* [PATCH] mm/mmu_notifier: silence PROVE_RCU_LIST warnings
@ 2020-03-17 17:56 Qian Cai
2020-03-17 18:06 ` Paul E. McKenney
2020-03-18 12:13 ` Jason Gunthorpe
0 siblings, 2 replies; 3+ messages in thread
From: Qian Cai @ 2020-03-17 17:56 UTC (permalink / raw)
To: akpm; +Cc: jgg, paulmck, linux-mm, linux-kernel, Qian Cai
It is safe to traverse mm->notifier_subscriptions->list either under
SRCU read lock or mm->notifier_subscriptions->lock using
hlist_for_each_entry_rcu(). Silence the PROVE_RCU_LIST false positives,
for example,
WARNING: suspicious RCU usage
-----------------------------
mm/mmu_notifier.c:484 RCU-list traversed in non-reader section!!
other info that might help us debug this:
rcu_scheduler_active = 2, debug_locks = 1
3 locks held by libvirtd/802:
#0: ffff9321e3f58148 (&mm->mmap_sem#2){++++}, at: do_mprotect_pkey+0xe1/0x3e0
#1: ffffffff91ae6160 (mmu_notifier_invalidate_range_start){+.+.}, at: change_p4d_range+0x5fa/0x800
#2: ffffffff91ae6e08 (srcu){....}, at: __mmu_notifier_invalidate_range_start+0x178/0x460
stack backtrace:
CPU: 7 PID: 802 Comm: libvirtd Tainted: G I 5.6.0-rc6-next-20200317+ #2
Hardware name: HP ProLiant BL460c Gen8, BIOS I31 11/02/2014
Call Trace:
dump_stack+0xa4/0xfe
lockdep_rcu_suspicious+0xeb/0xf5
__mmu_notifier_invalidate_range_start+0x3ff/0x460
change_p4d_range+0x746/0x800
change_protection+0x1df/0x300
mprotect_fixup+0x245/0x3e0
do_mprotect_pkey+0x23b/0x3e0
__x64_sys_mprotect+0x51/0x70
do_syscall_64+0x91/0xae8
entry_SYSCALL_64_after_hwframe+0x49/0xb3
Signed-off-by: Qian Cai <cai@lca.pw>
---
mm/mmu_notifier.c | 27 ++++++++++++++++++---------
1 file changed, 18 insertions(+), 9 deletions(-)
diff --git a/mm/mmu_notifier.c b/mm/mmu_notifier.c
index ef3973a5d34a..06852b896fa6 100644
--- a/mm/mmu_notifier.c
+++ b/mm/mmu_notifier.c
@@ -307,7 +307,8 @@ static void mn_hlist_release(struct mmu_notifier_subscriptions *subscriptions,
* ->release returns.
*/
id = srcu_read_lock(&srcu);
- hlist_for_each_entry_rcu(subscription, &subscriptions->list, hlist)
+ hlist_for_each_entry_rcu(subscription, &subscriptions->list, hlist,
+ srcu_read_lock_held(&srcu))
/*
* If ->release runs before mmu_notifier_unregister it must be
* handled, as it's the only way for the driver to flush all
@@ -370,7 +371,8 @@ int __mmu_notifier_clear_flush_young(struct mm_struct *mm,
id = srcu_read_lock(&srcu);
hlist_for_each_entry_rcu(subscription,
- &mm->notifier_subscriptions->list, hlist) {
+ &mm->notifier_subscriptions->list, hlist,
+ srcu_read_lock_held(&srcu)) {
if (subscription->ops->clear_flush_young)
young |= subscription->ops->clear_flush_young(
subscription, mm, start, end);
@@ -389,7 +391,8 @@ int __mmu_notifier_clear_young(struct mm_struct *mm,
id = srcu_read_lock(&srcu);
hlist_for_each_entry_rcu(subscription,
- &mm->notifier_subscriptions->list, hlist) {
+ &mm->notifier_subscriptions->list, hlist,
+ srcu_read_lock_held(&srcu)) {
if (subscription->ops->clear_young)
young |= subscription->ops->clear_young(subscription,
mm, start, end);
@@ -407,7 +410,8 @@ int __mmu_notifier_test_young(struct mm_struct *mm,
