Re: [PATCH 05/19] irqchip/mmp: do not use of_address_to_resource() to get mux regs

[Marc Zyngier <maz@kernel.org>]

On Fri, 16 Aug 2019 20:41:22 +0200
Lubomir Rintel <lkundrak@v3.sk> wrote:

> On Fri, 2019-08-09 at 13:12 +0100, Marc Zyngier wrote:
> > On 09/08/2019 10:31, Lubomir Rintel wrote:  
> > > The "regs" property of the "mrvl,mmp2-mux-intc" devices are silly. They
> > > are offsets from intc's base, not addresses on the parent bus. At this
> > > point it probably can't be fixed.
> > > 
> > > On an OLPC XO-1.75 machine, the muxes are children of the intc, not the
> > > axi bus, and thus of_address_to_resource() won't work. We should treat
> > > the values as mere integers as opposed to bus addresses.
> > > 
> > > Signed-off-by: Lubomir Rintel <lkundrak@v3.sk>
> > > Acked-by: Pavel Machek <pavel@ucw.cz>
> > > 
> > > ---
> > >  drivers/irqchip/irq-mmp.c | 20 +++++++++++---------
> > >  1 file changed, 11 insertions(+), 9 deletions(-)
> > > 
> > > diff --git a/drivers/irqchip/irq-mmp.c b/drivers/irqchip/irq-mmp.c
> > > index 14618dc0bd396..af9cba4a51c2e 100644
> > > --- a/drivers/irqchip/irq-mmp.c
> > > +++ b/drivers/irqchip/irq-mmp.c
> > > @@ -424,9 +424,9 @@ IRQCHIP_DECLARE(mmp2_intc, "mrvl,mmp2-intc", mmp2_of_init);
> > >  static int __init mmp2_mux_of_init(struct device_node *node,
> > >  				   struct device_node *parent)
> > >  {
> > > -	struct resource res;
> > >  	int i, ret, irq, j = 0;
> > >  	u32 nr_irqs, mfp_irq;
> > > +	u32 reg[4];
> > >  
> > >  	if (!parent)
> > >  		return -ENODEV;
> > > @@ -438,18 +438,20 @@ static int __init mmp2_mux_of_init(struct device_node *node,
> > >  		pr_err("Not found mrvl,intc-nr-irqs property\n");
> > >  		return -EINVAL;
> > >  	}
> > > -	ret = of_address_to_resource(node, 0, &res);
> > > +
> > > +	/*
> > > +	 * For historical reasonsm, the "regs" property of the
> > > +	 * mrvl,mmp2-mux-intc is not a regular * "regs" property containing
> > > +	 * addresses on the parent bus, but offsets from the intc's base.
> > > +	 * That is why we can't use of_address_to_resource() here.
> > > +	 */
> > > +	ret = of_property_read_u32_array(node, "reg", reg, ARRAY_SIZE(reg));  
> > 
> > This will return 0 even if you've read less than your expected 4 u32s.
> > You may want to try of_property_read_variable_u32_array instead.  
> 
> Will it? Unless I'm reading the of_property_read_u32_array()
> documentation wrong, it suggests that would return -EOVERFLOW in that
> case.

You're appear to be right, and I read it wrong.

> 
> It ignores the extra values it the property is larger. I guess that is
> not a good thing and we still want to use
> of_property_read_variable_u32_array() though.

It doesn't hurt to check for all possible problems, specially given
that this machine doesn't appear to have a mainline DT (and its OF
implementation looks a bit buggy).

Thanks,

	M.
-- 
Without deviation from the norm, progress is not possible.