On Tuesday 10 February 2004 01:14, threadhead@web.de wrote: > > dev_base is a global variable of type struct net_device * > > and should be accessed using a read/write lock, dev_base_lock. > > (If in uniprocessor machine lock doesn't matter). > > > > Parsing through the list u can find the appropriate network > > device ( one way is identifying by the name such eth0,.. > > in the name field). > > > > And then can make use of the function pointer > > hard_start_xmit in the structure net_device. > > ok when i have obtained the hard_start_xmit function that way, > i can call it just like that with the two required arguments? > > i tried that once but as i said earlier that resulted in a kernel oops. > (i did the function call only, not anything locking related. is it actually > necessarry to lock the kernel for the moment the function is called?) > > another question would be, how can i build my struct sk_buff and struct > net_device the correct way? for example if i want to send a packet with the > ack flag set from that level. passing empty structs doesnt make sense, > obviously. ;) "struct netdevice" is created when the driver is registered to the kernel. There are some Kernel API's for that such as register_netdev,.. You can make use any of the network device driver code to understand Eg:- 8139too.c Or Linux Device Drivers, Alessandro Rubini ( just google for free e-version) Filling a sk_buff is much related to the protocols used. I think we need to make use of the protocol specific kernel API's. To know how to create a sk_buff, just trace out the packet reception part in any of the network driver. Driver allocates a sk_buff using the kernel API dev_alloc_skb and fill appropriate fields. It places the received data in one of the field ( "data" I suppose ) of the sk_buff, ... Hope this may help you.... Regards, Shine Mohamed -- Kernelnewbies: Help each other learn about the Linux kernel. Archive: http://mail.nl.linux.org/kernelnewbies/ FAQ: http://kernelnewbies.org/faq/
