CS 498 Lecture 6 Packet Management

1
CS 498 Lecture 6Packet Management

• Jennifer Hou
• Department of Computer Science
• University of Illinois at Urbana-Champaign
• Reading Chapters 3-4, The Linux Networking
  Architecture Design and Implementation of
  Network Protocols in the Linux Kernel

2
Outline

• Socket buffer sk_buff structure
• APIs for creating, releasing, and duplicating
  socket buffers.
• APIs for manipulating parameters within the
  sk_buff structure
• APIs for managing the socket buffer queue.

3
Structure of sk_buff
sk_buff
sk_buff_head
next
sk_buff
prev
list
stamp
net_device
dev
h
nh
mac
Paketdatenbereich
dst
len
MAC-Header
...
IP-Header
head
UDP-Header
data
UDP-Data
tail
end
datarefp 1
4
Socket Buffer

• The structure used to manage a packet during its
  lifetime.
• Created when an application passes data to a
  socket or when a packet arrives at the network
  adaptor (dev_alloc_skb() is invoked).
• Packet headers of each layer are inserted in
  front of the payload.
• The packet is copied only twice once from the
  user address space to the kernel address space,
  and once when the packet is passed onto the
  network adaptor.

5
Change of sk_buff as a Packet Traverses Across
the Stack
sk_buff
sk_buff
sk_buff
next
next
next
prev
prev
prev
...
...
...
head
head
head
data
data
data
tail
tail
tail
end
end
end
Paketdatenbereich
Paketdatenbereich
Paketdatenbereich
IP-Header
UDP-Header
UDP-Header
UDP-Daten
UDP-Daten
UDP-Daten
datarefp 1
datarefp 1
datarefp 1
6
Parameters of sk_buff Structure

• list points to the queue where the socket buffer
  is currently located.
• sk points to the socket that created the packet.
• stamp specifies the time when the packet arrived
  in the Linux (in jiffies)
• dev states the current network device on which
  the socket buffer operates. Once the routing
  decision is made, dev points to the network
  adapter on which the packet leaves.

7
Parameters of sk_buff Structure

• input_dev always points to the network device
  that received the packet.
• h, nh, mac are pointers to packet headers in the
  transport layer (h), the network layer (nh), and
  the MAC layer (mac).
• dst a reference to the adapter on which the
  packet leaves the computer, or a reference to a
  MAC header stored in the hard header cache.
• cloned indicates if a packet was cloned.

8
Parameters of sk_buff Structure

• pkt_type specifies the type of a packet
• PACKET_HOST a packet sent to the local host
• PACKET_BROADCAST a broadcast packet
• PACKET_MULTICAST a multicast packet
• PACKET_OTHERHOSTa packet not destined for the
  local host, but received in the promiscuous mode.
• PACKET_OTGOING a packet leaving the host
• PACKET_LOOKBACK a packet sent by the local host
  to itself.

9
Parameters of sk_buff Structure

• len specifies the length of a packet.
• data, tail point to currently valid packet data.
• head, end point to the total location that can
  be used for packet data.
• The space between head and data is called
  headroom, and the space between tail and end is
  called tailroom
• datarefp a referece counter that indicates how
  many times the packet has been referenced. Not
  part of the sk_buff structure

10
Operations on Socket Buffers

• Create, release, and duplicate socket buffers.
• Manipulate parameters within the sk_buffer
  structure
• Manage socket buffer queues.

11
1. Creating and Releasing Socket Buffers

• alloc_skb(size,gfp_mask)
• No immediate attempt is made to allocate the
  memory with kmalloc() for the sk_buff structure.
• Tries to reuse a sk_buff in the skb_head_cache
  queue if not successful, tries to obtain a
  packet from the central socket-buffer cache
  (skbuff_head_cache) with kmem_cache_alloc().
• If neither is successful, the invoke kmalloc() to
  reserve memory.
• The packet data space is reserved by use of
  kmalloc().

12
1. Creating and Releasing Socket Buffers

• skb_copy(skb,gfp_mask) creates a copy of the
  socket buffer skb, copying both the sk_buff
  structure and the packet data.
• skb_copy_expand(skb,newheadroom, newtailroom,
  gfp_mask) creates a new copy of the socket
  buffer and packet data, and in addition, reserves
  a larger space before and after the packet data.

