[PATCH 0/2] Add dev_is_amba and replace the usage in platform

[PATCH 0/2] Add dev_is_amba and replace the usage in platform

[Kunwu Chan]

Add dev_is_amba function and use dev_is_platform()/dev_is_amba() 
instead of checking bus type directly.

Kunwu Chan (2):
  amba: Add dev_is_amba() function and export it for modules
  of/platform: PCI: Use dev_is_platform/dev_is_amba to identify
    platform/amba devices

 drivers/amba/bus.c       | 6 ++++++
 drivers/of/platform.c    | 4 ++--
 include/linux/amba/bus.h | 5 +++++
 3 files changed, 13 insertions(+), 2 deletions(-)

-- 
2.43.0

[PATCH 1/2] amba: Add dev_is_amba() function and export it for modules

[Kunwu Chan]

Add dev_is_amba() function to determine
whether the device is a AMBA device.

Suggested-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
Signed-off-by: Kunwu Chan <chentao@kylinos.cn>
---
 drivers/amba/bus.c       | 6 ++++++
 include/linux/amba/bus.h | 5 +++++
 2 files changed, 11 insertions(+)

diff --git a/drivers/amba/bus.c b/drivers/amba/bus.c
index 0230c43377c1..8ef259b4d037 100644
--- a/drivers/amba/bus.c
+++ b/drivers/amba/bus.c
@@ -449,6 +449,12 @@ const struct bus_type amba_bustype = {
 };
 EXPORT_SYMBOL_GPL(amba_bustype);
 
+bool dev_is_amba(const struct device *dev)
+{
+	return dev->bus == &amba_bustype;
+}
+EXPORT_SYMBOL_GPL(dev_is_amba);
+
 static int __init amba_init(void)
 {
 	return bus_register(&amba_bustype);
diff --git a/include/linux/amba/bus.h b/include/linux/amba/bus.h
index dda2f3ea89cb..9946276aff73 100644
--- a/include/linux/amba/bus.h
+++ b/include/linux/amba/bus.h
@@ -121,6 +121,7 @@ extern const struct bus_type amba_bustype;
 #ifdef CONFIG_ARM_AMBA
 int __amba_driver_register(struct amba_driver *, struct module *);
 void amba_driver_unregister(struct amba_driver *);
+bool dev_is_amba(const struct device *dev);
 #else
 static inline int __amba_driver_register(struct amba_driver *drv,
 					 struct module *owner)
@@ -130,6 +131,10 @@ static inline int __amba_driver_register(struct amba_driver *drv,
 static inline void amba_driver_unregister(struct amba_driver *drv)
 {
 }
+static inline bool dev_is_amba(const struct device *dev)
+{
+	return false;
+}
 #endif
 
 struct amba_device *amba_device_alloc(const char *, resource_size_t, size_t);
-- 
2.43.0

Re: [PATCH 1/2] amba: Add dev_is_amba() function and export it for modules

[Andy Shevchenko]

On Mon, Sep 23, 2024 at 05:42:47PM +0800, Kunwu Chan wrote:
> Add dev_is_amba() function to determine
> whether the device is a AMBA device.

Thanks, but it needs at least a user to show the conversion.
I think I suggested in the last discussion to convert all
straightforward cases (which are not the direct users of the
given bus type).

-- 
With Best Regards,
Andy Shevchenko

Re: [PATCH 1/2] amba: Add dev_is_amba() function and export it for modules

[Andy Shevchenko]

On Mon, Sep 23, 2024 at 02:11:40PM +0300, Andy Shevchenko wrote:
> On Mon, Sep 23, 2024 at 05:42:47PM +0800, Kunwu Chan wrote:
> > Add dev_is_amba() function to determine
> > whether the device is a AMBA device.
> 
> Thanks, but it needs at least a user to show the conversion.
> I think I suggested in the last discussion to convert all
> straightforward cases (which are not the direct users of the
> given bus type).

Okay, I see now, I was only Cc'ed to the first in the series...
Scratch that comment then, sorry for the noise.

-- 
With Best Regards,
Andy Shevchenko

Re: [PATCH 1/2] amba: Add dev_is_amba() function and export it for modules

[Rob Herring]

On Mon, Sep 23, 2024 at 05:42:47PM +0800, Kunwu Chan wrote:
> Add dev_is_amba() function to determine
> whether the device is a AMBA device.
> 
> Suggested-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
> Signed-off-by: Kunwu Chan <chentao@kylinos.cn>
> ---
>  drivers/amba/bus.c       | 6 ++++++
>  include/linux/amba/bus.h | 5 +++++
>  2 files changed, 11 insertions(+)

Russell, Can I get an ack for this to take it with patch #2.

Rob

Re: [PATCH 1/2] amba: Add dev_is_amba() function and export it for modules

[Russell King (Oracle)]

On Tue, Sep 24, 2024 at 05:28:57PM -0500, Rob Herring wrote:
> On Mon, Sep 23, 2024 at 05:42:47PM +0800, Kunwu Chan wrote:
> > Add dev_is_amba() function to determine
> > whether the device is a AMBA device.
> > 
> > Suggested-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
> > Signed-off-by: Kunwu Chan <chentao@kylinos.cn>
> > ---
> >  drivers/amba/bus.c       | 6 ++++++
> >  include/linux/amba/bus.h | 5 +++++
> >  2 files changed, 11 insertions(+)
> 
> Russell, Can I get an ack for this to take it with patch #2.

