®
80C286
High Performance Microprocessor with Memory Management and Protection
Description
The Intersil 80C286 is a static CMOS version of the NMOS 80286 microprocessor. The 80C286 is an advanced, highperformance microprocessor with specially optimized capabilities for multiple user and multi-tasking systems. The 80C286 has built-in memory protection that supports operating system and task isolation as well as program and data privacy within tasks. A 25MHz 80C286 provides up to nineteen times the throughput of a standard 5MHz 8086. The 80C286 includes memory management capabilities that map 230 (one gigabyte) of virtual address space per task into 224 bytes (16 megabytes) of physical memory. The 80C286 is upwardly compatible with 80C86 and 80C88 software (the 80C286 instruction set is a superset of the 80C86/80C88 instruction set). Using the 80C286 real address mode, the 80C286 is object code compatible with existing 80C86 and 80C88 software. In protected virtual address mode, the 80C286 is source code compatible with 80C86 and 80C88 software but may require upgrading to use virtual address as supported by the 80C286’s integrated memory management and protection mechanism. Both modes operate at full 80C286 performance and execute a superset of the 80C86 and 80C88 instructions. The 80C286 provides special operations to support the efficient implementation and execution of operating systems. For example, one instruction can end execution of one task, save its state, switch to a new task, load its state, and start execution of the new task. The 80C286 also supports virtual memory systems by providing a segment-not-present exception and restartable instructions.
March 1997
Features
• Compatible with NMOS 80286 • Wide Range of Clock Rates - DC to 25MHz (80C286-25) - DC to 20MHz (80C286-20) - DC to 16MHz (80C286-16) - DC to 12.5MHz (80C286-12) - DC to 10MHz (80C286-10) • Static CMOS Design for Low Power Operation - ICCSB = 5mA Maximum - ICCOP = 185mA Maximum (80C286-10) 220mA Maximum (80C286-12) 260mA Maximum (80C286-16) 310mA Maximum (80C286-20) 410mA Maximum (80C286-25) • High Performance Processor (Up to 19 Times the 8086 Throughput) • Large Address Space • 16 Megabytes Physical/1 Gigabyte Virtual per Task • Integrated Memory Management, Four-Level Memory Protection and Support for Virtual Memory and Operating Systems • Two 80C86 Upward Compatible Operating Modes - 80C286 Real Address Mode - PVAM • Compatible with 80287 Numeric Data Co-Processor • High Bandwidth Bus Interface (25 Megabyte/Sec) • Available In - 68 Pin PGA (Commercial, Industrial, and Military) - 68 Pin PLCC (Commercial and Industrial)
Ordering Information
PACKAGE PGA TEMP. RANGE 0oC to +70oC -40oC to +85oC -55oC to +125oC PLCC 0oC to +70oC -40oC to +85oC 10MHz IG80C286-10 59629067801MXC IS80C286-10 12.5MHz CG80C286-12 IG80C286-12 59629067802MXC CS80C286-12 IS80C286-12 16MHz CG80C286-16 CS80C286-16 IS80C286-16 20MHz CG80C286-20 CS80C286-20 IS80C286-20 25MHz CS80C286-25 PKG. NO. G68.B G68.B G68.B N68.95 N68.95
CAUTION: These devices are sensitive to electrostatic discharge; follow proper IC Handling Procedures. 1-888-INTERSIL or 321-724-7143 | Intersil (and design) is a registered trademark of Intersil Americas Inc. Copyright © Intersil Americas Inc. 2003. All Rights Reserved 1 All other trademarks mentioned are the property of their respective owners.
FN2947.2
80C286 Pinouts
68 LEAD PGA Component Pad View - As viewed from underside of the component when mounted on the board.
VSS D10 D12 D13 D14 49 50 D7 D11 D15 51 53 55 57 59 61 63 65 67 16 15 A15 14 13 A17 12 11 A19 10 9 A21 8 7 A22 6 5 PEACK 4 3 S1 2 1 NC 52 54 56 58 60 62 64 66 68 ERROR NC INTR NMI PEREQ READY HLDA M/IO NC NC BUSY NC NC VSS VCC HOLD COD/INTA LOCK ERROR D8 D1 D9 D2
D0
D3
D4
D5 45 46
35 A0 A2 VCC A3 A5 A7 A9 A11 A13 D0 A1 CLK RESET A4 A6 A8 A10 A12 34 32 30 28 26 24 22 20 18 36 33 31 29 27 25 23 21 19 17 A12
37 38
39 40
41 42
43 44
47 48
D6
PIN 1 INDICATOR
VSS
S0
NC
68 LEAD PGA P.C. Board View - As viewed from the component side of the P.C. board.
VSS 35 36 33 31 29 27 25 23 21 4 3 S1 6 5 PEACK 8 7 A22 10 9 A21 12 11 A19 14 13 A17 16 15 A15 19 17 A12 34 32 30 28 26 24 22 20 18 D0 A1 CLK RESET A4 A6 A8 A10 A12 A0 A2 VCC A3 A5 A7 A9 A11 A13 A14 D0 D15 D14 D13 D12 D11 D10 D9 D2 39 40 A18 D8 A16 D1 37 38
ERROR
D7
D6
D5
D4
51 NC BUSY NC NC VSS VCC HOLD COD/INTA LOCK ERROR NC INTR NMI PEREQ READY HLDA M/IO NC 52 54 56 58 60 62 64 66 68 53 55 57 59 61 63 65 67 2 1 NC
49 50
47 48
45 46
43 44
PIN 1 INDICATOR
S0
A23
BHE
2
VSS
A20
NC
D3
41 42
BHE
A14
A16
A18
A20
A23
80C286 Pinouts
(Continued) 68 LEAD PLCC P.C. Board View - As viewed from the component side of the P.C. board.
LOCK M/IO COD/INTA HLDA HOLD READY VCC PEREQ VSS NMI NC INTR NC NC BUSY ERROR NC 68 67 66 65 64 63 62 61 60 59 58 57 56 55 54 53 52 BHE NC NC S1 S0 PEACK A23 A22 VSS A21 A20 A19 A18 A17 A16 A15 A14 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 A13 A12 A11 A10 A9 A8 A7 A6 A5 A4 A3 RESET VCC CLK A2 A1 A0 51 50 49 48 47 46 45 44 43 42 41 40 39 38 37 36 35 D15 D7 D14 D6 D13 D5 D12 D4 D11 D3 D10 D2 D9 D1 D8 D0 VSS
Functional Diagram
ADDRESS UNIT (AU)
ADDRESS LATCHES AND DRIVERS PHYSICAL ADDRESS ADDER PREFETCHER PROCESSOR EXTENSION INTERFACE
A23 - A0, BHE, M/IO PEACK PEREQ READY, HOLD, S1, S0, COD/INTA, LOCK, HLDA D15 - D0
SEGMENT BASES OFFSET ADDER SEGMENT LIMIT SEGMENT CHECKER SIZES
BUS CONTROL
DATA TRANSCEIVERS 6-BYTE PREFETCH QUEUE
ALU
BUS UNIT (BU) RESET
REGISTERS CONTROL
3 DECODED INSTRUCTION INSTRUCTION DECODER QUEUE
INSTRUCTION UNIT (IU)
CLK VSS VCC
EXECUTION UNIT (EU)
NMI
BUSY
INTR ERROR
3
80C286 Pin Descriptions
SYMBOL CLK PIN NUMBER 31 The following pin function descriptions are for the 80C286 microprocessor.
TYPE I
DESCRIPTION SYSTEM CLOCK: provides the fundamental timing for the 80C286 system. It is divided by two inside the 80C286 to generate the processor clock. The internal divide-by-two circuitry can be synchronized to an external clock generator by a LOW to HIGH transition on the RESET input. DATA BUS: inputs data during memory, I/O, and interrupt acknowledge read cycles; outputs data during memory and I/O write cycles. The data bus is active HIGH and is held at high impedance to the last valid logic level during bus hold acknowledge. ADDRESS BUS: outputs physical memory and I/O port addresses. A23 - A16 are LOW during I/O transfers. A0 is LOW when data is to be transferred on pins D7 - D0 (see table below). The address bus is active High and floats to three-state off during bus hold acknowledge. BUS HIGH ENABLE: indicates transfer of data on the upper byte of the data bus, D15 - D8. Eight-bit oriented devices assigned to the upper byte of the data bus would normally use BHE to condition chip select functions. BHE is active LOW and floats to three-state OFF during bus hold acknowledge. BHE AND A0 ENCODINGS BHE VALUE 0 0 1 1 A0 VALUE 0 1 0 1 Word transfer Byte transfer on upper half of data bus (D15 - D8) Byte transfer on lower half of data bus (D7 - D0) Reserved FUNCTION
D15 - D0
36 - 51
I/O
A23 - A0
7-8 10 - 28 32 - 43 1
O
BHE
O
S1, S0
4, 5
O
BUS CYCLE STATUS: indicates initiation of a bus cycle and along with M/IO and COD/lNTA, defines the type of bus cycle. The bus is in a TS state whenever one or both are LOW. S1 and S0 are active LOW and are held at a high impedance logic one during bus hold acknowledge. 80C286 BUS CYCLE STATUS DEFINITION COD/INTA 0(LOW) 0 0 0 0 0 0 0 1(HIGH) 1 1 1 1 1 1 1 M/IO 0 0 0 0 1 1 1 1 0 0 0 0 1 1 1 1 S1 0 0 1 1 0 0 1 1 0 0 1 1 0 0 1 1 S0 0 1 0 1 0 1 0 1 0 1 0 1 0 1 0 1 BUS CYCLE INITIATED Interrupt acknowledge Reserved Reserved None; not a status cycle If A1 = 1 then halt; else shutdown Memory data read Memory data write None; not a status cycle Reserved I/O read I/O write None; not a status cycle Reserved Memory instruction read Reserved None; not a status cycle
4
80C286 Pin Descriptions
SYMBOL M/IO PIN NUMBER 67 The following pin function descriptions are for the 80C286 microprocessor. (Continued)
TYPE O
DESCRIPTION MEMORY I/O SELECT: distinguishes memory access from I/O access. If HIGH during TS, a memory cycle or a halt/shutdown cycle is in progress. If LOW, an I/O cycle or an interrupt acknowledge cycle is in progress. M/IO is held at high impedance to the last valid logic state during bus hold acknowledge. CODE/INTERRUPT ACKNOWLEDGE: distinguishes instruction fetch cycles from memory data read cycles. Also distinguishes interrupt acknowledge cycles from I/O cycles. COD/lNTA is held at high impedance to the last valid logic state during bus hold acknowledge. Its timing is the same as M/IO. BUS LOCK: indicates that other system bus masters are not to gain control of the system bus for the current and following bus cycles. The LOCK signal may be activated explicitly by the “LOCK” instruction prefix or automatically by 80C286 hardware during memory XCHG instructions, interrupt acknowledge, or descriptor table access. LOCK is active LOW and is held at a high impedance logic one during bus hold acknowledge. BUS READY: terminates a bus cycle. Bus cycles are extended without limit until terminated by READY LOW. READY is an active LOW synchronous input requiring setup and hold times relative to the system clock be met for correct operation. READY is ignored during bus hold acknowledge. (See Note 1) BUS HOLD REQUEST AND HOLD ACKNOWLEDGE: control ownership of the 80C286 local bus. The HOLD input allows another local bus master to request control of the local bus. When control is granted, the 80C286 will float its bus drivers and then activate HLDA, thus entering the bus hold acknowledge condition. The local bus will remain granted to the requesting master until HOLD becomes inactive which results in the 80C286 deactivating HLDA and regaining control of the local bus. This terminates the bus hold acknowledge condition. HOLD may be asynchronous to the system clock. These signals are active HIGH. Note that HLDA never floats. INTERRUPT REQUEST: requires the 80C286 to suspend its current program execution and service a pending external request. Interrupt requests are masked whenever the interrupt enable bit in the flag word is cleared. When the 80C286 responds to an interrupt request, it performs two interrupt acknowledge bus cycles to read an 8-bit interrupt vector that identifies the source of the interrupt. To ensure program interruption, INTR must remain active until an interrupt acknowledge bus cycle is initiated. INTR is sampled at the beginning of each processor cycle and must be active HIGH at least two processor cycles before the current instruction ends in order to interrupt before the next instruction. INTR is level sensitive, active HIGH, and may be asynchronous to the system clock. NON-MASKABLE INTERRUPT REQUEST: interrupts the 80C286 with an internally supplied vector value of two. No interrupt acknowledge cycles are performed. The interrupt enable bit in the 80C286 flag word does not affect this input. The NMI input is active HIGH, may be asynchronous to the system clock, and is edge triggered after internal synchronization. For proper recognition, the input must have been previously LOW for at least four system clock cycles and remain HIGH for at least four system clock cycles. PROCESSOR EXTENSION OPERAND REQUEST AND ACKNOWLEDGE: extend the memory management and protection capabilities of the 80C286 to processor extensions. The PEREQ input requests the 80C286 to perform a data operand transfer for a processor extension. The PEACK output signals the processor extension when the requested operand is being transferred. PEREQ is active HIGH. PEACK is active LOW and is held at a high impedance logic one during bus hold acknowledge. PEREQ may be asynchronous to the system clock. PROCESSOR EXTENSION BUSY AND ERROR: indicates the operating condition of a processor extension to the 80C286. An active BUSY input stops 80C286 program execution on WAIT and some ESC instructions until BUSY becomes inactive (HIGH). The 80C286 may be interrupted while waiting for BUSY to become inactive. An active ERROR input causes the 80C286 to perform a processor extension interrupt when executing WAIT or some ESC instructions. These inputs are active LOW and may be asynchronous to the system clock.
