linux-kernel - Re: [PATCH] perf/bpf: Don't call bpf_overflow

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: <20240715111208.GB14400@noisy.programming.kicks-ass.net>
Date: Mon, 15 Jul 2024 13:12:08 +0200
From: Peter Zijlstra <peterz@...radead.org>
To: Jiri Olsa <olsajiri@...il.com>
Cc: Kyle Huey <me@...ehuey.com>, khuey@...ehuey.com,
	Ingo Molnar <mingo@...hat.com>, Namhyung Kim <namhyung@...nel.org>,
	Linus Torvalds <torvalds@...ux-foundation.org>,
	robert@...llahan.org, Joe Damato <jdamato@...tly.com>,
	Arnaldo Carvalho de Melo <acme@...nel.org>,
	Mark Rutland <mark.rutland@....com>,
	Alexander Shishkin <alexander.shishkin@...ux.intel.com>,
	Ian Rogers <irogers@...gle.com>,
	Adrian Hunter <adrian.hunter@...el.com>,
	"Liang, Kan" <kan.liang@...ux.intel.com>,
	Andrii Nakryiko <andrii@...nel.org>, Song Liu <song@...nel.org>,
	linux-perf-users@...r.kernel.org, linux-kernel@...r.kernel.org,
	bpf@...r.kernel.org
Subject: Re: [PATCH] perf/bpf: Don't call bpf_overflow_handler() for tracing
 events

On Sat, Jul 13, 2024 at 10:32:07PM +0200, Jiri Olsa wrote:
> On Fri, Jul 12, 2024 at 09:46:45PM -0700, Kyle Huey wrote:
> > The regressing commit is new in 6.10. It assumed that anytime event->prog
> > is set bpf_overflow_handler() should be invoked to execute the attached bpf
> > program. This assumption is false for tracing events, and as a result the
> > regressing commit broke bpftrace by invoking the bpf handler with garbage
> > inputs on overflow.
> > 
> > Prior to the regression the overflow handlers formed a chain (of length 0,
> > 1, or 2) and perf_event_set_bpf_handler() (the !tracing case) added
> > bpf_overflow_handler() to that chain, while perf_event_attach_bpf_prog()
> > (the tracing case) did not. Both set event->prog. The chain of overflow
> > handlers was replaced by a single overflow handler slot and a fixed call to
> > bpf_overflow_handler() when appropriate. This modifies the condition there
> > to include !perf_event_is_tracing(), restoring the previous behavior and
> > fixing bpftrace.
> > 
> > Signed-off-by: Kyle Huey <khuey@...ehuey.com>
> > Reported-by: Joe Damato <jdamato@...tly.com>
> > Fixes: f11f10bfa1ca ("perf/bpf: Call BPF handler directly, not through overflow machinery")
> > Tested-by: Joe Damato <jdamato@...tly.com> # bpftrace
> > Tested-by: Kyle Huey <khuey@...ehuey.com> # bpf overflow handlers
> > ---
> >  kernel/events/core.c | 11 ++++++++++-
> >  1 file changed, 10 insertions(+), 1 deletion(-)
> > 
> > diff --git a/kernel/events/core.c b/kernel/events/core.c
> > index 8f908f077935..f0d7119585dc 100644
> > --- a/kernel/events/core.c
> > +++ b/kernel/events/core.c
> > @@ -9666,6 +9666,8 @@ static inline void perf_event_free_bpf_handler(struct perf_event *event)
> >   * Generic event overflow handling, sampling.
> >   */
> >  
> > +static bool perf_event_is_tracing(struct perf_event *event);
> > +
> >  static int __perf_event_overflow(struct perf_event *event,
> >  				 int throttle, struct perf_sample_data *data,
> >  				 struct pt_regs *regs)
> > @@ -9682,7 +9684,9 @@ static int __perf_event_overflow(struct perf_event *event,
> >  
> >  	ret = __perf_event_account_interrupt(event, throttle);
> >  
> > -	if (event->prog && !bpf_overflow_handler(event, data, regs))
> > +	if (event->prog &&
> > +	    !perf_event_is_tracing(event) &&
> > +	    !bpf_overflow_handler(event, data, regs))
> >  		return ret;
> 
> ok makes sense, it's better to follow the perf_event_set_bpf_prog condition
> 
> Reviewed-by: Jiri Olsa <jolsa@...nel.org>

Urgh, so wth does event_is_tracing do with event->prog? And can't we
clean this up?

That whole perf_event_is_tracing() is a pretty gross function.

Also, I think the default return value of bpf_overflow_handler() is
wrong -- note how if !event->prog we won't call bpf_overflow_handler(),
but if we do call it, but then have !event->prog on the re-read, we
still return 0.