On Fri, Aug 1, 2025 at 8:19 AM Steven Rostedt <rost...@goodmis.org> wrote: > > On Fri, 1 Aug 2025 08:12:08 -0700 > Alexei Starovoitov <alexei.starovoi...@gmail.com> wrote: > > > > > You can use it in kernel/trace/trace_output.c, of course, > > I guess that means I should just use the DEFINE_FREE() in that file > directly and not in the btf.h header file?
yes. I don't want this to proliferate and people spam us with this kind of "cleanups". > > but I really think it's a step back in maintainability. > > All this cleanup.h is not a silver bullet. It needs to be used sparingly. > > I have my reservations about the cleanup.h code too. But the more I use it, > the more I like it. My biggest worry is guard() leak. That is, having a > lock or interrupt/preemption disabled for longer than they need to be, > because code was added at the end of the function after the protection is > needed. guard() is ok. We use it too, but __free() is imo garbage. It's essence of what's wrong with C++
