From: Oliver Hartkopp <socketcan@hartkopp.net>
To: Paolo Abeni <pabeni@redhat.com>, Florian Westphal <fw@strlen.de>,
Marc Kleine-Budde <mkl@pengutronix.de>,
Vincent Mailhol <mailhol@kernel.org>,
Robin van der Gracht <robin@protonic.nl>,
Oleksij Rempel <o.rempel@pengutronix.de>,
kernel@pengutronix.de, "David S. Miller" <davem@davemloft.net>,
Eric Dumazet <edumazet@google.com>,
Jakub Kicinski <kuba@kernel.org>, Simon Horman <horms@kernel.org>
Cc: linux-can@vger.kernel.org, linux-kernel@vger.kernel.org,
netdev@vger.kernel.org, kernel test robot <lkp@intel.com>
Subject: Re: [PATCH net-next v8 0/6] move CAN skb headroom content to skb extensions
Date: Tue, 3 Feb 2026 20:19:31 +0100 [thread overview]
Message-ID: <c5929461-8da5-4c77-bf3d-796c7645d7f9@hartkopp.net> (raw)
In-Reply-To: <7d544645-8699-409f-89c0-6fc606113627@redhat.com>
Hello Paolo,
On 03.02.26 15:40, Paolo Abeni wrote:
> On 2/1/26 3:33 PM, Oliver Hartkopp via B4 Relay wrote:
>> CAN bus related skbuffs (ETH_P_CAN/ETH_P_CANFD/ETH_P_CANXL) simply contain
>> CAN frame structs for CAN CC/FD/XL of skb->len length at skb->data. Those
>> CAN skbs do not have network/mac/transport headers nor other such
>> references for encapsulated protocols like ethernet/IP protocols.
>>
>> To store data for CAN specific use-cases all CAN bus related skbuffs are
>> created with a 16 byte private skb headroom (struct can_skb_priv). Using
>> the skb headroom and accessing skb->head for this private data led to
>> several problems in the past likely due to "The struct can_skb_priv
>> business is highly unconventional for the networking stack." [1]
>>
>> This patch set aims to remove the unconventional skb headroom usage for CAN
>> bus related skbuffs and use the common skb extensions instead.
>>
>> [1] https://lore.kernel.org/linux-can/20260104074222.29e660ac@kernel.org/
>
> Could you please share how skb_ext size change with this series?
> (possibly breaking down the actual size to each separate extension).
>
> Ideally/hopefully the skbuff_ext_cache size is not going to change, and
> that would ensure that this change will not cause any indirect regressions.
>
> /P
I'm not really sure what your question is about and how I could actively
change the impact of this series.
When CONFIG_CAN is enabled the skbuff_ext_cache element would increase
in size by 8 bytes (sizeof(struct can_skb_ext)) on my machine (see
pahole output below).
So when everything is enabled it would be
CONFIG_SKB_EXTENSIONS
8 bytes sizeof(struct skb_ext)
CONFIG_BRIDGE_NETFILTER
32 bytes sizeof(struct nf_bridge_info)
CONFIG_XFRM
88 bytes sizeof(struct sec_path)
CONFIG_NET_TC_SKB_EXT
16 bytes sizeof(struct tc_skb_ext)
CONFIG_MPTCP
32 bytes sizeof(struct mptcp_ext)
CONFIG_MCTP_FLOWS
8 bytes sizeof(struct mctp_flow)
CONFIG_INET_PSP
8 bytes sizeof(struct psp_skb_ext)
CONFIG_CAN
8 bytes sizeof(struct can_skb_ext)
---------
200 bytes total skbuff_ext_cache element size
(255 * 8 = 2040 bytes max space for skb extension users).
Does this answer your question?
