linux-kernel - Re: [PATCH v3 1/8] locking/lockdep: Decrement irq context counters when removing lock chain

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: <20200116153212.GS2827@hirez.programming.kicks-ass.net>
Date:   Thu, 16 Jan 2020 16:32:12 +0100
From:   Peter Zijlstra <peterz@...radead.org>
To:     Waiman Long <longman@...hat.com>
Cc:     Ingo Molnar <mingo@...hat.com>, Will Deacon <will.deacon@....com>,
        linux-kernel@...r.kernel.org, Bart Van Assche <bvanassche@....org>
Subject: Re: [PATCH v3 1/8] locking/lockdep: Decrement irq context counters
 when removing lock chain

On Wed, Jan 15, 2020 at 04:43:06PM -0500, Waiman Long wrote:
> There are currently three counters to track the irq context of a lock
> chain - nr_hardirq_chains, nr_softirq_chains and nr_process_chains.
> They are incremented when a new lock chain is added, but they are
> not decremented when a lock chain is removed. That causes some of the
> statistic counts reported by /proc/lockdep_stats to be incorrect.
> 
> Fix that by decrementing the right counter when a lock chain is removed.
> 
> Fixes: a0b0fd53e1e6 ("locking/lockdep: Free lock classes that are no longer in use")
> Signed-off-by: Waiman Long <longman@...hat.com>
> ---
>  kernel/locking/lockdep.c           | 36 +++++++++++++++++++++---------
>  kernel/locking/lockdep_internals.h |  6 +++++
>  2 files changed, 32 insertions(+), 10 deletions(-)
> 
> diff --git a/kernel/locking/lockdep.c b/kernel/locking/lockdep.c
> index 32282e7112d3..b20fa6236b2a 100644
> --- a/kernel/locking/lockdep.c
> +++ b/kernel/locking/lockdep.c
> @@ -2299,16 +2299,24 @@ static int check_irq_usage(struct task_struct *curr, struct held_lock *prev,
>  	return 0;
>  }
>  
> -static void inc_chains(void)
> +static void inc_chains(int irq_context)
>  {
> -	if (current->hardirq_context)
> +	if (irq_context & LOCK_CHAIN_HARDIRQ_CONTEXT)
>  		nr_hardirq_chains++;
> -	else {
> -		if (current->softirq_context)
> -			nr_softirq_chains++;
> -		else
> -			nr_process_chains++;
> -	}
> +	else if (irq_context & LOCK_CHAIN_SOFTIRQ_CONTEXT)
> +		nr_softirq_chains++;
> +	else
> +		nr_process_chains++;
> +}
> +
> +static void dec_chains(int irq_context)
> +{
> +	if (irq_context & LOCK_CHAIN_HARDIRQ_CONTEXT)
> +		nr_hardirq_chains--;
> +	else if (irq_context & LOCK_CHAIN_SOFTIRQ_CONTEXT)
> +		nr_softirq_chains--;
> +	else
> +		nr_process_chains--;
>  }
>  
>  #else
> @@ -2324,6 +2332,10 @@ static inline void inc_chains(void)
>  	nr_process_chains++;
>  }
>  
> +static void dec_chains(int irq_context)
> +{
> +	nr_process_chains--;
> +}
>  #endif /* CONFIG_TRACE_IRQFLAGS */
>  

Is there really need for two versions of those functions? Would the
@irq_context argument not always be 0 in the CONFIG_TRACE_IRQFLAGS=n
case?