COD/lNTA
66
O
LOCK
68
O
READY
63
l
HOLD HLDA
64 65
I O
INTR
57
I
NMI
59
l
PEREQ PEACK
61 6
l O
BUSY ERROR
54 53
l I
5
80C286 Pin Descriptions
SYMBOL RESET PIN NUMBER 29 The following pin function descriptions are for the 80C286 microprocessor. (Continued)
TYPE l
DESCRIPTION SYSTEM RESET: clears the internal logic of the 80C286 and is active HIGH. The 80C286 may be reinitialize at any time with a LOW to HIGH transition on RESET which remains active for more than 16 system clock cycles. During RESET active, the output pins of the 80C286 enter the state shown below. 80C286 PIN STATE DURING RESET PIN VALUE 1 (HIGH) 0 (LOW) HIGH IMPEDANCE PIN NAMES S0, S1, PEACK, A23 - A0, BHE, LOCK M/IO, COD/lNTA, HLDA (Note 2) D15 - D0
Operation of the 80C286 begins after a HlGH to LOW transition on RESET. The HIGH to LOW transition of RESET must be synchronous to the system clock. Approximately 50 system clock cycles are required by the 80C286 for internal initializations before the first bus cycle to fetch code from the power-on execution address is performed. A LOW to HIGH transition of RESET synchronous to the system clock will end a processor cycle at the second HIGH to LOW transition of the system clock. The LOW to HIGH transition of RESET may be asynchronous to the system clock; however, in this case it cannot be predetermined which phase of the processor clock will occur during the next system clock period. Synchronous LOW to HIGH transitions of RESET are required only for systems where the processor clock must be phase synchronous to another clock. VSS VCC NOTES: 1. READY is an open-collector signal and should be pulled inactive with an appropriate resistor (620Ω at 10MHz and 12.5MHz, 470Ω at 16MHz, 390Ω at 20MHz, 270Ω at 25MHz). 2. HLDA is only Low if HOLD is inactive (Low). 3. All unused inputs should be pulled to their inactive state with pull up/down resistors. 9, 35, 60 30, 62 l l SYSTEM GROUND: are the ground pins (all must be connected to system ground). SYSTEM POWER: +5V power supply pins. A 0.1µF capacitor between pins 60 and 62 is recommended.
Functional Description
Introduction The Intersil 80C286 microprocessor is a static CMOS version of the NMOS 80286 microprocessor. The 80C286 is an advanced, high-performance microprocessor with specially optimized capabilities for multiple user and multi-tasking systems. Depending on the application, the 80C286's performance is up to nineteen times faster than the standard 5MHz 8086's, while providing complete upward software compatibility with Intersil 80C86 and 80C88 CPU family. The 80C286 operates in two modes: 80C286 real address mode and protected virtual address mode. Both modes execute a superset of the 80C86 and 80C88 instruction set. In 80C286 real address mode programs use real addresses with up to one megabyte of address space. Programs use virtual addresses in protected virtual address mode, also called protected mode. In protected mode, the 80C286 CPU automatically maps 1 gigabyte of virtual addresses per task into a 16 megabyte real address space. This mode also provides memory protection to isolate the operating system and ensure privacy of each tasks' programs and data. Both modes provide the same base instruction set, registers and addressing modes. The Functional Description describes the following: Static operation, the base 80C286 architecture common to both modes, 80C286 real address mode, and finally, protected mode. Static Operation The 80C286 is comprised of completely static circuitry. Internal registers, counters, and latches are static and require no refresh as with dynamic circuit design. This eliminates the minimum operating frequency restriction typically placed on microprocessors. The CMOS 80C286 can operate from DC to the specified upper frequency limit. The clock to the processor may be stopped at any point (either phase one or phase two of the processor clock cycle) and held there indefinitely. There is, however, a significant decrease in power requirement if the clock is stopped in phase two of the processor clock cycle. Details on the clock relationships will be discussed in the Bus Operation section. The ability to stop the clock to the processor is especially useful for system debug or power critical applications.
6
80C286
The 80C286 can be single-stepped using only the CPU clock. This state can be maintained as long as necessary. Single step clock information allows simple interface circuitry to provide critical information for system debug. Static design also allows very low frequency operation (down to DC). In a power critical situation, this can provide low power operation since 80C286 power dissipation is directly related to operating frequency. As the system frequency is reduced, so is the operating power until, ultimately, with the clock stopped in phase two of the processor clock cycle, the 80C286 power requirement is the standby current (5mA maximum).
16-BIT REGISTER NAME 7 AX BYTE ADDRESSABLE (8-BIT REGISTER NAMES SHOWN) DX CX BX BP SI AH DH CH BH SPECIAL REGISTER FUNCTIONS 07 AL DL CL BL BASE REGISTERS 0 MULTIPLY/DIVIDE I/O INSTRUCTIONS LOOP/SHIFT/REPEAT COUNT
80C286 Base Architecture
DI
INDEX REGISTERS
The 80C86, 80C88, and 80C286 CPU family all contain the same basic set of registers, instructions, and addressing modes. The 80C286 processor is upwardly compatible with the 80C86 and 80C88 CPU's. Register Set The 80C286 base architecture has fifteen registers as shown in Figure 1. These registers are grouped into the following four categories. GENERAL REGISTERS: Eight 16-bit general purpose registers used to contain arithmetic and logical operands. Four of these (AX, BX, CX and DX) can be used either in their entirety as 16-bit words or split into pairs of separate 8-bit registers. SEGMENT REGISTERS: Four 16-bit special purpose registers select, at any given time, the segments of memory that are immediately addressable for code, stack and data. (For usage, refer to Memory Organization.) BASE AND INDEX REGISTERS: Four of the general purpose registers may also be used to determine offset addresses of operands in memory. These registers may contain base addresses or indexes to particular locations within a segment. The addressing mode determines the specific registers used for operand address calculations. STATUS AND CONTROL REGISTERS: Three 16-bit special purpose registers record or control certain aspects of the 80C286 processor state. These include the Flags register and Machine Status Word register shown in Figure 2, and the Instruction Pointer, which contains the offset address of the next sequential instruction to be executed.
SP 15 GENERAL REGISTERS 0
STACK POINTER
15 CS DS SS ES SEGMENT REGISTERS 15 F IP MSW
0 CODE SEGMENT SELECTOR DATA SEGMENT SELECTOR STACK SEGMENT SELECTOR EXTRA SEGMENT SELECTOR
0 FLAGS INSTRUCTION POINTER MACHINE STATUS WORD
STATUS AND CONTROL REGISTERS
FIGURE 1. REGISTER SET
Flags Word Description The Flags word (Flags) records specific characteristics of the result of logical and arithmetic instructions (bits 0, 2, 4, 6, 7 and 11) and controls the operation of the 80C286 within a given operating mode (bits 8 and 9). Flags is a 16-bit register. The function of the flag bits is given in Table 1.
7
80C286
STATUS FLAGS: CARRY PARITY AUXILIARY CARRY ZERO SIGN OVERFLOW 15 FLAGS: 14 NT 13 IOPL 12 11 OF 10 DF 9 IF 8 TF 7 SF 6 ZF 5 4 AF 3 2 PF 1 CF 0
CONTROL FLAGS: TRAP FLAG INTERRUPT ENABLE DIRECTION FLAG
SPECIAL FIELDS: I/O PRIVILEGE LEVEL NESTED TASK FLAG 15 MSW: 3 TS 2 EM 1 MP 0 PE
RESERVED
TASK SWITCH PROCESSOR EXTENSION EMULATED MONITOR PROCESSOR EXTENSION PROTECTION ENABLE
FIGURE 2. STATUS AND CONTROL REGISTER BIT FUNCTIONS
TABLE 1. FLAGS WORD BIT FUNCTIONS BIT POSITION 0 2 4 6 7 11 NAME CF PF AF ZF SF OF FUNCTION Carry Flag - Set on high-order bit carry or borrow; cleared otherwise. Parity Flag - Set if low-order 8 bits of result contain an even number of 1 bits; cleared otherwise. Set on carry from or borrow to the low order four bits of AL; cleared otherwise. Zero Flag - Set if result is zero; cleared otherwise. Sign Flag - Set equal to high-order bit of result (0 if positive, 1 if negative). Overflow Flag - Set if result is a too-large positive number or a too-small negative number (excluding sign-bit) to fit in destination operand; cleared otherwise. Single Step Flag - Once set, a single step interrupt occurs after the next instruction executes. TF is cleared by the single step interrupt. Interrupt-Enable Flag - When set, maskable interrupts will cause the CPU to transfer control to an interrupt vector specified location. Direction Flag - Causes string instructions to auto decrement the appropriate index registers when set. Clearing DF causes auto increment.
