Skip to contentSkip to conversation

tracing: ipv6: Replace deprecated strcpy() with strscpy()

PATCH In review Thorsten Blum 3 participants 8 messages Jul 14, 2025 – Jul 23, 2025 Tracing

lore
Share & tools
lore
#0 lore raw

PATCH 1/1 tracing: ipv6: Replace deprecated strcpy() with strscpy()

strcpy() is deprecated; use strscpy() instead.

Since the destination buffer has a fixed length, strscpy() automatically determines its size using sizeof() when the size argument is omitted. This makes the explicit size argument unnecessary - remove it.

Now, combine both if-else branches using strscpy() and the same buffer into a single statement to simplify the code.

No functional changes intended.

---
 include/trace/events/fib6.h | 8 +++-----
 1 file changed, 3 insertions(+), 5 deletions(-)
include/trace/events/fib6.h +3 -5
index 8d22b2e98d48..903a18836bc6 100644
@@ -64,11 +64,9 @@ TRACE_EVENT(fib6_table_lookup,
 			__entry->dport = 0;
 		}
 
-		if (res->nh && res->nh->fib_nh_dev) {
-			strscpy(__entry->name, res->nh->fib_nh_dev->name, IFNAMSIZ);
-		} else {
-			strcpy(__entry->name, "-");
-		}
+		strscpy(__entry->name, res->nh && res->nh->fib_nh_dev ?
+				       res->nh->fib_nh_dev->name : "-");
+
 		if (res->f6i == net->ipv6.fib6_null_entry) {
 			in6 = (struct in6_addr *)__entry->gw;
 			*in6 = in6addr_any;
Signature · 1 line
2.50.0
Acked-by on v1 1/1
#1 lore raw
Quoted 23 lines · Thorsten Blum

strcpy() is deprecated; use strscpy() instead.

Since the destination buffer has a fixed length, strscpy() automatically determines its size using sizeof() when the size argument is omitted. This makes the explicit size argument unnecessary - remove it.

Now, combine both if-else branches using strscpy() and the same buffer into a single statement to simplify the code.

No functional changes intended.

---
 include/trace/events/fib6.h | 8 +++-----
 1 file changed, 3 insertions(+), 5 deletions(-)
include/trace/events/fib6.h +0 -0
index 8d22b2e98d48..903a18836bc6 100644
@@ -64,11 +64,9 @@ TRACE_EVENT(fib6_table_lookup,
 			__entry->dport = 0;

include/trace/events/fib6.h TRACE_EVENT(fib6_table_lookup, · line 69

Quoted 10 lines
include/trace/events/fib6.h +3 -5
 		}
 
-		if (res->nh && res->nh->fib_nh_dev) {
-			strscpy(__entry->name, res->nh->fib_nh_dev->name, IFNAMSIZ);
-		} else {
-			strcpy(__entry->name, "-");
-		}
+		strscpy(__entry->name, res->nh && res->nh->fib_nh_dev ?
+				       res->nh->fib_nh_dev->name : "-");
+
maintainer
#2 lore raw
In reply to Thorsten Blum
Quoted 9 lines · On Mon, 14 Jul 2025 09:54:33 +0200 Thorsten Blum <[email protected]>…
include/trace/events/fib6.h +0 -1
index 8d22b2e98d48..903a18836bc6 100644
@@ -64,11 +64,9 @@ TRACE_EVENT(fib6_table_lookup,
 			__entry->dport = 0;
 		}
 
-		if (res->nh && res->nh->fib_nh_dev) {

include/trace/events/fib6.h TRACE_EVENT(fib6_table_lookup, · line 72

Quoted 10 lines
include/trace/events/fib6.h +3 -4
-			strscpy(__entry->name, res->nh->fib_nh_dev->name, IFNAMSIZ);
-		} else {
-			strcpy(__entry->name, "-");
-		}
+		strscpy(__entry->name, res->nh && res->nh->fib_nh_dev ?
+				       res->nh->fib_nh_dev->name : "-");
+
 		if (res->f6i == net->ipv6.fib6_null_entry) {
 			in6 = (struct in6_addr *)__entry->gw;
 			*in6 = in6addr_any;

Hmm, why is that string hard coded to 16 bytes and doesn't use the dynamic string facility? Perhaps something like this?

