bus_get_dev_root(&platform_bus_type) always returns NULL.
This makes mbigen_of_create_domain always return -ENODEV.

Fixes: fea087fc291b ("irqchip/mbigen: move to use bus_get_dev_root()")
Signed-off-by: Chen Jun <chenjun102@huawei.com>
---
 drivers/irqchip/irq-mbigen.c | 8 +-------
 1 file changed, 1 insertion(+), 7 deletions(-)

diff --git a/drivers/irqchip/irq-mbigen.c b/drivers/irqchip/irq-mbigen.c
index 5101a3fb11df..58881d313979 100644
--- a/drivers/irqchip/irq-mbigen.c
+++ b/drivers/irqchip/irq-mbigen.c
@@ -235,22 +235,17 @@ static const struct irq_domain_ops mbigen_domain_ops = {
 static int mbigen_of_create_domain(struct platform_device *pdev,
 				   struct mbigen_device *mgn_chip)
 {
-	struct device *parent;
 	struct platform_device *child;
 	struct irq_domain *domain;
 	struct device_node *np;
 	u32 num_pins;
 	int ret = 0;
 
-	parent = bus_get_dev_root(&platform_bus_type);
-	if (!parent)
-		return -ENODEV;
-
 	for_each_child_of_node(pdev->dev.of_node, np) {
 		if (!of_property_read_bool(np, "interrupt-controller"))
 			continue;
 
-		child = of_platform_device_create(np, NULL, parent);
+		child = of_platform_device_create(np, NULL, NULL);
 		if (!child) {
 			ret = -ENOMEM;
 			break;
@@ -273,7 +268,6 @@ static int mbigen_of_create_domain(struct platform_device *pdev,
 		}
 	}
 
-	put_device(parent);
 	if (ret)
 		of_node_put(np);
 
-- 
2.17.1

On Mon, Jan 29 2024 at 21:00, Chen Jun wrote:
> bus_get_dev_root(&platform_bus_type) always returns NULL.
> This makes mbigen_of_create_domain always return -ENODEV.
>
> Fixes: fea087fc291b ("irqchip/mbigen: move to use bus_get_dev_root()")

Amazing that it took 9 month to figure that out ...

On Mon, Jan 29 2024 at 21:00, Chen Jun wrote:

That's not about an incorrect NULL pointer check. That's about using a
function which is guaranteed to return NULL on your platform, no?

> bus_get_dev_root(&platform_bus_type) always returns NULL.

Lacks an explanation why this always returns NULL.

Thanks

        tglx

在 2024/2/13 18:09, Thomas Gleixner 写道:
> On Mon, Jan 29 2024 at 21:00, Chen Jun wrote:
> 
> That's not about an incorrect NULL pointer check. That's about using a
> function which is guaranteed to return NULL on your platform, no?
> 
>> bus_get_dev_root(&platform_bus_type) always returns NULL.
> 
> Lacks an explanation why this always returns NULL.
> 

Thank you for the reply.

bus_get_dev_root returns sp->dev_root set in subsys_register.
And subsys_register is not called by platform_bus_init.

So, for platform_bus_type, bus_get_dev_root always returns NULL.

> Thanks
> 
>          tglx
> 
> 

On Sun, Feb 18 2024 at 03:13, chenjun (AM) wrote:
> 在 2024/2/13 18:09, Thomas Gleixner 写道:
>> On Mon, Jan 29 2024 at 21:00, Chen Jun wrote:
>> 
>> That's not about an incorrect NULL pointer check. That's about using a
>> function which is guaranteed to return NULL on your platform, no?
>> 
>>> bus_get_dev_root(&platform_bus_type) always returns NULL.
>> 
>> Lacks an explanation why this always returns NULL.
>> 
>
> Thank you for the reply.
>
> bus_get_dev_root returns sp->dev_root set in subsys_register.
> And subsys_register is not called by platform_bus_init.
>
> So, for platform_bus_type, bus_get_dev_root always returns NULL.

Please add this information to the change log.

Thanks,

        tglx