netdev - Re: [RFC PATCH v3 1/5] irq: add tracepoint to softirq

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: <20100721135654.GE21259@hmsreliant.think-freely.org>
Date:	Wed, 21 Jul 2010 09:56:54 -0400
From:	Neil Horman <nhorman@...driver.com>
To:	KOSAKI Motohiro <kosaki.motohiro@...fujitsu.com>
Cc:	Koki Sanagi <sanagi.koki@...fujitsu.com>, netdev@...r.kernel.org,
	linux-kernel@...r.kernel.org, davem@...emloft.net,
	kaneshige.kenji@...fujitsu.com, izumi.taku@...fujitsu.com,
	laijs@...fujitsu.com, scott.a.mcmillan@...el.com,
	rostedt@...dmis.org, eric.dumazet@...il.com, fweisbec@...il.com,
	mathieu.desnoyers@...ymtl.ca
Subject: Re: [RFC PATCH v3 1/5] irq: add tracepoint to softirq_raise

On Wed, Jul 21, 2010 at 10:01:34PM +0900, KOSAKI Motohiro wrote:
> > > >>  #endif /*  _TRACE_IRQ_H */
> > > >> diff --git a/kernel/softirq.c b/kernel/softirq.c
> > > >> index 825e112..6790599 100644
> > > >> --- a/kernel/softirq.c
> > > >> +++ b/kernel/softirq.c
> > > >> @@ -215,9 +215,9 @@ restart:
> > > >>  			int prev_count = preempt_count();
> > > >>  			kstat_incr_softirqs_this_cpu(h - softirq_vec);
> > > >>  
> > > >> -			trace_softirq_entry(h, softirq_vec);
> > > >> +			trace_softirq_entry(h - softirq_vec);
> > > >>  			h->action(h);
> > > >> -			trace_softirq_exit(h, softirq_vec);
> > > >> +			trace_softirq_exit(h - softirq_vec);
> > > > 
> > > > You're loosing information here by reducing the numbers of parameters in this
> > > > tracepoint.  How many other tracepoint scripts rely on having both pointers
> > > > handy?  Why not just do the pointer math inside your tracehook instead?
> > > 
> > > In __raise_softirq_irqoff macro there is no method to refer softirq_vec, so it
> > > can't use softirq DECLARE_EVENT_CLASS as is.
> > > Currently,  there is no script using softirq_entry or softirq_exit.
> > > 
> > That shouldn't matter, just pass in NULL for softirq_vec in
> > __raise_softirq_irqoff as the second argument to the trace function.  You may
> > need to fix up the class definition so that the assignment or printk doesn't try
> > to dereference that pointer when its NULL, but thats easy enough, and it avoids
> > breaking any other perf scripts floating out there.
> 
> please see 5 lines above. we already have 'h - softirq_vec' calculation in
> this function. so, Sanagi-san's change don't makes any overhead.
> 
> So, if the overhead is zero, I'd prefer simplest tracepoint definition :)
> 
I never complained about performance here, I complained about information loss.
You have a tracepoint that provides two arguments here, and you're eliminating
one of them.  That will potentially break other users of this tracepoint.  I
understand we don't normally care about that with tracepoints as much, but if we
can avoid it, why don't we?
Neil

> 
> 
> --
> To unsubscribe from this list: send the line "unsubscribe netdev" in
> the body of a message to majordomo@...r.kernel.org
> More majordomo info at  http://vger.kernel.org/majordomo-info.html
> 
--
To unsubscribe from this list: send the line "unsubscribe netdev" in
the body of a message to majordomo@...r.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html