[PATCH v1 1/1] amba: bus: balance firmware node reference counting

[PATCH v1 1/1] amba: bus: balance firmware node reference counting

[Andy Shevchenko]

Currently the ACPI code doesn't bump the reference count of
the firmware node, while OF counter part does. Not that it's
a problem right now, since ACPI doesn't really use the reference
counting for firmware nodes, it still makes sense to make code
robust against any changes done there. For this,
 - switch ACPI case to use device_set_node() to be unified with OF
 - move reference counting to amba_device_add()
 - switch to use firmware nodes instead of OF ones

In the result we will have reference counting done in the same module
for all callers independently on the nature of firmware node behind.

Signed-off-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
---
 drivers/acpi/arm64/amba.c | 2 +-
 drivers/amba/bus.c        | 5 ++++-
 drivers/of/platform.c     | 2 +-
 3 files changed, 6 insertions(+), 3 deletions(-)

diff --git a/drivers/acpi/arm64/amba.c b/drivers/acpi/arm64/amba.c
index 60be8ee1dbdc..171b5c2c7edd 100644
--- a/drivers/acpi/arm64/amba.c
+++ b/drivers/acpi/arm64/amba.c
@@ -101,7 +101,7 @@ static int amba_handler_attach(struct acpi_device *adev,
 	if (parent)
 		dev->dev.parent = acpi_get_first_physical_node(parent);
 
-	ACPI_COMPANION_SET(&dev->dev, adev);
+	device_set_node(&dev->dev, acpi_fwnode_handle(adev));
 
 	ret = amba_device_add(dev, &iomem_resource);
 	if (ret) {
diff --git a/drivers/amba/bus.c b/drivers/amba/bus.c
index 09e72967b8ab..05d0209a6347 100644
--- a/drivers/amba/bus.c
+++ b/drivers/amba/bus.c
@@ -18,6 +18,7 @@
 #include <linux/limits.h>
 #include <linux/clk/clk-conf.h>
 #include <linux/platform_device.h>
+#include <linux/property.h>
 #include <linux/reset.h>
 #include <linux/of_irq.h>
 #include <linux/of_device.h>
@@ -528,7 +529,7 @@ static void amba_device_release(struct device *dev)
 {
 	struct amba_device *d = to_amba_device(dev);
 
-	of_node_put(d->dev.of_node);
+	fwnode_handle_put(dev_fwnode(&d->dev);
 	if (d->res.parent)
 		release_resource(&d->res);
 	mutex_destroy(&d->periphid_lock);
@@ -548,6 +549,8 @@ int amba_device_add(struct amba_device *dev, struct resource *parent)
 {
 	int ret;
 
+	fwnode_handle_get(dev_fwnode(&dev->dev));
+
 	ret = request_resource(parent, &dev->res);
 	if (ret)
 		return ret;
diff --git a/drivers/of/platform.c b/drivers/of/platform.c
index f235ab55b91e..126d265aa7d8 100644
--- a/drivers/of/platform.c
+++ b/drivers/of/platform.c
@@ -273,7 +273,7 @@ static struct amba_device *of_amba_device_create(struct device_node *node,
 	dev->dev.dma_mask = &dev->dev.coherent_dma_mask;
 
 	/* setup generic device info */
-	device_set_node(&dev->dev, of_fwnode_handle(of_node_get(node)));
+	device_set_node(&dev->dev, of_fwnode_handle(node));
 	dev->dev.parent = parent ? : &platform_bus;
 	dev->dev.platform_data = platform_data;
 	if (bus_id)
-- 
2.40.0.1.gaa8946217a0b

Re: [PATCH v1 1/1] amba: bus: balance firmware node reference counting

[kernel test robot]

Hi Andy,

kernel test robot noticed the following build errors:

[auto build test ERROR on next-20230824]
[cannot apply to rafael-pm/linux-next robh/for-next soc/for-next linus/master v6.5-rc7 v6.5-rc6 v6.5-rc5 v6.5-rc7]
[If your patch is applied to the wrong git tree, kindly drop us a note.
And when submitting patch, we suggest to use '--base' as documented in
https://git-scm.com/docs/git-format-patch#_base_tree_information]

url:    https://github.com/intel-lab-lkp/linux/commits/Andy-Shevchenko/amba-bus-balance-firmware-node-reference-counting/20230825-003007
base:   next-20230824
patch link:    https://lore.kernel.org/r/20230824162654.2890992-1-andriy.shevchenko%40linux.intel.com
patch subject: [PATCH v1 1/1] amba: bus: balance firmware node reference counting
config: arm-defconfig (https://download.01.org/0day-ci/archive/20230825/202308250317.taQ9kshJ-lkp@intel.com/config)
compiler: arm-linux-gnueabi-gcc (GCC) 13.2.0
reproduce: (https://download.01.org/0day-ci/archive/20230825/202308250317.taQ9kshJ-lkp@intel.com/reproduce)