8
TF
9
IF
10
DF
8
80C286
Instruction Set The instruction set is divided into seven categories: data transfer, arithmetic, string manipulation, shift/rotate/logical, high level, processor control and control transfer instructions. These categories are summarized in Table 2. An 80C286 instruction can reference zero, one, or two operands; where an operand may reside in a register, in the instruction itself, or in memory. Zero-operand instructions (e.g. NOP and HLT) are usually one byte long. One-operand instructions (e.g. INC and DEC) are usually two bytes long but some are encoded in only one byte. One-operand instructions may reference a register or memory location. Two-operand instructions permit the following six types of instruction operations: • Register to Register • Memory to Register • Immediate to Register • Memory to Memory • Register to Memory • Immediate to Memory
ADDITION ADD ADC INC AAA DAA SUBTRACTION SUB SBB DEC NEG CMP AAS DAS MULTIPLICATION MUL lMUL AAM DIVISION DlV lDlV AAD CBW CWD Divide byte or word unsigned Integer divide byte or word ASClI adjust for division Convert byte to word Convert word to doubleword Multiply byte or word unsigned Integer multiply byte or word ASClI adjust for multiply Subtract byte or word Subtract byte or word with borrow Decrement byte or word by 1 Negate byte or word Compare byte or word ASClI adjust for subtraction Decimal adjust for subtraction Add byte or word Add byte or word with carry Increment byte or word by 1 ASClI adjust for addition Decimal adjust for addition TABLE 2B. ARITHMETIC INSTRUCTIONS
Two-operand instructions (e.g. MOV and ADD) are usually three to six bytes long. Memory to memory operations are provided by a special class of string instructions requiring one to three bytes. For detailed instruction formats and encodings refer to the instruction set summary at the end of this document.
TABLE 2A. DATA TRANSFER INSTRUCTIONS GENERAL PURPOSE MOV PUSH POP PUSHA POPA XCHG XLAT INPUT/OUTPUT IN OUT Input byte or word Output byte or word Move byte or word Push word onto stack Pop word off stack Push all registers on stack Pop all registers from stack Exchange byte or word Translate byte
TABLE 2C. STRING INSTRUCTIONS MOVS INS OUTS CMPS SCAS LODS STOS Load AH register from flags Store AH register in flags Push flags onto stack Pop flags off stack REP REPE/REPZ REPNE/REPNZ Move byte or word string Input bytes or word string Output bytes or word string Compare byte or word string Scan byte or word string Load byte or word string Store byte or word string Repeat Repeat while equal/zero Repeat while not equal/not zero
ADDRESS OBJECT LEA LDS LES FLAG TRANSFER LAHF SAHF PUSHF POPF Load effective address Load pointer using DS Load pointer using ES
9
80C286
TABLE 2D. SHIFT/ROTATE LOGICAL INSTRUCTIONS LOGICALS NOT AND OR XOR TEST SHIFTS SHL/SAL SHR SAR ROTATES ROL ROR RCL RCR Rotate left byte or word Rotate right byte or word Rotate through carry left byte or word Rotate through carry right byte or word Shift logical/arithmetic left byte or word Shift logical right byte or word Shift arithmetic right byte or word “Not” byte or word “And” byte or word “Inclusive or” byte or word “Exclusive or” byte or word “Test” byte or word TABLE 2F. PROCESSOR CONTROL INSTRUCTIONS FLAG OPERATIONS STC CLC CMC STD CLD STl CLl Set carry flag Clear carry flag Complement carry flag Set direction flag Clear direction flag Set interrupt enable flag Clear interrupt enable flag
EXTERNAL SYNCHRONIZATION HLT WAIT ESC LOCK NO OPERATION NOP No operation Halt until interrupt or reset Wait for TEST pin active Escape to extension processor Lock bus during next instruction
EXECUTION ENVIRONMENT CONTROL TABLE 2E. HIGH LEVEL INSTRUCTIONS ENTER LEAVE BOUND Format stack for procedure entry Restore stack for procedure exit Detects values outside prescribed range TABLE 2G. PROGRAM TRANSFER INSTRUCTIONS CONDITIONAL TRANSFERS JA/JNBE JAE/JNB JB/JNAE JBE/JNA JC JE/JZ JG/JNLE JGE/JNL JL/JNGE JLE/JNG JNC JNE/JNZ JNO JNP/JPO JNS JO JP/JPE JS Jump if above/not below nor equal Jump if above or equal/not below Jump if below/not above nor equal Jump if below or equal/not above Jump if carry Jump if equal/zero Jump if greater/not less nor equal Jump if greater or equal/not less Jump if less/not greater nor equal Jump if less or equal/not greater Jump if not carry Jump if not equal/not zero Jump if not overflow Jump if not parity/parity odd Jump if not sign Jump if overflow Jump if parity/parity even Jump if sign INTO lRET Interrupt if overflow Interrupt return INTERRUPTS INT Interrupt LOOPE/LOOPZ LOOPNE/LOOPNZ JCXZ Loop if equal/zero Loop if not equal/not zero Jump if register CX = 0 ITERATION CONTROLS LOOP Loop UNCONDITIONAL TRANSFERS CALL RET JMP Call procedure Return from procedure Jump LMSW SMSW Load machine status word Store machine status word
10
80C286
Memory Organization Memory is organized as sets of variable-length segments. Each segment is a linear contiguous sequence of up to 64K (216) 8bit bytes. Memory is addressed using a two-component address (a pointer) that consists of a 16-bit segment selector and a 16-bit offset. The segment selector indicates the desired segment in memory. The offset component indicates the desired byte address within the segment. (See Figure 3). All instructions that address operands in memory must specify the segment and the offset. For speed and compact instruction encoding, segment selectors are usually stored in the high speed segment registers. An instruction need specify only the desired segment register and offset in order to address a memory operand.
POINTER SEGMENT 31 16 15 OFFSET 0 OPERAND SELECTED SELECTED SEGMENT
Addressing Modes The 80C286 provides a total of eight addressing modes for instructions to specify operands. Two addressing modes are provided for instructions that operate on register or immediate operands: REGISTER OPERAND MODE: The operand is located in one of the 8 or 16-bit general registers. IMMEDIATE OPERAND MODE: The operand is included in the instruction. Six modes are provided to specify the location of an operand in a memory segment. A memory operand address consists of two 16-bit components: segment selector and offset. The segment selector is supplied by a segment register either implicitly chosen by the addressing mode or explicitly chosen by a segment override prefix. The offset is calculated by summing any combination of the following three address elements: the displacement (an 8 or 16-bit immediate value contained in the instruction) the base (contents of either the BX or BP base registers) the index (contents of either the SI or Dl index registers)
MEMORY
FIGURE 3. TWO COMPONENT ADDRESS
CODE MODULE A DATA
Most instructions need not explicitly specify which segment register is used. The correct segment register is automatically chosen according to the rules of Table 3. These rules follow the way programs are written (see Figure 4) as independent modules that require areas for code and data, a stack, and access to external data areas. Special segment override instruction prefixes allow the implicit segment register selection rules to be overridden for special cases. The stack, data and extra segments may coincide for simple programs. To access operands not residing in one of the four immediately available segments, a full 32-bit pointer or a new segment selector must be loaded.
TABLE 3. SEGMENT REGISTER SELECTION RULES MEMORY REFERENCE NEEDED Instructions Stack SEGMENT REGISTER USED Code (CS)
CODE MODULE B DATA
CPU CODE DATA STACK
PROCESS STACK
EXTRA SEGMENT REGISTERS
IMPLICIT SEGMENT SELECTION RULE Automatic with instruction prefetch
PROCESS DATA BLOCK 1
Stack (SS) All stack pushes and pops. Any memory reference which uses BP as a base register. Data (DS) All data references except when relative to stack or string destination Alternate data segment and destination of string operation
PROCESS DATA BLOCK 2
Local Data
MEMORY
External (Global) Data
Extra (ES)
FIGURE 4. SEGMENTED MEMORY HELPS STRUCTURE SOFTWARE
Any carry out from the 16-bit addition is ignored. Eight-bit displacements are sign extended to 16-bit values.
11
80C286
Combinations of these three address elements define the six memory addressing modes, described below. DIRECT MODE: The operand's offset is contained in the instruction as an 8 or 16-bit displacement element. REGISTER INDIRECT MODE: The operand's offset is in one of the registers SI, Dl, BX or BP. BASED MODE: The operand's offset is the sum of an 8 or 16-bit displacement and the contents of a base register (BX or BP). INDEXED MODE: The operand's offset is the sum of an 8 or 16bit displacement and the contents of an index register (SI or Dl). BASED INDEXED MODE: The operand's offset is the sum of the contents of a base register and an index register. BASED INDEXED MODE WITH DISPLACEMENT: The operand's offset is the sum of a base register's contents, an index register's contents, and an 8 or 16-bit displacement. Data Types The 80C286 directly supports the following data types: Integer: A signed binary numeric value contained in an 8bit byte or a 16-bit word. All operations assume a 2's complement representation. Signed 32 and 64-bit integers are supported using the 80287 Numeric Data Processor. An unsigned binary numeric value contained in an 8-bit byte or 16-bit word. A 32-bit quantity, composed of a segment selector component and an offset component. Each component is a 16-bit word. A contiguous sequence of bytes or words. A string may contain from 1 byte to 64K bytes. A byte representation of alphanumeric and control characters using the ASClI standard of character representation. A byte (unpacked) representation of the decimal digits 0-9. A byte (packed) representation of two decimal digits 0-9 storing one digit in each nibble of the byte. A signed 32, 64 or 80-bit real number representation. (Floating point operands are supported using the 80287 Numeric Processor extension).
STRING PACKED BCD SIGNED WORD SIGN BIT SIGNED 31 DOUBLE WORD (NOTE) SIGN BIT +7 MSB MAGNITUDE +3 +2 16 15 +1 0 0 7 SIGNED BYTE SIGN BIT MAGNITUDE 7 UNSIGNED BYTE MSB MAGNITUDE 15 14 +1 87 0 0 0 0
MSB MAGNITUDE
SIGNED 63 QUAD WORD (NOTE) SIGN BIT
+6 +5 48 47
+4 +3 32 31
+2 +1 16 15
0 0
MSB MAGNITUDE +1 0 0
15 UNSIGNED WORD MSB MAGNITUDE BINARY 7 CODED DECIMAL (BCD) +N 0 ••• BCD DIGIT N +N 0 ••• ASCII CHARACTERN 7 +N 0 •••
Ordinal: Pointer:
7
+1
07
0
0
BCD DIGIT 1 7 +1 07
BCD DIGIT 0 0 0
String: ASClI:
7 ASCII
ASCII CHARACTER1 7 +1 07
ASCII CHARACTER0 0 0
BCD: Packed BCD: Floating Point:
MOST SIGNIFICANT DIGIT 7/15 +N 0 ••• 7/15
LEAST SIGNIFICANT DIGIT +1 0 7/15 0 0
BYTE/WORD N BYTE/WORD 1 31 POINTER SELECTOR 79 +9 FLOATING POINT (NOTE) SIGN BIT EXPONENT +8 +7 +6 +5 +3 +1 16 15 +1
BYTE/WORD 0 0 0
Figure 5 graphically represents the data types supported by the 80C286.