Would be nice to discuss "how shall we merge this cross-subsystem
patch series" first, hmm?

The reason I didn't take patch 1 originally is because it was submitted
to me without any users, and the general principle is not to accept
patches without users. Too many times, I've merged code where there's
been a "promise" that it will be used, only to have the author go
silent and users never come along. So now, my rule is... any code that
adds something must also come with its user.

-- 
RMK's Patch system: https://www.armlinux.org.uk/developer/patches/
FTTP is here! 80Mbps down 10Mbps up. Decent connectivity at last!

Re: [PATCH 1/2] amba: Add dev_is_amba() function and export it for modules

[Kunwu Chan]

Thanks for the reply.

On 2024/9/25 06:43, Russell King (Oracle) wrote:
> On Tue, Sep 24, 2024 at 05:28:57PM -0500, Rob Herring wrote:
>> On Mon, Sep 23, 2024 at 05:42:47PM +0800, Kunwu Chan wrote:
>>> Add dev_is_amba() function to determine
>>> whether the device is a AMBA device.
>>>
>>> Suggested-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
>>> Signed-off-by: Kunwu Chan <chentao@kylinos.cn>
>>> ---
>>>   drivers/amba/bus.c       | 6 ++++++
>>>   include/linux/amba/bus.h | 5 +++++
>>>   2 files changed, 11 insertions(+)
>> Russell, Can I get an ack for this to take it with patch #2.
> Would be nice to discuss "how shall we merge this cross-subsystem
> patch series" first, hmm?
>
> The reason I didn't take patch 1 originally is because it was submitted
> to me without any users, and the general principle is not to accept
> patches without users. Too many times, I've merged code where there's
> been a "promise" that it will be used, only to have the author go
> silent and users never come along. So now, my rule is... any code that
> adds something must also come with its user.
>
Just one user in drivers/of/platform.c right now.

Actually, I don't know if dev_is_amba will use it more in the future.

So we keep the usage of  "dev->bus == &platform_bus_type " too?
Although here come a wrapper function "dev_is_platform".

-- 
Thanks,
   Kunwu.Chan

Re: [PATCH 1/2] amba: Add dev_is_amba() function and export it for modules

[Rob Herring]

On Tue, Sep 24, 2024 at 5:44 PM Russell King (Oracle)
<linux@armlinux.org.uk> wrote:
>
> On Tue, Sep 24, 2024 at 05:28:57PM -0500, Rob Herring wrote:
> > On Mon, Sep 23, 2024 at 05:42:47PM +0800, Kunwu Chan wrote:
> > > Add dev_is_amba() function to determine
> > > whether the device is a AMBA device.
> > >
> > > Suggested-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
> > > Signed-off-by: Kunwu Chan <chentao@kylinos.cn>
> > > ---
> > >  drivers/amba/bus.c       | 6 ++++++
> > >  include/linux/amba/bus.h | 5 +++++
> > >  2 files changed, 11 insertions(+)
> >
> > Russell, Can I get an ack for this to take it with patch #2.
>
> Would be nice to discuss "how shall we merge this cross-subsystem
> patch series" first, hmm?

Sure. IMO and what seems to be typical, since the user is in
drivers/of/ and changing that code is the overall reason for this
series, I think merging it via the DT tree makes the most sense. But
either way is fine with me. I'm happy to either ack it or apply it and
move on to the next thing.

> The reason I didn't take patch 1 originally is because it was submitted
> to me without any users, and the general principle is not to accept
> patches without users. Too many times, I've merged code where there's
> been a "promise" that it will be used, only to have the author go
> silent and users never come along. So now, my rule is... any code that
> adds something must also come with its user.

The user is in drivers/of/ in patch 2 of this series. So no issue there.

Rob

[PATCH 2/2] of/platform: PCI: Use dev_is_platform/dev_is_amba to identify platform/amba devices

[Kunwu Chan]

Use dev_is_platform()/dev_is_amba() instead of checking bus type directly.

Signed-off-by: Kunwu Chan <chentao@kylinos.cn>
---
 drivers/of/platform.c | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/drivers/of/platform.c b/drivers/of/platform.c
index 9bafcff3e628..1937cbef55ee 100644
--- a/drivers/of/platform.c
+++ b/drivers/of/platform.c
@@ -619,10 +619,10 @@ int of_platform_device_destroy(struct device *dev, void *data)
 	of_node_clear_flag(dev->of_node, OF_POPULATED);
 	of_node_clear_flag(dev->of_node, OF_POPULATED_BUS);
 
-	if (dev->bus == &platform_bus_type)
+	if (dev_is_platform(dev))
 		platform_device_unregister(to_platform_device(dev));
 #ifdef CONFIG_ARM_AMBA
-	else if (dev->bus == &amba_bustype)
+	else if (dev_is_amba(dev))
 		amba_device_unregister(to_amba_device(dev));
 #endif
 
-- 
2.43.0