Physical:

• wl*: wireless (Wi-Fi, Bluetooth).
• e*: Ethernet.

systemd.net-naming-scheme
systemd's "Predictable Network Interface Names"
Chris Siebenmann's "Understanding something about udev's normal network device names"


Virtual:

• lo*: loopback.
• tun*: VPN tunnel.
• v*: virtual machine adapter.
• wg*: WireGuard VPN.

• TCP(?)-level: e.g. OpenVPN. Default device is tun*.
  
  Encrypted tunnel to a VPN server, which does user authentication and then implements a proxy (changing IP address) and goes to internet or LAN.
  
  
• IP-level: e.g. WireGuard. Default device is wg*.
  
  Key set-up is done in advance, when the connection is first defined. So there is no user authentication to do for each connection; just use the keys. Encrypted tunnel to a VPN server, which implements a proxy (changing IP address) and goes to internet or LAN.
  
  

[Where does IPSec/Strongswan fit in this ? IKEv2 ?]

• Dynamic code snippets inside kernel: eBPF.
  
  
• Kernel modules: e.g. iptables, nftables.
  
  Each kernel module usually has a user-space utility to configure it. Kernel module ip_tables is controlled by utility iptables, kernel module nftables is controlled by utility nft ("sudo nft list ruleset | less").
  
  

Also there are front-end apps: On top of iptables, there are ufw and gufw or firewalld (use one). On top of nftables, there is firewalld. There may be a GUI app on top of that, such as Fedora's firewall-config.

Application sandboxes (Firejail, AppArmor, SELinux) can disable networking or maybe filter it ?

"systemctl status NetworkManager.service"
"man nmcli"
"man nmtui"
"nmcli connection show"
"nmcli device"
Schkn's "Network Manager on Linux with Examples"
ArchWiki's "NetworkManager"
GNOME Developer's "NetworkManager"

"systemctl status systemd-networkd.service"
"networkctl list"
"ls /etc/systemd/network"
Derrik Diener's "How to set up systemd-networkd on Linux"
Sahitya Maruvada's "Working with systemd-networkd"

"systemctl status networking.service"
"ls /etc/network/interfaces" ?
"ls /etc/sysconfig/network-scripts" ?

"netplan is a Ubuntu tool to create a configuration file for networking. netplan can create either a NetworkManager configuration or a systemd-networkd configuration file at system boot time."

Egidio Docile's "Netplan network configuration tutorial for beginners"
Linux Shell Tips' "How to Set Static IP Address on Ubuntu Linux"
Canonical's "Netplan reference"
From someone on reddit 2/2021:

• ARP: Handles MAC addresses, knows which port of switch a given device is on.
  Linux Journey's "Link Layer"
  
  To see client's ARP cache, do "ip neigh" or "arp -n" or "arp".
  To see MAC addresses of client's network interfaces, do "ip link show".
  
  
• DHCP: Assigns IP addresses.
  Linux Journey's "DHCP Overview"
  
  To see where DHCP server is, on client do "ip route | grep dhcp".
  
  
• Routing: Routes IP traffic from the LAN switch out to WAN (Internet). AKA "gateway".
  
  
• DNS: Maps domain names to IP addresses.
  
  Normally the router does not do DNS, but your LAN may have a separate DNS server, if you're doing self-hosting or ad-blocking.
  
  On client: bind, dnsmasq, systemd-resolved, or openresolv. Also "cat /etc/hosts" and "cat /etc/resolv.conf". If systemd-resolved, see "ls -l /etc/resolv.conf" to tell which of 4 modes it is running in ? Also do "resolvectl status".
  
  [I'd like a command or logging that shows ALL the steps as an address gets resolved, including local config-file and cache lookups, but I can't find one.]
  
  See "DNS" section of "Connection Security and Privacy" page.
  
  

• Network interfaces: lo, eno1, wlp1s0, tun0.
• udev.
• Systemd.
• Network Manager.
• systemd-resolved (/etc/resolv.conf is a symlink to /run/systemd/resolve/stub-resolv.conf).
• Windscribe VPN connected via OpenVPN.
• Firewall stack: nftables / firewalld / firewall-config.
  (iptables is loaded but the tables always just say "accept".)

To do a network access, an application written in C would:

1. Construct an addrinfo struct specifying port number or service name, and host name or IP address. Could also specify IPv4 versus IPv6, stream versus datagram, more.
   
   
2. Call getaddrinfo() library-function to fill in address.
   
   I think this involves resolving via tables, services, and DNS. From source, on one path, getaddrinfo() is calling gethostbyname() [source] which maybe is calling getservent() to find a DNS service ? There's an indirection in the middle of the NSS source, making everything hard to follow.
   
