffffff83b71d8b>] start_kernel+0xf5/0x3f8
> [ 4248.579019] [<ffffffff83b71314>] x86_64_start_reservations+0xff/0x104
> [ 4248.579019] [<ffffffff83b7146e>] x86_64_start_kernel+0x155/0x164
> [ 4248.579019] }
> [ 4248.579019] ... key at: [<ffffffff834356b8>] pgd_lock+0x18/0x50
> [ 4248.579019] ... acquired at:
> [ 4248.579019] [<ffffffff8111a43c>] check_usage+0x1ac/0x1d0
> [ 4248.579019] [<ffffffff8111a4d1>] check_irq_usage+0x71/0xf0
> [ 4248.579019] [<ffffffff8111b7bb>] check_prev_add+0x16b/0x4d0
> [ 4248.579019] [<ffffffff8111c1be>] validate_chain+0x69e/0x790
> [ 4248.579019] [<ffffffff8111c6d3>] __lock_acquire+0x423/0x4c0
> [ 4248.579019] [<ffffffff8111c84c>] lock_acquire+0xdc/0x120
> [ 4248.579019] [<ffffffff8289823b>] _raw_spin_lock+0x3b/0x70
> [ 4248.579019] [<ffffffff81089476>] pgd_free+0x26/0xa0
> [ 4248.579019] [<ffffffff810b4d81>] __mmdrop+0x31/0xf0
> [ 4248.579019] [<ffffffff8107cacc>] apf_task_wake_one+0x4c/0x90
> [ 4248.579019] [<ffffffff8107cdf5>] kvm_async_pf_task_wake+0x75/0x130
> [ 4248.579019] [<ffffffff8107d1f6>] do_async_page_fault+0x96/0xb0
> [ 4248.579019] [<ffffffff82899695>] async_page_fault+0x25/0x30
> [ 4248.579019] [<ffffffff8289695e>] __schedule+0xce/0x880
> [ 4248.579019] [<ffffffff82897275>] schedule+0x55/0x60
> [ 4248.579019] [<ffffffff8115a3e2>] rcu_torture_reader+0x2a2/0x380
> [ 4248.579019] [<ffffffff810dada2>] kthread+0xb2/0xc0
> [ 4248.579019] [<ffffffff8289b0b4>] kernel_thread_helper+0x4/0x10
> [ 4248.579019]
> [ 4248.579019]
> [ 4248.579019] stack backtrace:
> [ 4248.579019] Pid: 1615, comm: rcu_torture_rea Tainted: G W 3.4.0-rc4-next-20120424-sasha #94
> [ 4248.579019] Call Trace:
> [ 4248.579019] [<ffffffff81119a82>] print_bad_irq_dependency+0x2f2/0x310
> [ 4248.579019] [<ffffffff8111a43c>] check_usage+0x1ac/0x1d0
> [ 4248.579019] [<ffffffff8107e4a8>] ? pvclock_clocksource_read+0x58/0xd0
> [ 4248.579019] [<ffffffff8111a4d1>] check_irq_usage+0x71/0xf0
> [ 4248.579019] [<ffffffff8111b7bb>] check_prev_add+0x16b/0x4d0
> [ 4248.579019] [<ffffffff8107d4d6>] ? kvm_clock_read+0x46/0x80
> [ 4248.579019] [<ffffffff8111c1be>] validate_chain+0x69e/0x790
> [ 4248.579019] [<ffffffff810f10a8>] ? sched_clock_cpu+0x108/0x120
> [ 4248.579019] [<ffffffff8111c6d3>] __lock_acquire+0x423/0x4c0
> [ 4248.579019] [<ffffffff8111c84c>] lock_acquire+0xdc/0x120
> [ 4248.579019] [<ffffffff81089476>] ? pgd_free+0x26/0xa0
> [ 4248.579019] [<ffffffff8289823b>] _raw_spin_lock+0x3b/0x70
> [ 4248.579019] [<ffffffff81089476>] ? pgd_free+0x26/0xa0
> [ 4248.579019] [<ffffffff81089476>] pgd_free+0x26/0xa0
> [ 4248.579019] [<ffffffff810b4d81>] __mmdrop+0x31/0xf0
> [ 4248.579019] [<ffffffff8107cacc>] apf_task_wake_one+0x4c/0x90
> [ 4248.579019] [<ffffffff8107cdf5>] kvm_async_pf_task_wake+0x75/0x130
> [ 4248.579019] [<ffffffff8107d1f6>] do_async_page_fault+0x96/0xb0
> [ 4248.579019] [<ffffffff82899695>] async_page_fault+0x25/0x30
> [ 4248.579019] [<ffffffff8111714d>] ? lock_is_held+0xfd/0x110
> [ 4248.579019] [<ffffffff8289695e>] __schedule+0xce/0x880
> [ 4248.579019] [<ffffffff82897275>] schedule+0x55/0x60
> [ 4248.579019] [<ffffffff8115a3e2>] rcu_torture_reader+0x2a2/0x380
> [ 4248.579019] [<ffffffff81158f20>] ? T.842+0x50/0x50
> [ 4248.579019] [<ffffffff8115a140>] ? rcu_torture_read_unlock+0x60/0x60
> [ 4248.579019] [<ffffffff810dada2>] kthread+0xb2/0xc0
> [ 4248.579019] [<ffffffff8289b0b4>] kernel_thread_helper+0x4/0x10
> [ 4248.579019] [<ffffffff810e820d>] ? finish_task_switch+0x8d/0x110
> [ 4248.579019] [<ffffffff828993b4>] ? retint_restore_args+0x13/0x13
> [ 4248.579019] [<ffffffff810dacf0>] ? __init_kthread_worker+0x70/0x70
> [ 4248.579019] [<ffffffff8289b0b0>] ? gs_change+0x13/0x13

