How Linux Boots: BIOS/UEFI to Login

When you press the power button on a Linux machine, five distinct things happen before you see a login prompt. Most people never think about this until something breaks. Understanding the chain is the difference between “I’ll just reinstall” and “I’ll boot into recovery and fix it.”

Stage 1: firmware (BIOS or UEFI)

The motherboard’s firmware (BIOS on old systems, UEFI on modern ones) runs first. It powers up hardware, runs the POST (Power-On Self-Test), and then looks for something to boot from — your hard drive, USB stick, or network.

• BIOS reads the first 512 bytes of the boot disk (the MBR) and runs that.
• UEFI reads an EFI System Partition (ESP) on the disk and runs an .efi file from it.

Stage 2: bootloader (GRUB)

The firmware hands off to a bootloader — almost always GRUB 2 on Linux. GRUB shows you a menu of operating systems and kernels. Pick one, and GRUB loads:

• The Linux kernel (a file like /boot/vmlinuz-6.6.10)
• The initramfs (a small temporary root filesystem like /boot/initrd.img-6.6.10)

Then it transfers control to the kernel.

Stage 3: the kernel boots

The kernel decompresses itself into RAM, initializes hardware drivers, mounts the initramfs as a temporary root filesystem, and looks for the real root filesystem (your /) on disk. Once it finds and mounts the real root, it executes the first userspace program: init.

Watch this happen on your next boot:

# Show kernel boot messages
dmesg | less

# Or with timestamps
dmesg -T

Stage 4: init (systemd)

The first process is PID 1. On almost every modern Linux system, that’s systemd. systemd reads its unit files (in /etc/systemd/system/ and /usr/lib/systemd/system/), figures out the target you want to boot into (usually multi-user.target or graphical.target), and starts services in dependency order.

This is when networking starts, when SSH starts, when your display manager starts.

# See what systemd is doing during boot
systemd-analyze
systemd-analyze blame    # which services took longest
systemd-analyze critical-chain

Stage 5: login

Either a text login prompt (getty) on a TTY or a graphical login screen (gdm, sddm, lightdm). After you log in, your shell starts (bash, zsh) or your desktop environment starts.

The whole chain in one picture

Power on
   ↓
Firmware (BIOS / UEFI)
   ↓
Bootloader (GRUB 2)
   ↓
Linux kernel + initramfs
   ↓
init (systemd, PID 1)
   ↓
Services (network, SSH, display manager)
   ↓
Login prompt

Where boot problems come from

• Black screen at power-on: hardware or firmware. Not Linux’s fault yet.
• “No boot device”: firmware can’t find the bootloader. Fix in BIOS settings or repair GRUB.
• GRUB rescue prompt: bootloader is broken. Boot a live USB and run grub-install.
• Kernel panic: kernel started but failed to mount root. Often missing kernel modules in initramfs.
• Hangs at “started X” service: a systemd unit is stuck. Boot with systemd.unit=rescue.target kernel parameter.

What to learn next

The filesystem layout — knowing where everything lives — is the natural next step. systemd itself gets its own deeper dive in the service management section of this roadmap.