irq: Invalid fwnode type (2) for irqdomain

irq: Invalid fwnode type (2) for irqdomain

[Paul Menzel]

Dear Linux folks,


On a Lenovo X60t with coreboot based firmware testing the latest master
branch from Linus, the message below is new, and isn’t there with Linux
4.12.

```
irq: Invalid fwnode type (2) for irqdomain
```

`git blame kernel/irq/irqdomain.c` shows, the message is added in
commit d59f661 (genirq: Allow fwnode to carry name information only).

The commit message describes the warning.

> Warn if an invalid node is supplied and treat it like no node. This
> happens e.g. with i2 devices on x86 which hand in an ACPI type node
> which has no interface for retrieving the name.

How can I find out, how the ACPI/ASL has to be fixed?


Kind regards,

Paul

Re: irq: Invalid fwnode type (2) for irqdomain

[Thomas Gleixner]

[-- Attachment #1: Type: text/plain, Size: 943 bytes --]

On Wed, 5 Jul 2017, Paul Menzel wrote:
> On a Lenovo X60t with coreboot based firmware testing the latest master
> branch from Linus, the message below is new, and isn’t there with Linux
> 4.12.
> 
> ```
> irq: Invalid fwnode type (2) for irqdomain
> ```
> 
> `git blame kernel/irq/irqdomain.c` shows, the message is added in
> commit d59f661 (genirq: Allow fwnode to carry name information only).
> 
> The commit message describes the warning.
> 
> > Warn if an invalid node is supplied and treat it like no node. This
> > happens e.g. with i2 devices on x86 which hand in an ACPI type node
> > which has no interface for retrieving the name.
> 
> How can I find out, how the ACPI/ASL has to be fixed?

It's not an ACPI/ASL issue. Sorry, if the description and error message is
not clear. That's a kernel internal issue because we have no interface yet
to retrieve the required information. I'll have a look later today.

Thanks,

	tglx

Re: irq: Invalid fwnode type (2) for irqdomain

[Marc Zyngier]

On 05/07/17 11:41, Thomas Gleixner wrote:
> On Wed, 5 Jul 2017, Paul Menzel wrote:
>> On a Lenovo X60t with coreboot based firmware testing the latest master
>> branch from Linus, the message below is new, and isn’t there with Linux
>> 4.12.
>>
>> ```
>> irq: Invalid fwnode type (2) for irqdomain
>> ```
>>
>> `git blame kernel/irq/irqdomain.c` shows, the message is added in
>> commit d59f661 (genirq: Allow fwnode to carry name information only).
>>
>> The commit message describes the warning.
>>
>>> Warn if an invalid node is supplied and treat it like no node. This
>>> happens e.g. with i2 devices on x86 which hand in an ACPI type node
>>> which has no interface for retrieving the name.
>>
>> How can I find out, how the ACPI/ASL has to be fixed?
> 
> It's not an ACPI/ASL issue. Sorry, if the description and error message is
> not clear. That's a kernel internal issue because we have no interface yet
> to retrieve the required information. I'll have a look later today.

I've been reported an issue that is somewhat related to this, Any
chance you could give this[1] a go? I have no way to test it...

Thanks,

	M.

[1] https://git.kernel.org/pub/scm/linux/kernel/git/maz/arm-platforms.git/commit/?h=irq/fixes-4.13
-- 
Jazz is not dead. It just smells funny...

Re: irq: Invalid fwnode type (2) for irqdomain

[Paul Menzel]