--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majo...@vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/

[Sasha Levin]

On Fri, Apr 27, 2012 at 5:23 PM, Paul E. McKenney
<pau...@linux.vnet.ibm.com> wrote:
>
> On Thu, Apr 26, 2012 at 04:50:35PM +0200, Sasha Levin wrote:
> > Hi all.
> >
> > I got a "INFO: SOFTIRQ-safe -> SOFTIRQ-unsafe lock order detected"
> > warning
> > while running LTP inside a KVM guest using the recent -next kernel.
> >
> > It seems that it was initially originated from rcu_torture_rea(), but I
> > don't think that it's a problem with RCU itself (I'll cc. Paul just in
> > case). RCU torture was indeed configured to run during the testing.
>
> I might be missing something, but I don't see how RCU is involved in
> this one.

It happened due to a rcutorture thread:

[ 4248.579019] rcu_torture_rea/1615 [HC0[0]:SC0[0]:HE0:SE1] is trying
to acquire:
[ 4248.579019] (pgd_lock){+.+...}, at: [<ffffffff81089476>] pgd_free+0x26/0xa0

rcutorture has never triggered anything on my tests up until now, and
is probably not responsible for the actual warning in this case, just
figured I should cc you in case I'm wrong :)

[Paul E. McKenney]

On Fri, Apr 27, 2012 at 08:05:15PM +0200, Sasha Levin wrote:
> On Fri, Apr 27, 2012 at 5:23 PM, Paul E. McKenney
> <pau...@linux.vnet.ibm.com> wrote:
> >
> > On Thu, Apr 26, 2012 at 04:50:35PM +0200, Sasha Levin wrote:
> > > Hi all.
> > >
> > > I got a "INFO: SOFTIRQ-safe -> SOFTIRQ-unsafe lock order detected"
> > > warning
> > > while running LTP inside a KVM guest using the recent -next kernel.
> > >
> > > It seems that it was initially originated from rcu_torture_rea(), but I
> > > don't think that it's a problem with RCU itself (I'll cc. Paul just in
> > > case). RCU torture was indeed configured to run during the testing.
> >
> > I might be missing something, but I don't see how RCU is involved in
> > this one.
>
> It happened due to a rcutorture thread:
>
> [ 4248.579019] rcu_torture_rea/1615 [HC0[0]:SC0[0]:HE0:SE1] is trying
> to acquire:
> [ 4248.579019] (pgd_lock){+.+...}, at: [<ffffffff81089476>] pgd_free+0x26/0xa0
>
> rcutorture has never triggered anything on my tests up until now, and
> is probably not responsible for the actual warning in this case, just
> figured I should cc you in case I'm wrong :)

No worries!

