Texas Instruments Incorporated and its subsidiaries (TI) reserve the right to make corrections, modifications,
enhancements, improvements, and other changes to its products and services at any time and to discontinue
any product or service without notice. Customers should obtain the latest relevant information before placing
orders and should verify that such information is current and complete. All products are sold subject to TI’s terms
and conditions of sale supplied at the time of order acknowledgment.
TI warrants performance of its hardware products to the specifications applicable at the time of sale in
accordance with TI’s standard warranty . Testing and other quality control techniques are used to the extent TI
deems necessary to support this warranty . Except where mandated by government requirements, testing of all
parameters of each product is not necessarily performed.
TI assumes no liability for applications assistance or customer product design. Customers are responsible for
their products and applications using TI components. T o minimize the risks associated with customer products
and applications, customers should provide adequate design and operating safeguards.
TI does not warrant or represent that any license, either express or implied, is granted under any TI patent right,
copyright, mask work right, or other TI intellectual property right relating to any combination, machine, or process
in which TI products or services are used. Information published by TI regarding third–party products or services
does not constitute a license from TI to use such products or services or a warranty or endorsement thereof.
Use of such information may require a license from a third party under the patents or other intellectual property
of the third party , or a license from TI under the patents or other intellectual property of TI.
Reproduction of information in TI data books or data sheets is permissible only if reproduction is without
alteration and is accompanied by all associated warranties, conditions, limitations, and notices. Reproduction
of this information with alteration is an unfair and deceptive business practice. TI is not responsible or liable for
such altered documentation.
Resale of TI products or services with statements different from or beyond the parameters stated by TI for that
product or service voids all express and any implied warranties for the associated TI product or service and
is an unfair and deceptive business practice. TI is not responsible or liable for any such statements.
Mailing Address:
Texas Instruments
Post Office Box 655303
Dallas, Texas 75265
The Texas Instruments PCI2050 PCI-to-PCI bridge provides a high-performance connection path between two
peripheral component interconnect (PCI) buses. Transactions occur between masters on one PCI bus and targets
on another PCI bus, and the PCI2050 allows bridged transactions to occur concurrently on both buses. The bridge
supports burst-mode transfers to maximize data throughput, and the two bus traffic paths through the bridge act
independently.
The PCI2050 bridge is compliant with the PCI Local Bus Specification, and can be used to overcome the electrical
loading limits of 10 devices per PCI bus and one PCI device per expansion slot by creating hierarchical buses. The
PCI2050 provides two-tier internal arbitration for up to nine secondary bus masters and may be implemented with
an external secondary PCI bus arbiter.
The CompactPCI hot-swap extended PCI capability is provided which makes the PCI2050 an ideal solution for
multifunction CompactPCI cards and for adapting single-function cards to hot-swap compliance.
The PCI2050 bridge is compliant with the PCI-to-PCI Bridge Specification 1.1. The PCI2050 provides compliance
for PCI Power Management 1.0 and 1.1. The PCI2050 has been designed to lead the industry in power conservation.
An advanced CMOS process is used to achieve low system power consumption while operating at PCI clock rates
up to 33 MHz.
The PCI2050I is an industrial version of the PCI2050 that has a larger operating temperature range. All references
to the PCI2050 also apply to the PCI2050I unless otherwise noted.
1.2Features
The PCI2050 supports the following features:
•Architecture configurable for PCI Bus Power Management Interface Specification
•CompactPCI hot-swap-friendly silicon
•3.3-V core logic with universal PCI interfaces compatible with 3.3-V and 5-V PCI signaling environments
•Two 32-bit, 33-MHz PCI buses
•Internal two-tier arbitration for up to nine secondary bus masters and supports an external secondary bus
arbiter
•Burst data transfers with pipeline architecture to maximize data throughput in both directions
•Independent read and write buffers for each direction
•Up to three delayed transactions in both directions
•Ten secondary PCI clock outputs
•Predictable latency per PCI Local Bus Specification
•Bus locking propagation
•Secondary bus is driven low during reset
•VGA/palette memory and I/O decoding options
•Advanced submicron, low-power CMOS technology
•208-terminal QFP or 209-terminal MicroStar BGA package
1–1
1.3Related Documents
•Advanced Configuration and Power Interface (ACPI) Specification (Revision 1.0)
•IEEE Standard Test Access Port and Boundary-Scan Architecture
•PCI Local Bus Specification (Revision 2.2)
•PCI-to-PCI Bridge Specification (Revision 1.1)
•PCI Bus Power Management Interface Specification (Revision 1.1)
CompactPCI is a trademark of PICMG – PCI Industrial Computer Manufacturers Group, Inc.
Intel is a trademark of Intel Corporation.
MicroStar BGA and TI are trademarks of Texas Instruments.
Other trademarks are the property of their respective owners.
1.5Ordering Information
ORDERING NUMBERVOLTAGETEMPERATUREPACKAGE
PCI2050PDV3.3 V , 5-V tolerant I/Os0 to 70°C208-terminal QFP
PCI2050IPDV3.3 V, 5-V tolerant I/Os–40 to 85°C208-terminal QFP
PCI2050GHK3.3 V, 5-V tolerant I/Os0 to 70°C209-terminal MicroStar BGA
PCI2050IGHK3.3 V, 5-V tolerant I/Os–40 to 85°C209-terminal MicroStar BGA
PCI2050ZHK3.3 V, 5-V tolerant I/Os0 to 70°C209-terminal MicroStar BGA
Leadfree
1–2
2 Terminal Descriptions
The PCI2050 device is packaged either in a 209-terminal GHK MicroStar BGA a 209-terminal ZHK MicroStar
BGA, or a 208-terminal PDV package. Figure 2–1 is a GHK-package terminal diagram. Figure 2–2 is a
ZHK-package terminal diagram. Figure 2–3 is a PDV-package terminal diagram. T able 2–1 lists terminals on the PDV
packaged device in increasing numerical order, with the signal name and corresponding GHK terminal number for
each. T able 2–2 lists terminals on the GHK packaged device in increasing alphanumerical order , with the signal name
and corresponding PDV terminal number for each. Table 2–3 lists signal names in alphabetical order, with
corresponding terminal numbers for both package types.
The terminals are grouped in tables by functionality, such as PCI system function and power-supply function (see
Table 2–4 through Table 2–12). The terminal numbers are listed for convenient reference.
Table 2–4. Primary PCI System Terminals
TERMINAL
PDV
NAME
P_CLK45N5I
P_RST
GHK/ZHK
NO.
43P1I
NO.
I/ODESCRIPTION
Primary PCI bus clock. P_CLK provides timing for all transactions on the primary PCI bus. All primary PCI
signals are sampled at rising edge of P_CLK.
PCI reset. When the primary PCI bus reset is asserted, P_RST causes the bridge to put all output buffers
in a high-impedance state and reset all internal registers. When asserted, the device is completely
nonfunctional. During P_RST, the secondary interface is driven low. After P_RST is deasserted, the
bridge is in its default state.
Primary address/data bus. These signals make up the multiplexed PCI address and data bus on the
primary interface. During the address phase of a primary bus PCI cycle, P_AD31–P_AD0 contain a
I/O
32-bit address or other destination information. During the data phase, P_AD31–P_AD0 contain data.
Primary bus commands and byte enables. These signals are multiplexed on the same PCI terminals.
During the address phase of a primary bus PCI cycle, P_C/BE3
During the data phase, this 4-bit bus is used as byte enables. The byte enables determine which byte
I/O
paths of the full 32-bit data bus carry meaningful data. P_C/BE0 applies to byte 0 (P_AD7–P_AD0),
P_C/BE1
P_C/BE3
applies to byte 1 (P_AD15–P_AD8), P_C/BE2 applies to byte 2 (P_AD23–P_AD16), and
applies to byte 3 (P_AD31–P_AD24).
–P_C/BE0 define the bus command.
2–9
Table 2–6. Primary PCI Interface Control Terminals
TERMINAL
PDV
NAME
P_DEVSEL84P11I/O
P_FRAME
P_GNT
P_IDSEL65V7I
P_IRDY82V11I/O
P_LOCK87V12I/O
P_PAR90R12I/O
P_PERR
P_REQ47R1O
P_SERR89P12O
P_STOP85R11I/O
P_TRDY83U11I/O
GHK/ZHK
NO.
80P10I/O
46P3I
88U12I/O
NO.
I/ODESCRIPTION
Primary device select. The bridge asserts P_DEVSEL to claim a PCI cycle as the target device. As
a PCI master on the primary bus, the bridge monitors P_DEVSEL
responds before time-out occurs, then the bridge terminates the cycle with a master abort.
Primary cycle frame. P_FRAME is driven by the master of a primary bus cycle. P_FRAME is asserted
to indicate that a bus transaction is beginning, and data transfers continue while this signal is
asserted. When P_FRAME is deasserted, the primary bus transaction is in the final data phase.
Primary bus grant to bridge. P_GNT is driven by the primary PCI bus arbiter to grant the bridge
access to the primary PCI bus after the current data transaction has completed. P_GNT
not follow a primary bus request, depending on the primary bus arbitration algorithm.
Primary initialization device select. P_IDSEL selects the bridge during configuration space
accesses. P_IDSEL can be connected to one of the upper 24 PCI address lines on the primary PCI
bus.
Note: There is no IDSEL signal interfacing the secondary PCI bus; thus, the entire configuration
space of the bridge can only be accessed from the primary bus.
Primary initiator ready. P_IRDY indicates ability of the primary bus master to complete the current
data phase of the transaction. A data phase is completed on a rising edge of P_CLK where both
P_IRDY and P_TRDY are asserted. Until P_IRDY and P_TRDY are both sampled asserted, wait
states are inserted.
Primary PCI bus lock. P_LOCK is used to lock the primary bus and gain exclusive access as a bus
master.
Primary parity. In all primary bus read and write cycles, the bridge calculates even parity across the
P_AD and P_C/BE
indicator with a one-P_CLK delay. As a target during PCI read cycles, the calculated parity is
compared to the parity indicator of the master; a miscompare can result in a parity error assertion
(P_PERR
Primary parity error indicator. P_PERR is driven by a primary bus PCI device to indicate that
calculated parity does not match P_PAR when P_PERR
register (PCI offset 04h, see Section 4.3).
Primary PCI bus request. Asserted by the bridge to request access to the primary PCI bus as a
master.
