Device Files, /proc & /sys

Devices, /proc and /sys

One of Linux's defining ideas is "everything is a file" - and hardware and kernel internals are no exception. Instead of needing a special API to ask the kernel "how much memory is free?", you just read a file. This means the ordinary tools you already know (cat, grep, redirection) work on system information too.

/dev - device files. sda/nvme0n1 (disks), tty (terminals), null, zero, random. These aren't "files about" devices - reading or writing them talks directly to the hardware. Managed dynamically by udev, a background service that creates and removes these entries as devices are plugged in and unplugged.
/proc - a virtual filesystem: it looks like ordinary files and directories, but nothing here actually exists on disk. The kernel generates its contents on the fly, live, the instant you read them. /proc/cpuinfo, /proc/meminfo, and a numbered directory per running process (/proc/<pid>/) all work this way.
/sys - sysfs: a more structured, more recent virtual filesystem exposing devices, drivers, and kernel tunables in an organized tree (versus /proc's older, looser layout).

Analogy: /proc and /sys are like a car's dashboard. The engine isn't inside the dashboard - but the dashboard gives you a live, always-current window into what the engine is doing right now. Read /proc/meminfo and you get this instant's memory numbers, not a cached snapshot from boot.

Hardware discovery commands - these ask the kernel to summarize what it has found, rather than making you read /proc//sys by hand:

Command	Shows
`lspci`	PCI devices (GPU, NIC)
`lsusb`	USB devices
`lsblk`	Block devices (disks/partitions)
`lscpu`	CPU architecture
`dmidecode`	Firmware/DMI hardware info
`dmesg`	Kernel ring buffer (boot & hardware messages)

$ lscpu | grep '^CPU(s):'
CPU(s):                  4

That one line answers "how many logical CPUs does this machine have?" - useful context later when you're reading load averages or deciding how many parallel jobs to run.

Try cat /proc/cpuinfo, lsblk, and dmesg | tail in the terminal below - notice that cating /proc/cpuinfo works exactly like reading a normal text file, even though there's no such file on any physical disk.