If you fix the issue in a separate patch/commit (i.e. not just a new version of
the same patch/commit), kindly add following tags
| Reported-by: kernel test robot <lkp@intel.com>
| Closes: https://lore.kernel.org/oe-kbuild-all/202308250317.taQ9kshJ-lkp@intel.com/

All errors (new ones prefixed by >>):

   drivers/amba/bus.c: In function 'amba_device_release':
>> drivers/amba/bus.c:532:46: error: expected ')' before ';' token
     532 |         fwnode_handle_put(dev_fwnode(&d->dev);
         |                          ~                   ^
>> drivers/amba/bus.c:536:18: error: expected ';' before '}' token
     536 |         kfree(d);
         |                  ^
         |                  ;
     537 | }
         | ~                 


vim +532 drivers/amba/bus.c

   527	
   528	static void amba_device_release(struct device *dev)
   529	{
   530		struct amba_device *d = to_amba_device(dev);
   531	
 > 532		fwnode_handle_put(dev_fwnode(&d->dev);
   533		if (d->res.parent)
   534			release_resource(&d->res);
   535		mutex_destroy(&d->periphid_lock);
 > 536		kfree(d);
   537	}
   538	

-- 
0-DAY CI Kernel Test Service
https://github.com/intel/lkp-tests/wiki

Re: [PATCH v1 1/1] amba: bus: balance firmware node reference counting

[kernel test robot]

Hi Andy,

kernel test robot noticed the following build errors:

[auto build test ERROR on next-20230824]
[cannot apply to rafael-pm/linux-next robh/for-next soc/for-next linus/master v6.5-rc7 v6.5-rc6 v6.5-rc5 v6.5-rc7]
[If your patch is applied to the wrong git tree, kindly drop us a note.
And when submitting patch, we suggest to use '--base' as documented in
https://git-scm.com/docs/git-format-patch#_base_tree_information]

url:    https://github.com/intel-lab-lkp/linux/commits/Andy-Shevchenko/amba-bus-balance-firmware-node-reference-counting/20230825-003007
base:   next-20230824
patch link:    https://lore.kernel.org/r/20230824162654.2890992-1-andriy.shevchenko%40linux.intel.com
patch subject: [PATCH v1 1/1] amba: bus: balance firmware node reference counting
config: arm-randconfig-001-20230825 (https://download.01.org/0day-ci/archive/20230825/202308250458.VwjoWF3t-lkp@intel.com/config)
compiler: clang version 17.0.0 (https://github.com/llvm/llvm-project.git 4a5ac14ee968ff0ad5d2cc1ffa0299048db4c88a)
reproduce: (https://download.01.org/0day-ci/archive/20230825/202308250458.VwjoWF3t-lkp@intel.com/reproduce)

If you fix the issue in a separate patch/commit (i.e. not just a new version of
the same patch/commit), kindly add following tags
| Reported-by: kernel test robot <lkp@intel.com>
| Closes: https://lore.kernel.org/oe-kbuild-all/202308250458.VwjoWF3t-lkp@intel.com/

All errors (new ones prefixed by >>):

>> drivers/amba/bus.c:532:39: error: expected ')'
     532 |         fwnode_handle_put(dev_fwnode(&d->dev);
         |                                              ^
   drivers/amba/bus.c:532:19: note: to match this '('
     532 |         fwnode_handle_put(dev_fwnode(&d->dev);
         |                          ^
   1 error generated.


vim +532 drivers/amba/bus.c

   527	
   528	static void amba_device_release(struct device *dev)
   529	{
   530		struct amba_device *d = to_amba_device(dev);
   531	
 > 532		fwnode_handle_put(dev_fwnode(&d->dev);
   533		if (d->res.parent)
   534			release_resource(&d->res);
   535		mutex_destroy(&d->periphid_lock);
   536		kfree(d);
   537	}
   538	

-- 
0-DAY CI Kernel Test Service
https://github.com/intel/lkp-tests/wiki

Re: [PATCH v1 1/1] amba: bus: balance firmware node reference counting

[Andy Shevchenko]

On Thu, Aug 24, 2023 at 07:26:54PM +0300, Andy Shevchenko wrote:
> Currently the ACPI code doesn't bump the reference count of
> the firmware node, while OF counter part does. Not that it's
> a problem right now, since ACPI doesn't really use the reference
> counting for firmware nodes, it still makes sense to make code
> robust against any changes done there. For this,
>  - switch ACPI case to use device_set_node() to be unified with OF
>  - move reference counting to amba_device_add()
>  - switch to use firmware nodes instead of OF ones
> 
> In the result we will have reference counting done in the same module
> for all callers independently on the nature of firmware node behind.

