Re: irq-disabled vs vmap vs text_poke

[Ingo Molnar <mingo@elte.hu>]

* Mathieu Desnoyers <mathieu.desnoyers@polymtl.ca> wrote:

> * Peter Zijlstra (peterz@infradead.org) wrote:
> > On Fri, 2009-02-13 at 09:25 -0500, Mathieu Desnoyers wrote:
> > > * Ingo Molnar (mingo@elte.hu) wrote:
> > > > 
> > > > * Peter Zijlstra <peterz@infradead.org> wrote:
> > > > 
> > > > > > I think I got a mostly working patch cobbled together sitting here
> > > > > > somewhere. I was waiting for some _really_ good use case before spending
> > > > > > more time on it. I would prefer if at all possible to do vmap operations
> > > > > > in sleepable, process context.
> > > > > 
> > > > > Agreed, I think we want to fix text_poke() and make the vmap/vunmap()
> > > > > ops yell louder at violations of these rules.
> > > > > 
> > > > > I'm just totally clueless wrt text_poke() hence this email ;-)
> > > > 
> > > > also, this started triggering yesterday for the first time - and never
> > > > saw it before. Has some commit caused this side-effect?
> > > > 
> > > > It triggers during kprobes self-test - has that been improved recently?
> > > > 
> > > 
> > > When is this self-test run ? If it's at early boot while still in UP
> > > with interrupts off, kprobes should probably use text_poke_early()
> > > rather than text_poke().
> > 
> > Looking at the dmesg it looks to be post smp-init, so its late init
> > calls.
> > 
> > I think its the do_initcalls() from do_basic_setup(). So the machine
> > should be mostly up and running.
> 
> Here with 2.6.29-rc4 (commit 8e4921515c1a379539607eb443d51c30f4f7f338).
> I don't seem to get any warning with a might_sleep() in text_poke.
> 
> Kprobe smoke test started
> Kprobe smoke test passed successfully

it doesnt always trigger. It triggers on a box that also generates a
lot of NMIs - maybe there's a connection?

	Ingo