Quoting David Gow (2024-05-01 00:55:10)

On Tue, 23 Apr 2024 at 07:24, Stephen Boyd sboyd@kernel.org wrote:

...

diff --git a/Documentation/dev-tools/kunit/api/index.rst b/Documentation/dev-tools/kunit/api/index.rst index 2d8f756aab56..282befa17edf 100644 --- a/Documentation/dev-tools/kunit/api/index.rst +++ b/Documentation/dev-tools/kunit/api/index.rst @@ -9,11 +9,15 @@ API Reference test resource functionredirection

•   of
  
  
  

This page documents the KUnit kernel testing API. It is divided into the following sections:

+Core KUnit API +==============

Documentation/dev-tools/kunit/api/test.rst

• Documents all of the standard testing API

@@ -25,3 +29,10 @@ Documentation/dev-tools/kunit/api/resource.rst Documentation/dev-tools/kunit/api/functionredirection.rst

• Documents the KUnit Function Redirection API

+Driver KUnit API +================

If we're adding a separate 'Driver' section here, it's probably sensible to move the existing device/driver helper documentation here, rather than leaving it in resource.rst as-is. I'm happy to do that in a follow-up patch, though.

To clarify, you're talking about "Managed Devices"? Looks like that can be a follow-up to split it into a new file and then put it here. If you're happy to do that then I'll leave it to you.

...

+Documentation/dev-tools/kunit/api/of.rst

• • Documents the KUnit device tree (OF) API

diff --git a/Documentation/dev-tools/kunit/api/of.rst b/Documentation/dev-tools/kunit/api/of.rst new file mode 100644 index 000000000000..8587591c3e78 --- /dev/null +++ b/Documentation/dev-tools/kunit/api/of.rst @@ -0,0 +1,13 @@ +.. SPDX-License-Identifier: GPL-2.0

+==================== +Device Tree (OF) API +====================

+The KUnit device tree API is used to test device tree (of_*) dependent code.

+.. kernel-doc:: include/kunit/of.h

• :internal:

+.. kernel-doc:: drivers/of/of_kunit.c

• :export:

diff --git a/drivers/of/Makefile b/drivers/of/Makefile index 251d33532148..0dfd05079313 100644 --- a/drivers/of/Makefile +++ b/drivers/of/Makefile @@ -19,6 +19,7 @@ obj-y += kexec.o endif endif

+obj-$(CONFIG_KUNIT) += of_kunit.o

I'm tempted to have this either live in lib/kunit, or be behind a separate Kconfig option, particularly since this will end up as a separate module, as-is.

Is the idea to have a single module that has all the kunit "stuff" in it so we can just load one module and be done? Is there any discussion on the list I can read to see the argument for this?

...

obj-$(CONFIG_OF_KUNIT_TEST) += of_test.o

obj-$(CONFIG_OF_UNITTEST) += unittest-data/ diff --git a/drivers/of/of_kunit.c b/drivers/of/of_kunit.c new file mode 100644 index 000000000000..f63527268a51 --- /dev/null +++ b/drivers/of/of_kunit.c @@ -0,0 +1,99 @@ +// SPDX-License-Identifier: GPL-2.0 +/*

• • Test managed device tree APIs
• */

+#include <linux/of.h> +#include <linux/of_fdt.h>

+#include <kunit/of.h> +#include <kunit/test.h> +#include <kunit/resource.h>

+static void of_overlay_fdt_apply_kunit_exit(void *ovcs_id) +{

•   of_overlay_remove(ovcs_id);
  

+}

+/**

• • of_overlay_fdt_apply_kunit() - Test managed of_overlay_fdt_apply()
• • @test: test context
• • @overlay_fdt: device tree overlay to apply
• • @overlay_fdt_size: size in bytes of @overlay_fdt
• • @ovcs_id: identifier of overlay, used to remove the overlay
• • Just like of_overlay_fdt_apply(), except the overlay is managed by the test
• • case and is automatically removed with of_overlay_remove() after the test
• • case concludes.
• • Return: 0 on success, negative errno on failure
• */

+int of_overlay_fdt_apply_kunit(struct kunit *test, void *overlay_fdt,

•                          u32 overlay_fdt_size, int *ovcs_id)
  

We're using kunit_ as a prefix for the device helpers (e.g. kunit_device_register()), so it may make sense to do that here, too. It's not as important as with the platform_device helpers, which are very similar to the existing device ones, but if we want to treat these as "part of KUnit which deals with of_overlays", rather than "part of "of_overlay which deals with KUnit", this may fit better.

Thoughts?

I'm fine either way with the name. I recall that last time we put a kunit postfix to make it easier to tab complete or something like that.

I find it hard to understand the distinction you're trying to make though. I guess you're saying the difference is what subsystem maintains the code, kunit or of. When they're simple wrappers it is easier to extract them out to lib/kunit and thus they can (should?) have the kunit prefix. Maybe that always holds true, because kunit wrappers are typically another API consumer, and if the API is exported either in a linux/ header or as an exported symbol it can be wrapped in lib/kunit easily. Did I follow correctly? When would of_overlay ever deal with KUnit?

...

diff --git a/include/kunit/of.h b/include/kunit/of.h new file mode 100644 index 000000000000..9981442ba578 --- /dev/null +++ b/include/kunit/of.h @@ -0,0 +1,94 @@ +/* SPDX-License-Identifier: GPL-2.0 */ +#ifndef _KUNIT_OF_H +#define _KUNIT_OF_H

+#include <kunit/test.h>

+struct device_node;

+#ifdef CONFIG_OF

Do we also need to check for CONFIG_OF_OVERLAY here?

Also, how useful is it to compile but skip tests without CONFIG_OF{,_OVERLAY} enabled? The other option is a compile error, which may make it more obvious that these are disabled if it's unexpected.

Thoughts?

I've tried to make it so that tests skip if an option isn't enabled. I suppose the CONFIG_OF_OVERLAY check can be hoisted up here as well so that the skip isn't buried in lower levels.

On Wed, May 1, 2024 at 3:55 AM 'David Gow' via KUnit Development kunit-dev@googlegroups.com wrote:

On Tue, 23 Apr 2024 at 07:24, Stephen Boyd sboyd@kernel.org wrote:

...

Describe a binding for an empty device node used by KUnit tests to confirm overlays load properly.

Reviewed-by: Rob Herring robh@kernel.org Cc: Krzysztof Kozlowski krzysztof.kozlowski+dt@linaro.org Cc: Conor Dooley conor+dt@kernel.org Cc: Brendan Higgins brendan.higgins@linux.dev Cc: David Gow davidgow@google.com Cc: Rae Moar rmoar@google.com Signed-off-by: Stephen Boyd sboyd@kernel.org

---

Looks good to me.

Reviewed-by: David Gow davidgow@google.com

Reviewed-by: Brendan Higgins brendanhiggins@google.com

-- David

...

