Re: [PATCH v2 11/11] lockdep,trace: Expose tracepoints

From: Guenter Roeck <linux@roeck-us.net>
To: peterz@infradead.org
Cc: linux-kernel@vger.kernel.org, mingo@kernel.org, will@kernel.org,
	npiggin@gmail.com, elver@google.com, jgross@suse.com,
	paulmck@kernel.org, rostedt@goodmis.org, rjw@rjwysocki.net,
	joel@joelfernandes.org, svens@linux.ibm.com, tglx@linutronix.de
Subject: Re: [PATCH v2 11/11] lockdep,trace: Expose tracepoints
Date: Wed, 2 Sep 2020 06:21:29 -0700	[thread overview]
Message-ID: <44e1d3a7-3cd5-8fa8-7fb2-f82b83a019d7@roeck-us.net> (raw)
In-Reply-To: <20200902072432.GI2674@hirez.programming.kicks-ass.net>

On 9/2/20 12:24 AM, peterz@infradead.org wrote:
> On Tue, Sep 01, 2020 at 08:51:46PM -0700, Guenter Roeck wrote:
>> On Fri, Aug 21, 2020 at 10:47:49AM +0200, Peter Zijlstra wrote:
>>> The lockdep tracepoints are under the lockdep recursion counter, this
>>> has a bunch of nasty side effects:
>>>
>>>  - TRACE_IRQFLAGS doesn't work across the entire tracepoint
>>>
>>>  - RCU-lockdep doesn't see the tracepoints either, hiding numerous
>>>    "suspicious RCU usage" warnings.
>>>
>>> Pull the trace_lock_*() tracepoints completely out from under the
>>> lockdep recursion handling and completely rely on the trace level
>>> recusion handling -- also, tracing *SHOULD* not be taking locks in any
>>> case.
>>>
>>
>> Wonder what is worse - the problem or its fix. This patch results in
>> a number of WARNING backtraces for several archtectures/platforms.
>> Reverting it fixes the problems.
> 
> Without all this there was a recursion that could crash. But yes,
> tedious.
> 
> OTOH the warnings are about real bugs that were pre-existing, we now see
> them and can fix them.
> 
> I'll reply to ARM separately, but let's have a peek at s390.
> 
>> s390:
>>
>> [   19.490586] =============================
>> [   19.490752] WARNING: suspicious RCU usage
>> [   19.490921] 5.9.0-rc3 #1 Not tainted
>> [   19.491086] -----------------------------
>> [   19.491253] include/trace/events/lock.h:37 suspicious rcu_dereference_check() usage!
> 
>> [   19.493147]  [<00000000001d5de2>] lock_acquire+0x41a/0x498
>> [   19.493320]  [<0000000000103b72>] enabled_wait+0xca/0x198
>> [   19.493493]  [<0000000000103f80>] arch_cpu_idle+0x20/0x38
> 
> Does this help?
> 
> ---
> 
> diff --git a/arch/s390/kernel/idle.c b/arch/s390/kernel/idle.c
> index c73f50649e7e..f7f1e64e0d98 100644
> --- a/arch/s390/kernel/idle.c
> +++ b/arch/s390/kernel/idle.c
> @@ -39,14 +39,13 @@ void enabled_wait(void)
>  	local_irq_restore(flags);
>  
>  	/* Account time spent with enabled wait psw loaded as idle time. */
> -	/* XXX seqcount has tracepoints that require RCU */
> -	write_seqcount_begin(&idle->seqcount);
> +	raw_write_seqcount_begin(&idle->seqcount);
>  	idle_time = idle->clock_idle_exit - idle->clock_idle_enter;
>  	idle->clock_idle_enter = idle->clock_idle_exit = 0ULL;
>  	idle->idle_time += idle_time;
>  	idle->idle_count++;
>  	account_idle_time(cputime_to_nsecs(idle_time));
> -	write_seqcount_end(&idle->seqcount);
> +	raw_write_seqcount_end(&idle->seqcount);
>  }
>  NOKPROBE_SYMBOL(enabled_wait);
>  

Yes, it does.

Tested-by: Guenter Roeck <linux@roeck-us.net>

Guenter