[PATCH net] netdevsim: use cond_resched() in nsim_dev_trap_report_work()

[PATCH net] netdevsim: use cond_resched() in nsim_dev_trap_report_work()

[Eric Dumazet]

I am still seeing many syzbot reports hinting that syzbot
might fool nsim_dev_trap_report_work() with hundreds of ports [1]

Lets use cond_resched(), and system_unbound_wq
instead of implicit system_wq.

[1]
INFO: task syz-executor:20633 blocked for more than 143 seconds.
      Not tainted 6.12.0-rc2-syzkaller-00205-g1d227fcc7222 #0
"echo 0 > /proc/sys/kernel/hung_task_timeout_secs" disables this message.
task:syz-executor    state:D stack:25856 pid:20633 tgid:20633 ppid:1      flags:0x00004006
...
NMI backtrace for cpu 1
CPU: 1 UID: 0 PID: 16760 Comm: kworker/1:0 Not tainted 6.12.0-rc2-syzkaller-00205-g1d227fcc7222 #0
Hardware name: Google Google Compute Engine/Google Compute Engine, BIOS Google 09/13/2024
Workqueue: events nsim_dev_trap_report_work
 RIP: 0010:__sanitizer_cov_trace_pc+0x0/0x70 kernel/kcov.c:210
Code: 89 fb e8 23 00 00 00 48 8b 3d 04 fb 9c 0c 48 89 de 5b e9 c3 c7 5d 00 0f 1f 00 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 <f3> 0f 1e fa 48 8b 04 24 65 48 8b 0c 25 c0 d7 03 00 65 8b 15 60 f0
RSP: 0018:ffffc90000a187e8 EFLAGS: 00000246
RAX: 0000000000000100 RBX: ffffc90000a188e0 RCX: ffff888027d3bc00
RDX: ffff888027d3bc00 RSI: 0000000000000000 RDI: 0000000000000000
RBP: ffff88804a2e6000 R08: ffffffff8a4bc495 R09: ffffffff89da3577
R10: 0000000000000004 R11: ffffffff8a4bc2b0 R12: dffffc0000000000
R13: ffff88806573b503 R14: dffffc0000000000 R15: ffff8880663cca00
FS:  0000000000000000(0000) GS:ffff8880b8700000(0000) knlGS:0000000000000000
CS:  0010 DS: 0000 ES: 0000 CR0: 0000000080050033
CR2: 00007fc90a747f98 CR3: 000000000e734000 CR4: 00000000003526f0
DR0: 0000000000000000 DR1: 000000000000002b DR2: 0000000000000000
DR3: 0000000000000000 DR6: 00000000ffff0ff0 DR7: 0000000000000400
Call Trace:
 <NMI>
 </NMI>
 <TASK>
  __local_bh_enable_ip+0x1bb/0x200 kernel/softirq.c:382
  spin_unlock_bh include/linux/spinlock.h:396 [inline]
  nsim_dev_trap_report drivers/net/netdevsim/dev.c:820 [inline]
  nsim_dev_trap_report_work+0x75d/0xaa0 drivers/net/netdevsim/dev.c:850
  process_one_work kernel/workqueue.c:3229 [inline]
  process_scheduled_works+0xa63/0x1850 kernel/workqueue.c:3310
  worker_thread+0x870/0xd30 kernel/workqueue.c:3391
  kthread+0x2f0/0x390 kernel/kthread.c:389
  ret_from_fork+0x4b/0x80 arch/x86/kernel/process.c:147
  ret_from_fork_asm+0x1a/0x30 arch/x86/entry/entry_64.S:244
 </TASK>

