On Wed, Jul 13, 2022 at 12:12 AM Alan Maguire <alan.maguire@xxxxxxxxxx> wrote: > > On 12/07/2022 05:24, Andrii Nakryiko wrote: > > > Sounds good! I'll add that to bpf_program__attach_ksyscall() doc > > comment (and to commit message). I'll implement those new virtual > > __kconfig variables that I mentioned in another thread and post it as > > v1, hopefully some time this week. > > > > This is really useful, thanks for doing it! I tested on arm64, > only issue was the tracefs path issue that I think was already > mentioned, i.e. for me it took > > diff --git a/tools/lib/bpf/libbpf.c b/tools/lib/bpf/libbpf.c > index 4749fb84e33d..a27f21619cfc 100644 > --- a/tools/lib/bpf/libbpf.c > +++ b/tools/lib/bpf/libbpf.c > @@ -4685,7 +4685,7 @@ static int probe_kern_syscall_wrapper(void) > * kernel was built with CONFIG_ARCH_HAS_SYSCALL_WRAPPER and uses > * syscall wrappers > */ > - static const char *kprobes_file = "/sys/kernel/tracing/available_filter_functions"; > + static const char *kprobes_file = "/sys/kernel/debug/tracing/available_filter_functions"; > char func_name[128], syscall_name[128]; > const char *ksys_pfx; > FILE *f; > > ...to get the feature probing - and hence auto-attach to the > right arch-specific probes - to work. > So I remember there were some patches generalizing this, but I can't remember why it hasn't landed yet. I'll dig it up a bit later, but I think this will be fixed separately and orthogonally. And also the non-RFC v1 doesn't rely on available_filter_functions anymore anyways. It would be nice if you can try v1 as well, and if it looks good give your ack/tested-by. Thanks! > Alan