Best regards,
Oliver
static const u8 skb_ext_type_len[] = {
#if IS_ENABLED(CONFIG_BRIDGE_NETFILTER)
[SKB_EXT_BRIDGE_NF] = SKB_EXT_CHUNKSIZEOF(struct nf_bridge_info),
#endif
#ifdef CONFIG_XFRM
[SKB_EXT_SEC_PATH] = SKB_EXT_CHUNKSIZEOF(struct sec_path),
#endif
#if IS_ENABLED(CONFIG_NET_TC_SKB_EXT)
[TC_SKB_EXT] = SKB_EXT_CHUNKSIZEOF(struct tc_skb_ext),
#endif
#if IS_ENABLED(CONFIG_MPTCP)
[SKB_EXT_MPTCP] = SKB_EXT_CHUNKSIZEOF(struct mptcp_ext),
#endif
#if IS_ENABLED(CONFIG_MCTP_FLOWS)
[SKB_EXT_MCTP] = SKB_EXT_CHUNKSIZEOF(struct mctp_flow),
#endif
#if IS_ENABLED(CONFIG_INET_PSP)
[SKB_EXT_PSP] = SKB_EXT_CHUNKSIZEOF(struct psp_skb_ext),
#endif
#if IS_ENABLED(CONFIG_CAN)
[SKB_EXT_CAN] = SKB_EXT_CHUNKSIZEOF(struct can_skb_ext),
#endif
};
CONFIG_SKB_EXTENSIONS
struct skb_ext {
refcount_t refcnt; /* 0 4 */
u8 offset[3]; /* 4 3 */
u8 chunks; /* 7 1 */
char data[]
__attribute__((__aligned__(8))); /* 8 0 */
/* size: 8, cachelines: 1, members: 4 */
/* forced alignments: 1 */
/* last cacheline: 8 bytes */
} __attribute__((__aligned__(8)));
CONFIG_BRIDGE_NETFILTER
struct nf_bridge_info {
enum {
BRNF_PROTO_UNCHANGED = 0,
BRNF_PROTO_8021Q = 1,
BRNF_PROTO_PPPOE = 2,
} orig_proto:8; /* 0: 0
4 */
/* Bitfield combined with next fields */
u8 pkt_otherhost:1; /* 1: 0 1 */
u8 in_prerouting:1; /* 1: 1 1 */
u8 bridged_dnat:1; /* 1: 2 1 */
u8 sabotage_in_done:1; /* 1: 3 1 */
/* XXX 4 bits hole, try to pack */
__u16 frag_max_size; /* 2 2 */
int physinif; /* 4 4 */
struct net_device * physoutdev; /* 8 8 */
union {
__be32 ipv4_daddr; /* 16 4 */
struct in6_addr ipv6_daddr; /* 16 16 */
char neigh_header[8]; /* 16 8 */
}; /* 16 16 */
/* size: 32, cachelines: 1, members: 9 */
/* sum members: 30 */
/* sum bitfield members: 12 bits, bit holes: 1, sum bit holes:
4 bits */
/* last cacheline: 32 bytes */
};
CONFIG_XFRM
struct sec_path {
int len; /* 0 4 */
int olen; /* 4 4 */
int verified_cnt; /* 8 4 */
/* XXX 4 bytes hole, try to pack */
struct xfrm_state * xvec[6]; /* 16 48 */
/* --- cacheline 1 boundary (64 bytes) --- */
struct xfrm_offload ovec[1]; /* 64 24 */
/* size: 88, cachelines: 2, members: 5 */
/* sum members: 84, holes: 1, sum holes: 4 */
/* last cacheline: 24 bytes */
};
CONFIG_NET_TC_SKB_EXT
struct tc_skb_ext {
union {
u64 act_miss_cookie; /* 0 8 */
__u32 chain; /* 0 4 */
}; /* 0 8 */
__u16 mru; /* 8 2 */
__u16 zone; /* 10 2 */
u8 post_ct:1; /* 12: 0 1 */
u8 post_ct_snat:1; /* 12: 1 1 */
u8 post_ct_dnat:1; /* 12: 2 1 */
u8 act_miss:1; /* 12: 3 1 */
u8 l2_miss:1; /* 12: 4 1 */
/* size: 16, cachelines: 1, members: 8 */
/* padding: 3 */
/* bit_padding: 3 bits */
/* last cacheline: 16 bytes */
};
CONFIG_MPTCP
struct mptcp_ext {
union {
u64 data_ack; /* 0 8 */
u32 data_ack32; /* 0 4 */
}; /* 0 8 */
u64 data_seq; /* 8 8 */
u32 subflow_seq; /* 16 4 */
u16 data_len; /* 20 2 */
