linux-kernel - Re: [PATCH v1 3/8] tracing/perf: guard syscall probe with preempt

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: <20241003182508.6ca76abc@gandalf.local.home>
Date: Thu, 3 Oct 2024 18:25:08 -0400
From: Steven Rostedt <rostedt@...dmis.org>
To: Mathieu Desnoyers <mathieu.desnoyers@...icios.com>
Cc: Masami Hiramatsu <mhiramat@...nel.org>, linux-kernel@...r.kernel.org,
 Peter Zijlstra <peterz@...radead.org>, Alexei Starovoitov <ast@...nel.org>,
 Yonghong Song <yhs@...com>, "Paul E . McKenney" <paulmck@...nel.org>, Ingo
 Molnar <mingo@...hat.com>, Arnaldo Carvalho de Melo <acme@...nel.org>, Mark
 Rutland <mark.rutland@....com>, Alexander Shishkin
 <alexander.shishkin@...ux.intel.com>, Namhyung Kim <namhyung@...nel.org>,
 Andrii Nakryiko <andrii.nakryiko@...il.com>, bpf@...r.kernel.org, Joel
 Fernandes <joel@...lfernandes.org>, linux-trace-kernel@...r.kernel.org,
 Michael Jeanson <mjeanson@...icios.com>, Frederic Weisbecker
 <fweisbec@...il.com>
Subject: Re: [PATCH v1 3/8] tracing/perf: guard syscall probe with
 preempt_notrace

On Thu,  3 Oct 2024 11:16:33 -0400
Mathieu Desnoyers <mathieu.desnoyers@...icios.com> wrote:

> In preparation for allowing system call enter/exit instrumentation to
> handle page faults, make sure that perf can handle this change by
> explicitly disabling preemption within the perf system call tracepoint
> probes to respect the current expectations within perf ring buffer code.
> 
> This change does not yet allow perf to take page faults per se within
> its probe, but allows its existing probes to adapt to the upcoming
> change.

Frederic,

Does the perf ring buffer expect preemption to be disabled when used?

In other words, is this patch needed?

