Re: [PATCH] irqchip/loongson-liointc: Fix an error handling path in liointc_init()

[Christophe JAILLET]

Le 24/05/2022 à 05:47, Huacai Chen a écrit :
> Hi, Christophe,
>
> On Tue, May 24, 2022 at 10:50 AM Huacai Chen <chenhuacai@xxxxxxxxxx> wrote:
> > Hi, Christophe,
> >
> > On Sun, May 22, 2022 at 9:44 PM Christophe JAILLET
> > <christophe.jaillet@xxxxxxxxxx> wrote:
> > > If a of_property_match_string() call fails, we still need to release some
> > > resources.
> > > Add the corresponding goto instead of a direct return.
> > Your patch is correct, but 807e93d0ecbb hasn't been upstream, I don't
> > know how to handle it.
> >
> > Huacai
> > > Fixes: 807e93d0ecbb ("irqchip/loongson-liointc: Add ACPI init support")
> > > Signed-off-by: Christophe JAILLET <christophe.jaillet@xxxxxxxxxx>
> > > ---
> > >   drivers/irqchip/irq-loongson-liointc.c | 6 ++++--
> > >   1 file changed, 4 insertions(+), 2 deletions(-)
> > >
> > > diff --git a/drivers/irqchip/irq-loongson-liointc.c b/drivers/irqchip/irq-loongson-liointc.c
> > > index ff3cb5b05710..2227b702a81d 100644
> > > --- a/drivers/irqchip/irq-loongson-liointc.c
> > > +++ b/drivers/irqchip/irq-loongson-liointc.c
> > > @@ -185,8 +185,10 @@ static int liointc_init(phys_addr_t addr, unsigned long size, int revision,
> > >                          int index = of_property_match_string(node,
> > >                                          "reg-names", core_reg_names[i]);
> > >
> > > -                       if (index < 0)
> > > -                               return -EINVAL;
> > > +                       if (index < 0) {
> > > +                               err = -EINVAL;
> > > +                               goto out_iounmap;
> > > +                       }
> Just goto out_iounmap is OK, because it returns -EINVAL at last.

Yes, agreed.

> I've squash your patch to the original one and add a Co-developed-by:,
> not sure it is the best solution. Thanks.

Squashing in the original patch is fine for me. It is what is usually 
done in such cases.

The Co-developed-by: is maybe a bit to much. I've just fixed a bug that 
can't happen (IMHO) in real life. I'll let you decide what is more relevant.
With or without the tag is fine for me as well.

CJ


> Huacai
> > >                          priv->core_isr[i] = of_iomap(node, index);
> > >                  }
> > > --
> > > 2.34.1