On 29/08/17 19:43, Florian Fainelli wrote: > Initialization of irq_fwd_mask was done using __raw_readl() which > happens to work for all cases except when using ARM BE8 which requires > readl() (with the proper swapping). Move the initialization of the
s/readl/writel/, I presume? > irq_fwd_mask till later when we have correctly defined our I/O > accessors. > > Signed-off-by: Florian Fainelli <[email protected]> > --- > drivers/irqchip/irq-bcm7120-l2.c | 10 ++++------ > 1 file changed, 4 insertions(+), 6 deletions(-) > > diff --git a/drivers/irqchip/irq-bcm7120-l2.c > b/drivers/irqchip/irq-bcm7120-l2.c > index 64c2692070ef..983640eba418 100644 > --- a/drivers/irqchip/irq-bcm7120-l2.c > +++ b/drivers/irqchip/irq-bcm7120-l2.c > @@ -250,12 +250,6 @@ static int __init bcm7120_l2_intc_probe(struct > device_node *dn, > if (ret < 0) > goto out_free_l1_data; > > - for (idx = 0; idx < data->n_words; idx++) { > - __raw_writel(data->irq_fwd_mask[idx], > - data->pair_base[idx] + > - data->en_offset[idx]); > - } > - > for (irq = 0; irq < data->num_parent_irqs; irq++) { > ret = bcm7120_l2_intc_init_one(dn, data, irq, valid_mask); > if (ret) > @@ -297,6 +291,10 @@ static int __init bcm7120_l2_intc_probe(struct > device_node *dn, > gc->reg_base = data->pair_base[idx]; > ct->regs.mask = data->en_offset[idx]; > > + /* gc->reg_base is defined and so is gc->writel */ > + irq_reg_writel(gc, data->irq_fwd_mask[idx], > + data->en_offset[idx]); > + > ct->chip.irq_mask = irq_gc_mask_clr_bit; > ct->chip.irq_unmask = irq_gc_mask_set_bit; > ct->chip.irq_ack = irq_gc_noop; > Thanks, M. -- Jazz is not dead. It just smells funny...
