Re: [PATCH 0/9] TRACE_IRQFLAGS wreckage

[Marco Elver]

On Fri, 21 Aug 2020 at 08:54, <peterz@xxxxxxxxxxxxx> wrote:
>
> On Thu, Aug 20, 2020 at 07:20:46PM +0200, Marco Elver wrote:
> > From 4ec9dd472c978e1eba622fb22bc04e4357f10421 Mon Sep 17 00:00:00 2001
> > From: Marco Elver <elver@xxxxxxxxxx>
> > Date: Thu, 20 Aug 2020 19:06:09 +0200
> > Subject: [PATCH] sched: Turn inline into __always_inline due to noinstr use
> >
> > is_idle_task() may be used from noinstr functions such as
> > irqentry_enter(). Since the compiler is free to not inline regular
> > inline functions, switch to using __always_inline.
> >
> > Signed-off-by: Marco Elver <elver@xxxxxxxxxx>
>
> Fair enough I suppose; didn't that generate an objtool warning on your
> build? (of course that requires PARAVIRT=n, I should get back to fixing
> that)

It does now after turning on VMLINUX_VALIDATION.

Although note that this patch (and the other, which I'll send
separately) wasn't required after the rcuidle fix. I suppose there is
still a small chance things can go wrong though, but syzkaller wasn't
able to find such a case (yet).

Thanks,
-- Marco