id = srcu_read_lock(&srcu);
hlist_for_each_entry_rcu(subscription,
- &mm->notifier_subscriptions->list, hlist) {
+ &mm->notifier_subscriptions->list, hlist,
+ srcu_read_lock_held(&srcu)) {
if (subscription->ops->test_young) {
young = subscription->ops->test_young(subscription, mm,
address);
@@ -428,7 +432,8 @@ void __mmu_notifier_change_pte(struct mm_struct *mm, unsigned long address,
id = srcu_read_lock(&srcu);
hlist_for_each_entry_rcu(subscription,
- &mm->notifier_subscriptions->list, hlist) {
+ &mm->notifier_subscriptions->list, hlist,
+ srcu_read_lock_held(&srcu)) {
if (subscription->ops->change_pte)
subscription->ops->change_pte(subscription, mm, address,
pte);
@@ -476,7 +481,8 @@ static int mn_hlist_invalidate_range_start(
int id;
id = srcu_read_lock(&srcu);
- hlist_for_each_entry_rcu(subscription, &subscriptions->list, hlist) {
+ hlist_for_each_entry_rcu(subscription, &subscriptions->list, hlist,
+ srcu_read_lock_held(&srcu)) {
const struct mmu_notifier_ops *ops = subscription->ops;
if (ops->invalidate_range_start) {
@@ -528,7 +534,8 @@ mn_hlist_invalidate_end(struct mmu_notifier_subscriptions *subscriptions,
int id;
id = srcu_read_lock(&srcu);
- hlist_for_each_entry_rcu(subscription, &subscriptions->list, hlist) {
+ hlist_for_each_entry_rcu(subscription, &subscriptions->list, hlist,
+ srcu_read_lock_held(&srcu)) {
/*
* Call invalidate_range here too to avoid the need for the
* subsystem of having to register an invalidate_range_end
@@ -582,7 +589,8 @@ void __mmu_notifier_invalidate_range(struct mm_struct *mm,
id = srcu_read_lock(&srcu);
hlist_for_each_entry_rcu(subscription,
- &mm->notifier_subscriptions->list, hlist) {
+ &mm->notifier_subscriptions->list, hlist,
+ srcu_read_lock_held(&srcu)) {
if (subscription->ops->invalidate_range)
subscription->ops->invalidate_range(subscription, mm,
start, end);
@@ -714,7 +722,8 @@ find_get_mmu_notifier(struct mm_struct *mm, const struct mmu_notifier_ops *ops)
spin_lock(&mm->notifier_subscriptions->lock);
hlist_for_each_entry_rcu(subscription,
- &mm->notifier_subscriptions->list, hlist) {
+ &mm->notifier_subscriptions->list, hlist,
+ lockdep_is_held(&mm->notifier_subscriptions->lock)) {
if (subscription->ops != ops)
continue;
--
2.21.0 (Apple Git-122.2)
^ permalink raw reply related [flat|nested] 3+ messages in thread
* Re: [PATCH] mm/mmu_notifier: silence PROVE_RCU_LIST warnings
2020-03-17 17:56 [PATCH] mm/mmu_notifier: silence PROVE_RCU_LIST warnings Qian Cai
@ 2020-03-17 18:06 ` Paul E. McKenney
2020-03-18 12:13 ` Jason Gunthorpe
1 sibling, 0 replies; 3+ messages in thread
From: Paul E. McKenney @ 2020-03-17 18:06 UTC (permalink / raw)
To: Qian Cai; +Cc: akpm, jgg, linux-mm, linux-kernel
On Tue, Mar 17, 2020 at 01:56:40PM -0400, Qian Cai wrote:
> It is safe to traverse mm->notifier_subscriptions->list either under
> SRCU read lock or mm->notifier_subscriptions->lock using
> hlist_for_each_entry_rcu(). Silence the PROVE_RCU_LIST false positives,
> for example,
>
> WARNING: suspicious RCU usage
> -----------------------------
> mm/mmu_notifier.c:484 RCU-list traversed in non-reader section!!
