TEXT   46

netdevices.txt

Guest on 5th August 2021 01:49:17 AM

  1.  
  2. Network Devices, the Kernel, and You!
  3.  
  4.  
  5. Introduction
  6. ============
  7. The following is a random collection of documentation regarding
  8. network devices.
  9.  
  10. struct net_device allocation rules
  11. ==================================
  12. Network device structures need to persist even after module is unloaded and
  13. must be allocated with kmalloc.  If device has registered successfully,
  14. it will be freed on last use by free_netdev.  This is required to handle the
  15. pathologic case cleanly (example: rmmod mydriver </sys/class/net/myeth/mtu )
  16.  
  17. There are routines in net_init.c to handle the common cases of
  18. alloc_etherdev, alloc_netdev.  These reserve extra space for driver
  19. private data which gets freed when the network device is freed. If
  20. separately allocated data is attached to the network device
  21. (netdev_priv(dev)) then it is up to the module exit handler to free that.
  22.  
  23. MTU
  24. ===
  25. Each network device has a Maximum Transfer Unit. The MTU does not
  26. include any link layer protocol overhead. Upper layer protocols must
  27. not pass a socket buffer (skb) to a device to transmit with more data
  28. than the mtu. The MTU does not include link layer header overhead, so
  29. for example on Ethernet if the standard MTU is 1500 bytes used, the
  30. actual skb will contain up to 1514 bytes because of the Ethernet
  31. header. Devices should allow for the 4 byte VLAN header as well.
  32.  
  33. Segmentation Offload (GSO, TSO) is an exception to this rule.  The
  34. upper layer protocol may pass a large socket buffer to the device
  35. transmit routine, and the device will break that up into separate
  36. packets based on the current MTU.
  37.  
  38. MTU is symmetrical and applies both to receive and transmit. A device
  39. must be able to receive at least the maximum size packet allowed by
  40. the MTU. A network device may use the MTU as mechanism to size receive
  41. buffers, but the device should allow packets with VLAN header. With
  42. standard Ethernet mtu of 1500 bytes, the device should allow up to
  43. 1518 byte packets (1500 + 14 header + 4 tag).  The device may either:
  44. drop, truncate, or pass up oversize packets, but dropping oversize
  45. packets is preferred.
  46.  
  47.  
  48. struct net_device synchronization rules
  49. =======================================
  50. dev->open:
  51.         Synchronization: rtnl_lock() semaphore.
  52.         Context: process
  53.  
  54. dev->stop:
  55.         Synchronization: rtnl_lock() semaphore.
  56.         Context: process
  57.         Note1: netif_running() is guaranteed false
  58.         Note2: dev->poll() is guaranteed to be stopped
  59.  
  60. dev->do_ioctl:
  61.         Synchronization: rtnl_lock() semaphore.
  62.         Context: process
  63.  
  64. dev->get_stats:
  65.         Synchronization: dev_base_lock rwlock.
  66.         Context: nominally process, but don't sleep inside an rwlock
  67.  
  68. dev->hard_start_xmit:
  69.         Synchronization: netif_tx_lock spinlock.
  70.  
  71.         When the driver sets NETIF_F_LLTX in dev->features this will be
  72.         called without holding netif_tx_lock. In this case the driver
  73.         has to lock by itself when needed. It is recommended to use a try lock
  74.         for this and return NETDEV_TX_LOCKED when the spin lock fails.
  75.         The locking there should also properly protect against
  76.         set_multicast_list. Note that the use of NETIF_F_LLTX is deprecated.
  77.         Don't use it for new drivers.
  78.  
  79.         Context: Process with BHs disabled or BH (timer),
  80.                  will be called with interrupts disabled by netconsole.
  81.  
  82.         Return codes:
  83.         o NETDEV_TX_OK everything ok.
  84.         o NETDEV_TX_BUSY Cannot transmit packet, try later
  85.           Usually a bug, means queue start/stop flow control is broken in
  86.           the driver. Note: the driver must NOT put the skb in its DMA ring.
  87.         o NETDEV_TX_LOCKED Locking failed, please retry quickly.
  88.           Only valid when NETIF_F_LLTX is set.
  89.  
  90. dev->tx_timeout:
  91.         Synchronization: netif_tx_lock spinlock.
  92.         Context: BHs disabled
  93.         Notes: netif_queue_stopped() is guaranteed true
  94.  
  95. dev->set_multicast_list:
  96.         Synchronization: netif_tx_lock spinlock.
  97.         Context: BHs disabled
  98.  
  99. struct napi_struct synchronization rules
  100. ========================================
  101. napi->poll:
  102.         Synchronization: NAPI_STATE_SCHED bit in napi->state.  Device
  103.                 driver's dev->close method will invoke napi_disable() on
  104.                 all NAPI instances which will do a sleeping poll on the
  105.                 NAPI_STATE_SCHED napi->state bit, waiting for all pending
  106.                 NAPI activity to cease.
  107.         Context: softirq
  108.                  will be called with interrupts disabled by netconsole.

Raw Paste


Login or Register to edit or fork this paste. It's free.