Re: [PATCH 2/3] irqchip: SigmaStar SSD20xD gpi

[Marc Zyngier <maz@kernel.org>]

On Mon, 20 Sep 2021 11:05:26 +0100,
Daniel Palmer <daniel@0x0f.com> wrote:
> 
> Hi Marc,
> 
> On Mon, 20 Sept 2021 at 18:45, Marc Zyngier <maz@kernel.org> wrote:
> > > +static void ssd20xd_gpi_unmask_irq(struct irq_data *data)
> > > +{
> > > +     irq_hw_number_t hwirq = irqd_to_hwirq(data);
> > > +     struct ssd20xd_gpi *gpi = irq_data_get_irq_chip_data(data);
> > > +     int offset_reg = REG_OFFSET(hwirq);
> > > +     int offset_bit = BIT_OFFSET(hwirq);
> > > +
> > > +     regmap_update_bits(gpi->regmap, REG_MASK + offset_reg, offset_bit, 0);
> >
> > Is this regmap call atomic? When running this, you are holding a
> > raw_spinlock already. From what I can see, this is unlikely to work
> > correctly with the current state of regmap.
> 
> I didn't even think about it. I will check.

You may want to enable lockdep to verify that.

> 
> > > +static void ssd20x_gpi_chainedhandler(struct irq_desc *desc)
> > > +{
> > > +     struct ssd20xd_gpi *intc = irq_desc_get_handler_data(desc);
> > > +     struct irq_chip *chip = irq_desc_get_chip(desc);
> > > +     unsigned int irqbit, hwirq, virq, status;
> > > +     int i;
> > > +
> > > +     chained_irq_enter(chip, desc);
> > > +
> > > +     for (i = 0; i < NUM_IRQ / IRQS_PER_REG; i++) {
> > > +             int offset_reg = STRIDE * i;
> > > +             int offset_irq = IRQS_PER_REG * i;
> > > +
> > > +             regmap_read(intc->regmap, REG_STATUS + offset_reg, &status);
> >
> > Does this act as an ACK in the HW?
> 
> Not that I'm aware of. The status registers have the interrupt bits
> set until the EOI callback is called from what I can tell.

Then this doesn't work for edge signalling, as you will lose
interrupts that occur between the handling and EOI.

> Technically I think the EOI callback should actually be ACK instead
> but from my fuzzy memory with the stack of IRQ controllers that
> resulted in a null pointer dereference.

That's probably because you are using the wrong flow handler. You
should turn this irq_eoi into an irq_ack, because that's really what
it is, and use handle_edge_irq() as the flow handler.

Thanks,

	M.

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