>
> other info that might help us debug this:
>
> rcu_scheduler_active = 2, debug_locks = 1
> 3 locks held by libvirtd/802:
> #0: ffff9321e3f58148 (&mm->mmap_sem#2){++++}, at: do_mprotect_pkey+0xe1/0x3e0
> #1: ffffffff91ae6160 (mmu_notifier_invalidate_range_start){+.+.}, at: change_p4d_range+0x5fa/0x800
> #2: ffffffff91ae6e08 (srcu){....}, at: __mmu_notifier_invalidate_range_start+0x178/0x460
>
> stack backtrace:
> CPU: 7 PID: 802 Comm: libvirtd Tainted: G I 5.6.0-rc6-next-20200317+ #2
> Hardware name: HP ProLiant BL460c Gen8, BIOS I31 11/02/2014
> Call Trace:
> dump_stack+0xa4/0xfe
> lockdep_rcu_suspicious+0xeb/0xf5
> __mmu_notifier_invalidate_range_start+0x3ff/0x460
> change_p4d_range+0x746/0x800
> change_protection+0x1df/0x300
> mprotect_fixup+0x245/0x3e0
> do_mprotect_pkey+0x23b/0x3e0
> __x64_sys_mprotect+0x51/0x70
> do_syscall_64+0x91/0xae8
> entry_SYSCALL_64_after_hwframe+0x49/0xb3
>
> Signed-off-by: Qian Cai <cai@lca.pw>
From an RCU perspective:
Reviewed-by: Paul E. McKenney <paulmck@kernel.org>
> ---
> mm/mmu_notifier.c | 27 ++++++++++++++++++---------
> 1 file changed, 18 insertions(+), 9 deletions(-)
>
> diff --git a/mm/mmu_notifier.c b/mm/mmu_notifier.c
> index ef3973a5d34a..06852b896fa6 100644
> --- a/mm/mmu_notifier.c
> +++ b/mm/mmu_notifier.c
> @@ -307,7 +307,8 @@ static void mn_hlist_release(struct mmu_notifier_subscriptions *subscriptions,
> * ->release returns.
> */
> id = srcu_read_lock(&srcu);
> - hlist_for_each_entry_rcu(subscription, &subscriptions->list, hlist)
> + hlist_for_each_entry_rcu(subscription, &subscriptions->list, hlist,
> + srcu_read_lock_held(&srcu))
> /*
> * If ->release runs before mmu_notifier_unregister it must be
> * handled, as it's the only way for the driver to flush all
> @@ -370,7 +371,8 @@ int __mmu_notifier_clear_flush_young(struct mm_struct *mm,
>
> id = srcu_read_lock(&srcu);
> hlist_for_each_entry_rcu(subscription,
> - &mm->notifier_subscriptions->list, hlist) {
> + &mm->notifier_subscriptions->list, hlist,
> + srcu_read_lock_held(&srcu)) {
> if (subscription->ops->clear_flush_young)
> young |= subscription->ops->clear_flush_young(
> subscription, mm, start, end);
> @@ -389,7 +391,8 @@ int __mmu_notifier_clear_young(struct mm_struct *mm,
>
> id = srcu_read_lock(&srcu);
> hlist_for_each_entry_rcu(subscription,
> - &mm->notifier_subscriptions->list, hlist) {
> + &mm->notifier_subscriptions->list, hlist,
> + srcu_read_lock_held(&srcu)) {
> if (subscription->ops->clear_young)
> young |= subscription->ops->clear_young(subscription,
> mm, start, end);
> @@ -407,7 +410,8 @@ int __mmu_notifier_test_young(struct mm_struct *mm,
>
> id = srcu_read_lock(&srcu);
> hlist_for_each_entry_rcu(subscription,
> - &mm->notifier_subscriptions->list, hlist) {
> + &mm->notifier_subscriptions->list, hlist,
> + srcu_read_lock_held(&srcu)) {
> if (subscription->ops->test_young) {
> young = subscription->ops->test_young(subscription, mm,
> address);
> @@ -428,7 +432,8 @@ void __mmu_notifier_change_pte(struct mm_struct *mm, unsigned long address,
>
> id = srcu_read_lock(&srcu);
> hlist_for_each_entry_rcu(subscription,
> - &mm->notifier_subscriptions->list, hlist) {
> + &mm->notifier_subscriptions->list, hlist,
> + srcu_read_lock_held(&srcu)) {
> if (subscription->ops->change_pte)
> subscription->ops->change_pte(subscription, mm, address,
> pte);
> @@ -476,7 +481,8 @@ static int mn_hlist_invalidate_range_start(
> int id;
>
> id = srcu_read_lock(&srcu);
> - hlist_for_each_entry_rcu(subscription, &subscriptions->list, hlist) {
> + hlist_for_each_entry_rcu(subscription, &subscriptions->list, hlist,
> + srcu_read_lock_held(&srcu)) {
> const struct mmu_notifier_ops *ops = subscription->ops;
>
> if (ops->invalidate_range_start) {
