summaryrefslogtreecommitdiff
path: root/Documentation/networking/driver.txt
diff options
context:
space:
mode:
authorLinus Torvalds <torvalds@ppc970.osdl.org>2005-04-17 02:20:36 +0400
committerLinus Torvalds <torvalds@ppc970.osdl.org>2005-04-17 02:20:36 +0400
commit1da177e4c3f41524e886b7f1b8a0c1fc7321cac2 (patch)
tree0bba044c4ce775e45a88a51686b5d9f90697ea9d /Documentation/networking/driver.txt
downloadlinux-1da177e4c3f41524e886b7f1b8a0c1fc7321cac2.tar.xz
Linux-2.6.12-rc2v2.6.12-rc2
Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
Diffstat (limited to 'Documentation/networking/driver.txt')
-rw-r--r--Documentation/networking/driver.txt94
1 files changed, 94 insertions, 0 deletions
diff --git a/Documentation/networking/driver.txt b/Documentation/networking/driver.txt
new file mode 100644
index 000000000000..11fd0ef5ff57
--- /dev/null
+++ b/Documentation/networking/driver.txt
@@ -0,0 +1,94 @@
+Documents about softnet driver issues in general can be found
+at:
+
+ http://www.firstfloor.org/~andi/softnet/
+
+Transmit path guidelines:
+
+1) The hard_start_xmit method must never return '1' under any
+ normal circumstances. It is considered a hard error unless
+ there is no way your device can tell ahead of time when it's
+ transmit function will become busy.
+
+ Instead it must maintain the queue properly. For example,
+ for a driver implementing scatter-gather this means:
+
+ static int drv_hard_start_xmit(struct sk_buff *skb,
+ struct net_device *dev)
+ {
+ struct drv *dp = dev->priv;
+
+ lock_tx(dp);
+ ...
+ /* This is a hard error log it. */
+ if (TX_BUFFS_AVAIL(dp) <= (skb_shinfo(skb)->nr_frags + 1)) {
+ netif_stop_queue(dev);
+ unlock_tx(dp);
+ printk(KERN_ERR PFX "%s: BUG! Tx Ring full when queue awake!\n",
+ dev->name);
+ return 1;
+ }
+
+ ... queue packet to card ...
+ ... update tx consumer index ...
+
+ if (TX_BUFFS_AVAIL(dp) <= (MAX_SKB_FRAGS + 1))
+ netif_stop_queue(dev);
+
+ ...
+ unlock_tx(dp);
+ ...
+ }
+
+ And then at the end of your TX reclaimation event handling:
+
+ if (netif_queue_stopped(dp->dev) &&
+ TX_BUFFS_AVAIL(dp) > (MAX_SKB_FRAGS + 1))
+ netif_wake_queue(dp->dev);
+
+ For a non-scatter-gather supporting card, the three tests simply become:
+
+ /* This is a hard error log it. */
+ if (TX_BUFFS_AVAIL(dp) <= 0)
+
+ and:
+
+ if (TX_BUFFS_AVAIL(dp) == 0)
+
+ and:
+
+ if (netif_queue_stopped(dp->dev) &&
+ TX_BUFFS_AVAIL(dp) > 0)
+ netif_wake_queue(dp->dev);
+
+2) Do not forget to update netdev->trans_start to jiffies after
+ each new tx packet is given to the hardware.
+
+3) Do not forget that once you return 0 from your hard_start_xmit
+ method, it is your driver's responsibility to free up the SKB
+ and in some finite amount of time.
+
+ For example, this means that it is not allowed for your TX
+ mitigation scheme to let TX packets "hang out" in the TX
+ ring unreclaimed forever if no new TX packets are sent.
+ This error can deadlock sockets waiting for send buffer room
+ to be freed up.
+
+ If you return 1 from the hard_start_xmit method, you must not keep
+ any reference to that SKB and you must not attempt to free it up.
+
+Probing guidelines:
+
+1) Any hardware layer address you obtain for your device should
+ be verified. For example, for ethernet check it with
+ linux/etherdevice.h:is_valid_ether_addr()
+
+Close/stop guidelines:
+
+1) After the dev->stop routine has been called, the hardware must
+ not receive or transmit any data. All in flight packets must
+ be aborted. If necessary, poll or wait for completion of
+ any reset commands.
+
+2) The dev->stop routine will be called by unregister_netdevice
+ if device is still UP.