Primary system error. Output pulsed from the bridge when enabled through the command register
(PCI offset 04h, see Section 4.3) indicating a system error has occurred. The bridge needs not be
the target of the primary PCI cycle to assert this signal. When bit 6 is enabled in the bridge control
register (PCI offset 3Eh, see Section 4.32), this signal also pulses, indicating that a system error has
occurred on one of the subordinate buses downstream from the bridge.
Primary cycle stop signal. This signal is driven by a PCI target to request that the master stop the
current primary bus transaction. This signal is used for target disconnects and is commonly asserted
by target devices which do not support burst data transfers.
Primary target ready. P_TRDY indicates the ability of the primary bus target to complete the current
data phase of the transaction. A data phase is completed upon a rising edge of P_CLK where both
P_IRDY and P_TRDY are asserted. Until both P_IRDY and P_TRDY are asserted, wait states are
inserted.
).
buses. As a bus master during PCI write cycles, the bridge outputs this parity
21J3ISecondary PCI bus clock input. This input synchronizes the PCI2050 to the secondary bus clocks.
NO.
N6
N3
N1
M5
M3
M2
L5
L6
L2
L1
Table 2–7. Secondary PCI System Terminals
I/ODESCRIPTION
Secondary PCI bus clocks. Provide timing for all transactions on the secondary PCI bus. Each
secondary bus device samples all secondary PCI signals at the rising edge of its corresponding
O
S_CLKOUT input.
Secondary external arbiter enable. When this signal is high, the secondary external arbiter is
enabled. When the external arbiter is enabled, the PCI2050 S_REQ0
a secondary bus grant input to the bridge and S_GNT0
request to the external arbiter on the secondary bus.
Secondary PCI reset. S_RST is a logical OR of P_RST and the state of the secondary bus reset
bit (bit 6) of the bridge control register (PCI offset 3Eh, see Section 4.32). S_RST
with respect to the state of the secondary interface CLK signal.
is reconfigured as a secondary bus master
terminal is reconfigured as
is asynchronous
2–11
NAME
TERMINAL
PDV
NO.
GHK/ZHK
NO.
Table 2–8. Secondary PCI Address and Data Terminals
Secondary address/data bus. These signals make up the multiplexed PCI address and data bus
on the secondary interface. During the address phase of a secondary bus PCI cycle,
S_AD31–S_AD0 contain a 32-bit address or other destination information. During the data phase,
S_AD31–S_AD0 contain data.
Secondary bus commands and byte enables. These signals are multiplexed on the same PCI
terminals. During the address phase of a secondary bus PCI cycle, S_C/BE3
bus command. During the data phase, this 4-bit bus is used as byte enables. The byte enables
determine which byte paths of the full 32-bit data bus carry meaningful data. S_C/BE0 applies to
byte 0 (S_AD7–S_AD0), S_C/BE1
2 (S_AD23–S_AD16), and S_C/BE3
Secondary device select. The bridge asserts S_DEVSEL to claim a PCI cycle as the target device.
As a PCI master on the secondary bus, the bridge monitors S_DEVSEL
If no target responds before time-out occurs, then the bridge terminates the cycle with a master
abort.
Secondary cycle frame. S_FRAME is driven by the master of a secondary bus cycle. S_FRAME
is asserted to indicate that a bus transaction is beginning and data transfers continue while
S_FRAME
data phase.
Secondary bus grant to the bridge. The bridge provides internal arbitration and these signals are
used to grant potential secondary PCI bus masters access to the bus. Ten potential masters
(including the bridge) can be located on the secondary PCI bus.
O
When the internal arbiter is disabled, S_GNT0
request signal for the bridge.
is asserted. When S_FRAME is deasserted, the secondary bus transaction is in the final
applies to byte 1 (S_AD15–S_AD8), S_C/BE2 applies to byte
applies to byte 3 (S_AD31–S_AD24).
is reconfigured as an external secondary bus
–S_C/BE0 define the
until a target responds.
2–12
Table 2–9. Secondary PCI Interface Control Terminals
Secondary initiator ready. S_IRDY indicates the ability of the secondary bus master to complete the
current data phase of the transaction. A data phase is completed on a rising edge of S_CLK where both
and S_TRDY are asserted; until S_IRDY and S_TRDY are asserted, wait states are inserted.
S_IRDY
Secondary PCI bus lock. S_LOCK is used to lock the secondary bus and gain exclusive access as a
master.
Secondary parity. In all secondary bus read and write cycles, the bridge calculates even parity across
the S_AD and S_C/BE
indicator with a one-S_CLK delay . As a target during PCI read cycles, the calculated parity is compared
to the master parity indicator. A miscompare can result in a parity error assertion (S_PERR
Secondary parity error indicator. S_PERR is driven by a secondary bus PCI device to indicate that
calculated parity does not match S_PAR when enabled through the command register (PCI offset 04h,
see Section 4.3).
Secondary PCI bus request signals. The bridge provides internal arbitration, and these signals are used
as inputs from secondary PCI bus masters requesting the bus. Ten potential masters (including the
bridge) can be located on the secondary PCI bus.
I
When the internal arbiter is disabled, the S_REQ0
grant for the bridge.
Secondary system error. S_SERR is passed through the primary interface by the bridge if enabled
through the bridge control register (PCI offset 3Eh, see Section 4.32). S_SERR
the bridge.
Secondary cycle stop signal. S_STOP is driven by a PCI target to request that the master stop the
current secondary bus transaction. S_STOP
by target devices that do not support burst data transfers.
Secondary target ready. S_TRDY indicates the ability of the secondary bus target to complete the
current data phase of the transaction. A data phase is completed on a rising edge of S_CLK where both
S_IRDY and S_TRDY are asserted; until S_IRDY and S_TRDY are asserted, wait states are inserted.
buses. As a master during PCI write cycles, the bridge outputs this parity
signal is reconfigured as an external secondary bus
is never asserted by
is used for target disconnects and is commonly asserted
).
2–13
Table 2–10. Miscellaneous Terminals
TERMINAL
PDV
NAME
BPCCE44P2I
GPIO3/HSSWITCH
GPIO2
GPIO1
GPIO0
HSENUM127L14OHot-swap ENUM
HSLED128K19OHot-swap LED output
MS0
MS1
NC
S_M66ENA153E18O
GHK/ZHK
NO.
24
25
27
28
155E17IMode select 0
106R17IMode select 1
102
125
NO.
K1
K2
K5
K6
R14
L17
I/ODESCRIPTION
Bus/power clock control management terminal. When signal BPCCE is tied high and when
the PCI2050 is placed in the D3 power state, it enables the PCI2050 to place the secondary
bus in the B2 power state. The PCI2050 disables the secondary clocks and drives them to
0. When tied low, placing the PCI2050 in the D3 power state has no effect on the secondary
bus clocks.
General-purpose I/O terminals
GPIO3 is HSSWITCH
I
HSSWITCH
NCThese terminals have no function on the PCI2050.
Secondary bus 66-MHz enable terminal. This terminal is always driven low to indicate that
the secondary bus speed is 33 MHz.
provides the status of the ejector handle switch to the cPCI logic.
in cPCI mode.
Table 2–11. JTAG Interface Terminals
TERMINAL
PDV
NAME
TCK133J19IJTAG boundary-scan clock. TCK is the clock controlling the JTAG logic.
TDI129K18I
TDO130K17O
TMS132K14IJTAG test mode select. TMS causes state transitions in the test access port controller.
TRST134J18I
NO.
GHK/ZHK
NO.
I/ODESCRIPTION
JTAG serial data in. TDI is the serial input through which JTAG instructions and test data enter the JT AG
interface. The new data on TDI is sampled on the rising edge of TCK.
JTAG serial data out. TDO is the serial output through which test instructions and data from the test logic
leave the PCI2050.
JTAG T AP reset. When TRST is asserted low , the TAP controller is asynchronously forced to enter a reset
state and initialize the test logic.
Primary bus-signaling environment supply. P_V
protection circuitry on primary bus I/O signals.
Secondary bus-signaling environment supply. S_V
protection circuitry on secondary bus I/O signals.
is used in
CCP
CCP
is used in
3 Feature/Protocol Descriptions
The following sections give an overview of the PCI2050 PCI-to-PCI bridge features and functionality. Figure 3–1
shows a simplified block diagram of a typical system implementation using the PCI2050.
CPU
Host Bus
PCI Bus 0
PCI2050
PCI Bus 1
Host
Bridge
Memory
PCI
Device
PCI Bus 2
PCI2050
PCI Option Slot
PCI
Device
PCI
Device
Figure 3–1. System Block Diagram
PCI
Device
PCI Option Card
PCI Option Card
(Option)
3.1Introduction to the PCI2050
The PCI2050 is a bridge between two PCI buses and is compliant with both the PCI Local Bus Specification and the
PCI-to-PCI Bridge Specification. The bridge supports two 32-bit PCI buses operating at a maximum of 33 MHz. The
primary and secondary buses operate independently in either a 3.3-V or 5-V signaling environment. The core logic
of the bridge, however, is powered at 3.3 V to reduce power consumption.
Host software interacts with the bridge through internal registers. These internal registers provide the standard PCI
status and control for both the primary and secondary buses. Many vendor-specific features that exist in the TI
extension register set are included in the bridge. The PCI configuration header of the bridge is only accessible from
the primary PCI interface.
The bridge provides internal arbitration for the nine possible secondary bus masters, and provides each with a
dedicated active-low request/grant pair (REQ
PCI2050 bridge defaulting to the highest priority tier.
Upon system power up, power-on self-test (POST) software configures the bridge according to the devices that exist
on subordinate buses, and enables performance-enhancing features of the PCI2050. In a typical system, this is the
only communication with the bridge internal register set.
/GNT). The arbiter features a two-tier rotational scheme with the
3–1
3.2PCI Commands
The bridge responds to PCI bus cycles as a PCI target device based on internal register settings and on the decoding
of each address phase. Table 3–1 lists the valid PCI bus cycles and their encoding on the command/byte enable
(C/BE
The bridge never responds as a PCI target to the interrupt acknowledge, special cycle, or reserved commands. The
bridge does, however, initiate special cycles on both interfaces when a type 1 configuration cycle issues the special
cycle request. The remaining PCI commands address either memory , I/O, or configuration space. The bridge accepts
PCI cycles by asserting DEVSEL
as a medium-speed device, i.e., DEVSEL is asserted two clock cycles after the
address phase.