Fixes: ba5e1272142d ("netdevsim: avoid potential loop in nsim_dev_trap_report_work()")
Signed-off-by: Eric Dumazet <edumazet@google.com>
Cc: Jiri Pirko <jiri@nvidia.com>
---
 drivers/net/netdevsim/dev.c | 12 +++++++-----
 1 file changed, 7 insertions(+), 5 deletions(-)

diff --git a/drivers/net/netdevsim/dev.c b/drivers/net/netdevsim/dev.c
index 92a7a36b93ac0cc1b02a551b974fb390254ac484..2f98443230895e8b6ee4cc36d5a2add8c2c0a00e 100644
--- a/drivers/net/netdevsim/dev.c
+++ b/drivers/net/netdevsim/dev.c
@@ -848,11 +848,12 @@ static void nsim_dev_trap_report_work(struct work_struct *work)
 			continue;
 
 		nsim_dev_trap_report(nsim_dev_port);
+		cond_resched();
 	}
 	devl_unlock(priv_to_devlink(nsim_dev));
-
-	schedule_delayed_work(&nsim_dev->trap_data->trap_report_dw,
-			      msecs_to_jiffies(NSIM_TRAP_REPORT_INTERVAL_MS));
+	queue_delayed_work(system_unbound_wq,
+			   &nsim_dev->trap_data->trap_report_dw,
+			   msecs_to_jiffies(NSIM_TRAP_REPORT_INTERVAL_MS));
 }
 
 static int nsim_dev_traps_init(struct devlink *devlink)
@@ -907,8 +908,9 @@ static int nsim_dev_traps_init(struct devlink *devlink)
 
 	INIT_DELAYED_WORK(&nsim_dev->trap_data->trap_report_dw,
 			  nsim_dev_trap_report_work);
-	schedule_delayed_work(&nsim_dev->trap_data->trap_report_dw,
-			      msecs_to_jiffies(NSIM_TRAP_REPORT_INTERVAL_MS));
+	queue_delayed_work(system_unbound_wq,
+			   &nsim_dev->trap_data->trap_report_dw,
+			   msecs_to_jiffies(NSIM_TRAP_REPORT_INTERVAL_MS));
 
 	return 0;
 
-- 
2.47.0.rc1.288.g06298d1525-goog

Re: [PATCH net] netdevsim: use cond_resched() in nsim_dev_trap_report_work()

[Jakub Kicinski]

On Fri, 11 Oct 2024 13:18:43 +0000 Eric Dumazet wrote:
> --- a/drivers/net/netdevsim/dev.c
> +++ b/drivers/net/netdevsim/dev.c
> @@ -848,11 +848,12 @@ static void nsim_dev_trap_report_work(struct work_struct *work)
>  			continue;
>  
>  		nsim_dev_trap_report(nsim_dev_port);
> +		cond_resched();
>  	}
>  	devl_unlock(priv_to_devlink(nsim_dev));
> -
> -	schedule_delayed_work(&nsim_dev->trap_data->trap_report_dw,
> -			      msecs_to_jiffies(NSIM_TRAP_REPORT_INTERVAL_MS));
> +	queue_delayed_work(system_unbound_wq,
> +			   &nsim_dev->trap_data->trap_report_dw,
> +			   msecs_to_jiffies(NSIM_TRAP_REPORT_INTERVAL_MS));

Makes sense, there's one more place which queues this work, in case we
couldn't grab the lock. Should it also be converted?

Re: [PATCH net] netdevsim: use cond_resched() in nsim_dev_trap_report_work()

[Eric Dumazet]

On Fri, Oct 11, 2024 at 5:59 PM Jakub Kicinski <kuba@kernel.org> wrote:
>
> On Fri, 11 Oct 2024 13:18:43 +0000 Eric Dumazet wrote:
> > --- a/drivers/net/netdevsim/dev.c
> > +++ b/drivers/net/netdevsim/dev.c
> > @@ -848,11 +848,12 @@ static void nsim_dev_trap_report_work(struct work_struct *work)
> >                       continue;
> >
> >               nsim_dev_trap_report(nsim_dev_port);
> > +             cond_resched();
> >       }
> >       devl_unlock(priv_to_devlink(nsim_dev));
> > -
> > -     schedule_delayed_work(&nsim_dev->trap_data->trap_report_dw,
> > -                           msecs_to_jiffies(NSIM_TRAP_REPORT_INTERVAL_MS));
> > +     queue_delayed_work(system_unbound_wq,
> > +                        &nsim_dev->trap_data->trap_report_dw,
> > +                        msecs_to_jiffies(NSIM_TRAP_REPORT_INTERVAL_MS));
>
> Makes sense, there's one more place which queues this work, in case we
> couldn't grab the lock. Should it also be converted?

Right of course, I will send a v2.