Skip to contentSkip to conversation

net/sched: sch_taprio: fix possible use-after-free

PATCH Applied Eric Dumazet 6 participants 9 messages Jan 13, 2023 – Jan 18, 2023

Subsystems (4)
lore
Share & tools
lore
#0 lore raw

PATCH 1/1 net/sched: sch_taprio: fix possible use-after-free

syzbot reported a nasty crash [1] in net_tx_action() which made little sense until we got a repro.

This repro installs a taprio qdisc, but providing an invalid TCA_RATE attribute.

qdisc_create() has to destroy the just initialized taprio qdisc, and taprio_destroy() is called.

However, the hrtimer used by taprio had already fired, therefore advance_sched() called __netif_schedule().

Then net_tx_action was trying to use a destroyed qdisc.

We can not undo the __netif_schedule(), so we must wait until one cpu serviced the qdisc before we can proceed.

Many thanks to Alexander Potapenko for his help.

[1]

BUG: KMSAN: uninit-value in queued_spin_trylock include/asm-generic/qspinlock.h:94 [inline]
BUG: KMSAN: uninit-value in do_raw_spin_trylock include/linux/spinlock.h:191 [inline]
BUG: KMSAN: uninit-value in __raw_spin_trylock include/linux/spinlock_api_smp.h:89 [inline]
BUG: KMSAN: uninit-value in _raw_spin_trylock+0x92/0xa0 kernel/locking/spinlock.c:138
 queued_spin_trylock include/asm-generic/qspinlock.h:94 [inline]
 do_raw_spin_trylock include/linux/spinlock.h:191 [inline]
 __raw_spin_trylock include/linux/spinlock_api_smp.h:89 [inline]
 _raw_spin_trylock+0x92/0xa0 kernel/locking/spinlock.c:138
 spin_trylock include/linux/spinlock.h:359 [inline]
 qdisc_run_begin include/net/sch_generic.h:187 [inline]
 qdisc_run+0xee/0x540 include/net/pkt_sched.h:125
 net_tx_action+0x77c/0x9a0 net/core/dev.c:5086
 __do_softirq+0x1cc/0x7fb kernel/softirq.c:571
 run_ksoftirqd+0x2c/0x50 kernel/softirq.c:934
 smpboot_thread_fn+0x554/0x9f0 kernel/smpboot.c:164
 kthread+0x31b/0x430 kernel/kthread.c:376
 ret_from_fork+0x1f/0x30

Uninit was created at:

 slab_post_alloc_hook mm/slab.h:732 [inline]
 slab_alloc_node mm/slub.c:3258 [inline]
 __kmalloc_node_track_caller+0x814/0x1250 mm/slub.c:4970
 kmalloc_reserve net/core/skbuff.c:358 [inline]
 __alloc_skb+0x346/0xcf0 net/core/skbuff.c:430
 alloc_skb include/linux/skbuff.h:1257 [inline]
 nlmsg_new include/net/netlink.h:953 [inline]
 netlink_ack+0x5f3/0x12b0 net/netlink/af_netlink.c:2436
 netlink_rcv_skb+0x55d/0x6c0 net/netlink/af_netlink.c:2507
 rtnetlink_rcv+0x30/0x40 net/core/rtnetlink.c:6108
 netlink_unicast_kernel net/netlink/af_netlink.c:1319 [inline]
 netlink_unicast+0xf3b/0x1270 net/netlink/af_netlink.c:1345
 netlink_sendmsg+0x1288/0x1440 net/netlink/af_netlink.c:1921
 sock_sendmsg_nosec net/socket.c:714 [inline]
 sock_sendmsg net/socket.c:734 [inline]
 ____sys_sendmsg+0xabc/0xe90 net/socket.c:2482
 ___sys_sendmsg+0x2a1/0x3f0 net/socket.c:2536
 __sys_sendmsg net/socket.c:2565 [inline]
 __do_sys_sendmsg net/socket.c:2574 [inline]
 __se_sys_sendmsg net/socket.c:2572 [inline]
 __x64_sys_sendmsg+0x367/0x540 net/socket.c:2572
 do_syscall_x64 arch/x86/entry/common.c:50 [inline]
 do_syscall_64+0x3d/0xb0 arch/x86/entry/common.c:80
 entry_SYSCALL_64_after_hwframe+0x63/0xcd