.../devicetree/bindings/test/test,empty.yaml | 30 +++++++++++++++++++ 1 file changed, 30 insertions(+) create mode 100644 Documentation/devicetree/bindings/test/test,empty.yaml

diff --git a/Documentation/devicetree/bindings/test/test,empty.yaml b/Documentation/devicetree/bindings/test/test,empty.yaml new file mode 100644 index 000000000000..20dc83b15bbf --- /dev/null +++ b/Documentation/devicetree/bindings/test/test,empty.yaml @@ -0,0 +1,30 @@ +# SPDX-License-Identifier: (GPL-2.0-only OR BSD-2-Clause) +%YAML 1.2 +--- +$id: http://devicetree.org/schemas/test/test,empty.yaml# +$schema: http://devicetree.org/meta-schemas/core.yaml#

+title: Empty node

+maintainers:

• • David Gow davidgow@google.com
• • Brendan Higgins brendanhiggins@google.com

Brendan: Do you want to use your linux.dev address here?

Ah, thanks for pointing that out, David. Yes, please use "Brendan Higgins brendan.higgins@linux.dev" here.

Otherwise it looks good to me.

Quoting Stephen Boyd (2024-04-22 16:23:58)

diff --git a/drivers/base/test/platform_kunit.c b/drivers/base/test/platform_kunit.c new file mode 100644 index 000000000000..54af6db2a6d8 --- /dev/null +++ b/drivers/base/test/platform_kunit.c @@ -0,0 +1,174 @@ +// SPDX-License-Identifier: GPL-2.0 +/*

• • Test managed platform driver
• */

[...]

+/**

• • platform_driver_register_kunit() - Register a KUnit test managed platform driver
• • @test: test context
• • @drv: platform driver to register
• • Register a test managed platform driver. This allows callers to embed the
• • @drv in a container structure and use container_of() in the probe function
• • to pass information to KUnit tests. It can be assumed that the driver has
• • probed when this function returns.
• • Example
• • .. code-block:: c

• • struct kunit_test_context {
    

• •         struct platform_driver pdrv;
    

• •         const char *data;
    

• • };
    

• • static inline struct kunit_test_context *
    

• • to_test_context(struct platform_device *pdev)
    

• • {
    

• •         return container_of(to_platform_driver(pdev->dev.driver),
    

• •                             struct kunit_test_context,
    

• •                             pdrv);
    

• • }
    

• • static int kunit_platform_driver_probe(struct platform_device *pdev)
    

• • {
    

• •         struct kunit_test_context *ctx;
    

• •         ctx = to_test_context(pdev);
    

• •         ctx->data = "test data";
    

• •         return 0;
    

• • }
    

• • static void kunit_platform_driver_test(struct kunit *test)
    

• • {
    

• •         struct kunit_test_context *ctx;
    

• •         ctx = kunit_kzalloc(test, sizeof(*ctx), GFP_KERNEL);
    

• •         KUNIT_ASSERT_NOT_ERR_OR_NULL(test, ctx);
    

• •         ctx->pdrv.probe = kunit_platform_driver_probe;
    

• •         ctx->pdrv.driver.name = "kunit-platform";
    

• •         ctx->pdrv.driver.owner = THIS_MODULE;
    

• •         KUNIT_EXPECT_EQ(test, 0, platform_driver_register_kunit(test, &ctx->pdrv));
    

• •         KUNIT_EXPECT_STREQ(test, ctx->data, "test data");
    

• • }
    

• • Return: 0 on success, negative errno on failure.
• */

+int platform_driver_register_kunit(struct kunit *test,

•                              struct platform_driver *drv)
  

+{

•   int ret;
  

•   ret = platform_driver_register(drv);
  

•   if (ret)
  

•           return ret;
  

•   /*
  

•    * Wait for the driver to probe (or at least flush out of the deferred
  

•    * workqueue)
  

•    */
  

•   wait_for_device_probe();
  

Should this be removed? I was thinking that this isn't a pure wrapper around platform_driver_register() because it has this wait call. Maybe it's better to have some other kunit API that can wait for a specific device to probe and timeout if it doesn't happen in that amount of time. That API would use the bus notifiers and look for BUS_NOTIFY_BOUND_DRIVER. Or maybe that function could setup a completion that the test can wait on.

•   return kunit_add_action_or_reset(test,
  

•                                    (kunit_action_t *)&platform_driver_unregister,
  

•                                    drv);
  

+} +EXPORT_SYMBOL_GPL(platform_driver_register_kunit);

On Tue, 23 Apr 2024 at 07:24, Stephen Boyd sboyd@kernel.org wrote:

Introduce KUnit resource wrappers around platform_driver_register(), platform_device_alloc(), and platform_device_add() so that test authors can register platform drivers/devices from their tests and have the drivers/devices automatically be unregistered when the test is done.

This makes test setup code simpler when a platform driver or platform device is needed. Add a few test cases at the same time to make sure the APIs work as intended.

Cc: Brendan Higgins brendan.higgins@linux.dev Cc: David Gow davidgow@google.com Cc: Rae Moar rmoar@google.com Cc: Greg Kroah-Hartman gregkh@linuxfoundation.org Cc: "Rafael J. Wysocki" rafael@kernel.org Signed-off-by: Stephen Boyd sboyd@kernel.org

---

I really like this: I think it'll definitely help make platform devices easier to use in tests. (And the handling of the unregistering is particularly much nicer than trying to do it by hand, IMO.)

I've got a few suggestions below, mostly around naming and that it's probably best to put this in lib/kunit/, given that the header is in include/kunit, and there's already the generic kunit_device functionality there. There's also a control-flow-integrity issue or two, as casting function pointers will trigger that.

Otherwise, this looks good.

-- David

Documentation/dev-tools/kunit/api/index.rst | 5 + .../dev-tools/kunit/api/platformdevice.rst | 10 + drivers/base/test/Makefile | 3 + drivers/base/test/platform_kunit-test.c | 140 ++++++++++++++ drivers/base/test/platform_kunit.c | 174 ++++++++++++++++++ include/kunit/platform_device.h | 15 ++ 6 files changed, 347 insertions(+) create mode 100644 Documentation/dev-tools/kunit/api/platformdevice.rst create mode 100644 drivers/base/test/platform_kunit-test.c create mode 100644 drivers/base/test/platform_kunit.c create mode 100644 include/kunit/platform_device.h

diff --git a/Documentation/dev-tools/kunit/api/index.rst b/Documentation/dev-tools/kunit/api/index.rst index 282befa17edf..02b26f5e8750 100644 --- a/Documentation/dev-tools/kunit/api/index.rst +++ b/Documentation/dev-tools/kunit/api/index.rst @@ -10,6 +10,7 @@ API Reference resource functionredirection of

•   platformdevice
  

A note (to myself, as much as anything): for the other device wrappers, we considered them 'resources' and so bundled the documentation in with the 'resource' documentation. Maybe it'd make sense to split it out into its own device.rst file. We could optionally include the platformdevice stuff in the same file if we wanted to consolidate documentation for "device helpers", though I'm not sure if it's worthwhile.

