USB System Architecture

The Split Token Packet

Figure 17-6 on page 297 illustrates the format of the SPLIT Token packet and defines the field names. This packet is sent at the beginning of all split transactions and is also used when completing a split transaction. The split packet contains the following fields:

•SPLIT PID — this 8-bit field is the packet identifier that contains the 4-bit code defining this packet as a SPLIT token packet.

•Hub Address — this 7-bit field specifies the device address of the hub that must decode and process the split transaction.

•SC (Start or Complete) — this bit field defines whether this packet is a Start SPLIT packet or a Complete SPLIT packet.

•Port — this 7-bit field identifies the port number that the split transaction is targeting. System software must be aware of the topology to identify the hub and port number to which the transaction must be delivered in order to reach the target device.

•S (Speed) — this bit field defines the speed of the transaction (either low speed or full speed) when the target endpoint is control, interrupt, or bulk. During Isochronous OUT start split transactions (i.e., endpoint type = isochronous, direction = OUT, and the SC bit = Start SPLIT) the Speed bit (S) and the End bit (E) define one of four types of Start SPLIT packet as listed in Figure 17-6 on page 297. For further details see “Isochronous OUT Split Transaction Sequence” on page 313.

•E (End) — this bit field, in conjunction with the S (speed) bit, specifies the type of Start SPLIT packet being issued by the host when an isochronous OUT transaction is to be performed to the target device. See section xx for details.

•ET (Endpoint Type) — this 2-bit field specifies the endpoint type: control, isochronous, bulk, or interrupt.

•CRC5 — this is the 5-bit CRC that covers the 19 bits of the packet that follow the split PID.

296