USB-A ports

How it tests

Watches USB devices appear port by port; with a port map it rings one drawn port at a time. Each device gets a check: a disk is read raw and read-only (O_DIRECT, never mounted, never written), 32 MiB twice and compared, or for the OmaBoot? stick checked against the image's chunk checksums; a mouse or keyboard has to send a move or a key within 20 s; anything else counts by enumerating. The stick's port and a tethered phone's count since power-on. Booted with copytoram, the stick itself moves from port to port and gets the verified read on each.

How it decides

Supported: at least one port worked, none marked dead, and no port's read failed. Partial: ports marked dead, or a read that errored, came up short, made no progress for 5 s, took over 150 s or didn't match, with nothing reading fine on that port after. Failed: nothing seen and a port marked dead.

How sure it is

Enumeration on a ringed port proves the data lines; a verified read proves data moves.

What could mislead it

• Could pass when it shouldn'tA device that can't be read (a phone charging, a hub) counts its port as working on enumeration alone; the evidence says which ports moved data.
• Could pass when it shouldn'tA USB 3 device at USB 2 speed on a USB 3 port (or reading under 40 MB/s at 5 Gb/s) is a finding in the evidence, not a verdict.
• Could fail when it shouldn'tPre-2012 Macs hand USB 1.1 devices to a companion controller, so one hole can look like two ports; with a map, the mouse is taken for the next port asked for.