OFFSET +4 +3 +2 +1 00
MAGNITUDE
FIGURE 5. 80C286 SUPPORTED DATA TYPES NOTE: Supported by 80C286/80C287 Numeric Data Processor Configuration
12
80C286
TABLE 4. INTERRUPT VECTOR ASSIGNMENTS DOES RETURN ADDRESS POINT TO INSTRUCTION CAUSING EXCEPTION? Yes
FUNCTION Divide Error Exception Single Step Interrupt NMI Interrupt Breakpoint Interrupt INTO Detected Overflow Exception BOUND Range Exceeded Exception Invalid Opcode Exception Processor Extension Not Available Exception Reserved - Do Not Use Processor Extension Error Interrupt Reserved User Defined
INTERRUPT NUMBER 0 1 2 3 4 s 6 7 8 - 15 16 17 - 31 32 - 255
RELATED INSTRUCTIONS DlV, lDlV All INT 2 or NMI Pin INT 3 INTO BOUND Any Undefined Opcode ESC or WAIT
No Yes Yes Yes
ESC or WAIT
I/O Space The I/O space consists of 64K 8-bit ports, 32K 16-bit ports, or a combination of the two. I/O instructions address the I/O space with either an 8-bit port address, specified in the instruction, or a 16-bit port address in the DX register. 8-bit port addresses are zero extended such that A15-A8 are LOW. I/O port addresses 00F8(H) through 00FF(H) are reserved. Interrupts An interrupt transfers execution to a new program location. The old program address (CS:lP) and machine state (Flags) are saved on the stack to allow resumption of the interrupted program. Interrupts fall into three classes: hardware initiated, INT instructions, and instruction exceptions. Hardware initiated interrupts occur in response to an external input and are classified as non-maskable or maskable. Programs may cause an interrupt with an INT instruction. Instruction exceptions occur when an unusual condition which prevents further instruction processing is detected while attempting to execute an instruction. The return address from an exception will always point to the instruction causing the exception and include any leading instruction prefixes. A table containing up to 256 pointers defines the proper interrupt service routine for each interrupt. Interrupts 0-31, some of which are used for instruction exceptions, are reserved. For each interrupt, an 8-bit vector must be supplied to the 80C286 which identifies the appropriate table entry. Exceptions supply the interrupt vector internally. INT instructions contain or imply the vector and allow access to all 256 interrupts. Maskable hardware initiated interrupts supply the 8-bit vector to the CPU during an interrupt acknowledge bus sequence. Nonmaskable hardware interrupts use a predefined internally supplied vector.
Maskable Interrupt (INTR) The 80C286 provides a maskable hardware interrupt request pin, INTR. Software enables this input by setting the interrupt flag bit (IF) in the flag word. All 224 user-defined interrupt sources can share this input, yet they can retain separate interrupt handlers. An 8-bit vector read by the CPU during the interrupt acknowledge sequence (discussed in System Interface section) identifies the source of the interrupt. The processor automatically disables further maskable interrupts internally by resetting the IF as part of the response to an interrupt or exception. The saved flag word will reflect the enable status of the processor prior to the interrupt. Until the flag word is restored to the flag register, the interrupt flag will be zero unless specifically set. The interrupt return instruction includes restoring the flag word, thereby restoring the original status of IF. Non-Maskable Interrupt Request (NMI) A non-maskable interrupt input (NMI) is also provided. NMI has higher priority than INTR. A typical use of NMI would be to activate a power failure routine. The activation of this input causes an interrupt with an internally supplied vector value of 2. No external interrupt acknowledge sequence is performed. While executing the NMI servicing procedure, the 80C286 will service neither further NMI requests, INTR requests, nor the processor extension segment overrun interrupt until an interrupt return (lRET) instruction is executed or the CPU is reset. If NMI occurs while currently servicing an NMI, its presence will be saved for servicing after executing the first IRET instruction. IF is cleared at the beginning of an NMI interrupt to inhibit INTR interrupts.
13
80C286
Single Step Interrupt The 80C286 has an internal interrupt that allows programs to execute one instruction at a time. It is called the single step interrupt and is controlled by the single step flag bit (TF) in the flag word. Once this bit is set, an internal single step interrupt will occur after the next instruction has been executed. The interrupt clears the TF bit and uses an internally supplied vector of 1. The lRET instruction is used to set the TF bit and transfer control to the next instruction to be single stepped. Interrupt Priorities When simultaneous interrupt requests occur, they are processed in a fixed order as shown in Table 5. Interrupt processing involves saving the flags, return address, and setting CS:lP to point at the first instruction of the interrupt handler. If another enabled interrupt should occur, it is processed before the next instruction of the current interrupt handler is executed. The last interrupt processed is therefore the first one serviced.
TABLE 5. INTERRUPT PROCESSING ORDER ORDER 1 2 3 4 5 6 INTERRUPT Instruction Exception Single Step NMI Processor Extension Segment Overrun INTR INT Instruction 3 TS
Machine Status Word Description The machine status word (MSW) records when a task switch takes place and controls the operating mode of the 80C286. It is a 16-bit register of which the lower four bits are used. One bit places the CPU into protected mode, while the other three bits, as shown in Table 7, control the processor extension interface. After RESET, this register contains FFF0(H) which places the 80C286 in 80C286 real address mode.
TABLE 7. MSW BIT FUNCTIONS BIT POSITION 0
NAME PE
FUNCTION Protected mode enable places the 80C286 into protected mode and cannot be cleared except by RESET. Monitor processor extension allows WAIT instructions to cause a processor extension not present exception (number 7). Emulate processor extension causes a processor extension not present exception (number 7) on ESC instructions to allow emulating a processor extension. Task switched indicates the next instruction using a processor extension will cause exception 7, allowing software to test whether the current processor extension context belongs to the current task.
1
MP
2
EM
The LMSW and SMSW instructions can load and store the MSW in real address mode. The recommended use of TS, EM, and MP is shown in Table 8. Halt The HLT instruction stops program execution and prevents the CPU from using the local bus until restarted. Either NMI, INTR with IF = 1, or RESET will force the 80C286 out of halt. If interrupted, the saved CS:IP will point to the next instruction after the HLT.
Initialization and Processor Reset Processor initialization or start up is accomplished by driving the RESET input pin HIGH. RESET forces the 80C286 to terminate all execution and local bus activity. No instruction or bus activity will occur as long as RESET is active. After RESET becomes inactive, and an internal processing interval elapses, the 80C286 begins execution in real address mode with the instruction at physical location FFFFF0(H). RESET also sets some registers to predefined values as shown in Table 6.
TABLE 6. 80C286 INITIAL REGISTER STATE AFTER RESET Flag Word Machine Status Word Instruction Pointer Code Segment Data Segment Extra Segment Stack Segment 0002(H) FFF0(H) FFF0(H) F000(H) 0000(H) 0000(H) 0000(H)
HOLD must not be active during the time from the leading edge of the initial RESET to 34 CLKs after the trailing edge of the initial RESET of an 80C286 system.
14
80C286
TABLE 8. RECOMMENDED MSW ENCODINGS FOR PROCESSOR EXTENSION CONTROL INSTRUCTION CAUSING EXCEPTION 7 None ESC ESC None ESC or WAIT
TS 0 0 1 0 1
MP 0 0 0 1 1
EM 0 1 1 0 0
RECOMMENDED USE Initial encoding after RESET. 80C286 operation is identical to 80C86/88. No processor extension is available. Software will emulate its function. No processor extension is available. Software will emulate its function. The current processor extension context may belong to another task. A processor extension exists. A processor extension exists. The current processor extension context may belong to another task. The exception 7 on WAIT allows software to test for an error pending from a previous processor extension operation.
TABLE 9. REAL ADDRESS MODE ADDRESSING INTERRUPTS INTERRUPT NUMBER 8 9 13 RETURN ADDRESS BEFORE INSTRUCTION Yes No Yes
FUNCTION Interrupt table limit too small exception Processor extension segment overrun interrupt Segment overrun exception
RELATED INSTRUCTIONS INT vector is not within table limit ESC with memory operand extending beyond offset FFFF(H) Word memory reference with offset = FFFF(H) or an attempt to execute past the end of a segment
80C286 Real Address Mode
The 80C286 executes a fully upward-compatible superset of the 80C86 instruction set in real address mode. In real address mode the 80C286 is object code compatible with 80C86 and 80C88 software. The real address mode architecture (registers and addressing modes) is exactly as described in the 80C286 Base Architecture section of this Functional Description. Memory Size Physical memory is a contiguous array of up to 1,048,576 bytes (one megabyte) addressed by pins A0 through A19 and BHE. A20 through A23 should be ignored. Memory Addressing In real address mode physical memory is a contiguous array of up to 1,048,576 bytes (one megabyte) addressed by pin A0 through A19 and BHE. Address bits A20-A23 may not always be zero in real mode. A20-A23 should not be used by the system while the 80C286 is operating in Real Mode. The selector portion of a pointer is interpreted as the upper 16-bits of a 20-bit segment address. The lower four bits of the 20-bit segment address are always zero. Segment addresses, therefore, begin on multiples of 16 bytes. See Figure 6 for a graphic representation of address information. All segments in real address mode are 64K bytes in size and may be read, written, or executed. An exception or interrupt can occur if data operands or instructions attempt to wrap around the end of a segment (e.g. a word with its low order byte at offset FFFF(H) and its high order byte at offset 0000(H)). If, in real address mode, the information contained
19 20-BIT PHYSICAL MEMORY ADDRESS 15 SEGMENT SELECTOR 0 0000 SEGMENT ADDRESS
in a segment does not use the full 64K bytes, the unused end of the segment may be overlaid by another segment to reduce physical memory requirements.
15 0000 OFFSET 0 OFFSET ADDRESS
ADDER
0
FIGURE 6. 80C286 REAL ADDRESS MODE ADDRESS CALCULATION
15
80C286
Reserved Memory Locations The 80C286 reserves two fixed areas of memory in real address mode (see Figure 7); system initialization area and interrupt table area. Locations from addresses FFFF0(H) through FFFFF(H) are reserved for system initialization. Initial execution begins at location FFFF0(H). Locations 00000(H) through 003FF(H) are reserved for interrupt vectors. Protected Mode Initialization To prepare the 80C286 for protected mode, the LIDT instruction is used to load the 24-bit interrupt table base and 16-bit limit for the protected mode interrupt table. This instruction can also set a base and limit for the interrupt vector table in real address mode. After reset, the interrupt table base is initialized to 000000(H) and its size set to 03FF(H). These values are compatible with 80C86 and 80C88 software. LIDT should only be executed in preparation for protected mode. Shutdown Shutdown occurs when a severe error is detected that prevents further instruction processing by the CPU. Shutdown and halt are externally signalled via a halt bus operation. They can be distinguished by A1 HIGH for halt and A1 LOW for shutdown. In real address mode, shutdown can occur under two conditions: • Exceptions 8 or 13 happen and the IDT limit does not include the interrupt vector. • A CALL INT or PUSH instruction attempts to wrap around the stack segment when SP is not even. An NMI input can bring the CPU out of shutdown if the IDT limit is at least 000F(H) and SP is greater than 0005(H), otherwise shutdown can only be exited via the RESET input.
RESET BOOTSTRAP PROGRAM JUMP • • • INTERRUPT POINTER FOR VECTOR 255 • • • INTERRUPT POINTER FOR VECTOR 1 INTERRUPT POINTER FOR VECTOR 0
FFFFFH FFFF0H
3FFH 3FCH
7H 4H 3H 0H
INITIAL CS:IP VALUE IS F000:FFF0
FIGURE 7. 80C286 REAL ADDRESS MODE INITIALLY RESERVED MEMORY LOCATIONS
Interrupts Table 9 shows the interrupt vectors reserved for exceptions and interrupts which indicate an addressing error. The exceptions leave the CPU in the state existing before attempting to execute the failing instruction (except for PUSH, POP, PUSHA, or POPA). Refer to the next section on protected mode initialization for a discussion on exception 8.
16
80C286 Protected Virtual Address Mode
The 80C286 executes a fully upward-compatible superset of the 80C86 instruction set in protected virtual address mode (protected mode). Protected mode also provides memory management and protection mechanisms and associated instructions. The 80C286 enters protected virtual address mode from real address mode by setting the PE (Protection Enable) bit of the machine status word with the Load Machine Status Word (LMSW) instruction. Protected mode offers extended physical and virtual memory address space, memory protection mechanisms, and new operations to support operating systems and virtual memory. All registers, instructions, and addressing modes described in the 80C286 Base Architecture section of this Functional Description remain the same. Programs for the 80C86, 80C88, and real address mode 80C286 can be run in protected mode; however, embedded constants for segment selectors are different. Memory Size The protected mode 80C286 provides a 1 gigabyte virtual address space per task mapped into a 16 megabyte physical address space defined by the address pins A23-A0 and BHE. The virtual address space may be larger than the physical address space since any use of an address that does not map to a physical memory location will cause a restartable exception.
