2 i/o space addressing, 3 configuration space addressing and transactions – Freescale Semiconductor MPC5200B User Manual

Page 346

Advertising
background image

Functional Description

MPC5200B Users Guide, Rev. 1

Freescale Semiconductor

10-49

For linear incrementing mode, the memory address is encoded/decoded using AD[31:2]. Thereafter, the address is incremented by 4 bytes
after each data phase completes until the transaction is terminated or completed (a 4 byte data width per data phase is implied). Note, the two
low-order bits of the address are still included in all the parity calculations.

MPC5200B supports both linear incrementing and cache wrap mode as an initiator. For memory transactions, when an XL Bus burst
transaction is wrapped, the cache wrap mode is automatically generated. For zero-word-aligned bursts and single-beat transactions,
MPC5200B drives AD[1:0] to 0b00. As a target, the MPC5200B treats cache wrap mode as a reserved memory mode. MPC5200B will return
the first beat of data and then signal a disconnect without data on the second data phase.

10.4.1.5.2

I/O space addressing

For PCI I/O accesses, all 32 address signals are used to provide an address with granularity of a single byte. Once a target has claimed an I/O
access, it must determine if it can complete the entire access as indicated by the byte enable signals. If all the selected bytes are not in the
address range of the target, the entire access cannot complete. In this case, the target does not transfer any data, and terminates the transaction
with a target-abort.

10.4.1.5.3

Configuration space addressing and transactions

PCI supports two types of configuration accesses. Their primary difference is the format of the address on the AD[31:0] signals during the
address phase. The two low-order bits of the address indicate the format used for the configuration address phase: type 0 (AD[1:0] = 0b00)
or type 1 (AD[1:0] = 0b01). Both address formats identify a specific device and a specific configuration register for that device.

Type 0 configuration accesses are used to select a device on the local PCI bus. They do not propagate beyond the local PCI bus and are either
claimed by a local device or terminated with a master-abort. Type 1 configuration accesses are used to target a device on a subordinate bus
through a PCI-to-PCI bridge. Type 1 accesses are ignored by all targets except PCI-to-PCI bridges that pass the configuration request to
another PCI bus.

When the controller initiates a configuration access on the PCI bus, it places the configuration address information on the AD bus and the
configuration command on the C/BE[3:0] bus. A Type 0 configuration transaction is indicated by setting AD[1:0] to 0b00 during the address
phase. The bit pattern tells the community of devices on the PCI bus that the bridge that “owns” that bus has already performed the bus number
comparison and verified that the request targets a device on its bus.

Figure 10-4

shows the contents of the AD bus during the address phase

of the Type 0 configuration access.

Table 10-6. PCI I/O space byte decoding

Access Size

AD[1:0]

C/BE[3:0]

Data

8-bit

00

xxx0

AD[7:0]

01

xx01

AD[15:8]

10

x011

AD[23:16]

11

0111

AD[31:24]

16-bit

00

xxx0

AD[15:0]

01

xx01

AD[23:8]

10

x011

AD[31:16]

24-bit

00

xxx0

AD[23:0]

01

xx01

AD[31:8]

32-bit

00

xxx0

AD[31:0]

Advertising