[ I didn't even compile the below, so it may have a syntax error ]

-- Steve

include/trace/events/fib6.h +4 -7
index 8d22b2e98d48..98d2edb02431 100644
@@ -32,7 +32,8 @@ TRACE_EVENT(fib6_table_lookup,
 		__field(        u16,	dport		)
 		__field(        u8,	proto		)
 		__field(        u8,	rt_type		)
-		__array(		char,	name,	IFNAMSIZ )
+		__string(	name,	res->nh && res->nh->fib_nh_dev ?
+					res->nh->fib_nh_dev->name : "-")
 		__array(		__u8,	gw,	16	 )
 	),
 
@@ -64,11 +65,7 @@ TRACE_EVENT(fib6_table_lookup,
 			__entry->dport = 0;
 		}
 
-		if (res->nh && res->nh->fib_nh_dev) {
-			strscpy(__entry->name, res->nh->fib_nh_dev->name, IFNAMSIZ);
-		} else {
-			strcpy(__entry->name, "-");
-		}
+		__assign_str(name);
 		if (res->f6i == net->ipv6.fib6_null_entry) {
 			in6 = (struct in6_addr *)__entry->gw;
 			*in6 = in6addr_any;
@@ -82,7 +79,7 @@ TRACE_EVENT(fib6_table_lookup,
 		  __entry->tb_id, __entry->oif, __entry->iif, __entry->proto,
 		  __entry->src, __entry->sport, __entry->dst, __entry->dport,
 		  __entry->flowlabel, __entry->tos, __entry->scope,
-		  __entry->flags, __entry->name, __entry->gw, __entry->err)
+		  __entry->flags, __get_str(name), __entry->gw, __entry->err)
 );
 
 #endif /* _TRACE_FIB6_H */
#3 lore raw

Hi Steve,

Quoted 3 lines · Steven Rostedt

Hmm, why is that string hard coded to 16 bytes and doesn't use the dynamic string facility? Perhaps something like this?

Your commit fca8300f68fe3 changed it from __dynamic_array() to __array() and __string() seems to be just a special version of __dynamic_array() with a length of -1.

In the commit description you wrote: "Since the size of the name is at most 16 bytes (defined by IFNAMSIZ), it is not worth spending the effort to determine the size of the string."

Thanks,
Thorsten

maintainer
#4 lore raw
Quoted 7 lines · On Wed, 23 Jul 2025 10:46:12 -0700 Thorsten Blum <[email protected]>…

Your commit fca8300f68fe3 changed it from __dynamic_array() to __array() and __string() seems to be just a special version of __dynamic_array() with a length of -1.

In the commit description you wrote: "Since the size of the name is at most 16 bytes (defined by IFNAMSIZ), it is not worth spending the effort to determine the size of the string."

So the original had:

__dynamic_array(char, name, IFNAMSIZ )

Which is not dynamic at all. A dynamic_array (like __string) saves the size in meta data within the event. So basically the above is wasting bytes to save a fixed size. If you are going to use a dynamic array, might as well make it dynamic!

I was doing various clean ups back then so I didn't look too deeply into this event when I made that change. I just saw the obvious waste of space in the ring buffer.

Just to explain it in more detail. A dynamic_array has in the ring buffer:

	short offset;
	short len;
	[..]
	char  name[len];

That is, 4 bytes are used to know the size of the array and where in the event it is located. Thus the __dynamic_array() usage basically had:

	short offset;
	short len = IFNAMSIZ;
	[..]
	char name[IFNAMSIZ];

Why have the offset and length? with just __array(char, name, IFNAMSIZ} it would be just:

	char name[IFNAMSIZ];

See why I changed it?

Now, the change I'm suggesting now would make the __string() be dynamic!

	short offset;
	short len = strlen(res->nh && res->nh->fib_nh_dev ? res->nh->fib_nh_dev->name : "-") + 1;
	[..]
	char name[len];

As IFNAMSIZ is 16, and the above adds 4 bytes to the name, if the name is less than 7 bytes or less, you save memory on the ring buffer.

	2 bytes: offset
	2 bytes: len;
	7 bytes + '\0'

total: 12 bytes

Note, if there's only one dynamic value, it is always at least 4 bytes aligned.

-- Steve

#5 lore raw
Quoted 55 lines · Steven Rostedt
Quoted 7 lines · On Wed, 23 Jul 2025 10:46:12 -0700 Thorsten Blum wrote:

Your commit fca8300f68fe3 changed it from __dynamic_array() to __array() and __string() seems to be just a special version of __dynamic_array() with a length of -1.

In the commit description you wrote: "Since the size of the name is at most 16 bytes (defined by IFNAMSIZ), it is not worth spending the effort to determine the size of the string."

So the original had:

__dynamic_array(char, name, IFNAMSIZ )

Which is not dynamic at all. A dynamic_array (like __string) saves the size in meta data within the event. So basically the above is wasting bytes to save a fixed size. If you are going to use a dynamic array, might as well make it dynamic!

I was doing various clean ups back then so I didn't look too deeply into this event when I made that change. I just saw the obvious waste of space in the ring buffer.

Just to explain it in more detail. A dynamic_array has in the ring buffer:

	short offset;
	short len;
	[..]
	char  name[len];

That is, 4 bytes are used to know the size of the array and where in the event it is located. Thus the __dynamic_array() usage basically had:

	short offset;
	short len = IFNAMSIZ;
	[..]
	char name[IFNAMSIZ];

Why have the offset and length? with just __array(char, name, IFNAMSIZ} it would be just:

	char name[IFNAMSIZ];

See why I changed it?