13
Copying Socket Buffers
skb_copy
sk_buff
sk_buff
sk_buff
next
next
next
prev
prev
prev
...
...
...
head
head
head
data
data
data
tail
tail
tail
end
end
end
Paket dat storage
Paket data storage
Paket data storage
IP-Header
IP-Header
IP-Header
UDP-Header
UDP-Header
UDP-Header
UDP-Daten
UDP-Daten
UDP-Daten
datarefp 1
datarefp 1
datarefp 1
14
1. Creating and Releasing Socket Buffers

• skb_clone() creates a new socket buffer sk_buff,
  but not the packet data. Pointers in both
  sk_buffs point to the same packet data space.
• Needed in multicast.

15
Cloning Socket Buffers
skb_clone
sk_buff
sk_buff
sk_buff
next
next
next
prev
prev
prev
...
...
...
head
head
head
data
data
data
tail
tail
tail
end
end
end
Paketdatenbereich
Paketdatenbereich
IP-Header
IP-Header
UDP-Header
UDP-Header
UDP-Daten
UDP-Daten
datarefp 1
datarefp 2
16
1. Creating and Releasing Socket Buffers

• kfree_skb() calls kfree_skbmem() to release the
  packet memory, and does all the sanity check
  (e.g., calls the destructor() for the socket
  buffer)
• kfree_skbmem() if skb_cloned() is tested for
  null, and datarefp is tested for one
• invokes kfree() to release the packet data
  memory.
• Invokes skb_head_to_pool() to insert the
  sk_buffer structure into the socket_buffer cache.

17
2. Manipulating Packet Data Space

• skb_unshare(skb) uses skb_cloned to check if the
  socket buffer is available for exclusive use. If
  not, then creates a copy of skb and returns the
  pointer.
• skb_put(skb,len) appends data to the end of the
  packet increments the pointer tail and skb?len
  by len need to ensure the tailroom is
  sufficient.
• skb_push(skb,len) inserts data in front of the
  packet data space decrements the pointer data by
  len, and increment skb?len by len need to check
  the headroom size.

18
2. Manipulating Packet Data Space

• skb_pull(skb,len) truncates len bytes at the
  beginning of a packet.
• skb_tailroom(skb) returns the size of the
  tailroom (in bytes).
• skb_headroom(skb) returns the size of the
  headroom (data-head)
• skb_realloc_headroom(skb,newheadroom) creates a
  new socket buffer with a headroom of size
  newheadroom.
• skb_reserve(skb,len) shifts the entire packet
  data backward by len bytes.

19
3. Managing the Socket Buffer Queue

• The socket buffers are arranged in a
  dual-concatenated ring structure.
• struct sk_buff_head
• struct sk_buff next
• struct sk_buff prev
• __u32 qlen
• spinlock_t lock

Used for atomic execution of operations on the
queue under critical conditions.
20
Packet Queue in the Linux Kernel
sk_buff_head
next
prev
qlen 3
sk_buff
sk_buff
sk_buff
next
next
next
prev
prev
prev
...
...
...
head
head
head
data
data
data
tail
tail
tail
end
end
end
Paketdaten
Paketdaten
Paketdaten
21
3. Managing the Socket Buffer Queue

• skb_queue_head_init(list) initializes an
  skb_queue_head structure
• skb_queue_empty(list) checks whether the queue
  list is empty list?qlen is returned.
• skb_queue_len(list) returns the actual length of
  the queue.
• skb_queue_head(list,skb) inserts the socket
  buffer skb at the head of the queue and increment
  list?qlen by one.
• skb_queue_tail(list,skb) appends the socket
  buffer skb to the end of the queue and increment
  list?qlen by one.

22
3. Managing the Socket Buffer Queue

• skb_dequeue(list) removes the top packet from
  the queue and returns the pointer to the packet.
• skb_dequene_tail(list) removes the last packet
  from the queue and returns the pointer to the
  packet.
• skb_queue_purge() empties the queue list all
  packets are removed via kfree_skb().
• skb_insert(oldskb,newskb) inserts newskb in
  front of oldskb in the queue of oldskb.
• skb_append(oldskb,newskb) inserts newskb behind
  oldskb in the queue of oldskb.

23
3. Managing the Socket Buffer Queue

• skb_unlink(skb) removes the socket buffer skb
  from its queue and decrement the queue length.
• skb_peek(list) returns a pointer to the first
  element of a list, if this list is not empty
  otherwise, returns NULL.
• skb_peek_tail(list) returns a pointer to the
  last element of a queue if the list is empty,
  returns NULL.