Dear Marc, dear John,


Am Donnerstag, den 06.07.2017, 15:21 +0100 schrieb Marc Zyngier:
> On 05/07/17 11:41, Thomas Gleixner wrote:
> > On Wed, 5 Jul 2017, Paul Menzel wrote:
> > > On a Lenovo X60t with coreboot based firmware testing the latest
> > > master
> > > branch from Linus, the message below is new, and isn’t there with
> > > Linux
> > > 4.12.
> > > 
> > > ```
> > > irq: Invalid fwnode type (2) for irqdomain
> > > ```
> > > 
> > > `git blame kernel/irq/irqdomain.c` shows, the message is added in
> > > commit d59f661 (genirq: Allow fwnode to carry name information
> > > only).
> > > 
> > > The commit message describes the warning.
> > > 
> > > > Warn if an invalid node is supplied and treat it like no node.
> > > > This
> > > > happens e.g. with i2 devices on x86 which hand in an ACPI type
> > > > node
> > > > which has no interface for retrieving the name.
> > > 
> > > How can I find out, how the ACPI/ASL has to be fixed?
> > 
> > It's not an ACPI/ASL issue. Sorry, if the description and error
> > message is
> > not clear. That's a kernel internal issue because we have no
> > interface yet
> > to retrieve the required information. I'll have a look later today.
> 
> I've been reported an issue that is somewhat related to this, Any
> chance you could give this[1] a go? I have no way to test it...

The commit below indeed made the warning go away.

> commit 6bf31c05f2a967e5351a3050ef51c18c8c9ee291
> Author: Marc Zyngier <marc.zyngier@arm.com>
> Date:   Thu Jul 6 14:51:31 2017 +0100
> 
>     irqdomain: Allow ACPI device nodes to be used as irqdomain identifiers
>     
>     A number of irqchip implementations are (ab)using the irqdomain
>     allocator by passing a fwnode that is neither a FWNODE_OF or
>     a FWNODE_IRQCHIP.
>     
>     This is pretty bad, but it also feels pretty crap to force these
>     drivers to allocate their own irqchip_fwid when they already have
>     a proper fwnode.
>     
>     Instead, let's teach the irqdomain allocator about ACPI device
>     nodes, and add some lovely name generation code... Tested on
>     an arm64 D05 system.
>     
>     Reported-by: John Garry <john.garry@huawei.com>
>     Signed-off-by: Marc Zyngier <marc.zyngier@arm.com>

I wonder if a warning/debug messages could and should still be shown,
if a fwnode is neither a FWNODE_OF or a FWNODE_IRQCHIP.


Kind regards,

Paul


> [1] https://git.kernel.org/pub/scm/linux/kernel/git/maz/arm-platforms.git/commit/?h=irq/fixes-4.13

Re: irq: Invalid fwnode type (2) for irqdomain

[Marc Zyngier]

On Sat, Jul 08 2017 at  9:28:17 am BST, Paul Menzel <pmenzel@molgen.mpg.de> wrote:

Hi Paul,


>> I've been reported an issue that is somewhat related to this, Any
>> chance you could give this[1] a go? I have no way to test it...
>
> The commit below indeed made the warning go away.
>
>> commit 6bf31c05f2a967e5351a3050ef51c18c8c9ee291
>> Author: Marc Zyngier <marc.zyngier@arm.com>
>> Date:   Thu Jul 6 14:51:31 2017 +0100
>> 
>>     irqdomain: Allow ACPI device nodes to be used as irqdomain identifiers
>>     
>>     A number of irqchip implementations are (ab)using the irqdomain
>>     allocator by passing a fwnode that is neither a FWNODE_OF or
>>     a FWNODE_IRQCHIP.
>>     
>>     This is pretty bad, but it also feels pretty crap to force these
>>     drivers to allocate their own irqchip_fwid when they already have
>>     a proper fwnode.
>>     
>>     Instead, let's teach the irqdomain allocator about ACPI device
>>     nodes, and add some lovely name generation code... Tested on
>>     an arm64 D05 system.
>>     
>>     Reported-by: John Garry <john.garry@huawei.com>
>>     Signed-off-by: Marc Zyngier <marc.zyngier@arm.com>
>
> I wonder if a warning/debug messages could and should still be shown,
> if a fwnode is neither a FWNODE_OF or a FWNODE_IRQCHIP.

With this patch, the code covers FWNODE_OF, FWNODE_IRQCHIP and
FWNODE_ACPI (the latter being the case you and John uncovered). Should
we get any other fwnode_type value, the irqdomain code will still warn
the user that this is not an object type we're ready to accept.

Thanks again for the report and the testing.

	M.
-- 
Jazz is not dead. It just smells funny.