From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1753315Ab3KSTTj (ORCPT ); Tue, 19 Nov 2013 14:19:39 -0500 Received: from terminus.zytor.com ([198.137.202.10]:51778 "EHLO terminus.zytor.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1752752Ab3KSTTf (ORCPT ); Tue, 19 Nov 2013 14:19:35 -0500 Date: Tue, 19 Nov 2013 11:18:46 -0800 From: tip-bot for Peter Zijlstra Message-ID: Cc: linux-kernel@vger.kernel.org, hpa@zytor.com, mingo@kernel.org, peterz@infradead.org, davej@redhat.com, vincent.weaver@maine.edu, fweisbec@gmail.com, rostedt@goodmis.org, tglx@linutronix.de Reply-To: mingo@kernel.org, hpa@zytor.com, linux-kernel@vger.kernel.org, peterz@infradead.org, davej@redhat.com, vincent.weaver@maine.edu, fweisbec@gmail.com, rostedt@goodmis.org, tglx@linutronix.de In-Reply-To: <20131114152304.GC5364@laptop.programming.kicks-ass.net> References: <20131114152304.GC5364@laptop.programming.kicks-ass.net> To: linux-tip-commits@vger.kernel.org Subject: [tip:perf/urgent] ftrace, perf: Avoid infinite event generation loop Git-Commit-ID: d5b5f391d434c5cc8bcb1ab2d759738797b85f52 X-Mailer: tip-git-log-daemon Robot-ID: Robot-Unsubscribe: Contact to get blacklisted from these emails MIME-Version: 1.0 Content-Transfer-Encoding: 8bit Content-Type: text/plain; charset=UTF-8 Content-Disposition: inline X-Greylist: Sender IP whitelisted, not delayed by milter-greylist-4.5.1 (terminus.zytor.com [127.0.0.1]); Tue, 19 Nov 2013 11:18:53 -0800 (PST) Sender: linux-kernel-owner@vger.kernel.org List-ID: X-Mailing-List: linux-kernel@vger.kernel.org Commit-ID: d5b5f391d434c5cc8bcb1ab2d759738797b85f52 Gitweb: http://git.kernel.org/tip/d5b5f391d434c5cc8bcb1ab2d759738797b85f52 Author: Peter Zijlstra AuthorDate: Thu, 14 Nov 2013 16:23:04 +0100 Committer: Ingo Molnar CommitDate: Tue, 19 Nov 2013 16:57:40 +0100 ftrace, perf: Avoid infinite event generation loop Vince's perf-trinity fuzzer found yet another 'interesting' problem. When we sample the irq_work_exit tracepoint with period==1 (or PERF_SAMPLE_PERIOD) and we add an fasync SIGNAL handler we create an infinite event generation loop: ,-> | irq_work_exit() -> | trace_irq_work_exit() -> | ... | __perf_event_overflow() -> (due to fasync) | irq_work_queue() -> (irq_work_list must be empty) '--------- arch_irq_work_raise() Similar things can happen due to regular poll() wakeups if we exceed the ring-buffer wakeup watermark, or have an event_limit. To avoid this, dis-allow sampling this particular tracepoint. In order to achieve this, create a special perf_perm function pointer for each event and call this (when set) on trying to create a tracepoint perf event. [ roasted: use expr... to allow for ',' in your expression ] Reported-by: Vince Weaver Tested-by: Vince Weaver Signed-off-by: Peter Zijlstra Cc: Steven Rostedt Cc: Dave Jones Cc: Frederic Weisbecker Link: http://lkml.kernel.org/r/20131114152304.GC5364@laptop.programming.kicks-ass.net Signed-off-by: Ingo Molnar --- arch/x86/include/asm/trace/irq_vectors.h | 11 +++++++++++ include/linux/ftrace_event.h | 16 ++++++++++++++++ include/linux/tracepoint.h | 4 ++++ include/trace/ftrace.h | 7 +++++++ kernel/trace/trace_event_perf.c | 6 ++++++ 5 files changed, 44 insertions(+) diff --git a/arch/x86/include/asm/trace/irq_vectors.h b/arch/x86/include/asm/trace/irq_vectors.h index 2874df2..4cab890 100644 --- a/arch/x86/include/asm/trace/irq_vectors.h +++ b/arch/x86/include/asm/trace/irq_vectors.h @@ -72,6 +72,17 @@ DEFINE_IRQ_VECTOR_EVENT(x86_platform_ipi); DEFINE_IRQ_VECTOR_EVENT(irq_work); /* + * We must dis-allow sampling irq_work_exit() because perf event sampling + * itself can cause irq_work, which would lead to an infinite loop; + * + * 1) irq_work_exit happens + * 2) generates perf sample + * 3) generates irq_work + * 4) goto 1 + */ +TRACE_EVENT_PERF_PERM(irq_work_exit, is_sampling_event(p_event) ? -EPERM : 0); + +/* * call_function - called when entering/exiting a call function interrupt * vector handler */ diff --git a/include/linux/ftrace_event.h b/include/linux/ftrace_event.h index 9abbe63..8c9b7a1 100644 --- a/include/linux/ftrace_event.h +++ b/include/linux/ftrace_event.h @@ -248,6 +248,9 @@ struct ftrace_event_call { #ifdef CONFIG_PERF_EVENTS int perf_refcount; struct hlist_head __percpu *perf_events; + + int (*perf_perm)(struct ftrace_event_call *, + struct perf_event *); #endif }; @@ -317,6 +320,19 @@ struct ftrace_event_file { } \ early_initcall(trace_init_flags_##name); +#define __TRACE_EVENT_PERF_PERM(name, expr...) \ + static int perf_perm_##name(struct ftrace_event_call *tp_event, \ + struct perf_event *p_event) \ + { \ + return ({ expr; }); \ + } \ + static int __init trace_init_perf_perm_##name(void) \ + { \ + event_##name.perf_perm = &perf_perm_##name; \ + return 0; \ + } \ + early_initcall(trace_init_perf_perm_##name); + #define PERF_MAX_TRACE_SIZE 2048 #define MAX_FILTER_STR_VAL 256 /* Should handle KSYM_SYMBOL_LEN */ diff --git a/include/linux/tracepoint.h b/include/linux/tracepoint.h index ebeab36..f16dc0a 100644 --- a/include/linux/tracepoint.h +++ b/include/linux/tracepoint.h @@ -267,6 +267,8 @@ static inline void tracepoint_synchronize_unregister(void) #define TRACE_EVENT_FLAGS(event, flag) +#define TRACE_EVENT_PERF_PERM(event, expr...) + #endif /* DECLARE_TRACE */ #ifndef TRACE_EVENT @@ -399,4 +401,6 @@ static inline void tracepoint_synchronize_unregister(void) #define TRACE_EVENT_FLAGS(event, flag) +#define TRACE_EVENT_PERF_PERM(event, expr...) + #endif /* ifdef TRACE_EVENT (see note above) */ diff --git a/include/trace/ftrace.h b/include/trace/ftrace.h index 52594b2..6b852f6 100644 --- a/include/trace/ftrace.h +++ b/include/trace/ftrace.h @@ -90,6 +90,10 @@ #define TRACE_EVENT_FLAGS(name, value) \ __TRACE_EVENT_FLAGS(name, value) +#undef TRACE_EVENT_PERF_PERM +#define TRACE_EVENT_PERF_PERM(name, expr...) \ + __TRACE_EVENT_PERF_PERM(name, expr) + #include TRACE_INCLUDE(TRACE_INCLUDE_FILE) @@ -140,6 +144,9 @@ #undef TRACE_EVENT_FLAGS #define TRACE_EVENT_FLAGS(event, flag) +#undef TRACE_EVENT_PERF_PERM +#define TRACE_EVENT_PERF_PERM(event, expr...) + #include TRACE_INCLUDE(TRACE_INCLUDE_FILE) /* diff --git a/kernel/trace/trace_event_perf.c b/kernel/trace/trace_event_perf.c index 78e27e3..630889f 100644 --- a/kernel/trace/trace_event_perf.c +++ b/kernel/trace/trace_event_perf.c @@ -24,6 +24,12 @@ static int total_ref_count; static int perf_trace_event_perm(struct ftrace_event_call *tp_event, struct perf_event *p_event) { + if (tp_event->perf_perm) { + int ret = tp_event->perf_perm(tp_event, p_event); + if (ret) + return ret; + } + /* The ftrace function trace is allowed only for root. */ if (ftrace_event_is_function(tp_event) && perf_paranoid_tracepoint_raw() && !capable(CAP_SYS_ADMIN))