The PCI2050 converts memory write and invalidate commands to memory write commands when forwarding
transactions from either the primary or secondary side of the bridge if the bridge cannot guarantee that an entire cache
line will be delivered.
3.3Configuration Cycles
PCI Local Bus Specification defines two types of PCI configuration read and write cycles: type 0 and type 1. The
bridge decodes each type differently . T ype 0 configuration cycles are intended for devices on the primary bus, while
type 1 configuration cycles are intended for devices on some hierarchically subordinate bus. The difference between
these two types of cycles is the encoding of the primary PCI (P_AD) bus during the address phase of the cycle.
Figure 3–2 shows the P_AD bus encoding during the address phase of a type 0 configuration cycle. The 6-bit register
number field represents an 8-bit address with the two lower bits masked to 0, indicating a doubleword boundary . This
results in a 256-byte configuration address space per function per device. Individual byte accesses may be selected
within a doubleword by using the P_C/BE
31111087210
Reserved
Figure 3–2. PCI AD31–AD0 During Address Phase of a Type 0 Configuration Cycle
The bridge claims only type 0 configuration cycles when its P_IDSEL terminal is asserted during the address phase
of the cycle and the PCI function number encoded in the cycle is 0. If the function number is 1 or greater, then the
signals during the data phase of the cycle.
Function
Number
Register
Number
00
3–2
bridge does not recognize the configuration command. In this case, the bridge does not assert DEVSEL, and the
configuration transaction results in a master abort. The bridge services valid type 0 configuration read or write cycles
by accessing internal registers from the bridge configuration header (see Table 4–1).
Because type 1 configuration cycles are issued to devices on subordinate buses, the bridge claims type 1 cycles
based on the bus number of the destination bus. The P_AD bus encoding during the address phase of a type 1 cycle
is shown in Figure 3–3. The device number and bus number fields define the destination bus and device for the cycle.
3124231615111087210
ReservedBus Number
Device
Number
Function
Number
Register
Number
01
Figure 3–3. PCI AD31–AD0 During Address Phase of a Type 1 Configuration Cycle
Several bridge configuration registers shown in Table 4–1 are significant when decoding and claiming type 1
configuration cycles. The destination bus number encoded on the P_AD bus is compared to the values programmed
in the bridge configuration registers 18h, 19h, and 1Ah, which are the primary bus number, secondary bus number,
and subordinate bus number registers, respectively. These registers default to 00h and are programmed by host
software to reflect the bus hierarchy in the system (see Figure 3–4 for an example of a system bus hierarchy and how
the PCI2050 bus number registers would be programmed in this case).
When the PCI2050 claims a type 1 configuration cycle that has a bus number equal to its secondary bus number,
the PCI2050 converts the type 1 configuration cycle to a type 0 configuration cycle and asserts the proper S_AD line
as the IDSEL (see Table 3–2). All other type 1 transactions that access a bus number greater than the bridge
secondary bus number but less than or equal to its subordinate bus number are forwarded as type 1 configuration
cycles.
The bridge is designed to generate special cycles on both buses through a type 1 cycle conversion. During a type 1
configuration cycle, if the bus number field matches the bridge secondary bus number, the device number field is 1Fh,
and the function number field is 07h, then the bridge generates a special cycle on the secondary bus with a message
that matches the type 1 configuration cycle data. If the bus number is a subordinate bus and not the secondary , then
the bridge passes the type 1 special cycle request through to the secondary interface along with the proper message.
Special cycles are never passed through the bridge. Type 1 configuration cycles with a special cycle request can
propagate in both directions.
3.5Secondary Clocks
The PCI2050 provides 10 secondary clock outputs (S_CLKOUT[0:9]). Nine are provided for clocking secondary
devices. The tenth clock should be routed back into the PCI2050 S_CLK input to ensure all secondary bus devices
see the same clock. Figure 3–5 is a block diagram of the secondary clock function.
3–4
PCI2050
S_CLK
S_CLKOUT9
S_CLKOUT8
S_CLKOUT2
S_CLKOUT1
S_CLKOUT0
PCI
Device
PCI
Device
PCI
Device
PCI
Device
Figure 3–5. Secondary Clock Block Diagram
3.6Bus Arbitration
The PCI2050 implements bus request (P_REQ) and bus grant (P_GNT) terminals for primary PCI bus arbitration.
Nine secondary bus requests and nine secondary bus grants are provided on the secondary of the PCI2050. Ten
potential initiators, including the bridge, can be located on the secondary bus. The PCI2050 provides a two-tier
arbitration scheme on the secondary bus for priority bus-master handling.
The two-tier arbitration scheme improves performance in systems in which master devices do not all require the same
bandwidth. Any master that requires frequent use of the bus can be programmed to be in the higher priority tier.
3.6.1Primary Bus Arbitration
The PCI2050, acting as an initiator on the primary bus, asserts P_REQ when forwarding transactions upstream to
the primary bus. If a target disconnect, a target retry , or a target abort is received in response to a transaction initiated
on the primary bus by the PCI2050, then the device deasserts P_REQ
When the primary bus arbiter asserts P_GNT
in response to a P_REQ from the PCI2050, the device initiates a
for two PCI clock cycles.
transaction on the primary bus during the next PCI clock cycle after the primary bus is sampled idle.
When P_REQ
parking the P_AD31–P_AD0 bus, the C/BE3
is not asserted and the primary bus arbiter asserts P_GNT to the PCI2050, the device responds by
–C/BE0 bus, and primary parity (P_P AR) by driving them to valid logic
levels. If the PCI2050 is parking the primary bus and wants to initiate a transaction on the bus, then it can start the
transaction on the next PCI clock by asserting the primary cycle frame (P_FRAME
P_GNT
is deasserted, then the bridge must rearbitrate for the bus to initiate a transaction.
) while P_GNT is still asserted. If
3.6.2Internal Secondary Bus Arbitration
S_CFN controls the state of the secondary internal arbiter. The internal arbiter can be enabled by pulling S_CFN low
or disabled by pulling S_CFN
high. The PCI2050 provides nine secondary bus request terminals and nine secondary
3–5
bus grant terminals. Including the bridge, there are a total of ten potential secondary bus masters. These request and
grant signals are connected to the internal arbiter. When an external arbiter is implemented, S_REQ8
S_GNT8
–S_GNT1 are placed in a high-impedance mode.
–S_REQ1 and
3.6.3External Secondary Bus Arbitration
An external secondary bus arbiter can be used instead of the PCI2050 internal bus arbiter. When using an external
arbiter, the PCI2050 internal arbiter should be disabled by pulling S_CFN
high.
When an external secondary bus arbiter is used, the PCI2050 internally reconfigures the S_REQ0
signals so that S_REQ0 becomes the secondary bus grant for the bridge and S_GNT0 becomes the secondary bus
request for the bridge. This is done because S_REQ0
the bridge, and S_GNT0
When an external arbiter is used, all unused secondary bus grant outputs (S_GNT8
impedance mode. Any unused secondary bus request inputs (S_REQ8
the inputs from oscillating.
is an output and can thus provide the request output from the bridge.
is an input and can thus be used to provide the grant input to
–S_GNT1) are placed in a high
–S_REQ1) should be pulled high to prevent
and S_GNT0
3.7Decode Options
The PCI2050 supports positive decoding on the primary interface and negative decoding on the secondary interface.
Positive decoding is a method of address decoding in which a device responds only to accesses within an assigned
address range. Negative decoding is a method of address decoding in which a device responds only to accesses
outside of an assigned address range.
3.8System Error Handling
The PCI2050 can be configured to signal a system error (SERR) for a variety of conditions. The P_SERR event
disable register (offset 64h, see Section 5.4) and the P_SERR status register (offset 6Ah, see Section 5.9) provide
control and status bits for each condition for which the bridge can signal SERR
reporting for both downstream and upstream transactions.
By default, the PCI2050 will not signal SERR
command register (offset 04h, see Section 4.3), then the bridge signals SERR
P_SERR event disable register occur and that condition is enabled. By default, all error conditions are enabled in the
P_SERR event disable register . When the bridge signals SERR
see Section 4.19) is set.
. If the PCI2050 is configured to signal SERR by setting bit 8 in the
, bit 14 in the secondary status register (offset 1Eh,
. These individual bits enable SERR
if any of the error conditions in the
3.8.1Posted Write Parity Error
If bit 1 in the P_SERR event disable register (offset 64h, see Section 5.4) is 0, then parity errors on the target bus
during a posted write are passed to the initiating bus as a SERR
(offset 6Ah, see Section 5.9) is set. The status bit is cleared by writing a 1.
. When this occurs, bit 1 of the P_SERR status register
3.8.2Posted Write Time-Out
If bit 2 in the P_SERR event disable register (offset 64h, see Section 5.4) is 0 and the retry timer expires while
attempting to complete a posted write, then the PCI2050 signals SERR
of the P_SERR status register (offset 6Ah, see Section 5.9) is set. The status bit is cleared by writing a 1.
on the initiating bus. When this occurs, bit 2
3.8.3Target Abort on Posted Writes
If bit 3 in the P_SERR event disable register (of fset 64h, see Section 5.4) is 0 and the bridge receives a target abort
during a posted write transaction, then the PCI2050 signals SERR
the P_SERR status register (offset 6Ah, see Section 5.9) is set. The status bit is cleared by writing a 1.
3–6
on the initiating bus. When this occurs, bit 3 of
3.8.4Master Abort on Posted Writes
If bit 4 in the P_SERR event disable register (PCI offset 64h, see Section 5.4) is 0 and a posted write transaction
results in a master abort, then the PCI2050 signals SERR
status register (PCI offset 6Ah, see Section 5.9) is set. The status bit is cleared by writing a 1.
on the initiating bus. When this occurs, bit 4 of the P_SERR
3.8.5Master Delayed Write Time-Out
If bit 5 in the P_SERR event disable register (PCI offset 64h, see Section 5.4) is 0 and the retry timer expires while
attempting to complete a delayed write, then the PCI2050 signals SERR
of the P_SERR status register (PCI offset 6Ah, see Section 5.9) is set. The status bit is cleared by writing a 1.
on the initiating bus. When this occurs, bit 5
3.8.6Master Delayed Read Time-Out
If bit 6 in the P_SERR event disable register (offset 64h, see Section 5.4) is 0 and the retry timer expires while
attempting to complete a delayed read, then the PCI2050 signals SERR
of the P_SERR status register (offset 6Ah, see Section 5.9) is set. The status bit is cleared by writing a 1.
on the initiating bus. When this occurs, bit 6
3.8.7Secondary SERR
The PCI2050 passes SERR from the secondary bus to the primary bus if it is enabled for SERR response, that is,
if bit 8 in the command register (PCI offset 04h, see Section 4.3) is set, and if bit 1 in the bridge control register (PCI
offset 3Eh, see Section 4.32) is set.
