All of lore.kernel.org
 help / color / mirror / Atom feed
From: Peter Zijlstra <peterz@infradead.org>
To: Vince Weaver <vincent.weaver@maine.edu>
Cc: Ingo Molnar <mingo@kernel.org>,
	linux-kernel@vger.kernel.org, eranian@google.com,
	Paul Mackerras <paulus@samba.org>,
	Arnaldo Carvalho de Melo <acme@kernel.org>
Subject: Re: [patch] inherited events not signalling parent on overflow
Date: Thu, 11 Jun 2015 10:32:01 +0200	[thread overview]
Message-ID: <1434011521.1495.71.camel@twins> (raw)
In-Reply-To: <alpine.DEB.2.20.1506110025540.13723@vincent-weaver-1.umelst.maine.edu>

On Thu, 2015-06-11 at 00:30 -0400, Vince Weaver wrote:
> On Fri, 29 May 2015, Ingo Molnar wrote:
>  
> > * Vince Weaver <vincent.weaver@maine.edu> wrote:
> 
> > > If we inherit events, we inherit the signal state but not the fasync state, so 
> > > overflows in inherited children will never trigger the signal handler.
> > > 
> > > Signed-off-by: Vince Weaver <vincent.weaver@maine.edu>
> > > 
> > > diff --git a/kernel/events/core.c b/kernel/events/core.c
> > > index 1a3bf48..7df4cf5 100644
> > > --- a/kernel/events/core.c
> > > +++ b/kernel/events/core.c
> > > @@ -8626,6 +8630,8 @@ inherit_event(struct perf_event *parent_event,
> > >  	child_event->overflow_handler_context
> > >  		= parent_event->overflow_handler_context;
> > >  
> > > +	child_event->fasync = parent_event->fasync;
> > > +
> > >  	/*
> > >  	 * Precalculate sample_data sizes
> > >  	 */
> 
> This patch, while it does work well enough to enable self-monitored-sampling 
> of OpenMP programs, falls apart under fuzzing.
> 
> You end up with lots of
> 
> [25592.289382] kill_fasync: bad magic number in fasync_struct!
> 
> warnings and eventually I managed to lock up the system that way.

Right, I had a peek earlier at how fasync worked but came away confused.

Today I seem to have had better luck. Installing fasync allocates memory
and sets filp->f_flags |= FASYNC, which upon the demise of the file
descriptor ensures the allocation is freed.

Now for perf, we can have the events stick around for a while after the
original FD is dead because of references from child events. With the
above patch these events would still have a pointer into this free'd
fasync. This is bad.

A further problem with the patch is that if the parent changes its
fasync state the children might lag and again have pointers into dead
space.

All is not lost though; does something like the below work?

---
 kernel/events/core.c | 12 ++++++++++--
 1 file changed, 10 insertions(+), 2 deletions(-)

diff --git a/kernel/events/core.c b/kernel/events/core.c
index 1e33b9141f03..057f599ae0dc 100644
--- a/kernel/events/core.c
+++ b/kernel/events/core.c
@@ -4742,12 +4742,20 @@ static const struct file_operations perf_fops = {
  * to user-space before waking everybody up.
  */
 
+static inline struct fasync_struct **perf_event_fasync(struct perf_event *event)
+{
+	/* only the parent has fasync state */
+	if (event->parent)
+		event = event->parent;
+	return &event->fasync;
+}
+
 void perf_event_wakeup(struct perf_event *event)
 {
 	ring_buffer_wakeup(event);
 
 	if (event->pending_kill) {
-		kill_fasync(&event->fasync, SIGIO, event->pending_kill);
+		kill_fasync(perf_event_fasync(event), SIGIO, event->pending_kill);
 		event->pending_kill = 0;
 	}
 }
@@ -6126,7 +6134,7 @@ static int __perf_event_overflow(struct perf_event *event,
 	else
 		perf_event_output(event, data, regs);
 
-	if (event->fasync && event->pending_kill) {
+	if (*perf_event_fasync(event) && event->pending_kill) {
 		event->pending_wakeup = 1;
 		irq_work_queue(&event->pending);
 	}


  reply	other threads:[~2015-06-11  8:32 UTC|newest]

Thread overview: 9+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2015-05-28 19:06 [patch] inherited events not signalling parent on overflow Vince Weaver
2015-05-28 19:15 ` Peter Zijlstra
2015-05-29 16:45   ` Vince Weaver
2015-05-29  6:36 ` Ingo Molnar
2015-06-11  4:30   ` Vince Weaver
2015-06-11  8:32     ` Peter Zijlstra [this message]
2015-07-31  4:42       ` Vince Weaver
2015-07-31  9:26         ` Peter Zijlstra
2015-08-04  8:51       ` [tip:perf/urgent] perf: Fix fasync handling on inherited events tip-bot for Peter Zijlstra

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=1434011521.1495.71.camel@twins \
    --to=peterz@infradead.org \
    --cc=acme@kernel.org \
    --cc=eranian@google.com \
    --cc=linux-kernel@vger.kernel.org \
    --cc=mingo@kernel.org \
    --cc=paulus@samba.org \
    --cc=vincent.weaver@maine.edu \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.