CPU 31 16 15 0 ACCESS RIGHTS BYTE
from the tables in memory. The 16-bit offset is added to the segment base address to form the physical address as shown in Figure 8. The tables are automatically referenced by the CPU whenever a segment register is loaded with a selector. All 80C286 instructions which load a segment register will reference the memory based tables without additional software. The memory based tables contain 8 byte values called descriptors. Descriptors Descriptors define the use of memory. Special types of descriptors also define new functions for transfer of control and task switching. The 80C286 has segment descriptors for code, stack and data segments, and system control descriptors for special system data segments and control transfer operations. Descriptor accesses are performed as locked bus operations to assure descriptor integrity in multi-processor systems. Code and Data Segment Descriptors (S = 1) Besides segment base addresses, code and data descriptors contain other segment attributes including segment size (1 to 64K bytes), access rights (read only, read/write, execute only, and execute/read), and presence in memory (for virtual memory systems) (See Table 10). Any segment usage violating a segment attribute indicated by the segment descriptor will prevent the memory cycle and cause an exception or interrupt.
7 +7 +5 P DPL S TYPE A +3 +1 15 07 RESERVED † BASE 23 - 16 0 +6 +4 +2 0 0
POINTER SELECTOR OFFSET PHYSICAL MEMORY
BASE 15 - 0 LIMIT 15 - 0 87
† MUST BE SET TO 0 FOR COMPATIBILITY WITH FUTURE UPGRADES
FIGURE 9. CODE OR DATA SEGMENT DESCRIPTOR
PHYSICAL ADDRESS ADDER MEMORY OPERAND SEGMENT
SEGMENT BASE ADDRESS 23 0
SEGMENT DESCRIPTOR
FIGURE 8. PROTECTED MODE MEMORY ADDRESSING
Memory Addressing As in real address mode, protected mode uses 32-bit pointers, consisting of 16-bit selector and offset components. The selector, however, specifies an index into a memory resident table rather than the upper 16-bits of a real memory address. The 24-bit base address of the desired segment is obtained
Code and data (including stack data) are stored in two types of segments: code segments and data segments. Both types are identified and defined by segment descriptors (S = 1). Code segments are identified by the executable (E) bit set to 1 in the descriptor access rights byte. The access rights byte of both code and data segment descriptor types have three fields in common: present (P) bit, Descriptor Privilege Level (DPL), and accessed (A) bit. If P = 0, any attempted use of this segment will cause a not-present exception. DPL specifies the privilege level of the segment descriptor. DPL controls when the descriptor may be used by a task (refer to privilege discussion below). The A bit shows whether the segment has been previously accessed for usage profiling, a necessity for virtual memory systems. The CPU will always set this bit when accessing the descriptor.
SEGMENT DESCRIPTION TABLE
17
80C286
TABLE 10. CODE AND DATA SEGMENT DESCRIPTOR FORMATS - ACCESS RIGHTS BYTE DEFINITION BIT POSITION 7
NAME Present (P) P=1 P=0
FUNCTION Segment is mapped into physical memory. No mapping to physical memory exits, base and limit are not used. Segment privilege attribute used in privilege tests. Code or Data (includes stacks) segment descriptor System Segment Descriptor or Gate Descriptor Data segment descriptor type is: Expand up segment, offsets must be ≤ limit. Expand down segment, offsets must be > limit. Data segment may not be written into. Data segment may be written into. Code Segment Descriptor type is: Code segment may only be executed when CPL ≥ DPL and CPL remains unchanged. Code segment may not be read. Code segment may be read. Segment has not been accessed. Segment selector has been loaded into segment register or used by selector test instructions. If Data Segment (S = 1, E = 0)
6-5 4
Descriptor Privilege Level (DPL) Segment Descriptor (S) S = 1 S=0
3 2
Executable (E) Expansion Direction (ED) Writable (W)
E=0 ED = 0 ED = 1 W=0 W=1
1
Type Field Definition
3 2 1
Executable (E) Conforming (C) Readable (R)
E=1 C=1 R=0 R=1
If Code Segment (S = 1, E = 1)
0
Accessed (A)
A=0 A=1
18
80C286
Data segments (S = 1, E = 0) may be either read-only or readwrite as controlled by the W bit of the access rights byte. Read-only (W = 0) data segments may not be written into. Data segments may grow in two directions, as determined by the Expansion Direction (ED) bit: upwards (ED = 0) for data segments, and downwards (ED = 1) for a segment containing a stack. The limit field for a data segment descriptor is interpreted differently depending on the ED bit (see Table 10). A code segment (S = 1, E = 1) may be execute-only or execute/read as determined by the Readable (R) bit. Code segments may never be written into and execute-only code segments (R = 0) may not be read. A code segment may also have an attribute called conforming (C). A conforming code segment may be shared by programs that execute at different privilege levels. The DPL of a conforming code segment defines the range of privilege levels at which the segment may be executed (refer to privilege discussion below). The limit field identifies the last byte of a code segment. System Segment Descriptors (S = 0, Type = 1-3) In addition to code and data segment descriptors, the protected mode 80C286 defines System Segment Descriptors. These descriptors define special system data segments which contain a table of descriptors (Local Descriptor Table Descriptor) or segments which contain the execution state of a task (Task State Segment Descriptor). Table 11 gives the formats for the special system data segment descriptors. The descriptors contain a 24-bit base address of the segment and a 16-bit limit. The access byte defines the type of descriptor, its state and privilege level. The descriptor contents are valid and the segment is in physical memory if P = 1. If P = 0, the segment is not valid. The DPL field is only used in Task State Segment descriptors and indicates the privilege level at which the descriptor may be used (see Privilege). Since the Local Descriptor Table descriptor may only be used by a special privileged instruction, the DPL field is not used. Bit 4 of the access byte is 0 to indicate that it is a system control descriptor. The type field specifies the descriptor type as indicated in Table 11.
7 +7 +5 P DPL 0 +3 +1 15 07 RESERVED † TYPE BASE 23 - 16 0 +6 +4 +2 0 0
TABLE 11. SYSTEM SEGMENT DESCRIPTOR FORMAT FIELDS NAME TYPE VALUE 1 2 3 P 0 1 DPL BASE 0-3 24-Bit Number 16-Bit Number DESCRIPTION Available Task State Segment (TSS) Local Descriptor Table Busy Task State Segment (TSS) Descriptor contents are not valid Descriptor contents are valid Descriptor Privilege Level Base Address of special system data segment in real memory Offset of last byte in segment
LIMIT
Gate Descriptors (S = 0, Type = 4-7) Gates are used to control access to entry points within the target code segment. The gate descriptors are call gates, task gates, interrupt gates and trap gates. Gates provide a level of indirection between the source and destination of the control transfer. This indirection allows the CPU to automatically perform protection checks and control entry point of the destination. Call gates are used to change privilege levels (see Privilege), task gates are used to perform a task switch, and interrupt and trap gates are used to specify interrupt service routines. The interrupt gate disables interrupts (resets IF) while the trap gate does not. Table 12 shows the format of the gate descriptors. The descriptor contains a destination pointer that points to the descriptor of the target segment and the entry point offset. The destination selector in an interrupt gate, trap gate, and call gate must refer to a code segment descriptor. These gate descriptors contain the entry point to prevent a program from constructing and using an illegal entry point. Task gates may only refer to a task state segment. Since task gates invoke a task switch, the destination offset is not used in the task gate. Exception 13 is generated when the gate is used if a destination selector does not refer to the correct descriptor type. The word count field is used in the call gate descriptor to indicate the number of parameters (0-31 words) to be automatically copied from the caller’s stack to the stack of the called routine when a control transfer changes privilege levels. The word count field is not used by any other gate descriptor. The access byte format is the same for all descriptors. P = 1 indicates that the gate contents are valid. P = 0 indicates the contents are not valid and causes exception 11 if referenced. DPL is the descriptor privilege level and specifies when this descriptor may be used by a task (refer to privilege discussion below). Bit 4 must equal 0 to indicate a system control descriptor. The type field specifies the descriptor type as indicated in Table 12.
BASE 15 - 0 LIMIT 15 - 0 87
† MUST BE SET TO 0 FOR COMPATIBILITY WITH FUTURE UPGRADES
FIGURE 10. SYSTEM SEGMENT DESCRIPTOR
19
80C286
Segment Descriptor Cache Registers A segment descriptor cache register is assigned to each of the four segment registers (CS, SS, DS, ES). Segment descriptors are automatically loaded (cached) into a segment descriptor cache register (Figure 12) whenever the associated segment register is loaded with a selector. Only segment descriptors may be loaded into segment descriptor cache registers. Once loaded, all references to that segment of memory use the cached descriptor information instead of reaccessing the descriptor. The descriptor cache registers are not visible to programs. No instructions exist to store their contents. They only change when a segment register is loaded.
7 +7 +5 P DPL 0 +3 +1 07 RESERVED † TYPE 0 +6 PROGRAM VISIBLE SEGMENT SELECTORS CS DS SS ES 15 0
SEGMENT REGISTERS (LOADED BY PROGRAM) PROGRAM INVISIBLE ACCESS RIGHTS SEGMENT PHYSICAL BASE ADDRESS SEGMENT SIZE
X X X WORD COUNT +4 4-0 X X +2 DESTINATION SELECTOR 15 - 0 DESTINATION OFFSET 15 - 0 87 0 0
47
40 39
16 15
0
SEGMENT DESCRIPTOR CACHE REGISTERS (AUTOMATICALLY LOADED BY CPU)
15 † MUST BE SET TO 0 FOR COMPATIBILITY WITH FUTURE UPGRADES
FIGURE 12. DESCRIPTOR CACHE REGISTERS
SELECTOR INDEX TI RPL 21 0
FIGURE 11. GATE DESCRIPTOR TABLE 12. GATE DESCRIPTOR FORMAT FIELD NAME TYPE VALUE 4 5 6 7 P 0 1 DPL WORD COUNT 0-3 0 - 31 Call Gate Task Gate Interrupt Gate Trap Gate Descriptor Contents are not valid Descriptor Contents are valid Descriptor Privilege Level Number of words to copy from callers stack to called procedures stack. Only used with call gate. 15 - 3 Index 2 DESCRIPTION BITS 1-0
15
8
7
NAME Requested Privilege Level (RPL) Table Indicator (TI)
FUNCTION Indicates Selector Privilege Level Desired TI = 0 Use Global Descriptor Table (GDT) TI = 1 Use Local Descriptor Table (LDT) Select Descriptor Entry In Table
FIGURE 13. SELECTOR FIELDS
Local and Global Descriptor Tables Two tables of descriptors, called descriptor tables, contain all descriptors accessible by a task at any given time. A descriptor table is a linear array of up to 8192 descriptors. The upper 13 bits of the selector value are an index into a descriptor table. Each table has a 24-bit base register to locate the descriptor table in physical memory and a 16-bit limit register that confine descriptor access to the defined limits of the table as shown in Figure 14. A restartable exception (13) will occur if an attempt is made to reference a descriptor outside the table limits. One table, called the Global Descriptor table (GDT), contains descriptors available to all tasks. The other table, called the Local Descriptor Table (LDT), contains descriptors that can be private to a task. Each task may have its own private LDT. The GDT may contain all descriptor types except interrupt and trap descriptors. The LDT may contain only segment, task gate, and call gate descriptors. A segment cannot be accessed by a task if its segment descriptor does not exist in either descriptor table at the time of access.
