dokucama

User Tools

Site Tools

Trace:
network_stuff:linux_network_internals

Differences

This shows you the differences between two versions of the page.

Link to this comparison view

Both sides previous revisionPrevious revision
Next revision		Previous revision
network_stuff:linux_network_internals [2019/02/12 13:31] – external edit 127.0.0.1network_stuff:linux_network_internals [2023/11/02 14:38] (current) – external edit 127.0.0.1
Line 1: Line 1:
-Always use the -d flag when dealing with the netlink api: +LINUX KERNEL NETWORKING: 
-  ip -d link show wlp2s0 +\\
-  ip add show | egrep "^[0-9*]|inet "  # this is unrelated but useful for a more compact view.+
  
-  * Interaction user-space >> kernel +Linux has Two main APIs for networking: 
-    * Existing system call +  * Ethtool is for physical link management (speed, duplex, buffers etc.) 
-    * ioctl +  * Netlink is for the rest. Every network device is represented as a netdev object in the kernel. 
-    * /proc file +    * **[[https://medium.com/@mdlayher/linux-netlink-and-go-part-1-netlink-4781aaeeaca8|NETLINK]]** (successor of ioctl): api (no restapi), system call. Is the way ip route talks to the kernel. socket family is a Linux kernel interface used for inter-process communication (IPC) between both the kernel and userspace processes, and between different userspace processes, in a way similar to the Unix domain sockets. NETLINK_ROUTE , NETLINK_ARPD , NETLINK_NETFILTER .. 
-    * NETLINK socket? +    * Netlink in python: [[https://pypi.org/project/pyroute2/|External Link]] 
-**[[https://medium.com/@mdlayher/linux-netlink-and-go-part-1-netlink-4781aaeeaca8|NETLINK]]** (successor of ioctl): api (no restapi), system call. Is the way ip route talks to the kernel. socket family is a Linux kernel interface used for inter-process communication (IPC) between both the kernel and userspace processes, and between different userspace processes, in a way similar to the Unix domain sockets. NETLINK_ROUTE , NETLINK_ARPD , NETLINK_NETFILTER ... \\ + 
-Do not confuse it with NETFILTER (~iptables). \\ +{{ :network_stuff:network_os_models.pdf |}} 
-Netlink in python: [[https://pypi.org/project/pyroute2/|External Link]]+ 
 +  * Do not confuse it with NETFILTER (~iptables). 
 +  * iproute2 is more a set of **utililities** (arpd, bridge, devlink, ip, nstat, routef, rtacct, rtmon,  ss, tc)  
 + 
 + 
 +----
  
 Place to find all information about interface is this, they are netdev, not just dev:  Place to find all information about interface is this, they are netdev, not just dev: 
Line 19: Line 23:
 TODO\\ TODO\\
 ... kernel objects and udev, and details of implementing logical interfaces that are associated with ASIC front-panel physical ports. From this discussion: [[http://blog.ipspace.net/2018/03/linux-interfaces-on-software-gone-wild.html]]\\ ... kernel objects and udev, and details of implementing logical interfaces that are associated with ASIC front-panel physical ports. From this discussion: [[http://blog.ipspace.net/2018/03/linux-interfaces-on-software-gone-wild.html]]\\
 +
 +\\
 +
 Interface can or cannot be related to physical port.\\ Interface can or cannot be related to physical port.\\
 Linux labels them according to the PCI name, UUID can be used to rename them. udev, driver..\\ Linux labels them according to the PCI name, UUID can be used to rename them. udev, driver..\\
   CPU-PCI BUS (unique) driver enumerates > port1,2,3 ...   CPU-PCI BUS (unique) driver enumerates > port1,2,3 ...
- 
   Then udev/uuid is used to map the name/rename the port name on boot time   Then udev/uuid is used to map the name/rename the port name on boot time
-\\ 
  
 ---- ----
  
-  * BPF is the in-kernel bytecode machine that can be used for tracing, virtual networks, and more. Alexei Starovoitov is the lead developer (he's now at Facebook), and there are developers from several companies contributing, including myself at Netflix, Daniel Borkmann at Cisco, and Brenden Blanco at PLUMgrid.+Always use the -d flag when dealing with the netlink api: 
 +  ip -d link show wlp2s0 
 +  ip add show | egrep "^[0-9*]|inet "  # this is unrelated but useful for a more compact view. 
 + 
 +  * Interaction user-space >> kernel 
 +    * Existing system call 
 +    * ioctl 
 +    * /proc file 
 +    * NETLINK socket? 
 + 
 +---- 
 + 
 +  * **eBPF [[https://lwn.net/Articles/740157/]]** 
 +    * BPF is the in-kernel bytecode machine that can be used for tracing, virtual networks, and more. Alexei Starovoitov is the lead developer (he's now at Facebook), and there are developers from several companies contributing, including myself at Netflix, Daniel Borkmann at Cisco, and Brenden Blanco at PLUMgrid.
   * P4 is very similar but needs compiling   * P4 is very similar but needs compiling
  
 +----
 +
 +OVS (Open VSwitch):\\
 +[[https://therandomsecurityguy.com/openvswitch-cheat-sheet/]] \\
 +
 +  ovs-vswitchd --help
 +  
 +
 +----
 +Kernel structs 
 +\\
 +  * ''net_device'' structure, which represents a network device,
 +  * ''sk_buff'' structure represents an incoming or outgoing packet, including its headers
network_stuff/linux_network_internals.1549978308.txt.gz · Last modified: (external edit)

Page Tools

Except where otherwise noted, content on this wiki is licensed under the following license: CC Attribution-Share Alike 4.0 International
CC Attribution-Share Alike 4.0 International Donate Powered by PHP Valid HTML5 Valid CSS Driven by DokuWiki