Any comment on this? I would like to have this applied so I can do something
similar to the platform driver code.

-- 
With Best Regards,
Andy Shevchenko

Re: [PATCH v1 1/1] amba: bus: balance firmware node reference counting

[Andy Shevchenko]

On Tue, Sep 19, 2023 at 11:01:37PM +0300, Andy Shevchenko wrote:
> On Thu, Aug 24, 2023 at 07:26:54PM +0300, Andy Shevchenko wrote:
> > Currently the ACPI code doesn't bump the reference count of
> > the firmware node, while OF counter part does. Not that it's
> > a problem right now, since ACPI doesn't really use the reference
> > counting for firmware nodes, it still makes sense to make code
> > robust against any changes done there. For this,
> >  - switch ACPI case to use device_set_node() to be unified with OF
> >  - move reference counting to amba_device_add()
> >  - switch to use firmware nodes instead of OF ones
> > 
> > In the result we will have reference counting done in the same module
> > for all callers independently on the nature of firmware node behind.
> 
> Any comment on this? I would like to have this applied so I can do something
> similar to the platform driver code.

Ah, I see, I missed LKP run on this, I'll send a v2 perhaps later on this week.

-- 
With Best Regards,
Andy Shevchenko

Re: [PATCH v1 1/1] amba: bus: balance firmware node reference counting

[Rob Herring]

On Tue, Sep 19, 2023 at 3:02 PM Andy Shevchenko
<andriy.shevchenko@linux.intel.com> wrote:
>
> On Tue, Sep 19, 2023 at 11:01:37PM +0300, Andy Shevchenko wrote:
> > On Thu, Aug 24, 2023 at 07:26:54PM +0300, Andy Shevchenko wrote:
> > > Currently the ACPI code doesn't bump the reference count of
> > > the firmware node, while OF counter part does. Not that it's
> > > a problem right now, since ACPI doesn't really use the reference
> > > counting for firmware nodes, it still makes sense to make code
> > > robust against any changes done there. For this,
> > >  - switch ACPI case to use device_set_node() to be unified with OF
> > >  - move reference counting to amba_device_add()
> > >  - switch to use firmware nodes instead of OF ones
> > >
> > > In the result we will have reference counting done in the same module
> > > for all callers independently on the nature of firmware node behind.
> >
> > Any comment on this? I would like to have this applied so I can do something
> > similar to the platform driver code.
>
> Ah, I see, I missed LKP run on this, I'll send a v2 perhaps later on this week.

What's the relationship/dependency with this and the other patches
dealing with refcounting? Did the AMBA one land?

Rob

Re: [PATCH v1 1/1] amba: bus: balance firmware node reference counting

[Andy Shevchenko]

On Fri, Sep 22, 2023 at 09:18:47AM -0500, Rob Herring wrote:
> On Tue, Sep 19, 2023 at 3:02 PM Andy Shevchenko
> <andriy.shevchenko@linux.intel.com> wrote:
> > On Tue, Sep 19, 2023 at 11:01:37PM +0300, Andy Shevchenko wrote:
> > > On Thu, Aug 24, 2023 at 07:26:54PM +0300, Andy Shevchenko wrote:
> > > > Currently the ACPI code doesn't bump the reference count of
> > > > the firmware node, while OF counter part does. Not that it's
> > > > a problem right now, since ACPI doesn't really use the reference
> > > > counting for firmware nodes, it still makes sense to make code
> > > > robust against any changes done there. For this,
> > > >  - switch ACPI case to use device_set_node() to be unified with OF
> > > >  - move reference counting to amba_device_add()
> > > >  - switch to use firmware nodes instead of OF ones
> > > >
> > > > In the result we will have reference counting done in the same module
> > > > for all callers independently on the nature of firmware node behind.
> > >
> > > Any comment on this? I would like to have this applied so I can do something
> > > similar to the platform driver code.
> >
> > Ah, I see, I missed LKP run on this, I'll send a v2 perhaps later on this week.
> 
> What's the relationship/dependency with this and the other patches
> dealing with refcounting? Did the AMBA one land?

The OF refcounting is spread over platform code in unexpected places. So, we
have something like of_foo() bumps it, platform ->release() drops it. It sounds
to me as weird design to think of and AMBA is part of this mixture as it uses
platform_bus, BUT has it's own ->release(), besides the fact that it's used in
OF _and_ ACPI environments.

What I would expect to have is both are really agnostic to the fwnode type
and take and release reference in one place, but with OF <--> platform is
kinda more complicated that with AMBA.

-- 
With Best Regards,
Andy Shevchenko