CPU: 0 PID: 13 Comm: ksoftirqd/0 Not tainted 6.0.0-rc2-syzkaller-47461-gac3859c02d7f #0
Hardware name: Google Google Compute Engine/Google Compute Engine, BIOS Google 07/22/2022
Fixes 5a781ccbd19e ("tc: Add support for configuring the taprio scheduler")
Cc Alexander Potapenko <[email protected]>
Cc Vinicius Costa Gomes <[email protected]>
---
 include/net/sch_generic.h | 7 +++++++
 net/sched/sch_taprio.c    | 3 +++
 2 files changed, 10 insertions(+)
include/net/sch_generic.h +7 -0
index d5517719af4ef22282f0a15b132f8e8a07ae4179..af4aa66aaa4eba8f2eacdd00bc8fef31165c6a90 100644
@@ -1288,4 +1288,11 @@ void mq_change_real_num_tx(struct Qdisc *sch, unsigned int new_real_tx);
 
 int sch_frag_xmit_hook(struct sk_buff *skb, int (*xmit)(struct sk_buff *skb));
 
+/* Make sure qdisc is no longer in SCHED state. */
+static inline void qdisc_synchronize(const struct Qdisc *q)
+{
+	while (test_bit(__QDISC_STATE_SCHED, &q->state))
+		msleep(1);
+}
+
 #endif
net/sched/sch_taprio.c +3 -0
index 570389f6cdd7dbab5749dc06d886555305cbf623..9a11a499ea2df8d18c9c062496fdcbcf5a861391 100644
@@ -1700,6 +1700,8 @@ static void taprio_reset(struct Qdisc *sch)
 	int i;
 
 	hrtimer_cancel(&q->advance_timer);
+	qdisc_synchronize(sch);
+
 	if (q->qdiscs) {
 		for (i = 0; i < dev->num_tx_queues; i++)
 			if (q->qdiscs[i])
@@ -1720,6 +1722,7 @@ static void taprio_destroy(struct Qdisc *sch)
 	 * happens in qdisc_create(), after taprio_init() has been called.
 	 */
 	hrtimer_cancel(&q->advance_timer);
+	qdisc_synchronize(sch);
 
 	taprio_disable_offload(dev, q, NULL);
 
Signature · 1 line
2.39.0.314.g84b9a713c41-goog
#1 lore raw

Hi,

Quoted 98 lines · Eric Dumazet

syzbot reported a nasty crash [1] in net_tx_action() which made little sense until we got a repro.

This repro installs a taprio qdisc, but providing an invalid TCA_RATE attribute.

qdisc_create() has to destroy the just initialized taprio qdisc, and taprio_destroy() is called.

However, the hrtimer used by taprio had already fired, therefore advance_sched() called __netif_schedule().

Then net_tx_action was trying to use a destroyed qdisc.

We can not undo the __netif_schedule(), so we must wait until one cpu serviced the qdisc before we can proceed.

Many thanks to Alexander Potapenko for his help.

[1]

BUG: KMSAN: uninit-value in queued_spin_trylock include/asm-generic/qspinlock.h:94 [inline]
BUG: KMSAN: uninit-value in do_raw_spin_trylock include/linux/spinlock.h:191 [inline]
BUG: KMSAN: uninit-value in __raw_spin_trylock include/linux/spinlock_api_smp.h:89 [inline]
BUG: KMSAN: uninit-value in _raw_spin_trylock+0x92/0xa0 kernel/locking/spinlock.c:138
 queued_spin_trylock include/asm-generic/qspinlock.h:94 [inline]
 do_raw_spin_trylock include/linux/spinlock.h:191 [inline]
 __raw_spin_trylock include/linux/spinlock_api_smp.h:89 [inline]
 _raw_spin_trylock+0x92/0xa0 kernel/locking/spinlock.c:138
 spin_trylock include/linux/spinlock.h:359 [inline]
 qdisc_run_begin include/net/sch_generic.h:187 [inline]
 qdisc_run+0xee/0x540 include/net/pkt_sched.h:125
 net_tx_action+0x77c/0x9a0 net/core/dev.c:5086
 __do_softirq+0x1cc/0x7fb kernel/softirq.c:571
 run_ksoftirqd+0x2c/0x50 kernel/softirq.c:934
 smpboot_thread_fn+0x554/0x9f0 kernel/smpboot.c:164
 kthread+0x31b/0x430 kernel/kthread.c:376
 ret_from_fork+0x1f/0x30

