Re: Hit WARN_ON() in rcutorture.c:1055

[Paul E. McKenney]

On Tue, Mar 24, 2020 at 11:39:19AM +0000, Qais Yousef wrote:
> On 03/23/20 11:44, Paul E. McKenney wrote:
> > > > > {register, unregister}_pm_notifier() don't seem to be too hard to use.
> > > > 
> > > > That part is easy.  It would also be necessary to find all the affected
> > > > warnings in rcutorture and suppress them, not only during this time,
> > > > but also for some period of time afterwards.  Maybe this is the only one,
> > > > but that would be surprising.  ;-)
> > > 
> > > Wouldn't be easier to just deinit/init()? ie: treat it like unload/load module.
> > > 
> > > But you'll lose some info then that maybe you'd like to keep across
> > > suspend/resume cycles.
> > 
> > Hmmm...  Are you running rcutorture as a loadable module or built into
> > your kernel?  In the latter case, it starts up automatically shortly
> > after boot.
> 
> Built-in. Sorry maybe I wasn't clear. I meant something like the attached.
> Which still generates this warning, so yeah I get your point about suppressing
> some warnings in the right places now.

What my approach will be is to think through how to deal with this.
It might be that I can come up with a simple strategy, and if so, I will
make the changes.

But yes, I never had considered this in any of my rcutorture design or
implementation, so there is likely to be 15 years worth of bugs that
would need to be fixed.

Unless I can come up with something a bit more targeted.  Which might
happen.  It sometimes has in past encounters with this sort of thing.  ;-)

							Thanx, Paul

> [   48.898066] rcu-torture: Stopping rcu_torture_fwd_prog task
> [   48.908998] rcu-torture: Stopping rcu_torture_reader
> [   48.917052] rcu-torture: Stopping rcu_torture_fakewriter
> [   48.926987] ------------[ cut here ]------------
> [   48.927025] WARNING: CPU: 4 PID: 261 at kernel/rcu/rcutorture.c:1085 rcu_torture_writer+0x49c/0xa90
> [   48.927038] rcu-torture: Stopping rcu_torture_fakewriter
> [   48.927041] Modules linked in:
> [   48.927070] CPU: 4 PID: 261 Comm: rcu_torture_wri Not tainted 5.6.0-rc6-00002-g533440e608d2-dirty #540
> [   48.927085] Hardware name: ARM Juno development board (r2) (DT)
> [   48.927103] pstate: 00000005 (nzcv daif -PAN -UAO)
> [   48.927124] pc : rcu_torture_writer+0x49c/0xa90
> [   48.927144] lr : rcu_torture_writer+0x494/0xa90
> [   48.927157] sp : ffff800018bfbde0
> [   48.927172] x29: ffff800018bfbde0 x28: ffff80001426da58
> [   48.927198] x27: ffff80001426cf08 x26: ffff80001426cf08
> [   48.927224] x25: ffff8000121b5000 x24: 0000000000000001
> [   48.927250] x23: ffff80001426d650 x22: ffff80001426b000
> [   48.927275] x21: ffff800013439000 x20: 0000000000000001
> [   48.927301] x19: ffff80001426c000 x18: 0000000000000000
> [   48.927326] x17: 0000000000000000 x16: 0000000000000000
> [   48.927351] x15: 0000000000000000 x14: 0000000000000000
> [   48.927376] x13: 0000000000000000 x12: 0000000000000000
> [   48.927401] x11: 00000000000001fa x10: 0000000000000000
> [   48.927426] x9 : ffff800013a130a0 x8 : ffff0009757f0850
> [   48.927450] x7 : 0000000000000000 x6 : ffff800018bfbcc0
> [   48.927475] x5 : 0000000000000001 x4 : 0000000000000000
> [   48.927499] x3 : 0000000000000080 x2 : 0000000000000000
> [   48.927524] x1 : 0000000000000000 x0 : 0000000000000001
> [   48.927548] Call trace:
> [   48.927569]  rcu_torture_writer+0x49c/0xa90
> [   48.927590]  kthread+0x13c/0x140
> [   48.927611]  ret_from_fork+0x10/0x18
> [   48.927627] irq event stamp: 34170
> [   48.927654] hardirqs last  enabled at (34169): [<ffff800010114300>] __local_bh_enable_ip+0x98/0x148
> [   48.927680] hardirqs last disabled at (34170): [<ffff8000100a95d0>] do_debug_exception+0x1a8/0x258
> [   48.927702] softirqs last  enabled at (34168): [<ffff8000101b949c>] rcu_torture_free+0x84/0x98
> [   48.927725] softirqs last disabled at (34166): [<ffff8000101b9478>] rcu_torture_free+0x60/0x98
> [   48.927740] ---[ end trace 57f88d092825f447 ]---
> 
> Thanks
> 
> --
> Qais Yousef