I only responded on the off-chance that everyone else was waiting on
me to identify the bug and provide the fix. ;-)

Thanx, Paul

[Gleb Natapov]

On Thu, Apr 26, 2012 at 04:50:35PM +0200, Sasha Levin wrote:
> Hi all.
>
> I got a "INFO: SOFTIRQ-safe -> SOFTIRQ-unsafe lock order detected" warning
> while running LTP inside a KVM guest using the recent -next kernel.
>
> It seems that it was initially originated from rcu_torture_rea(), but I
> don't think that it's a problem with RCU itself (I'll cc. Paul just in
> case). RCU torture was indeed configured to run during the testing.
>

> The output is attached since it's a bit long.
>

The reason is mmdrop() called from irq context and in reality we do
not need to hold reference to mm at all, so the patch below drops it.
---

KVM: Do not take reference to mm during async #PF

It turned to be totally unneeded. The reason the code was introduced is
so that KVM can prefault swapped in page, but prefault can fail even
if mm is pinned since page table can change anyway. KVM handles this
situation correctly though and does not inject spurious page faults.

Signed-off-by: Gleb Natapov <gl...@redhat.com>
diff --git a/arch/x86/kernel/kvm.c b/arch/x86/kernel/kvm.c
index b8ba6e4..e554e5a 100644
--- a/arch/x86/kernel/kvm.c
+++ b/arch/x86/kernel/kvm.c
@@ -79,7 +79,6 @@ struct kvm_task_sleep_node {
u32 token;
int cpu;
bool halted;
- struct mm_struct *mm;
};

static struct kvm_task_sleep_head {
@@ -126,9 +125,7 @@ void kvm_async_pf_task_wait(u32 token)

n.token = token;
n.cpu = smp_processor_id();
- n.mm = current->active_mm;
n.halted = idle || preempt_count() > 1;
- atomic_inc(&n.mm->mm_count);
init_waitqueue_head(&n.wq);
hlist_add_head(&n.link, &b->list);
spin_unlock(&b->lock);
@@ -161,9 +158,6 @@ EXPORT_SYMBOL_GPL(kvm_async_pf_task_wait);
static void apf_task_wake_one(struct kvm_task_sleep_node *n)
{
hlist_del_init(&n->link);
- if (!n->mm)
- return;
- mmdrop(n->mm);
if (n->halted)
smp_send_reschedule(n->cpu);
else if (waitqueue_active(&n->wq))
@@ -207,7 +201,7 @@ again:
* async PF was not yet handled.
* Add dummy entry for the token.
*/
- n = kmalloc(sizeof(*n), GFP_ATOMIC);
+ n = kzalloc(sizeof(*n), GFP_ATOMIC);
if (!n) {
/*
* Allocation failed! Busy wait while other cpu
@@ -219,7 +213,6 @@ again:
}
n->token = token;
n->cpu = smp_processor_id();
- n->mm = NULL;
init_waitqueue_head(&n->wq);
hlist_add_head(&n->link, &b->list);
} else

--
Gleb.

[Avi Kivity]

On 05/02/2012 03:04 PM, Gleb Natapov wrote:
> KVM: Do not take reference to mm during async #PF
>
> It turned to be totally unneeded. The reason the code was introduced is
> so that KVM can prefault swapped in page, but prefault can fail even
> if mm is pinned since page table can change anyway. KVM handles this
> situation correctly though and does not inject spurious page faults.
>
>

Applied to master for 3.4, thanks.

--
error compiling committee.c: too many arguments to function