Accessing Video Memory

The linear framebuffer model stores rows sequentially in memory. The top-left pixel is near the beginning of the mapped region. Later rows come after earlier rows.

uint8_t *pixel = framebuffer + y * pitch + x * bytes_per_pixel;

Before writing through that pointer, confirm the framebuffer is mapped, confirm x and y are inside the visible area, and confirm color packing matches the current mode. The formula is short, but it sits on top of several assumptions. If any of those assumptions are false, the write can become memory corruption instead of drawing.

Color Packing

A pixel is not always a uint32_t in the layout you would like. Some modes use 15, 16, 24, or 32 bits per pixel. Some store color channels in different positions. In Machine Lab the supported modes are intentionally limited, but the habit should be the same as in a real driver: read the mode information and pack colors according to that format.

Clipping

Clipping means trimming a draw operation to the visible screen. It is safer than trying to write and hoping coordinates are valid. A rectangle that begins partly off-screen can still have a visible portion, and a sprite with negative coordinates may still need to draw the pixels that overlap the screen.

Every final project needs clipping, even if the first rectangle test does not look like it. Menus slide, sprites move, cursors reach edges, and camera systems show only part of a larger world.