Re: [PATCH v4] lib/spinlock_debug.c: prevent a recursive cycle in the debug code

[Peter Hurley <peter@hurleysoftware.com>]

On 01/27/2016 04:01 AM, Byungchul Park wrote:
> changes form v3 to v4
> - reuse a existing code as much as possible for preventing an infinite
>   recursive cycle.
> 
> changes from v2 to v3
> - avoid printk() only in case of lockup suspected, not real lockup in
>   which case it does not help at all.
> - consider not only console_sem.lock but also logbuf_lock which is used
>   by printk().
> 
> changes from v1 to v2
> - only change comment and commit message esp. replacing "deadlock" with
>   "infinite recursive cycle", since it is not an actual deadlock.
> 
> thanks,
> byungchul
> 
> -----8<-----
> From 7b0c6e48625632fa1732b619083dc550b5d924c6 Mon Sep 17 00:00:00 2001
> From: Byungchul Park <byungchul.park@lge.com>
> Date: Wed, 27 Jan 2016 18:11:55 +0900
> Subject: [PATCH v4] lib/spinlock_debug.c: prevent a recursive cycle in the
>  debug code
> 
> It causes an infinite recursive cycle when using CONFIG_DEBUG_SPINLOCK,
> in the spin_dump(). Backtrace prints printk() -> console_trylock() ->
> do_raw_spin_lock() -> spin_dump() -> printk()... infinitely.

printk() is potentially recursive in many situations.
What about spinlocks used by console drivers?

And we already have lockdep turned off to avoid triggering a recursive
lockdep report (which I think is a mistake).

I think we should be working toward properly handling recursion
in printk().

Regards,
Peter Hurley


> When the debug spinlock code is called from printk(), we should prevent
> calling spin_dump() and the like, calling printk() again in that context.
> 
> Signed-off-by: Byungchul Park <byungchul.park@lge.com>
> ---
>  include/linux/debug_locks.h     |  4 ++++
>  kernel/locking/spinlock_debug.c | 14 +++++++++++---
>  2 files changed, 15 insertions(+), 3 deletions(-)
> 
> diff --git a/include/linux/debug_locks.h b/include/linux/debug_locks.h
> index 822c135..b0f977e 100644
> --- a/include/linux/debug_locks.h
> +++ b/include/linux/debug_locks.h
> @@ -10,6 +10,10 @@ struct task_struct;
>  extern int debug_locks;
>  extern int debug_locks_silent;
>  
> +static inline void __debug_locks_on(void)
> +{
> +	debug_locks = 1;
> +}
>  
>  static inline int __debug_locks_off(void)
>  {
> diff --git a/kernel/locking/spinlock_debug.c b/kernel/locking/spinlock_debug.c
> index 0374a59..65177ba 100644
> --- a/kernel/locking/spinlock_debug.c
> +++ b/kernel/locking/spinlock_debug.c
> @@ -113,11 +113,19 @@ static void __spin_lock_debug(raw_spinlock_t *lock)
>  			return;
>  		__delay(1);
>  	}
> -	/* lockup suspected: */
> -	spin_dump(lock, "lockup suspected");
> +
> +	/*
> +	 * We should prevent calling printk() further, since it would cause
> +	 * an infinite recursive cycle if it's called from printk()!
> +	 */
> +	if (__debug_locks_off()) {
> +		/* lockup suspected: */
> +		spin_dump(lock, "lockup suspected");
>  #ifdef CONFIG_SMP
> -	trigger_all_cpu_backtrace();
> +		trigger_all_cpu_backtrace();
>  #endif
> +		__debug_locks_on();
> +	}
>  
>  	/*
>  	 * The trylock above was causing a livelock.  Give the lower level arch
>