On 04/30/2014 01:42 PM, Ezequiel Garcia wrote:

On Apr 29, Sebastian Hesselbarth wrote:

quoted

On 04/29/2014 09:49 PM, Ezequiel Garcia wrote:

quoted

The following commit:

commit 9ec36cafe43bf835f8f29273597a5b0cbc8267ef
Author: Rob Herring [off-list ref]
Date:   Wed Apr 23 17:57:41 2014 -0500

     of/irq: do irq resolution in platform_get_irq

changed platform_get_irq() which now returns ENODEV and EPROBE_DEFER,
in addition to ENXIO. If there's no interrupt for mvmdio, platform_get_irq()
returns ENODEV, but we currently check only for ENXIO.

Fix this by looking for a positive integer, which is the proper way of
validating a virtual interrupt number.

Signed-off-by: Ezequiel Garcia <redacted>
---
  drivers/net/ethernet/marvell/mvmdio.c | 2 +-
  1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/drivers/net/ethernet/marvell/mvmdio.c b/drivers/net/ethernet/marvell/mvmdio.c
index b161a52..eb2cabf 100644
--- a/drivers/net/ethernet/marvell/mvmdio.c
+++ b/drivers/net/ethernet/marvell/mvmdio.c

@@ -232,7 +232,7 @@ static int orion_mdio_probe(struct platform_device *pdev)
  		clk_prepare_enable(dev->clk);

  	dev->err_interrupt = platform_get_irq(pdev, 0);
-	if (dev->err_interrupt != -ENXIO) {
+	if (dev->err_interrupt > 0) {

Ezequiel,

I cannot find where Rob's mentioned patch set adds -ENODEV, but isn't

Well, I don't think it's not mentioned in the patch. The path is:

platform_get_irq -> of_irq_get -> of_irq_parse_one -> EINVAL.

So it's EINVAL, not ENODEV. But the lesson is to avoid checking for
a particular error (except EPROBE_DEFER which is special) because
it's a fragile practice.

Ok, thanks for the clarification.

quoted

the semantic for -EPROBE_DEFER: there *should* be an irq, but it is
not yet available. That basically means, we should also defer on that
error otherwise we would ignore that we have actually been given an irq
to work with, right?

Yes, I agree. Did another patch for that, but haven't send it yet.
AFAICS, mvebu platforms will never hit the deferred case as the irqchip
is the first driver registered (as per drivers/Makefile).

Not that we should count on that :)

It doesn't hit it _now_ because of the above. I read about proper
platform_device for early devices here and there over and over
again, so I guess some day it may become an issue.

As we know about the potential -EPROBE_DEFER now, I suggest to
deal with it now, too.

Can you resend this as v2 with the other patch you mentioned
squashed in?

Sebastian