neighbor: tracing: Have neigh_create event use __string()

Commit Message

Steven Rostedt July 5, 2022, 10:37 p.m. UTC

From: "Steven Rostedt (Google)" <[e-mail]>

The dev field of the neigh_create event uses __dynamic_array() with a
fixed size, which defeats the purpose of __dynamic_array(). Looking at the
logic, as it already uses __assign_str(), just use the same logic in
__string to create the size needed. It appears that because "dev" can be
NULL, it needs the check. But __string() can have the same checks as
__assign_str() so use them there too.

Cc: David Ahern <[e-mail]>
Cc: David S. Miller <[e-mail]>
Cc: [e-mail]
Signed-off-by: Steven Rostedt (Google) <[e-mail]>
---

[ This is simpler logic than the fib* events, so I figured just
  convert to __string() instead of a static __array() ]

 include/trace/events/neigh.h | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

Comments

David Ahern July 6, 2022, 1:43 a.m. UTC | #1

On 7/5/22 4:37 PM, Steven Rostedt wrote:
> From: "Steven Rostedt (Google)" <[e-mail]>
> 
> The dev field of the neigh_create event uses __dynamic_array() with a
> fixed size, which defeats the purpose of __dynamic_array(). Looking at the
> logic, as it already uses __assign_str(), just use the same logic in
> __string to create the size needed. It appears that because "dev" can be
> NULL, it needs the check. But __string() can have the same checks as
> __assign_str() so use them there too.
> 
> Cc: David Ahern <[e-mail]>
> Cc: David S. Miller <[e-mail]>
> Cc: [e-mail]
> Signed-off-by: Steven Rostedt (Google) <[e-mail]>
> ---
> 
> [ This is simpler logic than the fib* events, so I figured just
>   convert to __string() instead of a static __array() ]
> 

agreed.

Reviewed-by: David Ahern <[e-mail]>

Jakub Kicinski July 8, 2022, 12:21 a.m. UTC | #2

On Tue, 5 Jul 2022 18:37:41 -0400 Steven Rostedt wrote:
> From: "Steven Rostedt (Google)" <[e-mail]>
> 
> The dev field of the neigh_create event uses __dynamic_array() with a
> fixed size, which defeats the purpose of __dynamic_array(). Looking at the
> logic, as it already uses __assign_str(), just use the same logic in
> __string to create the size needed. It appears that because "dev" can be
> NULL, it needs the check. But __string() can have the same checks as
> __assign_str() so use them there too.
> 
> Cc: David Ahern <[e-mail]>
> Cc: David S. Miller <[e-mail]>
> Cc: [e-mail]
> Signed-off-by: Steven Rostedt (Google) <[e-mail]>
> ---
> 
> [ This is simpler logic than the fib* events, so I figured just
>   convert to __string() instead of a static __array() ]

This one is also going via your tree?

Steven Rostedt July 8, 2022, 12:56 a.m. UTC | #3

On Thu, 7 Jul 2022 17:21:01 -0700
Jakub Kicinski <[e-mail]> wrote:

> > 
> > [ This is simpler logic than the fib* events, so I figured just
> >   convert to __string() instead of a static __array() ]  
> 
> This one is also going via your tree?

Yep, I can pull it in too.

-- Steve

Jakub Kicinski July 8, 2022, 1:04 a.m. UTC | #4

On Thu, 7 Jul 2022 20:56:22 -0400 Steven Rostedt wrote:
> > > [ This is simpler logic than the fib* events, so I figured just
> > >   convert to __string() instead of a static __array() ]    
> > 
> > This one is also going via your tree?  
> 
> Yep, I can pull it in too.

Acked-by: Jakub Kicinski <[e-mail]>

Patch

diff --git a/include/trace/events/neigh.h b/include/trace/events/neigh.h
index 62bb17516713..5eaa1fa99171 100644
--- a/include/trace/events/neigh.h
+++ b/include/trace/events/neigh.h
@@ -30,7 +30,7 @@  TRACE_EVENT(neigh_create,
 
 	TP_STRUCT__entry(
 		__field(u32, family)
-		__dynamic_array(char,  dev,   IFNAMSIZ )
+		__string(dev, dev ? dev->name : "NULL")
 		__field(int, entries)
 		__field(u8, created)
 		__field(u8, gc_exempt)