Uninit was created at:

 slab_post_alloc_hook mm/slab.h:732 [inline]
 slab_alloc_node mm/slub.c:3258 [inline]
 __kmalloc_node_track_caller+0x814/0x1250 mm/slub.c:4970
 kmalloc_reserve net/core/skbuff.c:358 [inline]
 __alloc_skb+0x346/0xcf0 net/core/skbuff.c:430
 alloc_skb include/linux/skbuff.h:1257 [inline]
 nlmsg_new include/net/netlink.h:953 [inline]
 netlink_ack+0x5f3/0x12b0 net/netlink/af_netlink.c:2436
 netlink_rcv_skb+0x55d/0x6c0 net/netlink/af_netlink.c:2507
 rtnetlink_rcv+0x30/0x40 net/core/rtnetlink.c:6108
 netlink_unicast_kernel net/netlink/af_netlink.c:1319 [inline]
 netlink_unicast+0xf3b/0x1270 net/netlink/af_netlink.c:1345
 netlink_sendmsg+0x1288/0x1440 net/netlink/af_netlink.c:1921
 sock_sendmsg_nosec net/socket.c:714 [inline]
 sock_sendmsg net/socket.c:734 [inline]
 ____sys_sendmsg+0xabc/0xe90 net/socket.c:2482
 ___sys_sendmsg+0x2a1/0x3f0 net/socket.c:2536
 __sys_sendmsg net/socket.c:2565 [inline]
 __do_sys_sendmsg net/socket.c:2574 [inline]
 __se_sys_sendmsg net/socket.c:2572 [inline]
 __x64_sys_sendmsg+0x367/0x540 net/socket.c:2572
 do_syscall_x64 arch/x86/entry/common.c:50 [inline]
 do_syscall_64+0x3d/0xb0 arch/x86/entry/common.c:80
 entry_SYSCALL_64_after_hwframe+0x63/0xcd

CPU: 0 PID: 13 Comm: ksoftirqd/0 Not tainted 6.0.0-rc2-syzkaller-47461-gac3859c02d7f #0
Hardware name: Google Google Compute Engine/Google Compute Engine, BIOS Google 07/22/2022
Fixes 5a781ccbd19e ("tc: Add support for configuring the taprio scheduler")
Cc Alexander Potapenko <[email protected]>
Cc Vinicius Costa Gomes <[email protected]>
---
 include/net/sch_generic.h | 7 +++++++
 net/sched/sch_taprio.c    | 3 +++
 2 files changed, 10 insertions(+)
include/net/sch_generic.h +7 -0
index d5517719af4ef22282f0a15b132f8e8a07ae4179..af4aa66aaa4eba8f2eacdd00bc8fef31165c6a90 100644
@@ -1288,4 +1288,11 @@ void mq_change_real_num_tx(struct Qdisc *sch, unsigned int new_real_tx);
 
 int sch_frag_xmit_hook(struct sk_buff *skb, int (*xmit)(struct sk_buff *skb));
 
+/* Make sure qdisc is no longer in SCHED state. */
+static inline void qdisc_synchronize(const struct Qdisc *q)
+{
+	while (test_bit(__QDISC_STATE_SCHED, &q->state))
+		msleep(1);
+}
+
 #endif
net/sched/sch_taprio.c +0 -0
index 570389f6cdd7dbab5749dc06d886555305cbf623..9a11a499ea2df8d18c9c062496fdcbcf5a861391 100644

net/sched/sch_taprio.c static void taprio_reset(struct Qdisc *sch) · line 1704

Quoted 5 lines
net/sched/sch_taprio.c +2 -0
@@ -1700,6 +1700,8 @@ static void taprio_reset(struct Qdisc *sch)
 	int i;
 
 	hrtimer_cancel(&q->advance_timer);
+	qdisc_synchronize(sch);
+

From the commit message, I got the impression that only the one qdisc_synchronize() in taprio_destroy() would be needed.

