On 8/3/23 2:31 PM, Zhu Wang wrote:
> When platform_get_irq_byname() is called to get a interrupt number, it
> may return -EINVAL or -ENXIO when failed, while current code returned
You still don't mention -EPROBE_DEFER. Propagating it upstream is
the _main_ reason we must not override what platform_get_irq_byname()
returns!
> -ENODEV, so we replace it with the return value of
> platform_get_irq_byname(). And we found that platform_get_irq_byname()
> never returned zero by reading its code.
It used to return 0 (as both IRQ0 and the error indication).
Historically, there were several patches fixing the inconsistencies...
> Commit ce753ad1549c ("platform: finally disallow IRQ0 in
> platform_get_irq() and its ilk") makes sure IRQ0 is not returned.
>
> Signed-off-by: Zhu Wang <[email protected]>
>
> ---
> Changes in v2:
> - Update the commit message, present the reason of replacing the return
> value of the probe.
> ---
> drivers/usb/musb/musb_core.c | 4 ++--
> 1 file changed, 2 insertions(+), 2 deletions(-)
>
> diff --git a/drivers/usb/musb/musb_core.c b/drivers/usb/musb/musb_core.c
> index ecbd3784bec3..b24adb5b399f 100644
> --- a/drivers/usb/musb/musb_core.c
> +++ b/drivers/usb/musb/musb_core.c
> @@ -2610,8 +2610,8 @@ static int musb_probe(struct platform_device *pdev)
> int irq = platform_get_irq_byname(pdev, "mc");
> void __iomem *base;
>
> - if (irq <= 0)
> - return -ENODEV;
> + if (irq < 0)
> + return irq;
>
> base = devm_platform_ioremap_resource(pdev, 0);
> if (IS_ERR(base))
>