On Thu, 18 Sep 2025 12:38:56 +0100 Pedro Falcato <[email protected]> wrote:
> Isn't the usual problem with that approach, that of static functions/static > inline functions? I was tracing through a problem a few months ago, and man > I really did think "wouldn't it be nice to have a tracepoint instead of > fishing > around for kprobe spots". > > Not that I particularly think a tracepoint is super worth it in this case, > but, > y'know. Yes, it would be useful. The issue is that tracepoints are not free. They do increase the I$ hit and take up memory. If you're going to inject a tracepoint somewhere, at least extract some useful information from that spot. If you can't think of anything to track, then it's not worth a tracepoint. If one really wants a way to track something, they could add a static branch that would call a function when enabled that could be traced. They would also need a way to enable that static branch. -- Steve
