linux-kernel - Re: [PATCH v2 11/11] perf/uprobe: Add uretprobe timer

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Message-ID: <20240711160649.GA38270@noisy.programming.kicks-ass.net>
Date: Thu, 11 Jul 2024 18:06:49 +0200
From: Peter Zijlstra <peterz@...radead.org>
To: Oleg Nesterov <oleg@...hat.com>
Cc: mingo@...nel.org, andrii@...nel.org, linux-kernel@...r.kernel.org,
	linux-trace-kernel@...r.kernel.org, rostedt@...dmis.org,
	mhiramat@...nel.org, jolsa@...nel.org, clm@...a.com,
	paulmck@...nel.org
Subject: Re: [PATCH v2 11/11] perf/uprobe: Add uretprobe timer

On Thu, Jul 11, 2024 at 05:55:42PM +0200, Peter Zijlstra wrote:
> On Thu, Jul 11, 2024 at 05:00:54PM +0200, Peter Zijlstra wrote:
> 
> > Let me ponder that a little, I *can* make it work, but all 'solutions'
> > I've come up with so far are really rather vile.
> 
> This is the least horrible solution I could come up with...
> 
> ---
> --- a/include/linux/uprobes.h
> +++ b/include/linux/uprobes.h
> @@ -83,6 +83,7 @@ struct uprobe_task {
>  
>  	struct timer_list		ri_timer;
>  	struct callback_head		ri_task_work;
> +	bool				ri_work_pending;
>  	struct task_struct		*task;
>  };
>  
> --- a/kernel/events/uprobes.c
> +++ b/kernel/events/uprobes.c
> @@ -1797,9 +1797,8 @@ void uprobe_free_utask(struct task_struc
>  	t->utask = NULL;
>  }
>  
> -static void return_instance_task_work(struct callback_head *head)
> +static void __return_instance_put(struct uprobe_task *utask)
>  {
> -	struct uprobe_task *utask = container_of(head, struct uprobe_task, ri_task_work);
>  	struct return_instance *ri;
>  
>  	for (ri = utask->return_instances; ri; ri = ri->next) {
> @@ -1815,9 +1814,43 @@ static void return_instance_task_work(st
>  	}
>  }
>  
> +static void return_instance_task_work(struct callback_head *head)
> +{
> +	struct uprobe_task *utask = container_of(head, struct uprobe_task, ri_task_work);
> +	utask->ri_work_pending = false;
> +	__return_instance_put(utask);
> +}
> +
> +static int return_instance_blocked(struct task_struct *p, void *arg)
> +{
> +	unsigned int state = READ_ONCE(p->__state);
> +
> +	if (state == TASK_RUNNING || state == TASK_WAKING)
> +		return 0;
> +
> +	smp_rmb();
> +	if (p->on_rq)
> +		return 0;
> +
> +	/*
> +	 * Per __task_needs_rq_locked() we now have: !p->on_cpu and only hold
> +	 * p->pi_lock, and can consiter the task fully blocked.
> +	 */
> +
> +	__return_instance_put(p->utask);

PREEMPT_RT might not like this though, doing the full RI iteration under
a raw_spinlock_t...

I just can't think of anything saner just now. Oh well, let me go make
dinner, perhaps something will come to me.