> diff --git a/kernel/rcu/rcutorture.c b/kernel/rcu/rcutorture.c
> index 1aeecc165b21..defad882b236 100644
> --- a/kernel/rcu/rcutorture.c
> +++ b/kernel/rcu/rcutorture.c
> @@ -45,6 +45,7 @@
>  #include <linux/sched/sysctl.h>
>  #include <linux/oom.h>
>  #include <linux/tick.h>
> +#include <linux/suspend.h>
>  
>  #include "rcu.h"
>  
> @@ -1628,7 +1629,7 @@ static int rcu_torture_stall(void *args)
>  }
>  
>  /* Spawn CPU-stall kthread, if stall_cpu specified. */
> -static int __init rcu_torture_stall_init(void)
> +static int rcu_torture_stall_init(void)
>  {
>  	if (stall_cpu <= 0)
>  		return 0;
> @@ -2008,7 +2009,7 @@ static int rcu_torture_fwd_prog(void *args)
>  }
>  
>  /* If forward-progress checking is requested and feasible, spawn the thread. */
> -static int __init rcu_torture_fwd_prog_init(void)
> +static int rcu_torture_fwd_prog_init(void)
>  {
>  	struct rcu_fwd *rfp;
>  
> @@ -2359,7 +2360,7 @@ static void rcutorture_sync(void)
>  		cur_ops->sync();
>  }
>  
> -static int __init
> +static int
>  rcu_torture_init(void)
>  {
>  	long i;
> @@ -2549,5 +2550,47 @@ rcu_torture_init(void)
>  	return firsterr;
>  }
>  
> -module_init(rcu_torture_init);
> -module_exit(rcu_torture_cleanup);
> +static int rcu_torture_notify(struct notifier_block *nb,
> +			      unsigned long mode, void *_unused)
> +{
> +	switch (mode) {
> +	case PM_HIBERNATION_PREPARE:
> +	case PM_SUSPEND_PREPARE:
> +		pr_info("Shutdodwn rcu torture..");
> +		rcu_torture_cleanup();
> +		break;
> +
> +	case PM_POST_RESTORE:
> +		pr_info("Restart rcu torture..");
> +		rcu_torture_init();
> +		break;
> +	}
> +
> +	return 0;
> +}
> +
> +static struct notifier_block rcu_nb = {
> +	.notifier_call = rcu_torture_notify
> +};
> +
> +static int nb_failed;
> +
> +int rcu_torture_module_init(void)
> +{
> +	nb_failed = register_pm_notifier(&rcu_nb);
> +	if (nb_failed)
> +		pr_warn("Failed to register PM notifier");
> +
> +	return rcu_torture_init();
> +}
> +
> +void rcu_torture_module_exit(void)
> +{
> +	if (!nb_failed)
> +		unregister_pm_notifier(&rcu_nb);
> +
> +	rcu_torture_cleanup();
> +}
> +
> +module_init(rcu_torture_module_init);
> +module_exit(rcu_torture_module_exit);