Re: [PATCH v8 7/8] net-next: mvpp2: relax return value check for IRQ get

From: Matti Vaittinen
Date: Tue Aug 01 2023 - 08:44:26 EST

On 8/1/23 15:31, Russell King (Oracle) wrote:
On Tue, Aug 01, 2023 at 03:04:24PM +0300, Matti Vaittinen wrote:
fwnode_irq_get[_byname]() were changed to not return 0 anymore.

Drop check for return value 0.

Signed-off-by: Matti Vaittinen <mazziesaccount@xxxxxxxxx>

Sorry, but I don't think you've properly considered the effects of your

Don't be sorry.

@@ -5833,7 +5833,7 @@ static int mvpp2_multi_queue_vectors_init(struct mvpp2_port *port,
v->irq = of_irq_get_byname(port_node, irqname);
v->irq = fwnode_irq_get(port->fwnode, i);
- if (v->irq <= 0) {
+ if (v->irq < 0) {

You're making this change based on the assumption that fwnode_irq_get()
has changed its return values, but I really don't think you've looked
at the code and considered the return value behaviour of the DT function
above. Reading it's documentation, it states that of_irq_get_byname()
may return 0 on IRQ mapping failure.

You're correct. Thanks for spotting this! Seems like I really overlooked the behaviour of the of_irq_get_byname().

So, by making this change, you are allowing IRQ mapping failure in the
DT path to succeed rather than fail.

ret = -EINVAL;
goto err;
@@ -6764,7 +6764,7 @@ static int mvpp2_port_probe(struct platform_device *pdev,
goto err_deinit_qvecs;
- if (port->port_irq <= 0)
+ if (port->port_irq < 0)

Exactly the same problem here, but...

/* the link irq is optional */
port->port_irq = 0;

this is less critical... but still wrong.

So, given that this patch is basically incorrect...


Matti Vaittinen
Linux kernel developer at ROHM Semiconductors
Oulu Finland

~~ When things go utterly wrong vim users can always type :help! ~~