Re: [PATCH 13/14] lockdep: Prepare for NMI IRQ state tracking

[Steven Rostedt]

On Fri, 29 May 2020 23:27:41 +0200
Peter Zijlstra <peterz@xxxxxxxxxxxxx> wrote:

> There is no reason not to always, accurately, track IRQ state.
> 
> This change also makes IRQ state tracking ignore lockdep_off().
> 
> Signed-off-by: Peter Zijlstra (Intel) <peterz@xxxxxxxxxxxxx>
> ---
>  kernel/locking/lockdep.c |   33 ++++++++++++++++++++++++++++++---
>  1 file changed, 30 insertions(+), 3 deletions(-)
> 
> --- a/kernel/locking/lockdep.c
> +++ b/kernel/locking/lockdep.c
> @@ -3646,7 +3646,13 @@ static void __trace_hardirqs_on_caller(v
>   */
>  void lockdep_hardirqs_on_prepare(unsigned long ip)
>  {
> -	if (unlikely(!debug_locks || current->lockdep_recursion))

Why remove the check for debug_locks? Isn't that there to disable
everything at once to prevent more warnings to be printed?

Also, isn't there other ways that we could have recursion besides NMIs?
Say we do a printk inside here, or call something that may also enable
interrupts? I thought the recursion check was also to prevent lockdep
infrastructure calling something that lockdep monitors being a problem?

Or am I missing something?

-- Steve


> +	/*
> +	 * NMIs do not (and cannot) track lock dependencies, nothing to do.
> +	 */
> +	if (in_nmi())
> +		return;
> +
> +	if (DEBUG_LOCKS_WARN_ON(current->lockdep_recursion & LOCKDEP_RECURSION_MASK))
>  		return;
>  
>  	if (unlikely(current->hardirqs_enabled)) {
> @@ -3692,7 +3698,24 @@ void noinstr lockdep_hardirqs_on(unsigne
>  {
>  	struct task_struct *curr = current;
>  
> -	if (unlikely(!debug_locks || curr->lockdep_recursion))
> +	/*
> +	 * NMIs can happen in the middle of local_irq_{en,dis}able() where the
> +	 * tracking state and hardware state are out of sync.
> +	 *
> +	 * NMIs must save lockdep_hardirqs_enabled() to restore IRQ state from,
> +	 * and not rely on hardware state like normal interrupts.
> +	 */
> +	if (in_nmi()) {
> +		/*
> +		 * Skip:
> +		 *  - recursion check, because NMI can hit lockdep;
> +		 *  - hardware state check, because above;
> +		 *  - chain_key check, see lockdep_hardirqs_on_prepare().
> +		 */
> +		goto skip_checks;
> +	}
> +
> +	if (DEBUG_LOCKS_WARN_ON(curr->lockdep_recursion & LOCKDEP_RECURSION_MASK))
>  		return;
>  
>  	if (curr->hardirqs_enabled) {
> @@ -3720,6 +3743,7 @@ void noinstr lockdep_hardirqs_on(unsigne
>  	DEBUG_LOCKS_WARN_ON(current->hardirq_chain_key !=
>  			    current->curr_chain_key);
>  
> +skip_checks:
>  	/* we'll do an OFF -> ON transition: */
>  	curr->hardirqs_enabled = 1;
>  	curr->hardirq_enable_ip = ip;
> @@ -3735,7 +3759,10 @@ void noinstr lockdep_hardirqs_off(unsign
>  {
>  	struct task_struct *curr = current;
>  
> -	if (unlikely(!debug_locks || curr->lockdep_recursion))
> +	/*
> +	 * NMIs can happen in lockdep.
> +	 */
> +	if (!in_nmi() && DEBUG_LOCKS_WARN_ON(curr->lockdep_recursion & LOCKDEP_RECURSION_MASK))
>  		return;
>  
>  	/*
>