3.9Parity Handling and Parity Error Reporting
When forwarding transactions, the PCI2050 attempts to pass the data parity condition from one interface to the other
unchanged, whenever possible, to allow the master and target devices to handle the error condition.
3.9.1Address Parity Error
If the parity error response bit (bit 6) in the command register (PCI offset 04h, see Section 4.3) is set, then the PCI2050
signals SERR
on address parity errors and target abort transactions.
3.9.2Data Parity Error
If the parity error response bit (bit 6) in the command register (PCI offset 04h, see Section 4.3) is set, then the PCI2050
signals PERR
register (PCI offset 06h, see Section 4.4) is set.
If the bridge is configured to respond to parity errors via bit 6 in the command register (PCI offset 04h, see Section 4.3),
then bit 8 (data parity error detected) in the status register (PCI offset 06h, see Section 4.4) is set when the bridge
detects bad parity . The data parity error detected bit is also set when the bridge, as a bus master, asserts PERR
detects PERR
when it receives bad data. When the bridge detects bad parity , bit 15 (detected parity error) in the status
or
.
3.10 Master and Target Abort Handling
If the PCI2050 receives a target abort during a write burst, then it signals target abort back on the initiator bus. If it
receives a target abort during a read burst, then it provides all of the valid data on the initiator bus and disconnects.
T arget aborts for posted and nonposted transactions are reported as specified in the PCI-to-PCI Bridge Specification.
Master aborts for posted and nonposted transactions are reported as specified in the PCI-to-PCI Bridge Specification.
If a transaction is attempted on the primary bus after a secondary reset is asserted, then the PCI2050 follows bit 5
(master abort mode) in the bridge control register (PCI offset 3Eh, see Section 4.32) for reporting errors.
3.11 Discard Timer
The PCI2050 is free to discard the data or status of a delayed transaction that was completed with a delayed
transaction termination when a bus master has not repeated the request within 2
10
or 215 PCI clocks (approximately
3–7
30 µs and 993 µs, respectively). The PCI Local Bus Specification recommends that a bridge wait 215 PCI clocks
before discarding the transaction data or status.
The PCI2050 implements a discard timer for use in delayed transactions. After a delayed transaction is completed
on the destination bus, the bridge may discard it under two conditions. The first condition occurs when a read
transaction is made to a region of memory that is inside a defined prefetchable memory region, or when the command
is a memory read line or a memory read multiple, implying that the memory region is prefetchable. The other condition
occurs when the master originating the transaction (either a read or a write, prefetchable or nonprefetchable) has not
retried the transaction within 2
timer. When the discard timer expires, the bridge is required to discard the data. The PCI2050 default value for the
discard timer is 2
(offset 3Eh, see Section 4.32). For more information on the discard timer , see error conditions in the PCI Local BusSpecification.
15
clocks; however, this value can be set to 210 clocks by setting bit 9 in the bridge control register
10
or 215 clocks. The number of clocks is tracked by a timer referred to as the discard
3.12 Delayed Transactions
The bridge supports delayed transactions as defined in PCI Local Bus Specification. A target must be able to complete
the initial data phase in 16 PCI clocks or less from the assertion of the cycle frame (FRAME
phases must complete in eight PCI clocks or less. A delayed transaction consists of three phases:
•An initiator device issues a request.
•The target completes the request on the destination bus and signals the completion to the initiator.
•The initiator completes the request on the originating bus.
If the bridge is the target of a PCI transaction and it must access a slow device to write or read the requested data,
and the transaction takes longer than 16 clocks, then the bridge must latch the address, the command, and the byte
enables, and then issue a retry to the initiator. The initiator must end the transaction without any transfer of data and
is required to retry the transaction later using the same address, command, and byte enables. This is the first phase
of the delayed transaction.
), and subsequent data
During the second phase, if the transaction is a read cycle, the bridge fetches the requested data on the destination
bus, stores it internally, and obtains the completion status, thus completing the transaction on the destination bus.
If it is a write transaction, then the bridge writes the data and obtains the completion status, thus completing the
transaction on the destination bus. The bridge stores the completion status until the master on the initiating bus retries
the initial request.
During the third phase, the initiator rearbitrates for the bus. When the bridge sees the initiator retry the transaction,
it compares the second request to the first request. If the address, command, and byte enables match the values
latched in the first request, then the completion status (and data if the request was a read) is transferred to the initiator.
At this point, the delayed transaction is complete. If the second request from the initiator does not match the first
request exactly, then the bridge issues another retry to the initiator.
The PCI supports up to three delayed transactions in each direction at any given time.
3.13 Mode Selection
Table 3–3 shows the mode selection via MS0 (PDV terminal 155, GHK/ZHK terminal E17) and MS1 (PDV terminal
106, GHK/ZHK terminal R17).
3–8
Table 3–3. Configuration via MS0 and MS1
MS0
00CompactPCI hot-swap friendly
01CompactPCI hot-swap disabled
1XIntel compatible
MS1MODE
PCI Bus Power Management Interface Specification Revision 1.1
HSSWITCH/GPIO(3) functions as HSSWITCH
PCI Bus Power Management Interface Specification Revision 1.1
HSSWITCH/GPIO(3) functions as GPIO(3)
No cPCI hot swap
PCI Bus Power Management Interface Specification Revision 1.0
3.14 CompactPCI Hot-Swap Support
The PCI2050 is hot-swap friendly silicon that supports all of the hot-swap capable features, contains support for
software control, and integrates circuitry required by the PICMG CompactPCI Hot-Swap Specification. To be
hot-swap capable, the PCI2050 supports the following:
•Compliance with PCI Local Bus Specification
•Tolerance of V
•Asynchronous reset
•Tolerance of precharge voltage
•I/O buffers that meet modified V/I requirements
•Limited I/O terminal voltage at precharge voltage
from early power
CC
•Hot-swap control and status programming via extended PCI capabilities linked list
•Hot-swap terminals: HS_ENUM
, HS_SWITCH, and HS_LED
cPCI hot-swap defines a process for installing and removing PCI boards without adversely affecting a running system.
The PCI2050 provides this functionality such that it can be implemented on a board that can be removed and inserted
in a hot-swap system.
The PCI2050 provides three terminals to support hot-swap when configured to be in hot-swap mode: HS_ENUM
(output), HS_SWITCH (input), and HS_LED (output). The HS_ENUM output indicates to the system that an insertion
event occurred or that a removal event is about to occur. The HS_SWITCH
input indicates the state of a board ejector
handle, and the HS_LED output lights a blue LED to signal insertion- and removal-ready status.
3–9
3.15 JTAG Support
The PCI2050 implements a JTAG test port based on IEEE Standard 1149.1, IEEE Standard Test Access Port and
Boundary-Scan Architecture. The JTAG test port consists of the following:
•A 5-wire test access port
•A test access port controller
•An instruction register
•A bypass register
•A boundary-scan register
3.15.1 Test Port Instructions
The PCI2050 supports the following JTAG instructions:
•EXTEST, BYPASS, and SAMPLE
•HIGHZ and CLAMP
•Private (various private instructions used by TI for test purposes)
Table 3–4 lists and describes the different test port instructions, and gives the op code of each one. The information
in Table 3–5 is for implementation of boundary scan interface signals to permit in-circuit testing.
Table 3–4. JTAG Instructions and Op Codes
INSTRUCTIONOP CODEDESCRIPTION
EXTEST00000External test: drives terminals from the boundary scan register
SAMPLE00001Sample I/O terminals
CLAMP00100Drives terminals from the boundary scan register and selects the bypass register for shifts
HIGHZ00101Puts all outputs and I/O terminals except for the TDO terminal in a high-impedance state
The PCI2050 implements a four-terminal general-purpose I/O interface. Besides functioning as a general-purpose
I/O interface, the GPIO terminals can be used to read in the secondary clock mask and to stop the bridge from
accepting I/O and memory transactions.
3.16.1 Secondary Clock Mask
The PCI2050 uses GPIO0, GPIO2, and MSK_IN to shift in the secondary clock mask from an external shift register.
A secondary clock mask timing diagram is shown in Figure 3–6. Table 3–6 lists the format for clock mask data.
MSK_IN
GPIO2
GPIO0
P_RST
S_RST
Bit 15
Bit 14 Bit 13 Bit 12 Bit 11 Bit 10 Bit 9 Bit 8 Bit 7Bit 6 Bit 5 Bit 4Bit 3 Bit 2Bit 1 Bit 0
The PCI 2050 will stop forwarding I/O and memory transactions if bit 5 of the chip control register (offset 40h, see
Section 5.1) is set to 1 and GPIO3 is driven high. The bridge will complete all queued posted writes and delayed
requests, but delayed completions will not be returned until GPIO3 is driven low and transaction forwarding is
resumed. The bridge will continue to accept configuration cycles in this mode. This feature is not available when in
CompactPCI hot-swap mode because GPIO3 is used as the HS_SWITCH
3–14
input in this mode.
3.17 PCI Power Management
The PCI Power Management Specification establishes the infrastructure required to let the operating system control
the power of PCI functions. This is done by defining a standard PCI interface and operations to manage the power
of PCI functions on the bus. The PCI bus and the PCI functions can be assigned one of four software visible power
management states, which result in varying levels of power savings.
The four power management states of PCI functions are D0—fully on state, D1 and D2—intermediate states, and
D3—off state. Similarly , bus power states of the PCI bus are B0–B3. The bus power states B0 –B3 are derived from
the device power state of the originating PCI2050 device.
For the operating system to manage the device power states on the PCI bus, the PCI function supports four power
management operations:
•Capabilities reporting
•Power status reporting
•Setting the power state
•System wake-up
The operating system identifies the capabilities of the PCI function by traversing the new capabilities list. The
presence of the new capabilities list is indicated by a bit in the status register (offset 06h, see Section 4.4) which
provides access to the capabilities list.
