[PATCH net-next 1/2] devlink: fix build error for CONFIG_MODULES=n

A driver calling trace_devlink_hwmsg cannot be built when modules are disabled:

include/trace/events/devlink.h: In function 'trace_event_get_offsets_devlink_hwmsg':
include/trace/events/devlink.h:25:51: error: dereferencing pointer to incomplete type 'struct module'
   __string(owner_name, devlink->dev->driver->owner->name)

This changes the code to only print the module name when modules are actually
enabled, otherwise we hardcode the string "built-in".

Signed-off-by: Arnd Bergmann <arnd@arndb.de>
Fixes: e5224f0fe2ac ("devlink: add hardware messages tracing facility")
---
 include/trace/events/devlink.h | 8 ++++++++
 1 file changed, 8 insertions(+)

diff --git a/include/trace/events/devlink.h b/include/trace/events/devlink.h
index 333c32ac9bfa..26f92d3c7e9c 100644
--- a/include/trace/events/devlink.h
+++ b/include/trace/events/devlink.h

@@ -22,7 +22,11 @@ TRACE_EVENT(devlink_hwmsg,
 	TP_STRUCT__entry(
 		__string(bus_name, devlink->dev->bus->name)
 		__string(dev_name, dev_name(devlink->dev))
+#ifdef CONFIG_MODULES
 		__string(owner_name, devlink->dev->driver->owner->name)
+#else
+		__string(owner_name, "built-in")
+#endif
 		__field(bool, incoming)
 		__field(unsigned long, type)
 		__dynamic_array(u8, buf, len)

@@ -32,7 +36,11 @@ TRACE_EVENT(devlink_hwmsg,
 	TP_fast_assign(
 		__assign_str(bus_name, devlink->dev->bus->name);
 		__assign_str(dev_name, dev_name(devlink->dev));
+#ifdef CONFIG_MODULES
 		__assign_str(owner_name, devlink->dev->driver->owner->name);
+#else
+		__assign_str(owner_name, "built-in");
+#endif
 		__entry->incoming = incoming;
 		__entry->type = type;
 		memcpy(__get_dynamic_array(buf), buf, len);

-- 
2.9.0

Including devlink.h on ARM and probably other 32-bit architectures results in
a harmless warning:

In file included from ../include/trace/define_trace.h:95:0,
                 from ../include/trace/events/devlink.h:51,
                 from ../net/core/devlink.c:30:
include/trace/events/devlink.h: In function 'trace_raw_output_devlink_hwmsg':
include/trace/events/devlink.h:42:12: error: format '%lu' expects argument of type 'long unsigned int', but argument 10 has type 'size_t {aka unsigned int}' [-Werror=format=]

The correct format string for 'size_t' is %zu, not %lu, this works on all
architectures.

Signed-off-by: Arnd Bergmann <arnd@arndb.de>
Fixes: e5224f0fe2ac ("devlink: add hardware messages tracing facility")
---
 include/trace/events/devlink.h | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/include/trace/events/devlink.h b/include/trace/events/devlink.h
index 26f92d3c7e9c..4b75a6f986fc 100644
--- a/include/trace/events/devlink.h
+++ b/include/trace/events/devlink.h

@@ -47,7 +47,7 @@ TRACE_EVENT(devlink_hwmsg,
 		__entry->len = len;
 	),
 
-	TP_printk("bus_name=%s dev_name=%s owner_name=%s incoming=%d type=%lu buf=0x[%*phD] len=%lu",
+	TP_printk("bus_name=%s dev_name=%s owner_name=%s incoming=%d type=%lu buf=0x[%*phD] len=%zu",
 		  __get_str(bus_name), __get_str(dev_name),
 		  __get_str(owner_name), __entry->incoming, __entry->type,
 		  (int) __entry->len, __get_dynamic_array(buf), __entry->len)

-- 
2.9.0

Wed, Jul 13, 2016 at 11:03:37PM CEST, arnd@arndb.de wrote:

I think would be better to use driver->name. I looks like it is always
present. I will do some tests and send a patch.

On Thursday, July 14, 2016 8:21:11 AM CEST Jiri Pirko wrote:

quoted

diff --git a/include/trace/events/devlink.h b/include/trace/events/devlink.h
index 333c32ac9bfa..26f92d3c7e9c 100644
--- a/include/trace/events/devlink.h
+++ b/include/trace/events/devlink.h

@@ -22,7 +22,11 @@ TRACE_EVENT(devlink_hwmsg,
      TP_STRUCT__entry(
              __string(bus_name, devlink->dev->bus->name)
              __string(dev_name, dev_name(devlink->dev))
+#ifdef CONFIG_MODULES
              __string(owner_name, devlink->dev->driver->owner->name)

I think would be better to use driver->name. I looks like it is always
present. I will do some tests and send a patch.

Yes, good idea.

	Arnd

Thu, Jul 14, 2016 at 10:12:43AM CEST, arnd@arndb.de wrote:

On Thursday, July 14, 2016 8:21:11 AM CEST Jiri Pirko wrote:

quoted

quoted

diff --git a/include/trace/events/devlink.h b/include/trace/events/devlink.h
index 333c32ac9bfa..26f92d3c7e9c 100644
--- a/include/trace/events/devlink.h
+++ b/include/trace/events/devlink.h

@@ -22,7 +22,11 @@ TRACE_EVENT(devlink_hwmsg,
      TP_STRUCT__entry(
              __string(bus_name, devlink->dev->bus->name)
              __string(dev_name, dev_name(devlink->dev))
+#ifdef CONFIG_MODULES
              __string(owner_name, devlink->dev->driver->owner->name)

I think would be better to use driver->name. I looks like it is always
present. I will do some tests and send a patch.

Yes, good idea.

I will take your second path, rebase and send along. Thanks.

Arnd