This page documents the KUnit kernel testing API. It is divided into the @@ -36,3 +37,7 @@ Driver KUnit API Documentation/dev-tools/kunit/api/of.rst

• Documents the KUnit device tree (OF) API

+Documentation/dev-tools/kunit/api/platformdevice.rst

• • Documents the KUnit platform device API

diff --git a/Documentation/dev-tools/kunit/api/platformdevice.rst b/Documentation/dev-tools/kunit/api/platformdevice.rst new file mode 100644 index 000000000000..b228fb6558c2 --- /dev/null +++ b/Documentation/dev-tools/kunit/api/platformdevice.rst @@ -0,0 +1,10 @@ +.. SPDX-License-Identifier: GPL-2.0

+=================== +Platform Device API +===================

+The KUnit platform device API is used to test platform devices.

+.. kernel-doc:: drivers/base/test/platform_kunit.c

• :export:

diff --git a/drivers/base/test/Makefile b/drivers/base/test/Makefile index e321dfc7e922..740aef267fbe 100644 --- a/drivers/base/test/Makefile +++ b/drivers/base/test/Makefile @@ -1,8 +1,11 @@ # SPDX-License-Identifier: GPL-2.0 obj-$(CONFIG_TEST_ASYNC_DRIVER_PROBE) += test_async_driver_probe.o

+obj-$(CONFIG_KUNIT) += platform_kunit.o

Do we want this to be part of the kunit.ko module (and hence, probably, under lib/kunit), or to keep this as a separate module. I'm tempted, personally, to treat this as a part of KUnit, and have it be part of the same module. There are a couple of reasons for this: - It's nice to have CONFIG_KUNIT produce only one module. If we want this to be separate, I'd be tempted to put it behind its own kconfig entry. - The name platform_kunit.ko suggests (to me, at least) that this is the test for platform devices, not the implementation of the helper.

I probably can be persuaded otherwise if you've got a strong preference for it to stay as-is, though.

obj-$(CONFIG_DM_KUNIT_TEST) += root-device-test.o obj-$(CONFIG_DM_KUNIT_TEST) += platform-device-test.o +obj-$(CONFIG_DM_KUNIT_TEST) += platform_kunit-test.o

obj-$(CONFIG_DRIVER_PE_KUNIT_TEST) += property-entry-test.o CFLAGS_property-entry-test.o += $(DISABLE_STRUCTLEAK_PLUGIN) diff --git a/drivers/base/test/platform_kunit-test.c b/drivers/base/test/platform_kunit-test.c new file mode 100644 index 000000000000..ce545532d209 --- /dev/null +++ b/drivers/base/test/platform_kunit-test.c @@ -0,0 +1,140 @@ +// SPDX-License-Identifier: GPL-2.0 +/*

• • KUnit test for platform driver infrastructure.
• */

+#include <linux/platform_device.h>

+#include <kunit/platform_device.h> +#include <kunit/test.h>

+static const char * const kunit_devname = "kunit-platform";

+/*

• • Test that platform_device_alloc_kunit() creates a platform device.
• */

+static void platform_device_alloc_kunit_test(struct kunit *test) +{

•   KUNIT_EXPECT_NOT_ERR_OR_NULL(test,
  

•                   platform_device_alloc_kunit(test, kunit_devname, 1));
  

+}

+/*

• • Test that platform_device_add_kunit() registers a platform device on the
• • platform bus with the proper name and id.
• */

+static void platform_device_add_kunit_test(struct kunit *test) +{

•   struct platform_device *pdev;
  

•   const char *name = kunit_devname;
  

•   const int id = -1;
  

•   pdev = platform_device_alloc_kunit(test, name, id);
  

•   KUNIT_ASSERT_NOT_ERR_OR_NULL(test, pdev);
  

•   KUNIT_EXPECT_EQ(test, 0, platform_device_add_kunit(test, pdev));
  

•   KUNIT_EXPECT_TRUE(test, dev_is_platform(&pdev->dev));
  

•   KUNIT_EXPECT_STREQ(test, pdev->name, name);
  

•   KUNIT_EXPECT_EQ(test, pdev->id, id);
  

+}

+/*

• • Test that platform_device_add_kunit() called twice with the same device name
• • and id fails the second time and properly cleans up.
• */

+static void platform_device_add_kunit_twice_fails_test(struct kunit *test) +{

•   struct platform_device *pdev;
  

•   const char *name = kunit_devname;
  

•   const int id = -1;
  

•   pdev = platform_device_alloc_kunit(test, name, id);
  

•   KUNIT_ASSERT_NOT_ERR_OR_NULL(test, pdev);
  

•   KUNIT_ASSERT_EQ(test, 0, platform_device_add_kunit(test, pdev));
  

•   pdev = platform_device_alloc_kunit(test, name, id);
  

•   KUNIT_ASSERT_NOT_ERR_OR_NULL(test, pdev);
  

•   KUNIT_EXPECT_NE(test, 0, platform_device_add_kunit(test, pdev));
  

+}

+/*

• • Test suite for struct platform_device kunit APIs
• */

+static struct kunit_case platform_device_kunit_test_cases[] = {

•   KUNIT_CASE(platform_device_alloc_kunit_test),
  

•   KUNIT_CASE(platform_device_add_kunit_test),
  

•   KUNIT_CASE(platform_device_add_kunit_twice_fails_test),
  

•   {}
  

+};

+static struct kunit_suite platform_device_kunit_suite = {

•   .name = "platform_device_kunit",
  

•   .test_cases = platform_device_kunit_test_cases,
  

+};

+struct kunit_platform_driver_test_context {

•   struct platform_driver pdrv;
  

•   const char *data;
  

+};

+static const char * const test_data = "test data";

+static inline struct kunit_platform_driver_test_context * +to_test_context(struct platform_device *pdev) +{

•   return container_of(to_platform_driver(pdev->dev.driver),
  

•                       struct kunit_platform_driver_test_context,
  

•                       pdrv);
  

+}

+static int kunit_platform_driver_probe(struct platform_device *pdev) +{

•   struct kunit_platform_driver_test_context *ctx;
  

•   ctx = to_test_context(pdev);
  

•   ctx->data = test_data;
  

•   return 0;
  

+}

+/* Test that platform_driver_register_kunit() registers a driver that probes. */ +static void platform_driver_register_kunit_test(struct kunit *test) +{

•   struct platform_device *pdev;
  

•   struct kunit_platform_driver_test_context *ctx;
  

•   ctx = kunit_kzalloc(test, sizeof(*ctx), GFP_KERNEL);
  

•   KUNIT_ASSERT_NOT_ERR_OR_NULL(test, ctx);
  

•   pdev = platform_device_alloc_kunit(test, kunit_devname, -1);
  

•   KUNIT_ASSERT_NOT_ERR_OR_NULL(test, pdev);
  

•   KUNIT_ASSERT_EQ(test, 0, platform_device_add_kunit(test, pdev));
  

•   ctx->pdrv.probe = kunit_platform_driver_probe;
  

•   ctx->pdrv.driver.name = kunit_devname;
  

•   ctx->pdrv.driver.owner = THIS_MODULE;
  

•   KUNIT_EXPECT_EQ(test, 0, platform_driver_register_kunit(test, &ctx->pdrv));
  