3.17.1 Behavior in Low-Power States
The PCI2050 supports D0, D1, D2, and D3
power states when in TI mode. The PCI2050 only supports D0 and
hot
D3 power states when in Intel mode. The PCI2050 is fully functional only in D0 state. In the lower power states, the
bridge does not accept any memory or I/O transactions. These transactions are aborted by the master. The bridge
accepts type 0 configuration cycles in all power states except D3
. The bridge also accepts type 1 configuration
cold
cycles but does not pass these cycles to the secondary bus in any of the lower power states. Type 1 configuration
writes are discarded and reads return all 1s. All error reporting is done in the low power states. When in D2 and D3
hot
states, the bridge turns off all secondary clocks for further power savings.
When going from D3
to D0, an internal reset is generated. This reset initializes all PCI configuration registers to
hot
their default values. The TI specific registers (40h – FFh) are not reset. Power management registers also are not
reset.
3–15
4 Bridge Configuration Header
The PCI2050 bridge is a single-function PCI device. The configuration header is in compliance with the PCI-to-PCI
Bridge Specification 1.1. T able 4–1 shows the PCI configuration header, which includes the predefined portion of the
bridge configuration space. The PCI configuration offset is shown in the right column under the OFFSET heading.
Table 4–1. Bridge Configuration Header
REGISTER NAMEOFFSET
Device IDVendor ID00h
StatusCommand04h
Class codeRevision ID08h
BISTHeader typePrimary latency timerCache line size0Ch
Base address 010h
Base address 114h
Secondary bus latency timerSubordinate bus numberSecondary bus numberPrimary bus number18h
ReservedHot swap control statusHS next item pointerHS capability IDE4h
ReservedE8h–FFh
4–1
4.1Vendor ID Register
This 16-bit value is allocated by the PCI Special Interest Group (SIG) and identifies TI as the manufacturer of this
device. The vendor ID assigned to TI is 104Ch.
Bit1514131211109876543210
NameVendor ID
TypeRRRRRRRRRRRRRRRR
Default0001000001001100
Register:Vendor ID
Type:Read-only
Offset:00h
Default:104Ch
4.2Device ID Register
This 16-bit value is allocated by the vendor and identifies the PCI device. The device ID for the PCI2050 is AC28h.
Bit1514131211109876543210
NameDevice ID
TypeRRRRRRRRRRRRRRRR
Default1010110000101000
Register:Device ID
Type:Read-only
Offset:02h
Default:AC28h
4–2
4.3Command Register
The command register provides control over the bridge interface to the primary PCI bus. VGA palette snooping is
enabled through this register, and all other bits adhere to the definitions in the PCI Local Bus Specification. Table 4–2
describes the bit functions in the command register.
9R/WFast back-to-back enable. This bit defaults to 0.
8R/W
7R
6R/W
5R/W
4RMemory write and invalidate enable. In a PCI-to-PCI bridge, bit 4 must be read-only and return 0 when read.
3R
2R/W
1R/W
0R/W
System error (SERR) enable. Bit 8 controls the enable for the SERR driver on the primary interface.
0 = Disable SERR
1 = Enable the SERR
Wait cycle control. Bit 7 controls address/data stepping by the bridge on both interfaces. The bridge does not support
address/data stepping and this bit is hardwired to 0.
Parity error response enable. Bit 6 controls the bridge response to parity errors.
VGA palette snoop enable. When set, the bridge passes I/O writes on the primary PCI bus with addresses 3C6h, 3C8h,
and 3C9h inclusive of ISA aliases (that is, only bits AD9–AD0 are included in the decode).
Special cycle enable. A PCI-to-PCI bridge cannot respond as a target to special cycle transactions, so bit 3 is defined as
read-only and must return 0 when read.
Bus master enable. Bit 2 controls the ability of the bridge to initiate a cycle on the primary PCI bus. When bit 2 is 0, the bridge
does not respond to any memory or I/O transactions on the secondary interface since they cannot be forwarded to the
primary PCI bus.
0 = Bus master capability disabled (default)
1 = Bus master capability enabled
Memory space enable. Bit 1 controls the bridge response to memory accesses for both prefetchable and nonprefetchable
memory spaces on the primary PCI bus. Only when bit 1 is set will the bridge forward memory accesses to the secondary
bus from a primary bus initiator.
0 = Memory space disabled (default)
1 = Memory space enabled
I/O space enable. Bit 0 controls the bridge response to I/O accesses on the primary interface. Only when bit 0 is set will
the bridge forward I/O accesses to the secondary bus from a primary bus initiator.
0 = I/O space disabled (default)
1 = I/O space enabled
driver on primary interface (default)
driver on primary interface
4–3
4.4Status Register
The status register provides device information to the host system. Bits in this register are cleared by writing a 1 to
the respective bit; writing a 0 to a bit location has no effect. Table 4–3 describes the status register.
15R/WDetected parity error. Bit 15 is set when a parity error is detected.
Signaled system error (SERR). Bit 14 is set if SERR is enabled in the command register (offset 04h, see Section 4.3,
14R/W
13R/W
12R/W
11R/W
10–9R
8R/W
7R
6R
5R66-MHz capable. The PCI2050 operates at a maximum P_CLK frequency of 33 MHz; therefore, bit 5 is hardwired to 0.
4R
3–0RReserved. Bits 3–0 return 0s when read.
Command Register) and the bridge signals a system error (SERR). See Section 3.8, System Error Handling.
0 = No SERR signaled (default)
1 = Signals SERR
Received master abort. Bit 13 is set when a cycle initiated by the bridge on the primary bus has been terminated by a master
abort.
0 = No master abort received (default)
1 = Master abort received
Received target abort. Bit 12 is set when a cycle initiated by the bridge on the primary bus has been terminated by a target
abort.
0 = No target abort received (default)
1 = Target abort received
Signaled target abort. Bit 11 is set by the bridge when it terminates a transaction on the primary bus with a target abort.
0 = No target abort signaled by the bridge (default)
1 = T arget abort signaled by the bridge
DEVSEL timing. These read-only bits encode the timing of P_DEVSEL and are hardwired 01b, indicating that the bridge
asserts this signal at a medium speed.
Data parity error detected. Bit 8 is encoded as:
0 = The conditions for setting this bit have not been met. No parity error detected. (default)
1 = A data parity error occurred and the following conditions were met:
a. P_PERR
b. The bridge was the bus master during the data parity error.
c. The parity error response bit (bit 6) was set in the command register (offset 04h, see Section 4.3).
Fast back-to-back capable. The bridge supports fast back-to-back transactions as a target; therefore, bit 7 is hardwired to
1.
User-definable feature (UDF) support. The PCI2050 does not support the user-definable features; therefore, bit 6 is
hardwired to 0.
Capabilities list. Bit 4 is read-only and is hardwired to 1, indicating that capabilities additional to standard PCI are
implemented. The linked list of PCI power management capabilities is implemented by this function.
was asserted by any PCI device including the bridge.
4–4
4.5Revision ID Register
The revision ID register indicates the silicon revision of the PCI2050.
Bit76543210
NameRevision ID
TypeRRRRRRRR
Default00000000
Register:Revision ID
Type:Read-only
Offset:08h
Default:00h (reflects the current revision of the silicon)
4.6Class Code Register
This register categorizes the PCI2050 as a PCI-to-PCI bridge device (0604h) with a 00h programming interface.
The cache line size register is programmed by host software to indicate the system cache line size needed by the
bridge for memory read line, memory read multiple, and memory write and invalidate transactions. The PCI2050
supports cache line sizes up to and including 16 doublewords for memory write and invalidate. If the cache line size
is larger than 16 doublewords, the command is converted to a memory write command.
Bit76543210
NameCache line size
TypeR/WR/WR/WR/WR/WR/WR/WR/W
Default00000000
Register:Cache line size
Type:Read/Write
Offset:0Ch
Default:00h
4–5
4.8Primary Latency Timer Register
The latency timer register specifies the latency timer for the bridge in units of PCI clock cycles. When the bridge is
a primary PCI bus initiator and asserts P_FRAME
before the bridge transaction has terminated, then the bridge terminates the transaction when its P_GNT
, the latency timer begins counting from 0. If the latency timer expires
The header type register is read-only and returns 01h when read, indicating that the PCI2050 configuration space
adheres to the PCI-to-PCI bridge configuration. Only the layout for bytes 10h–3Fh of configuration space is
considered.
Bit76543210
NameHeader type
TypeRRRRRRRR
Default00000001
Register:Header type
Type:Read-only
Offset:0Eh
Default:01h
4.10 BIST Register
The PCI2050 does not support built-in self test (BIST). The BIST register is read-only and returns the value 00h when
read.
The primary bus number register indicates the primary bus number to which the bridge is connected. The bridge uses
this register, in conjunction with the secondary bus number and subordinate bus number registers, to determine when
to forward PCI configuration cycles to the secondary buses.
Bit76543210
NamePrimary bus number
TypeR/WR/WR/WR/WR/WR/WR/WR/W
Default00000000
Register:Primary bus number
Type:Read/Write
Offset:18h
Default:00h
4–7
4.14 Secondary Bus Number Register
The secondary bus number register indicates the secondary bus number to which the bridge is connected. The
PCI2050 uses this register, in conjunction with the primary bus number and subordinate bus number registers, to
determine when to forward PCI configuration cycles to the secondary buses. Configuration cycles directed to the
secondary bus are converted to type 0 configuration cycles.
Bit76543210
NameSecondary bus number
TypeR/WR/WR/WR/WR/WR/WR/WR/W
Default00000000
Register:Secondary bus number
Type:Read/Write
Offset:19h
Default:00h
4.15 Subordinate Bus Number Register
The subordinate bus number register indicates the bus number of the highest numbered bus beyond the primary bus
existing behind the bridge. The PCI2050 uses this register, in conjunction with the primary bus number and secondary
bus number registers, to determine when to forward PCI configuration cycles to the subordinate buses. Configuration
cycles directed to a subordinate bus (not the secondary bus) remain type 1 cycles as the cycle crosses the bridge.
Bit76543210
NameSubordinate bus number
TypeR/WR/WR/WR/WR/WR/WR/WR/W
Default00000000
Register:Subordinate bus number
Type:Read/write
Offset:1Ah
Default:00h
4.16 Secondary Bus Latency Timer Register
The secondary bus latency timer specifies the latency time for the bridge in units of PCI clock cycles. When the bridge
is a secondary PCI bus initiator and asserts S_FRAME
expires before the bridge transaction has terminated, then the bridge terminates the transaction when its S_GNT
deasserted. The PCI-to-PCI bridge S_GNT
is an internal signal and is removed when another secondary bus master
arbitrates for the bus.
