[PATCH net] net: add skb->data_len and (skb>end - skb->tail) to skb_dump()

[PATCH net] net: add skb->data_len and (skb>end - skb->tail) to skb_dump()

[Eric Dumazet]

While working on a syzbot report, I found that skb_dump()
is lacking two important parts :

- skb->data_len.

- (skb>end - skb->tail) tailroom is zero if skb is not linear.

Signed-off-by: Eric Dumazet <edumazet@google.com>
---
 net/core/skbuff.c | 7 ++++---
 1 file changed, 4 insertions(+), 3 deletions(-)

diff --git a/net/core/skbuff.c b/net/core/skbuff.c
index a56133902c0d9c47b45a4a19b228b151456e5051..61746c2b95f63e465c1f2cd05bf6a61bc5331d8f 100644
--- a/net/core/skbuff.c
+++ b/net/core/skbuff.c
@@ -1312,14 +1312,15 @@ void skb_dump(const char *level, const struct sk_buff *skb, bool full_pkt)
 	has_mac = skb_mac_header_was_set(skb);
 	has_trans = skb_transport_header_was_set(skb);
 
-	printk("%sskb len=%u headroom=%u headlen=%u tailroom=%u\n"
-	       "mac=(%d,%d) mac_len=%u net=(%d,%d) trans=%d\n"
+	printk("%sskb len=%u data_len=%u headroom=%u headlen=%u tailroom=%u\n"
+	       "end-tail=%u mac=(%d,%d) mac_len=%u net=(%d,%d) trans=%d\n"
 	       "shinfo(txflags=%u nr_frags=%u gso(size=%hu type=%u segs=%hu))\n"
 	       "csum(0x%x start=%u offset=%u ip_summed=%u complete_sw=%u valid=%u level=%u)\n"
 	       "hash(0x%x sw=%u l4=%u) proto=0x%04x pkttype=%u iif=%d\n"
 	       "priority=0x%x mark=0x%x alloc_cpu=%u vlan_all=0x%x\n"
 	       "encapsulation=%d inner(proto=0x%04x, mac=%u, net=%u, trans=%u)\n",
-	       level, skb->len, headroom, skb_headlen(skb), tailroom,
+	       level, skb->len, skb->data_len, headroom, skb_headlen(skb),
+	       tailroom, skb->end - skb->tail,
 	       has_mac ? skb->mac_header : -1,
 	       has_mac ? skb_mac_header_len(skb) : -1,
 	       skb->mac_len,
-- 
2.52.0.457.g6b5491de43-goog

Re: [PATCH net] net: add skb->data_len and (skb>end - skb->tail) to skb_dump()

[Willem de Bruijn]

Eric Dumazet wrote:
> While working on a syzbot report, I found that skb_dump()
> is lacking two important parts :
> 
> - skb->data_len.
> 
> - (skb>end - skb->tail) tailroom is zero if skb is not linear.
> 
> Signed-off-by: Eric Dumazet <edumazet@google.com>

Reviewed-by: Willem de Bruijn <willemb@google.com>

Good point on tailroom, I was not aware of that limitation on linear.

data_len is calculated from len and headlen, but may be nice to print.

> ---
>  net/core/skbuff.c | 7 ++++---
>  1 file changed, 4 insertions(+), 3 deletions(-)
> 
> diff --git a/net/core/skbuff.c b/net/core/skbuff.c
> index a56133902c0d9c47b45a4a19b228b151456e5051..61746c2b95f63e465c1f2cd05bf6a61bc5331d8f 100644
> --- a/net/core/skbuff.c
> +++ b/net/core/skbuff.c
> @@ -1312,14 +1312,15 @@ void skb_dump(const char *level, const struct sk_buff *skb, bool full_pkt)
>  	has_mac = skb_mac_header_was_set(skb);
>  	has_trans = skb_transport_header_was_set(skb);
>  
> -	printk("%sskb len=%u headroom=%u headlen=%u tailroom=%u\n"
> -	       "mac=(%d,%d) mac_len=%u net=(%d,%d) trans=%d\n"
> +	printk("%sskb len=%u data_len=%u headroom=%u headlen=%u tailroom=%u\n"

Maybe order len, headlen, datalen, headroom, tailroom.

And really no need to ever print skb_tailroom if end-tail always
captures that. Can just call that tailroom or tailroom* or so.

> +	       "end-tail=%u mac=(%d,%d) mac_len=%u net=(%d,%d) trans=%d\n"
>  	       "shinfo(txflags=%u nr_frags=%u gso(size=%hu type=%u segs=%hu))\n"
>  	       "csum(0x%x start=%u offset=%u ip_summed=%u complete_sw=%u valid=%u level=%u)\n"
>  	       "hash(0x%x sw=%u l4=%u) proto=0x%04x pkttype=%u iif=%d\n"
>  	       "priority=0x%x mark=0x%x alloc_cpu=%u vlan_all=0x%x\n"
>  	       "encapsulation=%d inner(proto=0x%04x, mac=%u, net=%u, trans=%u)\n",
> -	       level, skb->len, headroom, skb_headlen(skb), tailroom,
> +	       level, skb->len, skb->data_len, headroom, skb_headlen(skb),
> +	       tailroom, skb->end - skb->tail,
>  	       has_mac ? skb->mac_header : -1,
>  	       has_mac ? skb_mac_header_len(skb) : -1,
>  	       skb->mac_len,
> -- 
> 2.52.0.457.g6b5491de43-goog
> 

Re: [PATCH net] net: add skb->data_len and (skb>end - skb->tail) to skb_dump()

[patchwork-bot+netdevbpf]

Hello:

This patch was applied to netdev/net.git (main)
by Jakub Kicinski <kuba@kernel.org>:

On Mon, 12 Jan 2026 17:26:21 +0000 you wrote:
> While working on a syzbot report, I found that skb_dump()
> is lacking two important parts :
> 
> - skb->data_len.
> 
> - (skb>end - skb->tail) tailroom is zero if skb is not linear.
> 
> [...]

Here is the summary with links:
  - [net] net: add skb->data_len and (skb>end - skb->tail) to skb_dump()
    https://git.kernel.org/netdev/net/c/220d89df1da6

You are awesome, thank you!
-- 
Deet-doot-dot, I am a bot.
https://korg.docs.kernel.org/patchwork/pwbot.html