•   KUNIT_EXPECT_STREQ(test, ctx->data, test_data);
  

+}

+static struct kunit_case platform_driver_kunit_test_cases[] = {

•   KUNIT_CASE(platform_driver_register_kunit_test),
  

•   {}
  

+};

+/*

• • Test suite for struct platform_driver kunit APIs
• */

+static struct kunit_suite platform_driver_kunit_suite = {

•   .name = "platform_driver_kunit",
  

•   .test_cases = platform_driver_kunit_test_cases,
  

+};

+kunit_test_suites(

•   &platform_device_kunit_suite,
  

•   &platform_driver_kunit_suite,
  

+);

+MODULE_LICENSE("GPL"); diff --git a/drivers/base/test/platform_kunit.c b/drivers/base/test/platform_kunit.c new file mode 100644 index 000000000000..54af6db2a6d8 --- /dev/null +++ b/drivers/base/test/platform_kunit.c @@ -0,0 +1,174 @@ +// SPDX-License-Identifier: GPL-2.0 +/*

• • Test managed platform driver
• */

+#include <linux/device/driver.h> +#include <linux/platform_device.h>

+#include <kunit/platform_device.h> +#include <kunit/resource.h>

+/**

• • platform_device_alloc_kunit() - Allocate a KUnit test managed platform device
• • @test: test context
• • @name: device name of platform device to alloc
• • @id: identifier of platform device to alloc.
• • Allocate a test managed platform device. The device is put when the test completes.
• • Return: Allocated platform device on success, NULL on failure.
• */

+struct platform_device * +platform_device_alloc_kunit(struct kunit *test, const char *name, int id)

I'd prefer, personally, this be named something like kunit_platform_device_alloc(), to match the existing kunit_device_register() functions.

+{

•   struct platform_device *pdev;
  

•   pdev = platform_device_alloc(name, id);
  

•   if (!pdev)
  

•           return NULL;
  

•   if (kunit_add_action_or_reset(test, (kunit_action_t *)&platform_device_put, pdev))
  

Alas, casting function pointers to kunit_action_t* breaks CFI. It's worth using a wrapper, which can be created with the KUNIT_DEFINE_ACTION_WRAPPER() macro, e.g.

KUNIT_DEFINE_ACTION_WRAPPER(platform_device_put_wrapper, platform_device_put, struct platform_device *);

•           return NULL;
  

•   return pdev;
  

+} +EXPORT_SYMBOL_GPL(platform_device_alloc_kunit);

+static void platform_device_add_kunit_exit(struct kunit_resource *res) +{

•   struct platform_device *pdev = res->data;
  

•   platform_device_unregister(pdev);
  

+}

+static bool +platform_device_alloc_kunit_match(struct kunit *test,

•                             struct kunit_resource *res, void *match_data)
  

+{

•   struct platform_device *pdev = match_data;
  

•   return res->data == pdev;
  

+}

+/**

• • platform_device_add_kunit() - Register a KUnit test managed platform device
• • @test: test context
• • @pdev: platform device to add
• • Register a test managed platform device. The device is unregistered when the
• • test completes.
• • Return: 0 on success, negative errno on failure.
• */

+int platform_device_add_kunit(struct kunit *test, struct platform_device *pdev)

As above, I'd lean towards naming this kunit_platform_device_add() for consistency with the other KUnit device helpers.

+{

•   struct kunit_resource *res;
  

•   int ret;
  

•   ret = platform_device_add(pdev);
  

•   if (ret)
  

•           return ret;
  

•   res = kunit_find_resource(test, platform_device_alloc_kunit_match, pdev);
  

•   if (res) {
  

•           /*
  

•            * Transfer the reference count of the platform device if it was
  

•            * allocated with platform_device_alloc_kunit(). In that case,
  

•            * calling platform_device_put() leads to reference count
  

•            * underflow because platform_device_unregister() does it for
  

•            * us and we call platform_device_unregister() from
  

•            * platform_device_add_kunit_exit().
  

•            *
  

•            * Usually callers transfer the refcount from
  

•            * platform_device_alloc() to platform_device_add() and simply
  

•            * call platform_device_unregister() when done, but with kunit
  

•            * we have to keep this straight by redirecting the free
  

•            * routine for the resource.
  

•            */
  

•           res->free = platform_device_add_kunit_exit;
  

•           kunit_put_resource(res);
  

•   } else if (kunit_add_action_or_reset(test,
  

•                                        (kunit_action_t *)&platform_device_unregister,
  

•                                        pdev)) {
  

Nit: We don't want to cast directly to kunit_action_t *, as that breaks CFI. Can we use KUNIT_DEFINE_ACTION_WRAPPER()?

•           return -ENOMEM;
  

Nit: This is fine, as kunit_add_action_or_reset() only returns 0 or -ENOMEM at the moment, but it could cause problems down the line if we ever want to return a different error. I don't think that's particularly likely, but it might be nicer to properly propagate the error.

•   }
  

•   return 0;
  

+} +EXPORT_SYMBOL_GPL(platform_device_add_kunit);

+/**

• • platform_driver_register_kunit() - Register a KUnit test managed platform driver
• • @test: test context
• • @drv: platform driver to register
• • Register a test managed platform driver. This allows callers to embed the
• • @drv in a container structure and use container_of() in the probe function
• • to pass information to KUnit tests. It can be assumed that the driver has
• • probed when this function returns.
• • Example
• • .. code-block:: c

• • struct kunit_test_context {
    

• •         struct platform_driver pdrv;
    

• •         const char *data;
    

• • };
    

• • static inline struct kunit_test_context *
    

• • to_test_context(struct platform_device *pdev)
    

• • {
    

• •         return container_of(to_platform_driver(pdev->dev.driver),
    

• •                             struct kunit_test_context,
    

• •                             pdrv);
    

• • }
    

• • static int kunit_platform_driver_probe(struct platform_device *pdev)
    

• • {
    

• •         struct kunit_test_context *ctx;
    

• •         ctx = to_test_context(pdev);
    

• •         ctx->data = "test data";
    

• •         return 0;
    

• • }
    

• • static void kunit_platform_driver_test(struct kunit *test)
    

• • {
    

• •         struct kunit_test_context *ctx;
    

• •         ctx = kunit_kzalloc(test, sizeof(*ctx), GFP_KERNEL);
    

• •         KUNIT_ASSERT_NOT_ERR_OR_NULL(test, ctx);
    

• •         ctx->pdrv.probe = kunit_platform_driver_probe;
    

• •         ctx->pdrv.driver.name = "kunit-platform";
    

• •         ctx->pdrv.driver.owner = THIS_MODULE;
    

• •         KUNIT_EXPECT_EQ(test, 0, platform_driver_register_kunit(test, &ctx->pdrv));
    

• •         KUNIT_EXPECT_STREQ(test, ctx->data, "test data");
    

• • }
    

• • Return: 0 on success, negative errno on failure.
• */

+int platform_driver_register_kunit(struct kunit *test,

•                              struct platform_driver *drv)
  

As above, I'd prefer kunit_platform_driver_register()

+{

•   int ret;
  

•   ret = platform_driver_register(drv);
  

•   if (ret)
  

•           return ret;
  

•   /*
  

•    * Wait for the driver to probe (or at least flush out of the deferred
  

•    * workqueue)
  

•    */
  

•   wait_for_device_probe();
  

Personally, I don't mind if this wrapper waits here (even if it makes it less of a 'pure' wrapper), so long as we document it. Can you think of any cases where we explicitly want _not_ to wait in a test?