DESTINATION 16-Bit Selector to the target code segment SELECTOR Selector (call, interrupt or selector Trap Gate). Selector to the target task state segment (Task Gate). DESTINATION OFFSET 16-Bit Offset Entry point within the target code segment
Selector Fields A protected mode selector has three fields: descriptor entry index, local or global descriptor table indicator (TI), and selector privilege (RPL) as shown in Figure 13. These fields select one of two memory based tables of descriptors, select the appropriate table entry and allow high-speed testing of the selector's privilege attribute (refer to privilege discussion below).
20
80C286
CPU 15 23 GDT LIMIT 0 GDT MEMORY MEMORY GATE FOR INTERRUPT #n GATE FOR INTERRUPT #n-1 CPU 0 LDT DESCR SELECTOR 15 23 LDT LIMIT 0 CURRENT LDT 23 15 LDT1 IDT LIMIT GATE FOR INTERRUPT #0 IDT BASE 0 0 GATE FOR INTERRUPT #1
GDT BASE 24-BIT PHYS AD 15
INTERRUPT DESCRIPTOR TABLE (IDT) INCREASING MEMORY ADDRESS PL = 3 PL = 2
LDT BASE 24-BIT PHYS AD PROGRAM INVISIBLE (AUTOMATICALLY LOADED FROM LDT DESCR WITHIN GDT) LDTn INCREASING MEMORY ADDRESS
FIGURE 16. INTERRUPT DESCRIPTOR TABLE DEFINITION
Privilege The 80C286 has a four-level hierarchical privilege system which controls the use of privileged instructions and access to descriptors (and their associated segments) within a task. Four-level privilege, as shown in Figure 17, is an extension of the users/supervisor mode commonly found in minicomputers. The privilege levels are numbered 0 through 3. Level 0 is the most privileged level. Privilege levels provide protection within a task. (Tasks are isolated by providing private LDT’s for each task.) Operating system routines, interrupt handlers, and other system software can be included and protected within the virtual address space of each task using the four levels of privilege. Each task in the system has a separate stack for each of its privilege levels. Tasks, descriptors, and selectors have a privilege level attribute that determines whether the descriptor may be used. Task privilege affects the use of instructions and descriptors. Descriptor and selector privilege only affect access to the descriptor.
CPU ENFORCED SOFTWARE INTERFACES
FIGURE 14. LOCAL AND GLOBAL DESCRIPTOR TABLE DEFINITION
The LGDT and LLDT instructions load the base and limit of the global and local descriptor tables. LGDT and LLDT are privileged, i.e. they may only be executed by trusted programs operating at level 0. The LGDT instruction loads a six byte field containing the 16-bit table limit and 24-bit physical base address of the Global Descriptor Table as shown in Figure 15. The LDT instruction loads a selector which refers to a Local Descriptor Table descriptor containing the base address and limit for an LDT, as shown in Table 11.
7 +5 +3 +1 15 RESERVED † 07 BASE 23 - 16 BASE 15 - 0 LIMIT 15 - 0 87 0 0 +4 +2 0
APPLICATIONS
† MUST BE SET TO 0 FOR COMPATIBILITY WITH FUTURE UPGRADES
OS EXTENSIONS SYSTEM SERVICES PL = 1 KERNAL PL = 0 MOST PRIVILEGED
FIGURE 15. GLOBAL DESCRIPTOR TABLE AND INTERRUPT DESCRlPTOR TABLE DATA TYPE
Interrupt Descriptor Table The protected mode 80C286 has a third descriptor table, called the Interrupt Descriptor Table (IDT) (see Figure 16), used to define up to 256 interrupts. It may contain only task gates, interrupt gates and trap gates. The IDT (Interrupt Descriptor Table) has a 24-bit physical base and 16-bit limit register in the CPU. The privileged LlDT instruction loads these registers with a six byte value of identical form to that of the LGDT instruction (see Figure 16 and Protected Mode lnitialization). References to IDT entries are made via INT instructions, external interrupt vectors, or exceptions. The IDT must be at least 256 bytes in size to allocate space for all reserved interrupts.
HIGH SPEED OPERATING SYSTEM INTERFACE
NOTE: PL becomes numerically lower as privilege level increases. FIGURE 17. HIERARCHICAL PRIVILEGE LEVELS
21
80C286
Task Privilege A task always executes at one of the four privilege levels. The task privilege level at any specific instant is called the Current Privilege Level (CPL) and is defined by the lower two bits of the CS register. CPL cannot change during execution in a single code segment. A task's CPL may only be changed by control transfers through gate descriptors to a new code segment (See Control Transfer). Tasks begin executing at the CPL value specified by the code segment selector within TSS when the task is initiated via a task switch operation (See Figure 18). A task executing at Level 0 can access all data segments defined in the GDT and the task's LDT and is considered the most trusted level. A task executing a Level 3 has the most restricted access to data and is considered the least trusted level. Descriptor Privilege Descriptor privilege is specified by the Descriptor Privilege Level (DPL) field of the descriptor access byte. DPL specifies the least trusted task privilege level (CPL) at which a task may access the descriptor. Descriptors with DPL = 0 are the most protected. Only tasks executing at privilege level 0 (CPL = 0) may access them. Descriptors with DPL = 3 are the least protected (i.e. have the least restricted access) since tasks can access them when CPL = 0, 1, 2, or 3). This rule applies to all descriptors, except LDT descriptors. Selector Privilege Selector privilege is specified by the Requested Privilege Level (RPL) field in the least significant two bits of a selector. Selector RPL may establish a less trusted privilege level than the current privilege level for the use of a selector. This level is called the task's effective privilege level (EPL). RPL can only reduce the scope of a task's access to data with this selector. A task's effective privilege is the numeric maximum of RPL and CPL. A selector with RPL = 0 imposes no additional restriction on its use while a selector with RPL = 3 can only refer to segments at privilege Level 3 regardless of the task's CPL. RPL is generally used to verify that pointer parameters passed to a more trusted procedure are not allowed to use data at a more privileged level than the caller (refer to pointer testing instructions). Descriptor Access and Privilege Validation Determining the ability of a task to access a segment involves the type of segment to be accessed, the instruction used, the type of descriptor used and CPL, RPL, and DPL. The two basic types of segment accesses are control transfer (selectors loaded into CS) and data (selectors loaded into DS, ES or SS). Data Segment Access Instructions that load selectors into DS and ES must refer to a data segment descriptor or readable code segment descriptor. The CPL of the task and the RPL of the selector must be the same as or more privileged (numerically equal to or lower than) than the descriptor DPL. In general, a task can only access data segments at the same or less privileged levels than the CPL or RPL (whichever is numerically higher) to prevent a program from accessing data it cannot be trusted to use. An exception to the rule is a readable conforming code segment. This type of code segment can be read from any privilege level. If the privilege checks fail (e.g. DPL is numerically less than the maximum of CPL and RPL) or an incorrect type of descriptor is referenced (e.g. gate descriptor or execute only code segment) exception 13 occurs. If the segment is not present, exception 11 is generated. Instructions that load selectors into SS must refer to data segment descriptors for writable data segments. The descriptor privilege (DPL) and RPL must equal CPL. All other descriptor types or a privilege level violation will cause exception 13. A not present fault causes exception 12.
TABLE 13. DESCRlPTOR TYPES USED FOR CONTROL TRANSFER CONTROL TRANSFER TYPES Intersegment within the same privilege levels Intersegment to the same or higher privilege level interrupt within task may change CPL OPERATION TYPES JMP, CALL, RET, lRET (Note 4) CALL Interrupt Instruction, Exception External Interrupt DESCRIPTOR REFERENCED Code Segment Call Gate Trap or Interrupt Gate Code Segment Task State Segment Task Gate Task Gate DESCRIPTOR TABLE GDT/LDT GDT/LDT lDT GDT/LDT GDT GDT/LDT IDT
Intersegment to a lower privilege level (changes task CPL) RET, IRET (Note 4) Task Switch CALL, JMP CALL, JMP lRET (Note 5) Interrupt Instruction, Exception External Interrupt NOTES: 4. NT (Nested Task bit of flag word) = 0 5. NT (Nested Task bit of flag word) = 1
22
80C286
Control Transfer Four types of control transfer can occur when a selector is loaded into CS by a control transfer operation (see Table 13). Each transfer type can only occur if the operation which loaded the selector references the correct descriptor type. Any violation of these descriptor usage rules (e.g. JMP through a call gate or RET to a Task State Segment) will cause exception 13. The ability to reference a descriptor for control transfer is also subject to rules of privilege. A CALL or JUMP instruction may only reference a code segment descriptor with DPL equal to the task CPL or a conforming segment with DPL of equal or greater privilege than CPL. The RPL of the selector used to reference the code descriptor must have as much privilege as CPL. RET and IRET instructions may only reference code segment descriptors with descriptor privilege equal to or less privileged than the task CPL. The selector loaded into CS is the return address from the stack. After the return, the selector RPL is the task's new CPL. If CPL changes, the old stack pointer is popped after the return address. When a JMP or CALL references a Task State Segment descriptor, the descriptor DPL must be the same or less privileged than the task's CPL. Reference to a valid Task State Segment descriptor causes a task switch (see Task Switch Operation). Reference to a Task State Segment descriptor at a more privileged level than the task's CPL generates exception 13. When an instruction or interrupt references a gate descriptor, the gate DPL must have the same or less privilege than the task CPL. If DPL is at a more privileged level than CPL, exception 13 occurs. If the destination selector contained in the gate references a code segment descriptor, the code segment descriptor DPL must be the same or more privileged than the task CPL. If not, Exception 13 is issued. After the control transfer, the code segment descriptors DPL is the task's new CPL. If the destination selector in the gate references a task state segment, a task switch is automatically performed (see Task Switch Operation). The privilege rules on control transfer require: • JMP or CALL direct to a code segment (code segment descriptor) can only be a conforming segment with DPL of equal or greater privilege than CPL or a non-conforming segment at the same privilege level. • Interrupts within the task, or calls that may change privilege levels, can only transfer control through a gate at the same or a less privileged level than CPL to a code segment at the same or more privileged level than CPL. • Return instructions that don't switch tasks can only return control to a code segment at the same or less privileged level. • Task switch can be performed by a call, jump or interrupt which references either a task gate or task state segment at the same or less privileged level. Privilege Level Changes Any control transfer that changes CPL within the task, causes a change of stacks as part of the operation. Initial values of SS:SP for privilege levels 0, 1, and 2 are kept in the task state segment (refer to Task Switch Operation). During a JMP or CALL control transfer, the new stack pointer is loaded into the SS and SP registers and the previous stack pointer is pushed onto the new stack. When returning to the original privilege level, its stack is restored as part of the RET or IRET instruction operation. For subroutine calls that pass parameters on the stack and cross privilege levels, a fixed number of words, as specified in the gate, are copied from the previous stack to the current stack. The inter-segment RET instruction with a stack adjustment value will correctly restore the previous stack pointer upon return. Protection The 80C286 includes mechanisms to protect critical instructions that effect the CPU execution state (e.g. HLT) and code or data segments from improper usage. These protection mechanisms are grouped into three forms: • Restricted usage of segments (e.g. no write allowed to read-only data segments). The only segments available for use are defined by descriptors in the Local Descriptor Table (LDT) and Global Descriptor Table (GDT). • Restricted access to segments via the rules of privilege and descriptor usage. • Privileged instructions or operations that may only be executed at certain privilege levels as determined by the CPL and I/O Privilege Level (lOPL). The lOPL is defined by bits 14 and 13 of the flag word. These checks are performed for all instructions and can be split into three categories: segment load checks (Table 14), operand reference checks (Table 15), and privileged instruction checks (Table 16). Any violation of the rules shown will result in an exception. A not-present exception related to the stack segment causes exception 12.