Bit76543210
NameSecondary bus latency timer
TypeR/WR/WR/WR/WR/WR/WR/WR/W
Default00000000
Register:Secondary bus latency timer
Type:Read/Write
Offset:1Bh
Default:00h
, the latency timer begins counting from 0. If the latency timer
is
4–8
4.17 I/O Base Register
The I/O base register is used in decoding I/O addresses to pass through the bridge. The bridge supports 32-bit I/O
addressing; thus, bits 3–0 are read-only and default to 0001b. The upper four bits are writable and correspond to
address bits AD15–AD12. The lower 12 address bits of the I/O base address are considered 0. Thus, the bottom of
the defined I/O address range is aligned on a 4K-byte boundary. The upper 16 address bits of the 32-bit I/O base
address corresponds to the contents of the I/O base upper 16 bits register (offset 30h, see Section 4.26).
Bit76543210
NameI/O base
TypeR/WR/WR/WR/WRRRR
Default00000001
Register:I/O base
Type:Read-only, Read/Write
Offset:1Ch
Default:01h
4.18 I/O Limit Register
The I/O limit register is used in decoding I/O addresses to pass through the bridge. The bridge supports 32-bit I/O
addressing; thus, bits 3–0 are read-only and default to 0001b. The upper four bits are writable and correspond to
address bits AD15–AD12. The lower 12 address bits of the I/O limit address are considered FFFh. Thus, the top of
the defined I/O address range is aligned on a 4K-byte boundary. The upper 16 address bits of the 32-bit I/O limit
address corresponds to the contents of the I/O limit upper 16 bits register (offset 32h, see Section 4.27).
The secondary status register is similar in function to the status register (offset 06h, see Section 4.4); however, its
bits reflect status conditions of the secondary interface. Bits in this register are cleared by writing a 1 to the respective
bit.
Bit1514131211109876543210
NameSecondary status
TypeR/WR/WR/WR/WR/WRRR/WRRRRRRRR
Default0000001010000000
Register:Secondary status
Type:Read-only, Read/Write
Offset:1Eh
Default:0280h
Table 4–4. Secondary Status Register Description
BITTYPEFUNCTION
Detected parity error. Bit 15 is set when a parity error is detected on the secondary interface.
15R/W
14R/W
13R/W
12R/W
11R/W
10–9R
8R/W
7RFast back-to-back capable. Bit 7 is hardwired to 1.
6RUser-definable feature (UDF) support. Bit 6 is hardwired to 0.
5R66-MHz capable. Bit 5 is hardwired to 0.
4–0RReserved. Bits 4–0 return 0s when read.
0 = No parity error detected on the secondary bus (default)
1 = Parity error detected on the secondary bus
Received system error. Bit 14 is set when the secondary interface detects S_SERR asserted. Note that the bridge never
asserts S_SERR
0 = No S_SERR
1 = S_SERR
Received master abort. Bit 13 is set when a cycle initiated by the bridge on the secondary bus has been terminated by a
master abort.
0 = No master abort received (default)
1 = Bridge master aborted the cycle
Received target abort. Bit 12 is set when a cycle initiated by the bridge on the secondary bus has been terminated by a target
abort.
0 = No target abort received (default)
1 = Bridge received a target abort
Signaled target abort. Bit 1 1 is set by the bridge when it terminates a transaction on the secondary bus with a target abort.
0 = No target abort signaled (default)
1 = Bridge signaled a target abort
DEVSEL timing. These read-only bits encode the timing of S_DEVSEL and are hardwired to 01b, indicating that the bridge
asserts this signal at a medium speed.
Data parity error detected.
0 = The conditions for setting this bit have not been met
1 = A data parity error occurred and the following conditions were met:
a. S_PERR
b. The bridge was the bus master during the data parity error.
c. The parity error response bit (bit 1) was set in the bridge control register (offset 3Eh, see Section 4.32).
.
detected on the secondary bus (default)
detected on the secondary bus
was asserted by any PCI device including the bridge.
4–10
4.20 Memory Base Register
The memory base register defines the base address of a memory-mapped I/O address range used by the bridge to
determine when to forward memory transactions from one interface to the other. The upper 12 bits of this register
are read/write and correspond to the address bits AD31–AD20. The lower 20 address bits are considered 0s; thus,
the address range is aligned to a 1M-byte boundary. The bottom four bits are read-only and return 0s when read.
Bit1514131211109876543210
NameMemory base
TypeR/WR/WR/WR/WR/WR/WR/WR/WR/WR/WR/WR/WRRRR
Default0000000000000000
Register:Memory base
Type:Read-only, Read/Write
Offset:20h
Default:0000h
4.21 Memory Limit Register
The memory limit register defines the upper-limit address of a memory-mapped I/O address range used to determine
when to forward memory transactions from one interface to the other. The upper 12 bits of this register are read/write
and correspond to the address bits AD31–AD20. The lower 20 address bits are considered 1s; thus, the address
range is aligned to a 1M-byte boundary. The bottom four bits are read-only and return 0s when read.
The prefetchable memory base register defines the base address of a prefetchable memory address range used by
the bridge to determine when to forward memory transactions from one interface to the other. The upper 12 bits of
this register are read/write and correspond to the address bits AD31–AD20. The lower 20 address bits are considered
0; thus, the address range is aligned to a 1M-byte boundary. The bottom four bits are read-only and return 0s
when read.
Bit1514131211109876543210
NamePrefetchable memory base
TypeR/WR/WR/WR/WR/WR/WR/WR/WR/WR/WR/WR/WRRRR
Default0000000000000000
Register:Prefetchable memory base
Type:Read-only, Read/Write
Offset:24h
Default:0000h
4–11
4.23 Prefetchable Memory Limit Register
The prefetchable memory limit register defines the upper-limit address of a prefetchable memory address range used
to determine when to forward memory transactions from one interface to the other. The upper 12 bits of this register
are read/write and correspond to the address bits AD31–AD20. The lower 20 address bits are considered 1s; thus,
the address range is aligned to a 1M-byte boundary. The bottom four bits are read-only and return 0s when read.
The prefetchable base upper 32 bits register, plus the prefetchable memory base register, defines the base address
of the 64-bit prefetchable memory address range used by the bridge to determine when to forward memory
transactions from one interface to the other. The prefetchable base upper 32 bits register should be programmed to
all zeros when 32-bit addressing is being used.
Bit31302928272625242322212019181716
NamePrefetchable base upper 32 bits
TypeR/WR/WR/WR/WR/WR/WR/WR/WR/WR/WR/WR/WR/WR/WR/WR/W
Default0000000000000000
Bit1514131211109876543210
NamePrefetchable base upper 32 bits
TypeR/WR/WR/WR/WR/WR/WR/WR/WR/WR/WR/WR/WR/WR/WR/WR/W
Default0000000000000000
4–12
Register:Prefetchable base upper 32 bits
Type:Read/Write
Offset:28h
Default:0000 0000h
4.25 Prefetchable Limit Upper 32 Bits Register
The prefetchable limit upper 32 bits register plus the prefetchable memory limit register defines the base address of
the 64-bit prefetchable memory address range used by the bridge to determine when to forward memory transactions
from one interface to the other. The prefetchable limit upper 32 bits register should be programmed to all zeros when
32-bit addressing is being used.
The I/O base upper 16 bits register specifies the upper 16 bits corresponding to AD31–AD16 of the 32-bit address
that specifies the base of the I/O range to forward from the primary PCI bus to the secondary PCI bus.
Bit1514131211109876543210
NameI/O base upper 16 bits
TypeR/WR/WR/WR/WR/WR/WR/WR/WR/WR/WR/WR/WR/WR/WR/WR/W
Default0000000000000000
Register:I/O base upper 16 bits
Type:Read/Write
Offset:30h
Default:0000h
4.27 I/O Limit Upper 16 Bits Register
The I/O limit upper 16 bits register specifies the upper 16 bits corresponding to AD31–AD16 of the 32-bit address
that specifies the upper limit of the I/O range to forward from the primary PCI bus to the secondary PCI bus.
The capability pointer register provides the pointer to the PCI configuration header where the PCI power management
register block resides. The capability pointer provides access to the first item in the linked list of capabilities. The
capability pointer register is read-only and returns DCh when read, indicating the power management registers are
located at PCI header offset DCh.
The PCI2050 does not implement the expansion ROM remapping feature. The expansion ROM base address
register returns all 0s when read.
Bit31302928272625242322212019181716
NameExpansion ROM base address
TypeRRRRRRRRRRRRRRRR
Default0000000000000000
Bit1514131211109876543210
NameExpansion ROM base address
TypeRRRRRRRRRRRRRRRR
Default0000000000000000
Register:Expansion ROM base address
Type:Read-only
Offset:38h
Default:0000 0000h
4.30 Interrupt Line Register
The interrupt line register is read/write and is used to communicate interrupt line routing information. Since the bridge
does not implement an interrupt signal terminal, this register defaults to 00h.
Bit76543210
NameInterrupt line
TypeR/WR/WR/WR/WR/WR/WR/WR/W
Default00000000
Register:Interrupt line
Type:Read/Write
Offset:3Ch
Default:00h
4–14
4.31 Interrupt Pin Register
The bridge default state does not implement any interrupt terminals. Reads from bits 7–0 of this register return 0s.
The bridge control register provides many of the same controls for the secondary interface that are provided by the
command register for the primary interface. Some bits affect the operation of both interfaces.
Bit1514131211109876543210
NameBridge control
TypeRRRRR/WR/WR/WR/WRR/WR/WRR/WR/WR/WR/W
Default0000000000000000
Register:Bridge control
Type:Read-only, Read/Write
Offset:3Eh
Default:0000h
Table 4–5. Bridge Control Register Description
BITTYPEFUNCTION
15–12RReserved. Bits 15–12 return 0s when read.
Discard timer SERR enable.
11R/W
10R/W
9R/W
8R/W
7R
6R/W
0 = SERR
1 = SERR
Discard timer status. Once set, this bit must be cleared by writing 1 to this bit.