•   return kunit_add_action_or_reset(test,
  

•                                    (kunit_action_t *)&platform_driver_unregister,
  

•                                    drv);
  

+} +EXPORT_SYMBOL_GPL(platform_driver_register_kunit); diff --git a/include/kunit/platform_device.h b/include/kunit/platform_device.h new file mode 100644 index 000000000000..28d28abf15a4 --- /dev/null +++ b/include/kunit/platform_device.h @@ -0,0 +1,15 @@ +/* SPDX-License-Identifier: GPL-2.0 */ +#ifndef _KUNIT_PLATFORM_DRIVER_H +#define _KUNIT_PLATFORM_DRIVER_H

+struct kunit; +struct platform_device; +struct platform_driver;

+struct platform_device * +platform_device_alloc_kunit(struct kunit *test, const char *name, int id); +int platform_device_add_kunit(struct kunit *test, struct platform_device *pdev);

+int platform_driver_register_kunit(struct kunit *test, struct platform_driver *drv);

+#endif

https://git.kernel.org/pub/scm/linux/kernel/git/clk/linux.git/ https://git.kernel.org/pub/scm/linux/kernel/git/sboyd/spmi.git

-- You received this message because you are subscribed to the Google Groups "KUnit Development" group. To unsubscribe from this group and stop receiving emails from it, send an email to kunit-dev+unsubscribe@googlegroups.com. To view this discussion on the web visit https://groups.google.com/d/msgid/kunit-dev/20240422232404.213174-6-sboyd%40....

On Fri, 3 May 2024 at 09:04, Stephen Boyd sboyd@kernel.org wrote:

Quoting David Gow (2024-05-01 00:55:46)

...

On Tue, 23 Apr 2024 at 07:24, Stephen Boyd sboyd@kernel.org wrote:

...

diff --git a/Documentation/dev-tools/kunit/api/platformdevice.rst b/Documentation/dev-tools/kunit/api/platformdevice.rst new file mode 100644 index 000000000000..b228fb6558c2 --- /dev/null +++ b/Documentation/dev-tools/kunit/api/platformdevice.rst @@ -0,0 +1,10 @@ +.. SPDX-License-Identifier: GPL-2.0

+=================== +Platform Device API +===================

+The KUnit platform device API is used to test platform devices.

+.. kernel-doc:: drivers/base/test/platform_kunit.c

• :export:

diff --git a/drivers/base/test/Makefile b/drivers/base/test/Makefile index e321dfc7e922..740aef267fbe 100644 --- a/drivers/base/test/Makefile +++ b/drivers/base/test/Makefile @@ -1,8 +1,11 @@ # SPDX-License-Identifier: GPL-2.0 obj-$(CONFIG_TEST_ASYNC_DRIVER_PROBE) += test_async_driver_probe.o

+obj-$(CONFIG_KUNIT) += platform_kunit.o

Do we want this to be part of the kunit.ko module (and hence, probably, under lib/kunit), or to keep this as a separate module. I'm tempted, personally, to treat this as a part of KUnit, and have it be part of the same module. There are a couple of reasons for this:

• It's nice to have CONFIG_KUNIT produce only one module. If we want

this to be separate, I'd be tempted to put it behind its own kconfig entry.

• The name platform_kunit.ko suggests (to me, at least) that this is

the test for platform devices, not the implementation of the helper.

I was following *_kunit as "helpers" and *_test as the test. Only loosely based on the documentation that mentions to use _test or _kunit for test files. Maybe it should have _kunit_helpers postfix?

Yeah, the style guide currently suggests that *_test is the default for tests, but that _kunit may also be used for tests if _test is already used for non-KUnit tests: https://docs.kernel.org/dev-tools/kunit/style.html#test-file-and-module-name...

DRM has drm_kunit_helpers, so _kunit_helpers seems like a good suffix to settle on.

Following the single module design should I merge the tests for this code into kunit-test.c? And do the same sort of thing for clk helpers? That sounds like it won't scale very well if everything is in one module.

I don't think it's as important that the tests live in the same module. It's nice from an ergonomic point-of-view to only have to modprobe the one thing, but we've already let that ship sail somewhat with string-stream-test.

Either way, splitting up kunit-test.c is something we'll almost certainly want to do at some point, and we can always put them into the same module even if they're different source files if we have to.

Shouldn't the wrapper code for subsystems live in those subsystems like drm_kunit_helpers.c does? Maybe the struct device kunit wrappers should be moved out to drivers/base/? lib/kunit can stay focused on providing pure kunit code then.

I tend to agree that wrapper code for subsystems should live in those subsystems, especially if the subsystems are relatively self-contained (i.e., the helpers are used to test that subsystem itself, rather than exported for other parts of the kernel to use to test interactions with said subsystem). For 'core' parts of the kernel, I think it makes it easier to make these obviously part of KUnit (e.g. kunit_kzalloc() is easier to have within KUnit, rather than as a part of the allocators).

The struct device wrappers have the problem that they rely on the kunit_bus being registered, which is currently done when the kunit module is loaded. So it hooks more deeply into KUnit than is comfortable to do from drivers/base. So we've treated it as a 'core' part of the kernel.

Ultimately, it's a grey area, so I can live with this going either way, depending on the actual helpers, so long as we don't end up with lots of half-in/half-out helpers, which behave a bit like both. (For example, at the moment, helpers which live outside lib/kunit are documented and have headers in the respective subsystems' directories.)

FWIW, my gut feeling for what's "most consistent" with what we've done so far is: 1. platform_device helpers should live alongside the current managed device stuff, which is currently in lib/kunit 2. clk helpers should probably live in clk 3. of/of_overlay sits a bit in the middle, but having thought more about it, it'd probably lean towards having it be part of 'of', not 'kunit.

But all of this is, to some extent, just bikeshedding, so as long as we pick somewhere to put them, and don't mix things up too much, I don't think it matters exactly what side of this fuzzy line they end up on.

...

I probably can be persuaded otherwise if you've got a strong preference for it to stay as-is, though.

...

diff --git a/drivers/base/test/platform_kunit.c b/drivers/base/test/platform_kunit.c new file mode 100644 index 000000000000..54af6db2a6d8 --- /dev/null +++ b/drivers/base/test/platform_kunit.c @@ -0,0 +1,174 @@ +// SPDX-License-Identifier: GPL-2.0 +/*

• • Test managed platform driver
• */

+#include <linux/device/driver.h> +#include <linux/platform_device.h>

+#include <kunit/platform_device.h> +#include <kunit/resource.h>

+/**

• • platform_device_alloc_kunit() - Allocate a KUnit test managed platform device
• • @test: test context
• • @name: device name of platform device to alloc
• • @id: identifier of platform device to alloc.
• • Allocate a test managed platform device. The device is put when the test completes.
• • Return: Allocated platform device on success, NULL on failure.
• */

+struct platform_device * +platform_device_alloc_kunit(struct kunit *test, const char *name, int id)

I'd prefer, personally, this be named something like kunit_platform_device_alloc(), to match the existing kunit_device_register() functions.