Quoted 14 lines
net/sched/sch_taprio.c +1 -0
 	if (q->qdiscs) {
 		for (i = 0; i < dev->num_tx_queues; i++)
 			if (q->qdiscs[i])
@@ -1720,6 +1722,7 @@ static void taprio_destroy(struct Qdisc *sch)
 	 * happens in qdisc_create(), after taprio_init() has been called.
 	 */
 	hrtimer_cancel(&q->advance_timer);
+	qdisc_synchronize(sch);
 
 	taprio_disable_offload(dev, q, NULL);
 
Signature · 1 line
2.39.0.314.g84b9a713c41-goog

Cheers,

Signature · 1 line
Vinicius
#2 lore raw
Quoted 6 lines · Vinicius Costa Gomes

Hi,

From the commit message, I got the impression that only the one qdisc_synchronize() in taprio_destroy() would be needed.

This could be, but then why having hrtimer_cancel(&q->advance_timer); in taprio_reset(), since it is already in taprio_destroy() ?

#3 lore raw
In reply to Vinicius Costa Gomes
Quoted 7 lines · Vinicius Costa Gomes

Hi,

From the commit message, I got the impression that only the one qdisc_synchronize() in taprio_destroy() would be needed.

Hmm, I think you are right, qdisc_reset() is probably called while qdisc lock is held,
with BH disabled.

So calling msleep() from qdisc_reset() is a no go.

I will send a patch removing the change in taprio_reset(), thanks.

#4 lore raw
In reply to Eric Dumazet
Quoted 9 lines · Eric Dumazet

syzbot reported a nasty crash [1] in net_tx_action() which made little sense until we got a repro.

This repro installs a taprio qdisc, but providing an invalid TCA_RATE attribute.

qdisc_create() has to destroy the just initialized taprio qdisc, and taprio_destroy() is called.

Quoted 8 lines

However, the hrtimer used by taprio had already fired, therefore advance_sched() called __netif_schedule().

Then net_tx_action was trying to use a destroyed qdisc.

We can not undo the __netif_schedule(), so we must wait until one cpu serviced the qdisc before we can proceed.

This workaround looks a bit ugly. I think we _may_ be able to make hrtimer_start() as the last step of the initialization, IOW, move other validations and allocations before it.

Can you share your reproducer?

Thanks,

#5 lore raw
Quoted 19 lines · Cong Wang
Quoted 9 lines · Eric Dumazet

syzbot reported a nasty crash [1] in net_tx_action() which made little sense until we got a repro.

This repro installs a taprio qdisc, but providing an invalid TCA_RATE attribute.

qdisc_create() has to destroy the just initialized taprio qdisc, and taprio_destroy() is called.

Quoted 8 lines

However, the hrtimer used by taprio had already fired, therefore advance_sched() called __netif_schedule().

Then net_tx_action was trying to use a destroyed qdisc.

We can not undo the __netif_schedule(), so we must wait until one cpu serviced the qdisc before we can proceed.

Quoted 7 lines

This workaround looks a bit ugly. I think we _may_ be able to make hrtimer_start() as the last step of the initialization, IOW, move other validations and allocations before it.

Can you share your reproducer?

Thanks,

Maybe the issue is the same as https://syzkaller.appspot.com/bug?id=1ccb246eecb5114c440218336e4c7205aed5f2c8

#6 lore raw
Quoted 27 lines · shaozhengchao
Quoted 23 lines · Cong Wang
Quoted 9 lines · Eric Dumazet

syzbot reported a nasty crash [1] in net_tx_action() which made little sense until we got a repro.

