Messages in this thread | | | Date | Wed, 25 Feb 2009 09:40:00 +0100 | From | Ingo Molnar <> | Subject | Re: [PATCH 2/4] tracing: add event trace infrastructure |
| |
* Andrew Morton <akpm@linux-foundation.org> wrote:
> But we don't do that, because userspace is hard, because we > don't have a delivery process. But nobody has even tried! > [...]
Actually, you can do it right now, with the current model:
echo raw > /debug/tracing/trace_options echo bin > /debug/tracing/trace_options
Reality is that rarely does anyone use it because obviously self-sufficient kernel instrumentation is very convenient to kernel developers, to testers and to bugreporters alike. (Processing kernel details in user-space is also often slower and duplicates code that the kernel already has like symbol lookups.)
_Maybe_ some of it (but certainly not all) could be done via klibc - but that got nacked. If you can convince Linus to take klibc and to allow a userspace 'halo' of utilities that are intimate with kernel details into the kernel proper then maybe we could start providing such tools in the kernel proper.
Ingo
| |