linux-kernel - Re: [PATCH printk v5 03/40] printk: Prepare for SRCU console list protection

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: <20221201215650.GK4001@paulmck-ThinkPad-P17-Gen-1>
Date:   Thu, 1 Dec 2022 13:56:50 -0800
From:   "Paul E. McKenney" <paulmck@...nel.org>
To:     John Ogness <john.ogness@...utronix.de>
Cc:     Nathan Chancellor <nathan@...nel.org>,
        Petr Mladek <pmladek@...e.com>,
        Sergey Senozhatsky <senozhatsky@...omium.org>,
        Steven Rostedt <rostedt@...dmis.org>,
        Thomas Gleixner <tglx@...utronix.de>,
        linux-kernel@...r.kernel.org, Miguel Ojeda <ojeda@...nel.org>,
        Greg Kroah-Hartman <gregkh@...uxfoundation.org>,
        qiang1.zhang@...el.com
Subject: Re: [PATCH printk v5 03/40] printk: Prepare for SRCU console list
 protection

On Thu, Dec 01, 2022 at 10:42:25PM +0106, John Ogness wrote:
> Hi Nathan,
> 
> Thanks for reporting this. Patch below.
> 
> @paulmck: Please also take a look below.
> 
> On 2022-12-01, Nathan Chancellor <nathan@...nel.org> wrote:
> > bad: scheduling from the idle thread!
> > CPU: 0 PID: 0 Comm: swapper Not tainted 6.1.0-rc1+ #1
> > Hardware name: PowerMac3,1 7400 0xc0209 PowerMac
> > Call Trace:
> > [c0bc1db0] [c07f07e0] dump_stack_lvl+0x34/0x50 (unreliable)
> > [c0bc1dd0] [c008429c] dequeue_task_idle+0x34/0x5c
> > [c0bc1df0] [c0820924] __schedule+0x56c/0x5c4
> > [c0bc1e40] [c08209d0] schedule+0x54/0xfc
> > [c0bc1e60] [c0826034] schedule_timeout+0x13c/0x194
> > [c0bc1ea0] [c082134c] __wait_for_common+0xcc/0x1f4
> > [c0bc1ee0] [c00ac8ac] synchronize_srcu+0xc8/0x12c
> > [c0bc1f20] [c00a0230] unregister_console+0xc8/0x10c
> > [c0bc1f40] [c009e314] register_console+0x2f4/0x390
> > [c0bc1f60] [c0a17510] pmz_console_init+0x34/0x48
> > [c0bc1f70] [c0a0491c] console_init+0x9c/0xf0
> > [c0bc1fa0] [c09f5584] start_kernel+0x588/0x6ac
> > [c0bc1ff0] [00003540] 0x3540
> 
> This config is using TINY_RCU. Its srcu_synchronize() implementation
> does not check if it called before scheduling is ready. The following
> patch will fix it.
> 
> @paulmck: Should it check (system_state < SYSTEM_SCHEDULING) instead
> since TINY_RCU does not use @rcu_scheduler_active?

Thank you for chasing this down, John!

You are exactly right, and I therefore need to pull this into the
pile for the upcoming merge window:

dbc6ca150842 ("srcu: Make Tiny synchronize_srcu() check for readers")

And kudos to Zqiang for a proactive fix!  ;-)

I will add your (John's) Tested-by, but please let me know if this is
inappropriate.

							Thanx, Paul

> John Ogness
> 
> diff --git a/kernel/rcu/srcutiny.c b/kernel/rcu/srcutiny.c
> index 33adafdad261..35338e6e37e7 100644
> --- a/kernel/rcu/srcutiny.c
> +++ b/kernel/rcu/srcutiny.c
> @@ -197,6 +197,9 @@ void synchronize_srcu(struct srcu_struct *ssp)
>  {
>  	struct rcu_synchronize rs;
>  
> +	if (rcu_scheduler_active == RCU_SCHEDULER_INACTIVE)
> +		return;
> +
>  	init_rcu_head_on_stack(&rs.head);
>  	init_completion(&rs.completion);
>  	call_srcu(ssp, &rs.head, wakeme_after_rcu);