   
3. Call socket() system-call to get a file descriptor for an open socket.
   
   I think this involves using the routing table and picking a network interface, then choosing a client port on that interface. There may be a stack of interfaces (such as tun0 on top of ether0) ?
   
   
4. Call connect() system-call, giving the socket and address. If success, you have a live connection to the destination.
   
   
5. Call send() system-call, giving socket and a message to send (such as "GET / HTTP/1.0").
   
   

Beej's Guide to Network Programming

Mohd Sohail's "Share Folders On Local Network Between Ubuntu And Windows"
Jonathan Moeller's "Install & Configure Samba On Linux Mint 19"
Ubuntu Tutorials' "Install and Configure Samba"
Stephan Avenwedde's "Share files between Linux and Windows computers"

Share folder from Linux:
In Mint Cinnamon: Nemo-share extension to Nemo file explorer.
In KDE: in Dolphin file explorer, select a folder, right-click, Properties, Share.
Apparently case of letters does not matter.
Leave read-only for everyone. No need to check "Allow Guests".

Set up networking on Linux side:
Make sure gufw/firewall is set to allow incoming connections.
Apparently VPN can be left on.
Do "ip addr | grep 192" to get LAN IP address for Linux system.
Install Samba and run Samba (server ?).

Connect from Windows:
Right-click on "This PC" or "My Computer", select "Add a new connection", type location "\\IPADDRESS\\FOLDERNAME", click Next.

Create a file-share on Windows:
In Windows file explorer, select a folder, right-click, Properties, click Sharing, click Share, "Choose people to share with" == "all", click Share. Click Advanced Sharing, enable "Share this folder", set "Share name" to something simple with NO spaces in it, click on Permissions, make sure "All" have "Read" permission.
Start Menu, run Task Manager, click More Details, click Users, see logged-in username.
Windows-R, run Powershell, run "ipconfig" to get LAN IP address.

In Linux:
gufw can have "Incoming" set to "Deny".
VPN can be on.

In browser on Linux, go to address "smb://IPADDRESS/SHARENAME", login with Windows account username and password. DIDN'T WORK

In KDE/Dolphin, on left side click on Network. Double-click on "Add Network Folder". Select "Microsoft Windows network drive". Name == anything, Server == IPADDRESS, folder name == name shared from Windows. Get login dialog, use username and password from Windows.
Get success.
On left side, click on Network, see share's name as one of items in list.

WinSCP's "Installing SFTP/SSH Server on Windows using OpenSSH"

On Linux:
"sudo apt install openssh-server"
"systemctl status sshd"
Make sure gufw/firewall is set to allow incoming connections.
Could make a specific rule to allow incoming port 22 from Windows machine's IP address.
Apparently VPN can be left on.

On Windows:
Start Menu, search for "Manage optional features".
See "OpenSSH Client".

Windows-R to open PowerShell.

Do "ssh IPADDRESS".
Get messages about trusting the Linux system, say yes.
Get prompted to login to Linux system, give user password.
Get shell prompt.
Ctrl-D to log out.

Do "scp IPADDRESS:.profile profile"
Get prompted to login to Linux system, give user password.
See new file "profile" appear in Windows dir.

WinSCP's "Installing SFTP/SSH Server on Windows using OpenSSH"

On Windows:
Start Menu, search for "Manage optional features".
See "OpenSSH Server" ? I don't.
???

ip addr | grep 192		# get LAN IP address of Linux machine
cd DIRTOSHARE
python -m SimpleHTTPServer
# go to Windows machine, and in browser go to:
http://LANIPADDRESS:8000

Ext2Fsd Project

Windows must be fully shut down, not hibernated, to allow Linux to have read/write access to the Windows partition. If all you want is read-only access in Linux, ignore the rest of this section.

In Windows 10, normally if you select "Start / Shutdown", it hibernates, doesn't fully shut down.

Ways to make Windows fully shut down:

• Turn off "Fast Startup", and now "Start / Shutdown" will do a full shutdown.
• Hold down Shift key while selecting "Start / Shutdown", and it will do a full shutdown.

I think it's best to leave "Fast Startup" turned off. But Windows will start up slower.

Chris Hoffman's "How to Mount Your Windows 10 (or 8) System Drive on Linux"
Unix & Linux Stack Exchange's "How to mount the 'D:\' disk of Windows in linux mint?"
community.linuxmint.com's "gnome-disk-utility"

But: Ubuntu 18 / Mint Tara automatically recognizes Windows OS partition in a dual-boot system and mounts it; no package installation or other steps needed. It was read-only in my live session, maybe because I didn't shut down Windows fully.