TABLE 14. SEGMENT REGISTER LOAD CHECKS EXCEPTION NUMBER 13 11 or 12 13 13
ERROR DESCRIPTION Descriptor table limit exceeded Segment descriptor not-present Privilege rules violated Invalid descriptor/segment type segment register load: - Read only data segment load to SS - Special control descriptor load to DS, ES, SS - Execute only Segment load to DS, ES, SS - Data segment load to CS - Read/Execute code segment load SS
23
80C286
TABLE 15. OPERAND REFERENCE CHECKS EXCEPTION NUMBER 13 13 13 12 or 13
address will not point at the ESC instruction that caused the exception; however, the processor extension registers may contain the address of the failing instruction. These exceptions indicate a violation to privilege rules or usage rules has occurred. Restart is generally not attempted under those conditions. All these checks are performed for all instructions and can be split into three categories: segment load checks (Table 14), operand reference checks (Table 15), and privileged instruction checks (Table 16). Any violation of the rules shown will result in an exception. A not-present exception causes exception 11 or 12 and is restartable. SPECIAL OPERATIONS Task Switch Operation The 80C286 provides a built-in task switch operation which saves the entire 80C286 execution state (registers, address space, and a link to the previous task), loads a new execution state, and commences execution in the new task. Like gates, the task switch operation is invoked by executing an inter-segment JMP or CALL instruction which refers to a Task State Segment (TSS) or task gate descriptor in the GDT or LDT. An INT instruction, exception, or external interrupt may also invoke the task switch operation by selecting a task gate descriptor in the associated IDT descriptor entry. The TSS descriptor points at a segment (see Figure 18) containing the entire 80C286 execution state while a task gate descriptor contains a TSS selector. The limit field of the descriptor must be greater than 002B(H). Each task must have a TSS associated with it. The current TSS is identified by a special register in the 80C286 called the Task Register (TR). This register contains a selector referring to the task state segment descriptor that defines the current TSS. A hidden base and limit register associated with TR are loaded whenever TR is loaded with a new selector. The IRET instruction is used to return control to the task that called the current task or was interrupted. Bit 14 in the flag register is called the Nested Task (NT) bit. It controls the
ERROR DESCRIPTION Write into code segment Read from execute-only code segment Write to read-only data segment Segment limit exceeded (See Note) NOTE: Carry out in offset calculations is ignored.
TABLE 16. PRIVILEGED INSTRUCTION CHECKS EXCEPTION NUMBER 13 13
ERROR DESCRIPTION CPL ≠ 0 when executing the following instructions: LIDT, LLDT, LGDT, LTR, LMSW, CTS, HLT CPT > IOPL when executing the following instructions: INS, IN, OUTS, OUT, STI, CLI, LOCK
The lRET and POPF instructions do not perform some of their defined functions if CPL is not of sufficient privilege (numerically small enough). Precisely these are: • The IF bit is not changed if CPL is greater than IOPL. • The lOPL field of the flag word is not changed if CPL is greater than 0. No exceptions or other indication are given when these conditions occur. Exceptions The 80C286 detects several types of exceptions and interrupts in protected mode (see Table 17). Most are restartable after the exceptional condition is removed. Interrupt handlers for most exceptions can read an error code, pushed on the stack after the return address, that identifies the selector involved (0 if none). The return address normally points to the failing instruction including all leading prefixes. For a processor extension segment overrun exception, the return
TABLE 17. PROTECTED MODE EXCEPTIONS INTERRUPT VECTOR 8 9 10 11 12 13 NOTES: 6. When a PUSHA or POPA instruction attempts to wrap around the stack segment, the machine state after the exception will not be restartable because stack segment wrap around is not permitted. This condition is identified by the value of the saved SP being either 0000(H), 0001(H), FFFE(H), or FFFF(H). 7. These exceptions indicate a violation to privilege rules or usage rules has occurred. Restart is generally not attempted under those conditions. RETURN ADDRESS AT FALLING INSTRUCTION? Yes No Yes Yes Yes Yes ALWAYS RESTARTABLE? No (Note 7) No (Note 7) Yes Yes Yes (Note 6) No (Note 7) ERROR CODE ON STACK? Yes No Yes Yes Yes Yes
FUNCTION Double exception detected Processor extension segment overrun Invalid task state segment Segment not present Stack segment overrun or stack segment not present General protection
24
80C286
function of the IRET instruction. If NT = 0, the IRET instruction performs the regular current task by popping values off the stack; when NT = 1, IRET performs a task switch operation back to the previous task. When a CALL, JMP, or INT instruction initiates a task switch, the old (except for case of JMP) and new TSS will be marked busy and the back link field of the new TSS set to the old TSS selector. The NT bit of the new task is set by CALL or INT initiated task switches. An interrupt that does not cause a task switch will clear NT. NT may also be set or cleared by POPF or IRET instructions. The task state segment is marked busy by changing the descriptor type field from Type 1 to Type 3. Use of a selector that references a busy task state segment causes Exception 13. Processor Extension Context Switching The context of a processor extension is not changed by the task switch operation. A processor extension context need only be changed when a different task attempts to use the processor extension (which still contains the context of a previous task). The 80C286 detects the first use of a processor extension after a task switch by causing the processor extension not present exception (7). The interrupt handler may then decide whether a context change is necessary. Whenever the 80C286 switches tasks, it sets the Task Switched (TS) bit of the MSW. TS indicates that a processor extension context may belong to a different task than the current one. The processor extension not present exception (7) will occur when attempting to execute an ESC or WAIT instruction if TS = 1 and a processor extension is present (MP = 1 in MSW). Pointer Testing Instructions The 80C286 provides several instructions to speed pointer testing and consistency checks for maintaining system integrity (see Table 18). These instructions use the memory management hardware to verify that a selector value refers to an appropriate segment without risking an exception. A condition flag (ZF) indicates whether use of the selector or segment will cause an exception. Double Fault and Shutdown If two separate exceptions are detected during a single instruction execution, the 80C286 performs the double fault exception (8). If an exception occurs during processing of the double fault exception, the 80C286 will enter shutdown. During shutdown no further instructions or exceptions are processed. Either NMI (CPU remains in protected mode) or RESET (CPU exits protected mode) can force the 80C286 out of shutdown. Shutdown is externally signalled via a HALT bus operation with A1 LOW. Protected Mode lnitialization The 80C286 initially executes in real address mode after RESET. To allow initialization code to be placed at the top of physical memory. A23-20 will be HIGH when the 80C286 performs memory references relative to the CS register until CS is changed. A23-20 will be zero for references to the DS, ES, or SS segments. Changing CS in real address mode will force A23-20 LOW whenever CS is used again. The initial CS:lP value of F000:FFF0 provides 64K bytes of code space for initialization code without changing CS. Protected mode operation requires several registers to be initialized. The GDT and IDT base registers must refer to a valid GDT and IDT. After executing the LMSW instruction to set PE, the 80C286 must immediately execute an intrasegment JMP instruction to clear the instruction queue of instructions decoded in real address mode. To force the 80C286 CPU registers to match the initial protected mode state assumed by software, execute a JMP instruction with a selector referring to the initial TSS used in the system. This will load the task register, local descriptor table register, segment registers and initial general register state. The TR should point at a valid TSS since any task switch operation involves saving the current task state.
TABLE 18. 80C286 POINTER TEST INSTRUCTIONS INSTRUCTION ARPL OPERANDS Selector, Register FUNCTION Adjust Requested Privilege Level: adjusts the RPL of the selector to the numeric maximum of current selector RPL value and the RPL value in the register. Set zero flag if selector RPL was changed by ARPL. VERify for Read: sets the zero flag if the segment referred to by the selector can be read. VERify for Write: sets the zero flag if the segment referred to by the selector can be written. Load Segment Limit: reads the segment limit into the register if privilege rules and descriptor type allow. Set zero flag if successful. Load Access Rights: reads the descriptor access rights byte into the register if privilege rules allow. Set zero flag if successful.
VERR VERW LSL
Selector Selector Register, Selector Register, Selector
LAR
25
80C286
CPU RESERVED TASK REGISTER TR 15 0 SYSTEM SEGMENT DESCRIPTOR P D 0 TYPE P L
TYPE 1
DESCRIPTION An available task state segment. May be used as the destination of a task switch operation. A busy task state segment. Cannot be used as the destination of a task switch.
BASE 23 - 16
BASE 15 - 0
PROGRAM INVISIBLE 15 LIMIT BASE 23 0 15 TASK LDT SELECTOR DS SELECTOR SS SELECTOR CS SELECTOR ES SELECTOR DI SI BP SP BX TASK STATE SEGMENT DX CX AX FLAG WORD IP (ENTRY POINT) SS FOR CPL 2 SP FOR CPL 2 SS FOR CPL 1 SP FOR CPL 1 SS FOR CPL 0 SP FOR CPL 0 0 BYTE OFFSET 42 40 38 36 34 32 30 28 26 24 22 20 18 16 14 12 10 8 6 4 2 0 LIMIT 15 - 0
3
P 1 0
DESCRIPTION Base and Limit fields are valid. Segment is not present in memory, Base and Limit are not defined.
CURRENT TASK STATE
INITIAL STACKS FOR CPL 0, 1, 2
BACK LINK SELECTOR TO TSS 0
FIGURE 18. TASK STATE SEGMENT AND TSS REGISTERS
System Interface
The 80C286 system interface appears in two forms: a local bus and a system bus. The local bus consists of address, data, status, and control signals at the pins of the CPU. A system bus is any buffered version of the local bus. A system bus may also differ from the local bus in terms of coding of status and control lines and/or timing and loading of signals. Bus Interface Signals and Timing The 80C286 microsystems local bus interfaces the 80C286 to local memory and I/O components. The interface has 24 address lines, 16 data lines, and 8 status and control signals. The 80C286 CPU, 82C284 clock generator, 82C288 bus controller, 82289 bus arbiter, 82C86H/87H transceivers, and 82C82/83H latches provide a buffered and decoded system bus interface. The 82C284 generates the system clock and
26
80C286
synchronizes READY and RESET. The 82C288 converts bus operation status encoded by the 80C286 into command and bus control signals. The 82289 bus arbiter generates Multibus™ bus arbitration signals. These components can provide the critical timing required for most system bus interfaces including the Multibus. Bus Hold Circuitry To avoid high current conditions caused by floating inputs to CMOS devices, and to eliminate the need for pull-up/down resistors, “bus-hold” circuitry has been used on the 80C286 pins 4-6, 36-51 and 66-68 (See Figure 19A and 19B). The circuit shown in Figure 19A will maintain the last valid logic state if no driving source is present (i.e. an unconnected pin or a driving source which goes to a high impedance state). The circuit shown in Figure 19B will maintain a high impedance logic one state if no driving source is present. To overdrive the “bus-hold” circuits, an external driver must be capable of sinking or sourcing approximately 400 microamps at valid input voltage levels. Since this “bus-hold” circuitry is active and not a “resistive” type element, the associated power supply current is negligible, and power dissipation is significantly reduced when compared to the use of passive pull-up resistors.
