From: "Steven Rostedt (Google)" <[email protected]>
The trace event devlink_trap_report uses the __dynamic_array() macro to
determine the size of the input_dev_name field. This is because it needs
to test the dev field for NULL, and will use "NULL" if it is. But it also
has the size of the dynamic array as a fixed IFNAMSIZ bytes. This defeats
the purpose of the dynamic array, as this will reserve that amount of
bytes on the ring buffer, and to make matters worse, it will even save
that size in the event as the event expects it to be dynamic (for which it
is not).
Since IFNAMSIZ is just 16 bytes, just make it a static array and this will
remove the meta data from the event that records the size.
Link: https://lkml.kernel.org/r/[email protected]
Cc: Leon Romanovsky <[email protected]>
Cc: Jiri Pirko <[email protected]>
Cc: "David S. Miller" <[email protected]>
Cc: Eric Dumazet <[email protected]>
Cc: Paolo Abeni <[email protected]>
Cc: [email protected]
Acked-by: Jakub Kicinski <[email protected]>
Signed-off-by: Steven Rostedt (Google) <[email protected]>
---
include/trace/events/devlink.h | 7 +++----
1 file changed, 3 insertions(+), 4 deletions(-)
diff --git a/include/trace/events/devlink.h b/include/trace/events/devlink.h
index 2814f188d98c..24969184c534 100644
--- a/include/trace/events/devlink.h
+++ b/include/trace/events/devlink.h
@@ -186,7 +186,7 @@ TRACE_EVENT(devlink_trap_report,
__string(driver_name, devlink_to_dev(devlink)->driver->name)
__string(trap_name, metadata->trap_name)
__string(trap_group_name, metadata->trap_group_name)
- __dynamic_array(char, input_dev_name, IFNAMSIZ)
+ __array(char, input_dev_name, IFNAMSIZ)
),
TP_fast_assign(
@@ -197,15 +197,14 @@ TRACE_EVENT(devlink_trap_report,
__assign_str(driver_name, devlink_to_dev(devlink)->driver->name);
__assign_str(trap_name, metadata->trap_name);
__assign_str(trap_group_name, metadata->trap_group_name);
- __assign_str(input_dev_name,
- (input_dev ? input_dev->name : "NULL"));
+ strscpy(__entry->input_dev_name, input_dev ? input_dev->name : "NULL", IFNAMSIZ);
),
TP_printk("bus_name=%s dev_name=%s driver_name=%s trap_name=%s "
"trap_group_name=%s input_dev_name=%s", __get_str(bus_name),
__get_str(dev_name), __get_str(driver_name),
__get_str(trap_name), __get_str(trap_group_name),
- __get_str(input_dev_name))
+ __entry->input_dev_name)
);
#endif /* _TRACE_DEVLINK_H */
--
2.35.1
On Thu, Jul 14, 2022 at 12:42:59PM -0400, Steven Rostedt wrote:
> From: "Steven Rostedt (Google)" <[email protected]>
>
> The trace event devlink_trap_report uses the __dynamic_array() macro to
> determine the size of the input_dev_name field. This is because it needs
> to test the dev field for NULL, and will use "NULL" if it is. But it also
> has the size of the dynamic array as a fixed IFNAMSIZ bytes. This defeats
> the purpose of the dynamic array, as this will reserve that amount of
> bytes on the ring buffer, and to make matters worse, it will even save
> that size in the event as the event expects it to be dynamic (for which it
> is not).
>
> Since IFNAMSIZ is just 16 bytes, just make it a static array and this will
> remove the meta data from the event that records the size.
>
> Link: https://lkml.kernel.org/r/[email protected]
>
> Cc: Leon Romanovsky <[email protected]>
> Cc: Jiri Pirko <[email protected]>
> Cc: "David S. Miller" <[email protected]>
> Cc: Eric Dumazet <[email protected]>
> Cc: Paolo Abeni <[email protected]>
> Cc: [email protected]
> Acked-by: Jakub Kicinski <[email protected]>
> Signed-off-by: Steven Rostedt (Google) <[email protected]>
On the off chance that my tags weren't omitted on purpose:
Reviewed-by: Ido Schimmel <[email protected]>
Tested-by: Ido Schimmel <[email protected]>
s/even/event/ in subject
On Thu, 14 Jul 2022 21:40:40 +0300
Ido Schimmel <[email protected]> wrote:
> On the off chance that my tags weren't omitted on purpose:
Ah, I probably pulled it in from my patchwork before you sent the tags. I
usually let my internal patchwork add them for me. So, it was not on
purpose.
It's also a reason I post to the mailing list before I push to linux-next.
In case I missed anything.
>
> Reviewed-by: Ido Schimmel <[email protected]>
> Tested-by: Ido Schimmel <[email protected]>
Will add. Thanks!
>
> s/even/event/ in subject
I'll fix that too.
-- Steve