0 = No discard timer error (default)
1 = Discard timer error. Either primary or secondary discard timer expired and a delayed transaction was discarded from
Secondary discard timer. Selects the number of PCI clocks that the bridge will wait for a master on the secondary interface
to repeat a delayed transaction request.
Primary discard timer. Selects the number of PCI clocks that the bridge will wait for a master on the primary interface to
repeat a delayed transaction request.
Fast back-to-back capable. The bridge never generates fast back-to-back transactions to different secondary devices. Bit
7 returns 0 when read.
Secondary bus reset. When bit 6 is set, the secondary reset signal (S_RST) is asserted. S_RST is deasserted by resetting
this bit. Bit 6 is encoded as:
0 = Do not force the assertion of S_RST (default).
1 = Force the assertion of S_RST
signaling disabled for primary discard time-outs (default)
signaling enabled for primary discard time-outs
the queue in the bridge.
.
4–15
Table 4–5. Bridge Control Register Description (continued)
BITTYPEFUNCTION
Master abort mode. Bit 5 controls how the bridge responds to a master abort that occurs on either interface when the bridge
is the master. If this bit is set, the posted write transaction has completed on the requesting interface, and SERR
(bit 8) of the command register (offset 04h, see Section 4.3) is 1, then P_SERR
5R/W
4RReserved. Returns 0 when read. Writes have no effect.
3R/W
2R/W
1R/W
0R/W
If the transaction has not completed, then a target abort is signaled. If the bit is cleared, then all 1s are returned on reads
and write data is accepted and discarded when a transaction that crosses the bridge is terminated with master abort. The
default state of bit 5 after a reset is 0.
0 = Do not report master aborts (return FFFF FFFFh on reads and discard data on writes) (default).
1 = Report master aborts by signaling target abort if possible, or if SERR
asserting SERR
VGA enable. When bit 3 is set, the bridge positively decodes and forwards VGA-compatible memory addresses in the video
frame buffer range 000A 0000h–000BFFFFh, I/O addresses in the range 03B0h–03BBh, and 03C0–03DFh from the
primary to the secondary interface, independent of the I/O and memory address ranges. When this bit is set, the bridge
blocks forwarding of these addresses from the secondary to the primary. Reset clears this bit. Bit 3 is encoded as:
0 = Do not forward VGA-compatible memory and I/O addresses from the primary to the secondary interface
(default).
1 = Forward VGA-compatible memory and I/O addresses from the primary to the secondary, independent of the I/O
and memory address ranges and independent of the ISA enable bit.
ISA enable. When bit 2 is set, the bridge blocks the forwarding of ISA I/O transactions from the primary to the secondary,
addressing the last 768 bytes in each 1K-byte block. This applies only to the addresses (defined by the I/O window registers)
that are located in the first 64K bytes of PCI I/O address space. From the secondary to the primary, I/O transactions are
forwarded if they address the last 768 bytes in each 1K-byte block in the address range specified in the I/O window registers.
Bit 2 is encoded as:
0 = Forward all I/O addresses in the address range defined by the I/O base and I/O limit registers (default).
1 = Block forwarding of ISA I/O addresses in the address range defined by the I/O base and I/O limit registers when
these I/O addresses are in the first 64K bytes of PCI I/O address space and address the top 768 bytes of each
1K-byte block.
SERR enable. Bit 1 controls the forwarding of secondary interface SERR assertions to the primary interface. Only when
this bit is set will the bridge forward S_SERR
bit 8 of the command register (offset 04h, see Section 4.3) must be set.
0 = SERR disabled (default)
1 = SERR
Parity error response enable. Bit 0 controls the bridge response to parity errors on the secondary interface. When this bit
is set, the bridge asserts S_PERR
0 = Ignore address and parity errors on the secondary interface (default).
1 = Enable parity error reporting and detection on the secondary interface.
enabled
.
to the primary bus signal P_SERR. For the primary interface to assert SERR,
to report parity errors on the secondary interface.
is asserted when a master abort occurs.
enable
is enabled via bit 1 of this register, by
4–16
5 Extension Registers
The TI extension registers are those registers that lie outside the standard PCI-to-PCI bridge device configuration
space (i.e., registers 40h–FFh in PCI configuration space in the PCI2050). These registers can be accessed through
configuration reads and writes. The TI extension registers add flexibility and performance benefits to the standard
PCI-to-PCI bridge. Mapping of the extension registers is contained in Table 4–1.
5.1Chip Control Register
The chip control register contains read/write and read-only bits and has a default value of 00h. This register is used
to control the functionality of certain PCI transactions.
Bit76543210
NameChip control
TypeRRR/WR/WRRR/WR
Default00000000
Register:Chip control
Type:Read/Write, Read-only
Offset:40h
Default:00h
Table 5–1. Chip Control Register Description
BITTYPEFUNCTION
7–6RReserved. Bits 7–6 return 0s when read.
Transaction forwarding control for I/O and memory cycles.
5R/W
4R/W
3–2RReserved. Bits 3 and 2 return 0s when read.
1R/W
0RReserved. Bit 0 returns 0 when read.
0 = Transaction forwarding controlled by bits 0 and 1 of the command register (offset 04h, see Section 4.3) (default).
1 = Transaction forwarding will be disabled if GPIO3 is driven high.
Memory read prefetch. When set, bit 4 enables the memory read prefetch.
0 = Upstream memory reads are disabled (default).
1 = Upstream memory reads are enabled
Memory write and memory write and invalidate disconnect control.
0 = Disconnects on queue full or 4-KB boundaries (default)
1 = Disconnects on queue full, 4-KB boundaries and cacheline boundaries.
5–1
5.2Extended Diagnostic Register
The extended diagnostic register is read or write and has a default value of 00h. Bit 0 of this register is used to reset
both the PCI2050 and the secondary bus.
Writing a 1 to this bit causes the PCI2050 to set bit 6 of the bridge control register (offset 3Eh, see Section 4.32) and then
internally reset the PCI2050. Bit 6 of the bridge control register will not be reset by the internal reset. Bit 0 is self-clearing.
5–2
5.3Arbiter Control Register
The arbiter control register is used for the bridge internal arbiter. The arbitration scheme used is a two-tier rotational
arbitration. The PCI2050 bridge is the only secondary bus initiator that defaults to the higher priority arbitration tier.
Bit1514131211109876543210
NameArbiter control
TypeRRRRRRR/WR/WR/WR/WR/WR/WR/WR/WR/WR/W
Default0000001000000000
Register:Arbiter control
Type:Read-only, Read/Write
Offset:42h
Default:0200h
Table 5–3. Arbiter Control Register Description
BITTYPEFUNCTION
15–10RReserved. Bits 15–10 return 0s when read.
Bridge tier select. This bit determines in which tier the PCI2250 bridge is placed in the two-tier arbitration scheme.
Master abort on posted write transactions. When set, bit 4 enables P_SERR reporting on master aborts on posted write
transactions.
0 = Master aborts on posted writes enabled (default)
1 = Master aborts on posted writes disabled
Target abort on posted writes. When set, bit 3 enables P_SERR reporting on target aborts on posted write transactions.
0 = T arget aborts on posted writes enabled (default).
1 = Target aborts on posted writes disabled.
Master posted write time-out.
0 = P_SERR
1 = P_SERR
Posted write parity error.
0 = P_SERR
1 = P_SERR
signaled on a master time-out after 224 retries on a delayed read (default).
is not signaled on a master time-out.
signaled on a master time-out after 224 retries on a delayed write (default).
is not signaled on a master time-out.
signaled on a master time-out after 224 retries on a posted write (default).
is not signaled on a master time-out.
signaled on a posted write parity error (default).
is not signaled on a posted write parity error.
5–4
5.5GPIO Output Data Register
The GPIO output data register controls the data driven on the GPIO terminals configured as outputs. If both an
output-high bit and an output-low bit are set for the same GPIO terminal, the output-low bit takes precedence. The
output data bits have no effect on a GPIO terminal that is programmed as an input.
Bit76543210
NameGPIO output data
TypeR/WR/WR/WR/WR/WR/WR/WR/W
Default00000000
Register:GPIO output data
Type:Read/Write
Offset:65h
Default:00h
Table 5–5. GPIO Output Data Register Description
BITTYPEFUNCTION
7R/WGPIO3 output high. Writing a 1 to this bit causes the GPIO signal to be driven high. Writing a 0 has no ef fect.
6R/WGPIO2 output high. Writing a 1 to this bit causes the GPIO signal to be driven high. Writing a 0 has no ef fect.
5R/WGPIO1 output high. Writing a 1 to this bit causes the GPIO signal to be driven high. Writing a 0 has no ef fect.
4R/WGPIO0 output high. Writing a 1 to this bit causes the GPIO signal to be driven high. Writing a 0 has no ef fect.
3R/WGPIO3 output low. Writing a 1 to this bit causes the GPIO signal to be driven low. Writing a 0 has no ef fect.
2R/WGPIO2 output low. Writing a 1 to this bit causes the GPIO signal to be driven low. Writing a 0 has no ef fect.
1R/WGPIO1 output low. Writing a 1 to this bit causes the GPIO signal to be driven low. Writing a 0 has no ef fect.
0R/WGPIO0 output low. Writing a 1 to this bit causes the GPIO signal to be driven low. Writing a 0 has no ef fect.
5.6GPIO Output Enable Register
The GPIO output enable register controls the direction of the GPIO signal. By default all GPIO terminals are inputs.
If both an output-enable bit and an input-enable bit are set for the same GPIO terminal, the input-enable bit takes
precedence.
7R/WGPIO3 output enable. Writing a 1 to this bit causes the GPIO signal to be configured as an output. W riting a 0 has no effect.
6R/WGPIO2 output enable. Writing a 1 to this bit causes the GPIO signal to be configured as an output. W riting a 0 has no effect.
5R/WGPIO1 output enable. Writing a 1 to this bit causes the GPIO signal to be configured as an output. W riting a 0 has no effect.
4R/WGPIO0 output enable. Writing a 1 to this bit causes the GPIO signal to be configured as an output. W riting a 0 has no effect.
3R/WGPIO3 input enable. Writing a 1 to this bit causes the GPIO signal to be configured as an input. Writing a 0 has no ef fect.
2R/WGPIO3 input enable. Writing a 1 to this bit causes the GPIO signal to be configured as an input. Writing a 0 has no ef fect.
1R/WGPIO3 input enable. Writing a 1 to this bit causes the GPIO signal to be configured as an input. Writing a 0 has no ef fect.