...

+{

•   struct platform_device *pdev;
  

•   pdev = platform_device_alloc(name, id);
  

•   if (!pdev)
  

•           return NULL;
  

•   if (kunit_add_action_or_reset(test, (kunit_action_t *)&platform_device_put, pdev))
  

Alas, casting function pointers to kunit_action_t* breaks CFI. It's worth using a wrapper, which can be created with the KUNIT_DEFINE_ACTION_WRAPPER() macro, e.g.

KUNIT_DEFINE_ACTION_WRAPPER(platform_device_put_wrapper, platform_device_put, struct platform_device *);

Thanks. I missed that.

...

...

•           return NULL;
  

•   return pdev;
  

+} +EXPORT_SYMBOL_GPL(platform_device_alloc_kunit);

+static void platform_device_add_kunit_exit(struct kunit_resource *res) +{

•   struct platform_device *pdev = res->data;
  

•   platform_device_unregister(pdev);
  

+}

+static bool +platform_device_alloc_kunit_match(struct kunit *test,

•                             struct kunit_resource *res, void *match_data)
  

+{

•   struct platform_device *pdev = match_data;
  

•   return res->data == pdev;
  

+}

+/**

• • platform_device_add_kunit() - Register a KUnit test managed platform device
• • @test: test context
• • @pdev: platform device to add
• • Register a test managed platform device. The device is unregistered when the
• • test completes.
• • Return: 0 on success, negative errno on failure.
• */

+int platform_device_add_kunit(struct kunit *test, struct platform_device *pdev)

As above, I'd lean towards naming this kunit_platform_device_add() for consistency with the other KUnit device helpers.

...

+{

•   struct kunit_resource *res;
  

•   int ret;
  

•   ret = platform_device_add(pdev);
  

•   if (ret)
  

•           return ret;
  

•   res = kunit_find_resource(test, platform_device_alloc_kunit_match, pdev);
  

•   if (res) {
  

•           /*
  

•            * Transfer the reference count of the platform device if it was
  

•            * allocated with platform_device_alloc_kunit(). In that case,
  

•            * calling platform_device_put() leads to reference count
  

•            * underflow because platform_device_unregister() does it for
  

•            * us and we call platform_device_unregister() from
  

•            * platform_device_add_kunit_exit().
  

•            *
  

•            * Usually callers transfer the refcount from
  

•            * platform_device_alloc() to platform_device_add() and simply
  

•            * call platform_device_unregister() when done, but with kunit
  

•            * we have to keep this straight by redirecting the free
  

•            * routine for the resource.
  

•            */
  

•           res->free = platform_device_add_kunit_exit;
  

•           kunit_put_resource(res);
  

•   } else if (kunit_add_action_or_reset(test,
  

•                                        (kunit_action_t *)&platform_device_unregister,
  

•                                        pdev)) {
  

Nit: We don't want to cast directly to kunit_action_t *, as that breaks CFI. Can we use KUNIT_DEFINE_ACTION_WRAPPER()?

...

•           return -ENOMEM;
  

Nit: This is fine, as kunit_add_action_or_reset() only returns 0 or -ENOMEM at the moment, but it could cause problems down the line if we ever want to return a different error. I don't think that's particularly likely, but it might be nicer to properly propagate the error.

I will propagate the return value.

...

...

•   }
  

•   return 0;
  

+} +EXPORT_SYMBOL_GPL(platform_device_add_kunit);

+/**

• • platform_driver_register_kunit() - Register a KUnit test managed platform driver
• • @test: test context
• • @drv: platform driver to register
• • Register a test managed platform driver. This allows callers to embed the
• • @drv in a container structure and use container_of() in the probe function
• • to pass information to KUnit tests. It can be assumed that the driver has
• • probed when this function returns.
• • Example
• • .. code-block:: c

• • struct kunit_test_context {
    

• •         struct platform_driver pdrv;
    

• •         const char *data;
    

• • };
    

• • static inline struct kunit_test_context *
    

• • to_test_context(struct platform_device *pdev)
    

• • {
    

• •         return container_of(to_platform_driver(pdev->dev.driver),
    

• •                             struct kunit_test_context,
    

• •                             pdrv);
    

• • }
    

• • static int kunit_platform_driver_probe(struct platform_device *pdev)
    

• • {
    

• •         struct kunit_test_context *ctx;
    

• •         ctx = to_test_context(pdev);
    

• •         ctx->data = "test data";
    

• •         return 0;
    

• • }
    

• • static void kunit_platform_driver_test(struct kunit *test)
    

• • {
    

• •         struct kunit_test_context *ctx;
    

• •         ctx = kunit_kzalloc(test, sizeof(*ctx), GFP_KERNEL);
    

• •         KUNIT_ASSERT_NOT_ERR_OR_NULL(test, ctx);
    

• •         ctx->pdrv.probe = kunit_platform_driver_probe;
    

• •         ctx->pdrv.driver.name = "kunit-platform";
    

• •         ctx->pdrv.driver.owner = THIS_MODULE;
    

• •         KUNIT_EXPECT_EQ(test, 0, platform_driver_register_kunit(test, &ctx->pdrv));
    

• •         KUNIT_EXPECT_STREQ(test, ctx->data, "test data");
    

• • }
    

• • Return: 0 on success, negative errno on failure.
• */

+int platform_driver_register_kunit(struct kunit *test,

•                              struct platform_driver *drv)
  

As above, I'd prefer kunit_platform_driver_register()

...

+{

•   int ret;
  

•   ret = platform_driver_register(drv);
  

•   if (ret)
  

•           return ret;
  

•   /*
  

•    * Wait for the driver to probe (or at least flush out of the deferred
  

•    * workqueue)
  

•    */
  

•   wait_for_device_probe();
  

Personally, I don't mind if this wrapper waits here (even if it makes it less of a 'pure' wrapper), so long as we document it. Can you think of any cases where we explicitly want _not_ to wait in a test?

I don't like it because it's not deterministic. The function doesn't take any struct device to wait for. I've already written the code to use a completion, and it works well enough so I'll just do that. Then we don't have to worry if this API goes away, or that it doesn't actually determine if the driver has probed the device.

Sounds good!

Cheers, -- David

Quoting Stephen Boyd (2024-04-22 16:23:58)

diff --git a/drivers/base/test/platform_kunit.c b/drivers/base/test/platform_kunit.c new file mode 100644 index 000000000000..54af6db2a6d8 --- /dev/null +++ b/drivers/base/test/platform_kunit.c @@ -0,0 +1,174 @@

[...]