__sum16 csum; /* 22 2 */
u8 use_map:1; /* 24: 0 1 */
u8 dsn64:1; /* 24: 1 1 */
u8 data_fin:1; /* 24: 2 1 */
u8 use_ack:1; /* 24: 3 1 */
u8 ack64:1; /* 24: 4 1 */
u8 mpc_map:1; /* 24: 5 1 */
u8 frozen:1; /* 24: 6 1 */
u8 reset_transient:1; /* 24: 7 1 */
u8 reset_reason:4; /* 25: 0 1 */
u8 csum_reqd:1; /* 25: 4 1 */
u8 infinite_map:1; /* 25: 5 1 */
/* size: 32, cachelines: 1, members: 16 */
/* padding: 6 */
/* bit_padding: 2 bits */
/* last cacheline: 32 bytes */
};
CONFIG_MCTP_FLOWS
struct mctp_flow {
struct mctp_sk_key * key; /* 0 8 */
/* size: 8, cachelines: 1, members: 1 */
/* last cacheline: 8 bytes */
};
CONFIG_INET_PSP
struct psp_skb_ext {
__be32 spi; /* 0 4 */
u16 dev_id; /* 4 2 */
u8 generation; /* 6 1 */
u8 version; /* 7 1 */
/* size: 8, cachelines: 1, members: 4 */
/* last cacheline: 8 bytes */
};
CONFIG_CAN
struct can_skb_ext {
int can_iif; /* 0 4 */
u16 can_framelen; /* 4 2 */
u8 can_gw_hops; /* 6 1 */
u8 can_ext_flags; /* 7 1 */
/* size: 8, cachelines: 1, members: 4 */
/* last cacheline: 8 bytes */
};
next prev parent reply other threads:[~2026-02-03 19:19 UTC|newest]
Thread overview: 12+ messages / expand[flat|nested] mbox.gz Atom feed top
2026-02-01 14:33 Oliver Hartkopp via B4 Relay
2026-02-01 14:33 ` [PATCH net-next v8 1/6] can: use skb hash instead of private variable in headroom Oliver Hartkopp via B4 Relay
2026-02-01 14:33 ` [PATCH net-next v8 2/6] can: add CAN skb extension infrastructure Oliver Hartkopp via B4 Relay
2026-02-01 14:33 ` [PATCH net-next v8 3/6] can: move ifindex to CAN skb extensions Oliver Hartkopp via B4 Relay
2026-02-01 14:33 ` [PATCH net-next v8 4/6] can: move frame_len " Oliver Hartkopp via B4 Relay
2026-02-01 14:33 ` [PATCH net-next v8 5/6] can: remove private CAN skb headroom infrastructure Oliver Hartkopp via B4 Relay
2026-02-01 14:33 ` [PATCH net-next v8 6/6] can: gw: use can_gw_hops instead of sk_buff::csum_start Oliver Hartkopp via B4 Relay
2026-02-03 14:40 ` [PATCH net-next v8 0/6] move CAN skb headroom content to skb extensions Paolo Abeni
2026-02-03 19:19 ` Oliver Hartkopp [this message]
2026-02-05 10:54 ` Paolo Abeni
2026-02-05 12:37 ` Oliver Hartkopp
2026-02-05 11:10 ` patchwork-bot+netdevbpf
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=c5929461-8da5-4c77-bf3d-796c7645d7f9@hartkopp.net \
--to=socketcan@hartkopp.net \
--cc=davem@davemloft.net \
--cc=edumazet@google.com \
--cc=fw@strlen.de \
--cc=horms@kernel.org \
--cc=kernel@pengutronix.de \
--cc=kuba@kernel.org \
--cc=linux-can@vger.kernel.org \
--cc=linux-kernel@vger.kernel.org \
--cc=lkp@intel.com \
--cc=mailhol@kernel.org \
--cc=mkl@pengutronix.de \
--cc=netdev@vger.kernel.org \
--cc=o.rempel@pengutronix.de \
--cc=pabeni@redhat.com \
--cc=robin@protonic.nl \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox
all inboxes | Powered by JetHome®