-- Steve


> 
> Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@...icios.com>
> Cc: Michael Jeanson <mjeanson@...icios.com>
> Cc: Steven Rostedt <rostedt@...dmis.org>
> Cc: Masami Hiramatsu <mhiramat@...nel.org>
> Cc: Peter Zijlstra <peterz@...radead.org>
> Cc: Alexei Starovoitov <ast@...nel.org>
> Cc: Yonghong Song <yhs@...com>
> Cc: Paul E. McKenney <paulmck@...nel.org>
> Cc: Ingo Molnar <mingo@...hat.com>
> Cc: Arnaldo Carvalho de Melo <acme@...nel.org>
> Cc: Mark Rutland <mark.rutland@....com>
> Cc: Alexander Shishkin <alexander.shishkin@...ux.intel.com>
> Cc: Namhyung Kim <namhyung@...nel.org>
> Cc: Andrii Nakryiko <andrii.nakryiko@...il.com>
> Cc: bpf@...r.kernel.org
> Cc: Joel Fernandes <joel@...lfernandes.org>
> ---
>  include/trace/perf.h          | 41 +++++++++++++++++++++++++++++++----
>  kernel/trace/trace_syscalls.c | 12 ++++++++++
>  2 files changed, 49 insertions(+), 4 deletions(-)
> 
> diff --git a/include/trace/perf.h b/include/trace/perf.h
> index ded997af481e..5650c1bad088 100644
> --- a/include/trace/perf.h
> +++ b/include/trace/perf.h
> @@ -12,10 +12,10 @@
>  #undef __perf_task
>  #define __perf_task(t)	(__task = (t))
>  
> -#undef DECLARE_EVENT_CLASS
> -#define DECLARE_EVENT_CLASS(call, proto, args, tstruct, assign, print)	\
> +#undef __DECLARE_EVENT_CLASS
> +#define __DECLARE_EVENT_CLASS(call, proto, args, tstruct, assign, print) \
>  static notrace void							\
> -perf_trace_##call(void *__data, proto)					\
> +do_perf_trace_##call(void *__data, proto)				\
>  {									\
>  	struct trace_event_call *event_call = __data;			\
>  	struct trace_event_data_offsets_##call __maybe_unused __data_offsets;\
> @@ -55,8 +55,38 @@ perf_trace_##call(void *__data, proto)					\
>  				  head, __task);			\
>  }
>  
> +/*
> + * Define unused __count and __task variables to use @args to pass
> + * arguments to do_perf_trace_##call. This is needed because the
> + * macros __perf_count and __perf_task introduce the side-effect to
> + * store copies into those local variables.
> + */
> +#undef DECLARE_EVENT_CLASS
> +#define DECLARE_EVENT_CLASS(call, proto, args, tstruct, assign, print)	\
> +__DECLARE_EVENT_CLASS(call, PARAMS(proto), PARAMS(args), PARAMS(tstruct), \
> +		      PARAMS(assign), PARAMS(print))			\
> +static notrace void							\
> +perf_trace_##call(void *__data, proto)					\
> +{									\
> +	u64 __count __attribute__((unused));				\
> +	struct task_struct *__task __attribute__((unused));		\
> +									\
> +	do_perf_trace_##call(__data, args);				\
> +}
> +
>  #undef DECLARE_EVENT_SYSCALL_CLASS
> -#define DECLARE_EVENT_SYSCALL_CLASS DECLARE_EVENT_CLASS
> +#define DECLARE_EVENT_SYSCALL_CLASS(call, proto, args, tstruct, assign, print) \
> +__DECLARE_EVENT_CLASS(call, PARAMS(proto), PARAMS(args), PARAMS(tstruct), \
> +		      PARAMS(assign), PARAMS(print))			\
> +static notrace void							\
> +perf_trace_##call(void *__data, proto)					\
> +{									\
> +	u64 __count __attribute__((unused));				\
> +	struct task_struct *__task __attribute__((unused));		\
> +									\
> +	guard(preempt_notrace)();					\
> +	do_perf_trace_##call(__data, args);				\
> +}
>  
>  /*
>   * This part is compiled out, it is only here as a build time check
> @@ -76,4 +106,7 @@ static inline void perf_test_probe_##call(void)				\
>  	DEFINE_EVENT(template, name, PARAMS(proto), PARAMS(args))
>  
>  #include TRACE_INCLUDE(TRACE_INCLUDE_FILE)
> +
> +#undef __DECLARE_EVENT_CLASS
> +
>  #endif /* CONFIG_PERF_EVENTS */
> diff --git a/kernel/trace/trace_syscalls.c b/kernel/trace/trace_syscalls.c
> index ab4db8c23f36..edcfa47446c7 100644
> --- a/kernel/trace/trace_syscalls.c
> +++ b/kernel/trace/trace_syscalls.c
> @@ -596,6 +596,12 @@ static void perf_syscall_enter(void *ignore, struct pt_regs *regs, long id)
>  	int rctx;
>  	int size;
>  
> +	/*
> +	 * Syscall probe called with preemption enabled, but the ring
> +	 * buffer and per-cpu data require preemption to be disabled.
> +	 */
> +	guard(preempt_notrace)();
> +
>  	syscall_nr = trace_get_syscall_nr(current, regs);
>  	if (syscall_nr < 0 || syscall_nr >= NR_syscalls)
>  		return;
> @@ -698,6 +704,12 @@ static void perf_syscall_exit(void *ignore, struct pt_regs *regs, long ret)
>  	int rctx;
>  	int size;
>  
> +	/*
> +	 * Syscall probe called with preemption enabled, but the ring
> +	 * buffer and per-cpu data require preemption to be disabled.
> +	 */
> +	guard(preempt_notrace)();
> +
>  	syscall_nr = trace_get_syscall_nr(current, regs);
>  	if (syscall_nr < 0 || syscall_nr >= NR_syscalls)
>  		return;