0R/WGPIO3 input enable. Writing a 1 to this bit causes the GPIO signal to be configured as an input. Writing a 0 has no ef fect.
5–5
5.7GPIO Input Data Register
The GPIO input data register returns the current state of the GPIO terminals when read.
Bit76543210
NameGPIO input data
TypeRRRRRRRR
DefaultXXXX0000
Register:GPIO input data
Type:Read-only
Offset:67h
Default:X0h
Table 5–7. GPIO Input Data Register Description
BITTYPEFUNCTION
7–4RGPIO3–GPIO0 input data. These four bits return the current state of the GPIO terminals.
3–0RReserved. Bits 3–0 return 0s when read.
5–6
5.8Secondary Clock Control Register
The secondary clock control register is used to control the secondary clock outputs.
Bit1514131211109876543210
NameSecondary clock control
TypeRR/WR/WR/WR/WR/WR/WR/WR/WR/WR/WR/WR/WR/WR/WR/W
Default0000000000000000
Register:Secondary clock control
Type:Read-only, Read/Write
Offset:68h
Default:0000h
Table 5–8. Secondary Clock Control Register Description
00, 01, 10 = S_CLKOUT3 enabled (00 is the default).
11 = S_CLKOUT3 disabled and driven high.
S_CLKOUT2 disable.
00, 01, 10 = S_CLKOUT2 enabled (00 is the default).
11 = S_CLKOUT2 disabled and driven high.
S_CLKOUT1 disable.
00, 01, 10 = S_CLKOUT1 enabled (00 is the default).
11 = S_CLKOUT1 disabled and driven high.
S_CLKOUT0 disable.
00, 01, 10 = S_CLKOUT0 enabled (00 is the default).
11 = S_CLKOUT0 disabled and driven high.
5–7
5.9P_SERR Status Register
The P_SERR status register indicates what caused a SERR event on the primary interface.
Bit76543210
NameP_SERR status
TypeRR/WR/WR/WR/WR/WR/WR
Default00000000
Register:P_SERR status
Type:Read-only Read/Write
Offset:6Ah
Default:00h
Table 5–9. P_SERR Status Register Description
BITTYPEFUNCTION
7RReserved. Bit 7 returns 0 when read.
6R/W
5R/W
4R/W
3R/WTarget abort on posted writes. A 1 indicates that P_SERR was signaled because of a target abort on a posted write.
2R/W
1R/WPosted write parity error. A 1 indicates that P_SERR was signaled because of parity error on a posted write.
0RReserved. Bit 0 returns 0 when read.
Master delayed read time-out. A 1 indicates that P_SERR was signaled because of a master time-out after 224 retries on
a delayed read.
Master delayed write time-out. A 1 indicates that P_SERR was signaled because of a master time-out after 224 retries on
a delayed write.
Master abort on posted write transactions. A 1 indicates that P_SERR was signaled because of a master abort on a posted
write.
Master posted write time-out. A 1 indicates that P_SERR was signaled because of a master time-out after 224 retries on
a posted write.
5.10 Power-Management Capability ID Register
The power-management capability ID register identifies the linked list item as the register for PCI power management.
The power-management capability ID register returns 01h when read, which is the unique ID assigned by the PCI
SIG for the PCI location of the capabilities pointer and the value.
Bit76543210
NamePower-management capability ID
TypeRRRRRRRR
Default00000001
Register:Power-management capability ID
Type:Read-only
Offset:DCh
Default:01h
5–8
5.11 Power-Management Next-Item Pointer Register
The power-management next-item pointer register is used to indicate the next item in the linked list of PCI
power-management capabilities. The next-item pointer returns E4h in CompactPCI mode, indicating that the
PCI2050 supports more than one extended capability, but in all other modes returns 00h, indicating that only one
extended capability is provided.
The power management capabilities register contains information on the capabilities of the PCI2050 functions related
to power management. The PCI2050 function supports D0, D1, D2, and D3 power states when MS1 is low. The
PCI2050 does not support any power states when MS1 is high.
PME support. This five-bit field indicates the power states that the device supports asserting PME. A 0 for any of these bits
15–11R
10R
9R
8–6RReserved. Bits 8–6 return 0s when read.
5R
4RAuxiliary power source. This bit returns a 0 when read because the PCI2050 does not support PME signaling.
3RPMECLK. This bit returns a 0 when read because the PME signaling is not supported.
2–0R
indicates that the PCI2050 cannot assert PME
read, indicating that PME
D2 support. This bit returns 1 when MS0 is 0, indicating that the bridge function supports the D2 device power state. This
bit returns 0 when MS0 is 1, indicating that the bridge function does not support the D2 device power state.
D1 support. This bit returns 1 when MS0 is 0, indicating that the bridge function supports the D1 device power state. This
bit returns 0 when MS0 is 1, indicating that the bridge function does not support the D1 device power state.
Device specific initialization. This bit returns 0 when read, indicating that the bridge function does not require special
initialization (beyond the standard PCI configuration header) before the generic class device driver is able to use it.
Version. This three-bit register returns the PCI Bus Power Management Interface Specification revision.
from that power state. For the PCI2050, these five bits return 00000b when
5–9
5.13 Power-Management Control/Status Register
The power-management control/status register determines and changes the current power state of the PCI2050. The
contents of this register are not affected by the internally generated reset caused by the transition from D3
state.
15RPME status. This bit returns a 0 when read because the PCI2050 does not support PME.
14–13R
12–9R
8RPME enable. This bit returns a 0 when read because the PCI2050 does not support PME signaling.
7–2RReserved. Bits 7–2 return 0s when read.
1–0R/W
Data scale. This 2-bit read-only field indicates the scaling factor to be used when interpreting the value of the data
register. These bits return only 00b, because the data register is not implemented.
Data select. This 4-bit field is used to select which data is to be reported through the data register and data-scale
field. These bits return only 0000b, because the data register is not implemented.
Power state. This 2-bit field is used both to determine the current power state of a function and to set the function
into a new power state. The definition of this is given below:
00 – D0
01 – D1
10 – D2
11 – D3
hot
hot
to D0
5–10
5.14 PMCSR Bridge Support Register
The PMCSR bridge support register is required for all PCI bridges and supports PCI-bridge-specific functionality.
Bit76543210
Name PMCSR bridge support
TypeRRRRRRRR
DefaultXX000000
Register:PMCSR bridge support
Type:Read-only
Offset:E2h
Default:X0h
Table 5–12. PMCSR Bridge Support Register Description
BITTYPEFUNCTION
Bus power control enable. This bit returns the value of the MS1/BCC input.
7R
6R
5–0RReserved.
0 = Bus power/ clock control disabled.
1 = Bus power/clock control enabled.
B2/B3 support for D3
are stopped when the device is placed in D3
states.
Note: If the primary clock is stopped, then the secondary clocks will stop because the primary clock is used to
generate the secondary clocks.
. This bit returns the value of MS1/BCC input. When this bit is 1, the secondary clocks
hot
. When this bit is 0, the secondary clocks remain on in all device
hot
5.15 Data Register
The data register is an optional, 8-bit read-only register that provides a mechanism for the function to report
state-dependent operating data such as power consumed or heat dissipation. The PCI2050 does not implement the
data register.
The HS capability ID register identifies the linked list item as the register for cPCI hot-swap capabilities. The register
returns 06h when read, which is the unique ID assigned by the PICMG for PCI location of the capabilities pointer and
the value. In Intel-compatible mode, this register is read-only and defaults to 00h.
Bit76543210
NameHS capability ID
TypeRRRRRRRR
Default00000110
Register:HS capability ID
Type:Read-only
Offset:E4h
Default:06h TI mode
00h Intel-compatible mode
5.17 HS Next-Item Pointer Register
The HS next-item pointer register is used to indicate the next item in the linked list of cPCI hot swap capabilities.
Because this is the last extended capability that the PCI2050 supports, the next-item pointer returns all 0s.
The hot-swap control status register contains control and status information for cPCI hot swap resources.
Bit76543210
NameHot swap control status
TypeRRRRR/WRR/WR
Default00000000
Register:Hot-swap control status
Type:Read-only, Read/Write
Offset:E6h
Default:00h
Table 5–13. Hot-Swap Control Status Register Description
BITTYPEFUNCTION
ENUM insertion status. When set, the ENUM output is driven by the PCI2050. This bit defaults to 0, and will be set after
7R
6R
5–4RReserved. Bits 5 and 4 return 0s when read.
3R/W
2RReserved. Bit 2 returns 0 when read.
1R/W
0RReserved. Bit 0 returns 0 when read.
a PCI reset occurs, the pre-load of serial ROM is complete, the ejector handle is closed, and bit 6 is 0. Thus, this bit is set
following an insertion when the board implementing the PCI2050 is ready for configuration. This bit cannot be set under
software control.
ENUM extraction status. When set, the ENUM output is driven by the PCI2050. This bit defaults to 0, and is set when the
ejector handle is opened and bit 7 is 0. Thus, this bit is set when the board implementing the PCI2050 is about to be removed.
This bit cannot be set under software control.
LED ON/OFF . This bit defaults to 0, and controls the external LED indicator (HSLED) under normal conditions. However,
for a duration following a PCI_RST
is interpreted, a 1 causes HSLED high and a 0 causes HSLED low.
Following PCI_RST
conditions are met, the HSLED is under software control via this bit.
ENUM interrupt mask. This bit allows the HSENUM output to be masked by software. Bits 6 and 7 are set independently
from this bit.
0 = Enable HSENUM output
1 = Mask HSENUM
, the HSLED output is driven high by the PCI2050 until the ejector handle is closed. When these
output
, the HSLED output is driven high by the PCI2050 and this bit is ignored. When this bit
5–13
6 Electrical Characteristics
†
†
§
6.1Absolute Maximum Ratings Over Operating Temperature Ranges
Storage temperature range, T
Virtual junction temperature, T
†
Stresses beyond those listed under absolute maximum ratings may cause permanent damage to the device. These are stress ratings only, and
functional operation of the device at these or any other conditions beyond those indicated under recommended operating conditions is not implied.
Exposure to absolute-maximum-rated conditions for extended periods may affect device reliability.
NOTES: 1. Applies for external input and bidirectional buffers. VI > VCC does not apply to fail-safe terminals.
2. Applies to external output and bidirectional buffers. VO > VCC does not apply to fail-safe terminals.