> @@ -528,7 +534,8 @@ mn_hlist_invalidate_end(struct mmu_notifier_subscriptions *subscriptions,
> int id;
>
> id = srcu_read_lock(&srcu);
> - hlist_for_each_entry_rcu(subscription, &subscriptions->list, hlist) {
> + hlist_for_each_entry_rcu(subscription, &subscriptions->list, hlist,
> + srcu_read_lock_held(&srcu)) {
> /*
> * Call invalidate_range here too to avoid the need for the
> * subsystem of having to register an invalidate_range_end
> @@ -582,7 +589,8 @@ void __mmu_notifier_invalidate_range(struct mm_struct *mm,
>
> id = srcu_read_lock(&srcu);
> hlist_for_each_entry_rcu(subscription,
> - &mm->notifier_subscriptions->list, hlist) {
> + &mm->notifier_subscriptions->list, hlist,
> + srcu_read_lock_held(&srcu)) {
> if (subscription->ops->invalidate_range)
> subscription->ops->invalidate_range(subscription, mm,
> start, end);
> @@ -714,7 +722,8 @@ find_get_mmu_notifier(struct mm_struct *mm, const struct mmu_notifier_ops *ops)
>
> spin_lock(&mm->notifier_subscriptions->lock);
> hlist_for_each_entry_rcu(subscription,
> - &mm->notifier_subscriptions->list, hlist) {
> + &mm->notifier_subscriptions->list, hlist,
> + lockdep_is_held(&mm->notifier_subscriptions->lock)) {
> if (subscription->ops != ops)
> continue;
>
> --
> 2.21.0 (Apple Git-122.2)
>
^ permalink raw reply [flat|nested] 3+ messages in thread
* Re: [PATCH] mm/mmu_notifier: silence PROVE_RCU_LIST warnings
2020-03-17 17:56 [PATCH] mm/mmu_notifier: silence PROVE_RCU_LIST warnings Qian Cai
2020-03-17 18:06 ` Paul E. McKenney
@ 2020-03-18 12:13 ` Jason Gunthorpe
1 sibling, 0 replies; 3+ messages in thread
From: Jason Gunthorpe @ 2020-03-18 12:13 UTC (permalink / raw)
To: Qian Cai; +Cc: akpm, paulmck, linux-mm, linux-kernel
On Tue, Mar 17, 2020 at 01:56:40PM -0400, Qian Cai wrote:
> It is safe to traverse mm->notifier_subscriptions->list either under
> SRCU read lock or mm->notifier_subscriptions->lock using
> hlist_for_each_entry_rcu(). Silence the PROVE_RCU_LIST false positives,
> for example,
>
> WARNING: suspicious RCU usage
> -----------------------------
> mm/mmu_notifier.c:484 RCU-list traversed in non-reader section!!
>
> other info that might help us debug this:
>
> rcu_scheduler_active = 2, debug_locks = 1
> 3 locks held by libvirtd/802:
> #0: ffff9321e3f58148 (&mm->mmap_sem#2){++++}, at: do_mprotect_pkey+0xe1/0x3e0
> #1: ffffffff91ae6160 (mmu_notifier_invalidate_range_start){+.+.}, at: change_p4d_range+0x5fa/0x800
> #2: ffffffff91ae6e08 (srcu){....}, at: __mmu_notifier_invalidate_range_start+0x178/0x460
>
> stack backtrace:
> CPU: 7 PID: 802 Comm: libvirtd Tainted: G I 5.6.0-rc6-next-20200317+ #2
> Hardware name: HP ProLiant BL460c Gen8, BIOS I31 11/02/2014
> Call Trace:
> dump_stack+0xa4/0xfe
> lockdep_rcu_suspicious+0xeb/0xf5
> __mmu_notifier_invalidate_range_start+0x3ff/0x460
> change_p4d_range+0x746/0x800
> change_protection+0x1df/0x300
> mprotect_fixup+0x245/0x3e0
> do_mprotect_pkey+0x23b/0x3e0
> __x64_sys_mprotect+0x51/0x70
> do_syscall_64+0x91/0xae8
> entry_SYSCALL_64_after_hwframe+0x49/0xb3
>
> Signed-off-by: Qian Cai <cai@lca.pw>
> ---
> mm/mmu_notifier.c | 27 ++++++++++++++++++---------
> 1 file changed, 18 insertions(+), 9 deletions(-)
Looks right to me
Reviewed-by: Jason Gunthorpe <jgg@mellanox.com>
This was missed during testing because we don't run with
PROVE_RCU_LIST enabled
Jason
^ permalink raw reply [flat|nested] 3+ messages in thread
end of thread, other threads:[~2020-03-18 12:13 UTC | newest]
Thread overview: 3+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2020-03-17 17:56 [PATCH] mm/mmu_notifier: silence PROVE_RCU_LIST warnings Qian Cai
2020-03-17 18:06 ` Paul E. McKenney
2020-03-18 12:13 ` Jason Gunthorpe
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).