Writing an OS Kernel from Scratch | Keyboard — How Keystrokes Become Characters

Have you ever wondered about this process: you press a key → a character appears on screen. What does the OS experience in between? How do keyboard scan codes become scancodes? What’s the difference between USB HID and PS/2?

From the mechanical keyboard matrix to the Linux input subsystem, this article breaks down the complete “key to character” chain, giving you a debuggable map.

1. Keyboard Physical Layer: Matrix Scanning

A typical keyboard has an N×M row/column matrix, with each key being a switch at a matrix intersection.

        Column 0   Column 1   Column 2   Column 3
           │         │         │         │
   Row 0 ──○─────────○─────────○─────────○  ← ESC, 1, 2, 3...
   Row 1 ──○─────────○─────────○─────────○  ← Tab, Q, W, E, R...
   Row 2 ──○─────────○─────────○─────────○  ← CapsLock, A, S, D, F...
   ...

Scanning process:

1. Keyboard controller drives Row 0 low, other Rows high
2. Read Column levels to determine which Column has low level (pressed)
3. Intersection determined → key coordinate (Row, Column) confirmed
4. Look up table to convert (Row, Column) to Scan Code

PS/2 Scan Code Set 2 example:

• Press A: keyboard sends 0x1C
• Press Shift+A: first sends 0x12 (Left Shift Make), then 0x1C
• Release A: sends 0xF0 + 0x1C (Break Code)

# Linux: view current keyboard scan codes
$ showkey -s
# Press any key to see its scancode (decimal)
# Press ESC or wait 10 seconds to exit

# View keycode mapping
$ dumpkeys | head -20
keycode  1 = Escape
keycode  2 = one
keycode  3 = two

2. PS/2 Protocol: Legacy Keyboard Communication

PS/2 keyboards use a simple synchronous serial protocol: 11 bits per frame (1 Start + 8 Data + 1 Parity + 1 Stop), clock provided by the keyboard (~10kHz).