+struct platform_device * +platform_device_alloc_kunit(struct kunit *test, const char *name, int id) +{

•   struct platform_device *pdev;
  

•   pdev = platform_device_alloc(name, id);
  

•   if (!pdev)
  

•           return NULL;
  

•   if (kunit_add_action_or_reset(test, (kunit_action_t *)&platform_device_put, pdev))
  

•           return NULL;
  

•   return pdev;
  

+} +EXPORT_SYMBOL_GPL(platform_device_alloc_kunit);

+static void platform_device_add_kunit_exit(struct kunit_resource *res) +{

•   struct platform_device *pdev = res->data;
  

•   platform_device_unregister(pdev);
  

+}

+static bool +platform_device_alloc_kunit_match(struct kunit *test,

•                             struct kunit_resource *res, void *match_data)
  

+{

•   struct platform_device *pdev = match_data;
  

•   return res->data == pdev;
  

+}

+/**

• • platform_device_add_kunit() - Register a KUnit test managed platform device
• • @test: test context
• • @pdev: platform device to add
• • Register a test managed platform device. The device is unregistered when the
• • test completes.
• • Return: 0 on success, negative errno on failure.
• */

+int platform_device_add_kunit(struct kunit *test, struct platform_device *pdev) +{

•   struct kunit_resource *res;
  

•   int ret;
  

•   ret = platform_device_add(pdev);
  

•   if (ret)
  

•           return ret;
  

•   res = kunit_find_resource(test, platform_device_alloc_kunit_match, pdev);
  

This doesn't work because platform_device_alloc_kunit() used kunit_add_action_or_reset() which has a chained free routine and data pointer. I've added a test to make sure the platform device is removed from the bus. It's not super great though because when this code fails to find a match it will still remove the device by calling platform_device_unregister() when the test ends. It will follow that up with a call to platform_device_put(), which is the problem as that causes an underflow and operates on an already freed device.

I couldn't come up with anything better than searching the platform bus. Maybe if there was a way to allocate the memory or redirect where platform_device_alloc_kunit() got memory from we could hold the device memory around after it should have been freed and make sure the kref for the device kobject is 0. That seems pretty invasive to do though so I'm just going to leave it for now and add this test to make sure it cleans up.

Quoting Stephen Boyd (2024-04-22 16:23:59)

diff --git a/Documentation/devicetree/bindings/test/test,clk-fixed-rate.yaml b/Documentation/devicetree/bindings/test/test,clk-fixed-rate.yaml new file mode 100644 index 000000000000..b9f58cba944c --- /dev/null +++ b/Documentation/devicetree/bindings/test/test,clk-fixed-rate.yaml @@ -0,0 +1,35 @@ +# SPDX-License-Identifier: (GPL-2.0-only OR BSD-2-Clause) +%YAML 1.2 +--- +$id: http://devicetree.org/schemas/test/test,clk-fixed-rate.yaml# +$schema: http://devicetree.org/meta-schemas/core.yaml#

+title: KUnit clk fixed rate test clk consumer

Removed KUnit from the title.

+maintainers:

• • Stephen Boyd sboyd@kernel.org

+description:

• A clk consumer of a fixed rate clk used to test the fixed rate clk
• implementation.

+properties:

• compatible:
• const: test,clk-fixed-rate

None of this is really fixed rate clk specific. I'm going to rename this to "test,single-clk-consumer" so that it can be used anywhere we want to consume a single clock cell. Someone can introduce a double/multiple binding as needed.

• clocks:
• maxItems: 1

+required:

• • compatible
• • clocks

+additionalProperties: false

+examples:

• • |
• clock-consumer {

•  compatible = "test,clk-fixed-rate";
  

•  clocks = <&fixed_clk>;
  

• };

+...

Quoting David Gow (2024-05-01 01:08:11)

Thanks very much. I'm about halfway through reviewing these, and I like them a lot so far.

Most of my thoughts are just naming ideas. I fear some of them may be the reverse of previous suggestions, as we've since landed the KUnit device wrappers in include/kunit/device.h, which we decided would live as part of KUnit, not as part of the device infrastructure. I don't enormously mind if we make the opposite decision for these, though it does seem a bit inconsistent if we do 'devices' differently from 'platform_devices'. Thoughts?

Let's discuss on one of the patches.

The other thing I've noted so far is that the of_apply_kunit_platform_device and of_overlay_apply_kunit_cleanup tests fail (and BUG() with a NULL pointer) on powerpc:

...

[15:18:51] # of_overlay_apply_kunit_platform_device: EXPECTATION FAILED at drivers/of/overlay_test.c:47 [15:18:51] Expected pdev is not null, but is [15:18:51] BUG: Kernel NULL pointer dereference at 0x0000004c

This seems to be because pdev is NULL and we call put_device(&pdev->dev) on it. We could be nicer and have an 'if (pdev)' check there. I wonder if that fixes the other two below?

---8<--- diff --git a/drivers/of/overlay_test.c b/drivers/of/overlay_test.c index 223e5a5c23c5..85cfbe6bb132 100644 --- a/drivers/of/overlay_test.c +++ b/drivers/of/overlay_test.c @@ -45,7 +45,8 @@ static void of_overlay_apply_kunit_platform_device(struct kunit *test)

pdev = of_find_device_by_node(np); KUNIT_EXPECT_NOT_ERR_OR_NULL(test, pdev); - put_device(&pdev->dev); + if (pdev) + put_device(&pdev->dev); }

static int of_overlay_bus_match_compatible(struct device *dev, const void *data) @@ -77,8 +78,8 @@ static void of_overlay_apply_kunit_cleanup(struct kunit *test) KUNIT_ASSERT_NOT_ERR_OR_NULL(test, np);

pdev = of_find_device_by_node(np); - put_device(&pdev->dev); /* Not derefing 'pdev' after this */ KUNIT_ASSERT_NOT_ERR_OR_NULL(test, pdev); + put_device(&pdev->dev); /* Not derefing 'pdev' after this */

/* Remove overlay */ kunit_cleanup(&fake); @@ -91,7 +92,8 @@ static void of_overlay_apply_kunit_cleanup(struct kunit *test) dev = bus_find_device(&platform_bus_type, NULL, kunit_compatible, of_overlay_bus_match_compatible); KUNIT_EXPECT_PTR_EQ(test, NULL, dev); - put_device(dev); + if (dev) + put_device(dev); }

static struct kunit_case of_overlay_apply_kunit_test_cases[] = {

...

[15:18:51] # of_overlay_apply_kunit_platform_device: try faulted: last line seen lib/kunit/resource.c:99 [15:18:51] # of_overlay_apply_kunit_platform_device: internal error occurred preventing test case from running: -4 [15:18:51] [FAILED] of_overlay_apply_kunit_platform_device

...

[15:18:51] BUG: Kernel NULL pointer dereference at 0x0000004c [15:18:51] note: kunit_try_catch[698] exited with irqs disabled [15:18:51] # of_overlay_apply_kunit_cleanup: try faulted: last line seen drivers/of/overlay_test.c:77 [15:18:51] # of_overlay_apply_kunit_cleanup: internal error occurred preventing test case from running: -4 [15:18:51] [FAILED] of_overlay_apply_kunit_cleanup

I've not had a chance to dig into it any further, yet, but it appears to work on all of the other architectures I tried.

