Re: [PATCH] powerpc/powernv: Check for IRQHAPPENED before sleeping

[Michael Neuling <mikey@neuling.org>]

On Tue, 2014-07-01 at 10:52 +1000, Michael Ellerman wrote:
> On Mon, 2014-06-30 at 11:54 +0530, Preeti U Murthy wrote:
> > Commit 8d6f7c5a: "powerpc/powernv: Make it possible to skip the IRQHAPP=
ENED
> > check in power7_nap()" added code that prevents even cores which enter =
sleep
> > on idle, from checking for pending interrupts. Fix this.
>=20
> To be clear, it is a bug in the original commit. Mikey and I pair program=
med
> that code, and he was *definitely* typing at that point ;)

It's always nice to know that in battle a comrade will throw himself on
a grenade to save you. :-P

Actually, I'm pretty sure this happened when you were away (probably on
a beach somewhere), I was forward porting the patch set, hit a conflict
and fucked it up. =20

Mikey

>=20
> Can you please include in the changelog what the symptom of the bug is, a=
nd
> have you seen it in practice?
>=20
> The bad commit went into 3.16-rc1, so we don't need to backport this to a=
ny
> stable release.
>=20
> cheers
>=20
> > diff --git a/arch/powerpc/kernel/idle_power7.S b/arch/powerpc/kernel/id=
le_power7.S
> > index 2480256..5cf3d36 100644
> > --- a/arch/powerpc/kernel/idle_power7.S
> > +++ b/arch/powerpc/kernel/idle_power7.S
> > @@ -131,7 +131,7 @@ _GLOBAL(power7_nap)
> > =20
> >  _GLOBAL(power7_sleep)
> >  	li	r3,1
> > -	li	r4,0
> > +	li	r4,1
> >  	b	power7_powersave_common
> >  	/* No return */
> > =20
> >=20
>=20
>=20
>=20
>=20
>=20