This repro installs a taprio qdisc, but providing an invalid TCA_RATE attribute.

qdisc_create() has to destroy the just initialized taprio qdisc, and taprio_destroy() is called.

Quoted 8 lines

However, the hrtimer used by taprio had already fired, therefore advance_sched() called __netif_schedule().

Then net_tx_action was trying to use a destroyed qdisc.

We can not undo the __netif_schedule(), so we must wait until one cpu serviced the qdisc before we can proceed.

This workaround looks a bit ugly. I think we _may_ be able to make hrtimer_start() as the last step of the initialization, IOW, move other validations and allocations before it.

Quoted 5 lines
Quoted 3 lines

Can you share your reproducer?

Thanks,

Maybe the issue is the same as
https://syzkaller.appspot.com/bug?id=1ccb246eecb5114c440218336e4c7205aed5f2c8

Most certainly, yes.
I also think there were stall reports with the same stack trace where qdisc_run was unable to take a freed lock because its value was set to 1 by another task.

Signature · 10 lines
Alexander Potapenko
Software Engineer

Google Germany GmbH
Erika-Mann-Straße, 33
80636 München

Geschäftsführer: Paul Manicle, Liana Sebastian
Registergericht und -nummer: Hamburg, HRB 86891
Sitz der Gesellschaft: Hamburg
#7 lore raw
In reply to Cong Wang
Quoted 16 lines · Cong Wang
Quoted 14 lines · Eric Dumazet

syzbot reported a nasty crash [1] in net_tx_action() which made little sense until we got a repro.

This repro installs a taprio qdisc, but providing an invalid TCA_RATE attribute.

qdisc_create() has to destroy the just initialized taprio qdisc, and taprio_destroy() is called.

However, the hrtimer used by taprio had already fired, therefore advance_sched() called __netif_schedule().

Then net_tx_action was trying to use a destroyed qdisc.

Quoted 8 lines
Quoted 3 lines

We can not undo the __netif_schedule(), so we must wait until one cpu serviced the qdisc before we can proceed.

This workaround looks a bit ugly. I think we _may_ be able to make hrtimer_start() as the last step of the initialization, IOW, move other validations and allocations before it.

taprio_init() detects no error.

So moving around the hrtimer_start() inside it won't help.

The error comes later from a wrong TCA_RATE attempt can then:

static struct Qdisc *qdisc_create(...
...
err = gen_new_estimator(...);

if (err) {
    NL_SET_ERR_MSG(extack, "Failed to generate new estimator");
    goto err_out4;
}

...

err_out4:
qdisc_put_stab(rtnl_dereference(sch->stab));
if (ops->destroy)

     ops->destroy(sch);

goto err_out3;

This is why we need to make sure ->destroy will fully undo what ->init did, including the possible fact that the hrtimer already fired. This seems to be taprio specific.

Or we would need a new method, like ->post_init(), that should be
called once all steps have been a success.

Or call the hrtimer_start() at first taprio_enqueue(), adding a conditional in fast path...

Quoted 1 line

Can you share your reproducer?

Not publicly.

Although I think the bug is clear enough.

#8 lore raw
In reply to Eric Dumazet
This patch was applied to netdev/net.git (master) by David S. Miller <[email protected]>:

Hello:

This patch was applied to netdev/net.git (master)
by David S. Miller <[email protected]>:

Quoted 10 lines · On Fri, 13 Jan 2023 16:48:49 +0000 you wrote:

syzbot reported a nasty crash [1] in net_tx_action() which made little sense until we got a repro.

This repro installs a taprio qdisc, but providing an invalid TCA_RATE attribute.

qdisc_create() has to destroy the just initialized taprio qdisc, and taprio_destroy() is called.

[...]

Here is the summary with links:
- [net] net/sched: sch_taprio: fix possible use-after-free

    https://git.kernel.org/netdev/net/c/3a415d59c1db

You are awesome, thank you!

Signature · 2 lines
Deet-doot-dot, I am a bot.
https://korg.docs.kernel.org/patchwork/pwbot.html