Re: [PATCH v1 7/8] tracing/perf: Add might_fault check to syscall probes

[Mathieu Desnoyers]

On 2024-10-04 00:37, Steven Rostedt wrote:
> On Thu,  3 Oct 2024 11:16:37 -0400
> Mathieu Desnoyers <mathieu.desnoyers@xxxxxxxxxxxx> wrote:
>
> > Add a might_fault() check to validate that the perf sys_enter/sys_exit
> > probe callbacks are indeed called from a context where page faults can
> > be handled.
> >
> > Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@xxxxxxxxxxxx>
> > Cc: Michael Jeanson <mjeanson@xxxxxxxxxxxx>
> > Cc: Steven Rostedt <rostedt@xxxxxxxxxxx>
> > Cc: Masami Hiramatsu <mhiramat@xxxxxxxxxx>
> > Cc: Peter Zijlstra <peterz@xxxxxxxxxxxxx>
> > Cc: Alexei Starovoitov <ast@xxxxxxxxxx>
> > Cc: Yonghong Song <yhs@xxxxxx>
> > Cc: Paul E. McKenney <paulmck@xxxxxxxxxx>
> > Cc: Ingo Molnar <mingo@xxxxxxxxxx>
> > Cc: Arnaldo Carvalho de Melo <acme@xxxxxxxxxx>
> > Cc: Mark Rutland <mark.rutland@xxxxxxx>
> > Cc: Alexander Shishkin <alexander.shishkin@xxxxxxxxxxxxxxx>
> > Cc: Namhyung Kim <namhyung@xxxxxxxxxx>
> > Cc: Andrii Nakryiko <andrii.nakryiko@xxxxxxxxx>
> > Cc: bpf@xxxxxxxxxxxxxxx
> > Cc: Joel Fernandes <joel@xxxxxxxxxxxxxxxxx>
> > ---
> >   include/trace/perf.h          | 1 +
> >   kernel/trace/trace_syscalls.c | 2 ++
> >   2 files changed, 3 insertions(+)
> >
> > diff --git a/include/trace/perf.h b/include/trace/perf.h
> > index 5650c1bad088..321bfd7919f6 100644
> > --- a/include/trace/perf.h
> > +++ b/include/trace/perf.h
> > @@ -84,6 +84,7 @@ perf_trace_##call(void *__data, proto)					\
> >   	u64 __count __attribute__((unused));				\
> >   	struct task_struct *__task __attribute__((unused));		\
> >   									\
> > +	might_fault();							\
> >   	guard(preempt_notrace)();					\
> >   	do_perf_trace_##call(__data, args);				\
>
> Same for this. This is used for all tracepoints that perf hooks to.

You're also missing the context:

#undef DECLARE_EVENT_SYSCALL_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));             \
                                                                        \
        might_fault();                                                  \
        guard(preempt_notrace)();                                       \
        do_perf_trace_##call(__data, args);                             \
}

Not an issue.

Thanks,

Mathieu


> -- Steve
>
> >   }
> > diff --git a/kernel/trace/trace_syscalls.c b/kernel/trace/trace_syscalls.c
> > index 89d7e4c57b5b..0d42d6f293d6 100644
> > --- a/kernel/trace/trace_syscalls.c
> > +++ b/kernel/trace/trace_syscalls.c
> > @@ -602,6 +602,7 @@ static void perf_syscall_enter(void *ignore, struct pt_regs *regs, long id)
> >   	 * Syscall probe called with preemption enabled, but the ring
> >   	 * buffer and per-cpu data require preemption to be disabled.
> >   	 */
> > +	might_fault();
> >   	guard(preempt_notrace)();
> >   
> >   	syscall_nr = trace_get_syscall_nr(current, regs);
> > @@ -710,6 +711,7 @@ static void perf_syscall_exit(void *ignore, struct pt_regs *regs, long ret)
> >   	 * Syscall probe called with preemption enabled, but the ring
> >   	 * buffer and per-cpu data require preemption to be disabled.
> >   	 */
> > +	might_fault();
> >   	guard(preempt_notrace)();
> >   
> >   	syscall_nr = trace_get_syscall_nr(current, regs);

--
Mathieu Desnoyers
EfficiOS Inc.
https://www.efficios.com