Re: [BUGFIX PATCH] kprobes/x86: Remove IRQ disabling from jprobe handlers
From: Ingo Molnar
Date: Thu Oct 05 2017 - 03:57:13 EST
* Masami Hiramatsu <mhiramat@xxxxxxxxxx> wrote:
> On Wed, 4 Oct 2017 12:41:01 +0200
> Ingo Molnar <mingo@xxxxxxxxxx> wrote:
> > * Masami Hiramatsu <mhiramat@xxxxxxxxxx> wrote:
> > > Hmm, actually we can not disable jprobe, that has no separate Kconfig.
> > > So we need to introduce new kconfig for that.
> > >
> > > And, there are several network protocols using jprobe to trace events.
> > > (e.g. NET_DCCPPROBE and NET_TCPPROBE)
> > > I think they need to migrate to trace-event at first.
> > >
> > > So, how about below idea?
> > >
> > > 1. Introduce CONFIG_JPROBE_API which only separate jprobe general parts
> > > (no arch dependent code involves) and make it default n.
> > > 2. Mark break_handler and jprobe APIs deprecated so that no new user comes up.
> > > 3. migrate in-kernel jprobe user to trace-event or ftrace.
> > > (may take some time)
> > So my suggestion would be to just return from register_jprobe() and don't register
> > anything.
> with CONFIG_JPROBE_API=n, is that right?
No, unconditionally off with a WARN_ON_ONCE() warning in the registry function and
the deactivation of all in-kernel uses (such as self-tests).
The point is to make people that _truly_ rely on it complain - not just make them
silently turn on a Kconfig option ...
> > Yes, there are usecases of jprobes in the kernel, but they all look
> > pretty ancient and unused.
> Hmm, in that case, should we also remove those users? If we disable such way
> those features are just useless.
My hypothesis is that those features are not used (hence useless), but we should
first test whether there's any reliance before we remove code.