On Tue, 27 Oct 2026 09:12:43 +0900 "Masami Hiramatsu (Google)" <[email protected]> wrote:
> BTW, to check the correct input is encoded correctly, we may need > in kernel selftest or a feature to dump the interpreted pred after > "#" line via filter file. Latter one allows us to test it via tracefs > (ftracetest). Hmm, I guess we can update the filter file, although I'm a bit nervous about backward compatibility. IIRC, the libtracefs does self tests and clears filters. That may be a way to see if it breaks user space. -- Steve