Cool. I don't know why powerpc doesn't make devices. Maybe it has a similar design to sparc to create resources. I'll check it out.

On Tue, May 14, 2024 at 4:29 PM Stephen Boyd sboyd@kernel.org wrote:

Quoting Stephen Boyd (2024-05-02 18:27:42)

...

Quoting David Gow (2024-05-01 01:08:11)

...

The other thing I've noted so far is that the of_apply_kunit_platform_device and of_overlay_apply_kunit_cleanup tests fail (and BUG() with a NULL pointer) on powerpc:

...

[15:18:51] # of_overlay_apply_kunit_platform_device: EXPECTATION FAILED at drivers/of/overlay_test.c:47 [15:18:51] Expected pdev is not null, but is [15:18:51] BUG: Kernel NULL pointer dereference at 0x0000004c

This seems to be because pdev is NULL and we call put_device(&pdev->dev) on it. We could be nicer and have an 'if (pdev)' check there. I wonder if that fixes the other two below?

---8<--- diff --git a/drivers/of/overlay_test.c b/drivers/of/overlay_test.c index 223e5a5c23c5..85cfbe6bb132 100644 --- a/drivers/of/overlay_test.c +++ b/drivers/of/overlay_test.c @@ -91,7 +92,8 @@ static void of_overlay_apply_kunit_cleanup(struct kunit *test) dev = bus_find_device(&platform_bus_type, NULL, kunit_compatible, of_overlay_bus_match_compatible); KUNIT_EXPECT_PTR_EQ(test, NULL, dev);

•   put_device(dev);
  

•   if (dev)
  

•           put_device(dev);
  

}

This last hunk isn't needed.

...

static struct kunit_case of_overlay_apply_kunit_test_cases[] = {

...

...

[15:18:51] # of_overlay_apply_kunit_platform_device: try faulted: last line seen lib/kunit/resource.c:99 [15:18:51] # of_overlay_apply_kunit_platform_device: internal error occurred preventing test case from running: -4 [15:18:51] [FAILED] of_overlay_apply_kunit_platform_device

...

[15:18:51] BUG: Kernel NULL pointer dereference at 0x0000004c [15:18:51] note: kunit_try_catch[698] exited with irqs disabled [15:18:51] # of_overlay_apply_kunit_cleanup: try faulted: last line seen drivers/of/overlay_test.c:77 [15:18:51] # of_overlay_apply_kunit_cleanup: internal error occurred preventing test case from running: -4 [15:18:51] [FAILED] of_overlay_apply_kunit_cleanup

I've not had a chance to dig into it any further, yet, but it appears to work on all of the other architectures I tried.

Cool. I don't know why powerpc doesn't make devices. Maybe it has a similar design to sparc to create resources. I'll check it out.

powerpc doesn't mark the root node with OF_POPULATED_BUS. If I set that in of_platform_default_populate_init() then the overlays can be applied.

---8<---- diff --git a/drivers/of/platform.c b/drivers/of/platform.c index 389d4ea6bfc1..fa7b439e9402 100644 --- a/drivers/of/platform.c +++ b/drivers/of/platform.c @@ -565,6 +565,10 @@ static int __init of_platform_default_populate_init(void) of_platform_device_create(node, buf, NULL); }

•           node = of_find_node_by_path("/");
  

•           if (node)
  

•                   of_node_set_flag(node, OF_POPULATED_BUS);
  

I think you want to do this in of_platform_bus_probe() instead to mirror of_platform_populate(). These are supposed to be the same except that 'populate' only creates devices for nodes with compatible while 'probe' will create devices for all child nodes. Looks like we are missing some devlink stuff too. There may have been some issue for PPC with it.

•           of_node_put(node);
    } else {
            /*
             * Handle certain compatibles explicitly, since we don't want to create
  
  

I'm guessing this is wrong though, because I see bunch of powerpc specific code calling of_platform_bus_probe() which will set the flag on the actual platform bus nodes. Maybe we should just allow overlays to create devices at the root node regardless? Of course, the flag doc says "platform bus created for children" and if we never populated the root then that isn't entirely accurate.

Rob, can you point me in the right direction? Do we need to use simple-bus in the test overlays and teach overlay code to populate that bus?

Overlays adding things to the root node might be suspect, but probably there are some valid reasons to do so. If simple-bus makes sense here, then yes, you should use it. But if what's on it is not MMIO devices, don't. That's a warning in the schema now.

Rob

On Wed, May 15, 2024 at 4:15 PM Stephen Boyd sboyd@kernel.org wrote:

Quoting Rob Herring (2024-05-15 06:06:09)

...

On Tue, May 14, 2024 at 4:29 PM Stephen Boyd sboyd@kernel.org wrote:

...

powerpc doesn't mark the root node with OF_POPULATED_BUS. If I set that in of_platform_default_populate_init() then the overlays can be applied.

---8<---- diff --git a/drivers/of/platform.c b/drivers/of/platform.c index 389d4ea6bfc1..fa7b439e9402 100644 --- a/drivers/of/platform.c +++ b/drivers/of/platform.c @@ -565,6 +565,10 @@ static int __init of_platform_default_populate_init(void) of_platform_device_create(node, buf, NULL); }

•           node = of_find_node_by_path("/");
  

•           if (node)
  

•                   of_node_set_flag(node, OF_POPULATED_BUS);
  

I think you want to do this in of_platform_bus_probe() instead to mirror of_platform_populate(). These are supposed to be the same except that 'populate' only creates devices for nodes with compatible while 'probe' will create devices for all child nodes. Looks like we are missing some devlink stuff too. There may have been some issue for PPC with it.

Got it. So this patch?

---8<--- diff --git a/drivers/of/platform.c b/drivers/of/platform.c index 389d4ea6bfc1..acecefcfdba7 100644 --- a/drivers/of/platform.c +++ b/drivers/of/platform.c @@ -421,6 +421,7 @@ int of_platform_bus_probe(struct device_node *root, if (of_match_node(matches, root)) { rc = of_platform_bus_create(root, matches, NULL, parent, false); } else for_each_child_of_node(root, child) {

•           of_node_set_flag(root, OF_POPULATED_BUS);
  

No, the same spot as of_platform_populate has it. I guess this would be the same, but no reason to do this in the for_each_child_of_node loop...

            if (!of_match_node(matches, child))
                    continue;
            rc = of_platform_bus_create(child, matches, NULL, parent, false);

This doesn't work though. I see that prom_init() is called, which constructs a DTB and flattens it to be unflattened by unflatten_device_tree(). The powerpc machine type used by qemu is PLATFORM_PSERIES_LPAR. It looks like it never calls of_platform_bus_probe() from the pseries platform code.

Huh. Maybe pseries doesn't have any platform devices?

Ideally, we'd still do it in of_platform_default_populate_init(), but if you look at the history, you'll see that broke some PPC boards (damn initcall ordering).

What about skipping the OF_POPULATED_BUS check, or skipping the check when the parent is the root node? This is the if condition that's giving the headache.

I don't think we should just remove it, but a root node check seems fine.

Rob