On Tue, 2016-07-05 at 15:31 +1000, Michael Ellerman wrote:

On Tue, 2016-28-06 at 03:11:39 UTC, Sam Mendoza-Jonas wrote:

quoted

diff --git a/drivers/tty/hvc/hvc_opal.c b/drivers/tty/hvc/hvc_opal.c
index b7cd0ae..8c53f5b 100644
--- a/drivers/tty/hvc/hvc_opal.c
+++ b/drivers/tty/hvc/hvc_opal.c

@@ -214,7 +216,15 @@ static int hvc_opal_probe(struct platform_device *dev)
 		dev->dev.of_node->full_name,
 		boot ? " (boot console)" : "");
 
-	irq = opal_event_request(ilog2(OPAL_EVENT_CONSOLE_INPUT));
+	rc = of_property_read_u32(dev->dev.of_node, "interrupts", &prop);
+	if (rc) {
+		pr_info("hvc%d: No interrupts property, using OPAL event\n",
+				termno);
+		irq = opal_event_request(ilog2(OPAL_EVENT_CONSOLE_INPUT));
+	} else {
+		irq = irq_of_parse_and_map(dev->dev.of_node, 0);
+	}

That seems a bit backward.

Shouldn't we try irq_of_parse_and_map() and if that fails, then we go back to
opal_event_request() ?

cheers

I was unsure for a minute so I double checked this;
of_property_read_u32() returns of_property_read_u32_array(), which
returns 0 on success, unlike a bunch of other of_property helpers.

So if the 'interrupts' property does exist we get 0, and try
irq_of_parse_and_map(). But are you suggesting we try
irq_of_parse_and_map() regardless and then fall back to
opal_event_request()?

Cheers,
Sam