tech-kern archive

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index][Old Index]

Re: Hunting kernel lock and interrupt latency





On Fri, Nov 18, 2022 at 03:31 Taylor R Campbell <campbell+netbsd-tech-kern%mumble.net@localhost> wrote:
> Date: Fri, 18 Nov 2022 10:06:19 +0000
> From: nia <nia%NetBSD.org@localhost>
>
> This is neat, thanks for the work. Do we need a kernel
> compiled with any specific options to make the DTrace
> probes active?

Just `options KDTRACE_HOOKS', which is on by default in x86 and arm


Is not MKDTRACE a consideration?

/bch


GENERIC kernels.  dtrace itself, and the various probes, are in
modules:

dtrace.kmod
dtrace_fbt.kmod
dtrace_lockstat.kmod
dtrace_profile.kmod (must be loaded explicitly for tick-1s &c.)
dtrace_sdt.kmod
dtrace_syscall.kmod

dtrace(1) will autoload the module dtrace_FOO.kmod when you try to set
a probe on FOO:*:*:* like fbt::hardclock:entry.  For the tick-1s,
tick-10s, &c. probes, you need to load dtrace_profile.kmod explicitly.


Home | Main Index | Thread Index | Old Index