iptfs_alloc_skb(9)

💅 iptfs_alloc_skb — from the Kernel API Manual, bestie ✨

okay so you're reading the iptfs_alloc_skb(9) page, and honestly? it's a little baddie. it does exactly one thing and it does it cute: it hands you a fresh skb, no cap 🫶

---

🌸 NAME 🌸

iptfs_alloc_skb — allocate a new skb. that's it. that's the whole vibe 💖

---

✨ SYNOPSIS ✨

struct sk_buff * iptfs_alloc_skb (struct sk_buff *tpl , u32 len , bool
l3resv );

like, one call, three args, main character energy 👑

---

🎀 ARGUMENTS 🎀

• tpl — the skb you're copying the required meta-data from. think of it as the blueprint, the template, the moment 🦋
• len — the linear length of the head data. and yes, zero is fine. zero is valid. zero ate 💅
• l3resv — true if the skb reserve needs to support pushing L3 headers. lowkey the most important bool in this function ngl

---

🧚 DESCRIPTION 🧚

so here's the tea ☕ — a new skb gets allocated, and the required meta-data is copied over from tpl. the head data gets sized to len, and the reserved space is set according to the l3resv boolean. that's the whole thing, periodt ✨

now let's talk about the reserve, because this is where it gets interesting 👀

when l3resv is false, resv is XFRM_IPTFS_MIN_L2HEADROOM, which arranges for skb->data - 16 — and that's a good guess for good cache alignment, placing the to-be-pushed L2 header at the start of a cacheline. cute, efficient, iconic 🎀

otherwise, when l3resv is true, resv is set to the correct reserved space for dst->dev plus the calculated L3 overhead for the xfrm dst — or XFRM_IPTFS_MIN_L3HEADROOM, whichever is larger. and then it's cache aligned so that all the headers will commonly fall in a cacheline when possible. the way it just aligns 😭💖

and here's the why, bestie:

l3resv=true is used on tunnel ingress (tx), because we need to reserve for the new IPTFS packet (i.e., L2+L3 headers). on tunnel egress (rx) the data being copied into the skb includes the user L3 headers already, so we only need to reserve for L2. iykyk 🦋

---

🔥 RETURN 🔥

the new skb, or NULL. that's it. either she slays or she doesn't 💀

---

🌈 SEE ALSO 🌈

• Kernel file ./net/xfrm/xfrm_iptfs.c
• IPTFS_DEFAULT_DROP_TIME_USECS(9)
• IPTFS_DEFAULT_REORDER_WINDOW(9)
• IPTFS_DEFAULT_INIT_DELAY_USECS(9)
• IPTFS_DEFAULT_MAX_QUEUE_SIZE(9)
• struct xfrm_iptfs_config(9)
• struct xfrm_iptfs_data(9)
• iptfs_skb_head_to_frag(9)
• struct iptfs_skb_frag_walk(9)
• iptfs_skb_prepare_frag_walk(9)
• iptfs_skb_can_add_frags(9)
• iptfs_skb_add_frags(9)
• iptfs_pskb_add_frags(9)
• iptfs_pskb_extract_seq(9)
• iptfs_input_save_runt(9)
• __iptfs_iphlen(9)
• __iptfs_iplen(9)
• iptfs_complete_inner_skb(9)
• iptfs_reassem_abort(9)
• iptfs_reassem_done(9)
• iptfs_reassem_cont(9)
• iptfs_input_ordered(9)
• iptfs_drop_timer(9)
• iptfs_input(9)
• iptfs_enqueue(9)
• iptfs_copy_create_frag(9)
• iptfs_copy_create_frags(9)
• iptfs_first_skb(9)
• iptfs_encap_add_ipv4(9)
• iptfs_encap_add_ipv6(9)
• iptfs_prepare_output(9)
• __iptfs_get_inner_mtu(9)
• iptfs_get_inner_mtu(9)
• iptfs_user_init(9)

---

and that's the page, bestie!! 🌈✨ this lil gem lives in ./net/xfrm, it's dated October 2026, and it's filed under iptfs_alloc_skb(9) in the Kernel API Manual. go forth and allocate responsibly 💖🎀🦋