currently if skb does not have enough headroom skb_realloc_headrom is called.
It is not optimal because it creates new skb.

Unlike skb_realloc_headroom, new helper pskb_realloc_headroom 
does not allocate a new skb if possible; 
copies skb->sk on new skb when as needed
and frees original skb in case of failures.

This helps to simplify ip[6]_finish_output2(), ip6_xmit() and a few other
functions in vrf, ax25 and bpf.

There are few other cases where this helper can be used but they require
an additional investigations. 

NB: patch "ipv6: use pskb_realloc_headroom in ip6_finish_output2" depends on 
patch "ipv6: allocate enough headroom in ip6_finish_output2()" submitted separately
https://lkml.org/lkml/2021/7/12/732

Vasily Averin (7):
  skbuff: introduce pskb_realloc_headroom()
  ipv6: use pskb_realloc_headroom in ip6_finish_output2
  ipv6: use pskb_realloc_headroom in ip6_xmit
  ipv4: use pskb_realloc_headroom in ip_finish_output2
  vrf: use pskb_realloc_headroom in vrf_finish_output
  ax25: use pskb_realloc_headroom
  bpf: use pskb_realloc_headroom in bpf_out_neigh_v4/6

 drivers/net/vrf.c      | 14 +++------
 include/linux/skbuff.h |  2 ++
 net/ax25/ax25_out.c    | 13 +++------
 net/ax25/ax25_route.c  | 13 +++------
 net/core/filter.c      | 22 +++-----------
 net/core/skbuff.c      | 41 ++++++++++++++++++++++++++
 net/ipv4/ip_output.c   | 12 ++------
 net/ipv6/ip6_output.c  | 78 ++++++++++++++++++--------------------------------
 8 files changed, 89 insertions(+), 106 deletions(-)

-- 
1.8.3.1