[PATCH] ARM: rockchip: keep reset control around

[Heiko Stuebner]

Am Freitag, 22. Mai 2026, 10:20:06 Mitteleuropäische Sommerzeit schrieb Philipp Zabel:
> On Fr, 2026-05-22 at 09:20 +0200, Heiko Stuebner wrote:
> > Am Donnerstag, 21. Mai 2026, 23:09:15 Mitteleuropäische Sommerzeit schrieb Heiko Stuebner:
> > > Do not put the reset control, retain exclusive control over it.
> > > After turning on a CPU, the corresponding reset line must stay
> > > deasserted.
> > > 
> > > This also avoids calling reset_control_put() before workqueues
> > > are operational.
> > > 
> > > Fixes: 78ebbff6d1a0 ("reset: handle removing supplier before consumers")
> > > Signed-off-by: Philipp Zabel <p.zabel at pengutronix.de>
> > > Tested-by: Steven Price <steven.price at arm.com>
> > > Signed-off-by: Heiko Stuebner <heiko at sntech.de>
> > > ---
> > >  arch/arm/mach-rockchip/platsmp.c | 16 ++++++++++------
> > >  1 file changed, 10 insertions(+), 6 deletions(-)
> > > 
> > > diff --git a/arch/arm/mach-rockchip/platsmp.c b/arch/arm/mach-rockchip/platsmp.c
> > > index f432d22bfed8..f659d894bfae 100644
> > > --- a/arch/arm/mach-rockchip/platsmp.c
> > > +++ b/arch/arm/mach-rockchip/platsmp.c
> > > @@ -34,6 +34,7 @@ static int ncores;
> > >  
> > >  static struct regmap *pmu;
> > >  static int has_pmu = true;
> > > +static struct reset_control *cpu_rstc[4];
> > 
> > After sleeping on that, this should be cpu_rstc[5];
> > 
> > Coretx-A9 SoCs need to enable the SCU power-domain which thankfully
> > sits at index 4 of the power-domain register.
> > 
> > So while we (already) expect no reset control for that, we need at least
> > make sure, it's not reading into undefined memory
> 
> The access in pmu_set_power_domain() is gated by
> (pd < ARRAY_SIZE(cpu_rstc)). And ncores in rockchip_smp_prepare_cpus()
> can not be larger than 4.

argh ... you're right of course. So the patch can stay as it is.
Should not think about code before going to sleep ;-)


Heiko

> > and thus need that empty field in the array.
> 
> We could drop the ARRAY_SIZE check and add a placeholder to the array,
> but it would need to be set to ERR_PTR(-EINVAL). Otherwise we'd have to
> replace all the IS_ERR(rstc) checks as well. I think that would be a
> good follow-up change.
> 
> regards
> Philipp
>