Now, the change I'm suggesting now would make the __string() be dynamic!

	short offset;
	short len = strlen(res->nh && res->nh->fib_nh_dev ? res->nh->fib_nh_dev->name : "-") + 1;
	[..]
	char name[len];

As IFNAMSIZ is 16, and the above adds 4 bytes to the name, if the name is less than 7 bytes or less, you save memory on the ring buffer.

Quoted 7 lines
	2 bytes: offset
	2 bytes: len;
	7 bytes + '\0'

total: 12 bytes

Note, if there's only one dynamic value, it is always at least 4 bytes aligned.

Thanks for the detailed explanation.

I think the better change would be this then:

include/trace/events/fib6.h +0 -0
index 8d22b2e98d48..3f95df1fd155 100644
@@ -32,8 +32,9 @@ TRACE_EVENT(fib6_table_lookup,
		__field(        u16,	dport		)
		__field(        u8,	proto		)
		__field(        u8,	rt_type		)

- __array( char, name, IFNAMSIZ )
- __array( __u8, gw, 16 )
+ __string( name, res->nh && res->nh->fib_nh_dev ?
+ res->nh->fib_nh_dev->name : "-" )
+ __array( __u8, gw, 16 )

	),

	TP_fast_assign(
@@ -64,11 +65,8 @@ TRACE_EVENT(fib6_table_lookup,
			__entry->dport = 0;
		}

- if (res->nh && res->nh->fib_nh_dev) {
- strscpy(__entry->name, res->nh->fib_nh_dev->name, IFNAMSIZ);
- } else {
- strcpy(__entry->name, "-");
- }
+ __assign_str(name);
+

		if (res->f6i == net->ipv6.fib6_null_entry) {
			in6 = (struct in6_addr *)__entry->gw;
			*in6 = in6addr_any;
@@ -82,7 +80,7 @@ TRACE_EVENT(fib6_table_lookup,
		  __entry->tb_id, __entry->oif, __entry->iif, __entry->proto,
		  __entry->src, __entry->sport, __entry->dst, __entry->dport,
		  __entry->flowlabel, __entry->tos, __entry->scope,

- __entry->flags, __entry->name, __entry->gw, __entry->err)
+ __entry->flags, __get_str(name), __entry->gw, __entry->err)
);

#endif /* _TRACE_FIB6_H */

I'll submit a v2 if you agree that this is correct.

Thanks,
Thorsten

maintainer
#6 lore raw
Quoted 45 lines · On Wed, 23 Jul 2025 14:15:02 -0700 Thorsten Blum <[email protected]>…

Thanks for the detailed explanation.

I think the better change would be this then:

include/trace/events/fib6.h +6 -8
index 8d22b2e98d48..3f95df1fd155 100644
@@ -32,8 +32,9 @@ TRACE_EVENT(fib6_table_lookup,
 		__field(        u16,	dport		)
 		__field(        u8,	proto		)
 		__field(        u8,	rt_type		)
-		__array(		char,	name,	IFNAMSIZ )
-		__array(		__u8,	gw,	16	 )
+		__string(	name,	res->nh && res->nh->fib_nh_dev ?
+					res->nh->fib_nh_dev->name : "-"	)
+		__array(	__u8,	gw,	16	)
 	),
 
 	TP_fast_assign(
@@ -64,11 +65,8 @@ TRACE_EVENT(fib6_table_lookup,
 			__entry->dport = 0;
 		}
 
-		if (res->nh && res->nh->fib_nh_dev) {
-			strscpy(__entry->name, res->nh->fib_nh_dev->name, IFNAMSIZ);
-		} else {
-			strcpy(__entry->name, "-");
-		}
+		__assign_str(name);
+
 		if (res->f6i == net->ipv6.fib6_null_entry) {
 			in6 = (struct in6_addr *)__entry->gw;
 			*in6 = in6addr_any;
@@ -82,7 +80,7 @@ TRACE_EVENT(fib6_table_lookup,
 		  __entry->tb_id, __entry->oif, __entry->iif, __entry->proto,
 		  __entry->src, __entry->sport, __entry->dst, __entry->dport,
 		  __entry->flowlabel, __entry->tos, __entry->scope,
-		  __entry->flags, __entry->name, __entry->gw, __entry->err)
+		  __entry->flags, __get_str(name), __entry->gw, __entry->err)
 );
 
 #endif /* _TRACE_FIB6_H */
Quoted 1 line

I'll submit a v2 if you agree that this is correct.

Isn't the above pretty much what I suggested?

-- Steve

#7 lore raw
Quoted 1 line · Steven Rostedt

Isn't the above pretty much what I suggested?

Ah sorry, yes it's the same. I'll blame the jet lag from traveling :)

Thanks,
Thorsten