On 2024-10-04 00:38, Steven Rostedt wrote:
On Thu,  3 Oct 2024 11:16:38 -0400
Mathieu Desnoyers <[email protected]> wrote:

Add a might_fault() check to validate that the bpf sys_enter/sys_exit
probe callbacks are indeed called from a context where page faults can
be handled.

Signed-off-by: Mathieu Desnoyers <[email protected]>
Acked-by: Andrii Nakryiko <[email protected]>
Tested-by: Andrii Nakryiko <[email protected]> # BPF parts
Cc: Michael Jeanson <[email protected]>
Cc: Steven Rostedt <[email protected]>
Cc: Masami Hiramatsu <[email protected]>
Cc: Peter Zijlstra <[email protected]>
Cc: Alexei Starovoitov <[email protected]>
Cc: Yonghong Song <[email protected]>
Cc: Paul E. McKenney <[email protected]>
Cc: Ingo Molnar <[email protected]>
Cc: Arnaldo Carvalho de Melo <[email protected]>
Cc: Mark Rutland <[email protected]>
Cc: Alexander Shishkin <[email protected]>
Cc: Namhyung Kim <[email protected]>
Cc: Andrii Nakryiko <[email protected]>
Cc: [email protected]
Cc: Joel Fernandes <[email protected]>
---
  include/trace/bpf_probe.h | 1 +
  1 file changed, 1 insertion(+)

diff --git a/include/trace/bpf_probe.h b/include/trace/bpf_probe.h
index 211b98d45fc6..099df5c3e38a 100644
--- a/include/trace/bpf_probe.h
+++ b/include/trace/bpf_probe.h
@@ -57,6 +57,7 @@ __bpf_trace_##call(void *__data, proto)                       
                \
  static notrace void                                                   \
  __bpf_trace_##call(void *__data, proto)                                       
\
  {                                                                     \
+       might_fault();                                                  \

And I think this gets called at places that do not allow faults.

Context matters:

#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

        guard(preempt_notrace)();                                       \
        CONCATENATE(bpf_trace_run, COUNT_ARGS(args))(__data, 
CAST_TO_U64(args));        \
  }


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


Reply via email to