On Sat, 24 Oct 2020 10:31:12 +0900 Masami Hiramatsu <mhiramat@xxxxxxxxxx> wrote: > > Perhaps we should have: > > > > # older kernels have do_fork, but newer kernels have kernel_clone > > echo kernel_clone >> set_ftrace_filter || echo *do_fork >> set_ftrace_filter > > Good catch. BTW, can we check the filter-bility by grep the pattern from set_ftrace_filter? I think we need to use /proc/kallsyms, as the kprobe code should still work if function tracing is disabled, and the function filter files will not exist. -- Steve
