linux-kernel - Re: [PATCH 7/5] tracing: Make sure rcu_irq_enter() can work for trace_*

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]

Date:   Fri, 7 Apr 2017 13:26:13 -0400
From:   Steven Rostedt <rostedt@...dmis.org>
To:     Mathieu Desnoyers <mathieu.desnoyers@...icios.com>
Cc:     linux-kernel <linux-kernel@...r.kernel.org>,
        Ingo Molnar <mingo@...nel.org>,
        Andrew Morton <akpm@...ux-foundation.org>,
        "Paul E. McKenney" <paulmck@...ux.vnet.ibm.com>
Subject: Re: [PATCH 7/5] tracing: Make sure rcu_irq_enter() can work for
 trace_*_rcuidle() trace events

On Fri, 7 Apr 2017 17:19:05 +0000 (UTC)
Mathieu Desnoyers <mathieu.desnoyers@...icios.com> wrote:

> ----- On Apr 7, 2017, at 1:06 PM, rostedt rostedt@...dmis.org wrote:
> 
> > From: "Steven Rostedt (VMware)" <rostedt@...dmis.org>
> > 
> > Stack tracing discovered that there's a small location inside the RCU
> > infrastructure that calling rcu_irq_enter() does not work. As trace events  
> 
> that -> where

ok

> 
> Do you have a link to the lkml thread where this stack tracing discovery
> happened ?

Actually it's this thread. But here:

Version 1 of the patch series:
 http://lkml.kernel.org/r/20170406164237.874767449@goodmis.org

Version 2 of the patch series:
 http://lkml.kernel.org/r/20170407140106.051135969@goodmis.org

> 
> > use rcu_irq_enter() it must make sure that it is functionable. A check  
> 
> I don't think functionable is the word you are looking for here. Perhaps
> "must make sure that it can be invoked" ?
> 
> > against rcu_irq_enter_disabled() is added with a WARN_ON_ONCE() as no trace
> > event should ever be used in that part of RCU. If the warning is triggered,
> > then the trace event is ignored.
> > 
> > Cc: Mathieu Desnoyers <mathieu.desnoyers@...icios.com>
> > Signed-off-by: Steven Rostedt (VMware) <rostedt@...dmis.org>
> > ---
> > include/linux/tracepoint.h | 2 ++
> > 1 file changed, 2 insertions(+)
> > 
> > diff --git a/include/linux/tracepoint.h b/include/linux/tracepoint.h
> > index f72fcfe..8baef96 100644
> > --- a/include/linux/tracepoint.h
> > +++ b/include/linux/tracepoint.h
> > @@ -159,6 +159,8 @@ extern void syscall_unregfunc(void);
> > 				TP_PROTO(data_proto),			\
> > 				TP_ARGS(data_args),			\
> > 				TP_CONDITION(cond),			\
> > +				if (WARN_ON_ONCE(rcu_irq_enter_disabled())) \
> > +					return;				\  
> 
> I must admit that it's a bit odd to have:
> 
> if (WARN_ON_ONCE(rcu_irq_enter_disabled()))
>        return;
> rcu_irq_enter_irqson()

Welcome to MACRO MAGIC!

> 
> as one argument to the __DO_TRACE() macro. To me it's a bit unexpected
> coding-style wise. Am I the only one not comfortable with the proposed
> syntax ?

The entire TRACE_EVENT()/__DO_TRACE() is special.

I thought about add yet another parameter, but as it doesn't change
much, I figured this was good enough. We could beak it up if you like:

#define RCU_IRQ_ENTER_CHECK \
	if (WARN_ON_ONCE(rcu_irq_enter_disabled()) 	\
		return;					\
	rcu_irq_enter_irqson();

[..]
			__DO_TRACE(&__tracepoint_##name,		\
				TP_PROTO(data_proto),			\
				TP_ARGS(data_args),			\
				TP_CONDITION(cond),			\
				PARAMS(RCU_IRQ_ENTER_CHECK),		\
				rcu_irq_exit_irqson());			\


Would that make you feel more comfortable?

-- Steve