BOND PAD OUTPUT DRIVER EXTERNAL PIN
address. Byte transfers occur on either half of the 16-bit local data bus. Even bytes are accessed over D7-0 while odd bytes are transferred over D15-8. Even addressed words are transferred over D15-0 in one bus cycle, while odd addressed word require two bus operations. The first transfers data on D15-8, and the second transfers data on D7-0. Both byte data transfers occur automatically, transparent to software. Two bus signals, A0 and BHE, control transfers over the lower and upper halves of the data bus. Even address byte transfers are indicated by A0 LOW and BHE HIGH. Odd address byte transfers are indicated by A0 HlGH and BHE LOW. Both A0 and BHE are LOW for even address word transfers. The I/O address space contains 64K addresses in both modes. The I/O space is accessible as either bytes or words, as is memory. Byte wide peripheral devices may be attached to either the upper or lower byte of the data bus. Byte-wide I/O devices attached to the upper data byte (D15-8) are accessed with odd I/O addresses. Devices on the lower data byte are accessed with even I/O addresses. An interrupt controller such as Intersil's 82C59A must be connected to the lower data byte (D7-0) for proper return of the interrupt vector. Bus Operation The 80C286 uses a double frequency system clock (CLK input) to control bus timing. All signals on the local bus are measured relative to the system CLK input. The CPU divides the system clock by 2 to produce the internal processor clock, which determines bus state. Each processor clock is composed of two system clock cycles named phase 1 and phase 2. The 82C284 clock generator output (PCLK) identifies the next phase of the processor clock. (See Figure 20.)
INPUT DRIVER
INPUT PROTECTION CIRCUITRY
ONE PROCESSOR CLOCK CYCLE
FIGURE 19A. BUS HOLD CIRCUITRY, PINS 36-51, 66, 67
ONE BUS T STATE PHASE 1 OF PROCESSOR CLOCK CYCLE CLK PHASE 2 OF PROCESSOR CLOCK CYCLE
OUTPUT DRIVER BOND PAD VCC P EXTERNAL PIN
ONE SYSTEM CLK CYCLE PCLK
INPUT DRIVER
FIGURE 20. SYSTEM AND PROCESSOR CLOCK RELATIONSHIPS
INPUT PROTECTION CIRCUITRY
Six types of bus operations are supported; memory read, memory write, I/O read, I/O write, interrupt acknowledge, and halt/shutdown. Data can be transferred at a maximum rate of one word per two processor clock cycles. The 80C286 bus has three basic states: idle (TI), send status (TS), and perform command (TC). The 80C286 CPU also has a fourth local bus state called hold (TH). TH indicates that the 80C286 has surrendered control of the local bus to another bus master in response to a HOLD request. Each bus state is one processor clock long. Figure 21 shows the four 80C286 local bus states and allowed transitions.
FIGURE 19B. BUS HOLD CIRCUITRY, PINS 4-6, 68
Physical Memory and I/O Interface A maximum of 16 megabytes of physical memory can be addressed in protected mode. One megabyte can be addressed in real address mode. Memory is accessible as bytes or words. Words consist of any two consecutive bytes addressed with the least significant byte stored in the lowest
27
80C286
RESET HLDA HLDA HOLD TH
HLDA • NEW CYCLE
NEW CYCLE • HLDA
The address remains valid during phase 1 of the first TC to guarantee hold time, relative to ALE, for the address latch inputs. Bus Control Signals
IDLE TI READY • NEW CYCLE
NEW CYCLE
HLDA • NEW CYCLE
READY ALWAYS STATUS TS COMMAND TC
The 82C288 bus controller provides control signals; address latch enable (ALE), Read/Write commands, data transmit/receive (DT/R), and data enable (DEN) that control the address latches, data transceivers, write enable, and output enable for memory and I/O systems. The Address Latch Enable (ALE) output determines when the address may be latched. ALE provides at least one system CLK period of address hold time from the end of the previous bus operation until the address for the next bus operation appears at the latch outputs. This address hold time is required to support Multibus and common memory systems. The data bus transceivers are controlled by 82C288 outputs Data Enable (DEN) and Data Transmit/Receive (DT/R). DEN enables the data transceivers; while DT/R controls transceiver direction. DEN and DT/R are timed to prevent bus contention between the bus master, data bus transceivers, and system data bus transceivers. Command Timing Controls Two system timing customization options, command extension and command delay, are provided on the 80C286 local bus. Command extension allows additional time for external devices to respond to a command and is analogous to inserting wait states on the 80C86. External logic can control the duration of any bus operation such that the operation is only as long as necessary. The READY input signal can extend any bus operation for as long as necessary. Command delay allows an increase of address or write data setup time to system bus command active for any bus operation by delaying when the system bus command becomes active. Command delay is controlled by the 82C288 CMDLY input. After TS, the bus controller samples CMDLY at each failing edge of CLK. If CMDLY is HIGH, the 82C288 will not activate the command signal. When CMDLY is LOW, the 82C288 will activate the command signal. After the command becomes active, the CMDLY input is not sampled. When a command is delayed, the available response time from command active to return read data or accept write data is less. To customize system bus timing, an address decoder can determine which bus operations require delaying the command. The CMDLY input does not affect the timing of ALE, DEN or DT/R. Figure 23 illustrates four uses of CMDLY. Example 1 shows delaying the read command two system CLKs for cycle N-1 and no delay for cycle N, and example 2 shows delaying the read command one system CLK for cycle N-1 and one system CLK delay for cycle N.
READY • NEW CYCLE
FIGURE 21. 80C286 BUS STATES
Bus States The idle (TI) state indicates that no data transfers are in progress or requested. The first active state TS is signaled by status line S1 or S0 going LOW and identifying phase 1 of the processor clock. During TS, the command encoding, the address, and data (for a write operation) are available on the 80C286 output pins. The 82C288 bus controller decodes the status signals and generates Multibus compatible read/write command and local transceiver control signals. After TS, the perform command (TC) state is entered. Memory or I/O devices respond to the bus operation during TC , either transferring read data to the CPU or accepting write data. TC states may be repeated as often as necessary to ensure sufficient time for the memory or I/O device to respond. The READY signal determines whether TC is repeated. A repeated TC state is called a wait state. During hold (TH), the 80C286 will float all address, data, and status output drivers enabling another bus master to use the local bus. The 80C286 HOLD input signal is used to place the 80C286 into the TH state. The 80C286 HLDA output signal indicates that the CPU has entered TH. Pipelined Addressing The 80C286 uses a local bus interface with pipelined timing to allow as much time as possible for data access. Pipelined timing allows a new bus operation to be initiated every two processor cycles, while allowing each individual bus operation to last for three processor cycles. The timing of the address outputs is pipelined such that the address of the next bus operation becomes available during the current bus operation. Or, in other words, the first clock of the next bus operation is overlapped with the last clock of the current bus operation. Therefore, address decode and routing logic can operate in advance of the next bus operation. External address latches may hold the address stable for the entire bus operation, and provide additional AC and DC buffering. The 80C286 does not maintain the address of the current bus operation during all TC states. Instead, the address for the next bus operation may be emitted during phase 2 of any TC .
28
80C286
READ BUS CYCLE N TI φ1 CLK TS φ2 φ1 TC φ2 φ1
READ BUS CYCLE N + 1 TS φ2 φ1 TC φ2
PROC CLK 2 PCLK CYCLE TRANSFER 2 PCLK CYCLE TRANSFER 2.5 CLOCK CYCLE ADDRESS TO DATA VALID A23 - A0 VALID ADDR (N) VALID ADDR (N + 1)
S0 • S1
READY
D15 - D0 VALID READ DATA (N) PIPELINING: VALID ADDRESS (N + 1) AVAILABLE IN LAST PHASE OF BUS CYCLE (N). VALID READ DATA (N + 1)
FIGURE 22. BASIC BUS CYCLE
29
80C286
READ CYCLE N -1 TS φ1 CLK φ2 φ1 TC φ2 φ1 TC φ2 φ1 TS
READ CYCLE N TC φ2 φ1 φ2
PROC CLK
A23 - A0
VALID ADDR (N-1)
VALID ADDR N
S1 • S0
ALE
READY
RD EX1 CMDLY
RD EX2 CMDLY
FIGURE 23. CMDLY CONTROLS THE LEADING EDGE OF COMMAND SIGNAL
Bus Cycle Termination At maximum transfer rates, the 80C286 bus alternates between the status and command states. The bus status signals become inactive after TS so that they may correctly signal the start of the next bus operation after the completion of the current cycle. No external indication of TC exists on the 80C286 local bus. The bus master and bus controller enter TC directly after TS and continue executing TC cycles until terminated by the assertion of READY. READY Operation The current bus master and 82C288 bus controller terminate each bus operation simultaneously to achieve maximum bus operation bandwidth. Both are informed in advance by READY active (open-collector output from 82C284) which identifies the last TC cycle of the current bus operation. The bus master and bus controller must see the same sense of the READY signal, thereby requiring READY to be synchronous to the system clock. Synchronous Ready The 82C284 clock generator provides READY synchronization from both synchronous and asynchronous sources (see Figure 24). The synchronous ready input (SRDY) of the clock generator is sampled with the falling edge of CLK at
the end of phase 1 of each TC. The state of SRDY is then broadcast to the bus master and bus controller via the READY output line. Asynchronous Ready Many systems have devices or subsystems that are asynchronous to the system clock. As a result, their ready outputs cannot be guaranteed to meet the 82C284 SRDY setup and hold time requirements. But the 82C284 asynchronous ready input (ARDY) is designed to accept such signals. The ARDY input is sampled at the beginning of each TC cycle by 82C284 synchronization logic. This provides one system CLK cycle time to resolve its value before broadcasting it to the bus master and bus controller. ARDY or ARDYEN must be HIGH at the end of TS. ARDY cannot be used to terminate the bus cycle with no wait states. Each ready input of the 82C284 has an enable pin (SRDYEN and ARDYEN) to select whether the current bus operation will be terminated by the synchronous or asynchronous ready. Either of the ready inputs may terminate a bus operation. These enable inputs are active low and have the same timing as their respective ready inputs. Address decode logic usually selects whether the current bus operation should be terminated by ARDY or SRDY.
30
80C286
Data Bus Control Figures 25, 26, and 27 show how the DT/R, DEN, data bus, and address signals operate for different combinations of read, write, and idle bus operations. DT/R goes active (LOW) for a read operation. DT/R remains HIGH before, during, and between write operations. The data bus is driven with write data during the second phase of TS . The delay in write data timing allows the read data drivers, from a previous read cycle, sufficient time to enter three-state OFF before the 80C286 CPU begins driving the local data bus for write operations. Write data will always remain valid for one system clock past the last TC to provide sufficient hold time for Multibus or other similar memory or I/O systems. During write-read or write-idle sequences the data bus enters a high impedance state during the second phase of the processor cycle after the last TC . In a write-write sequence the data bus does not enter a high impedance state between TC and TS. Bus Usage The 80C286 local bus may be used for several functions: instruction data transfers, data transfers by other bus masters, instruction fetching, processor extension data transfers, interrupt acknowledge, and halt/shutdown. This section describes local bus activities which have special signals or requirements. Note that I/O transfers take place in exactly the same manner as memory transfers (i.e. to the 80C286 the timing, etc. of an I/O transfer is identical to a memory transfer). HOLD and HLDA HOLD and HLDA allow another bus master to gain control of the local bus by placing the 80C286 bus into the TH state. The sequence of events required to pass control between the 80C286 and another local bus master are shown in Figure 28. In this example, the 80C286 is initially in the TH state as signaled by HLDA being active. Upon leaving TH, as signaled by HLDA going inactive, a write operation is started. During the write operation another local bus master requests the local bus from the 80C286 as shown by the HOLD signal. After completing the write operation, the 80C286 performs one TI bus cycle, to guarantee write data hold time, then enters TH as signaled by HLDA going active. The CMDLY signal and ARDY ready are used to start and stop the write bus command, respectively. Note that SRDY must be inactive or disabled by SRDYEN to guarantee ARDY will terminate the cycle. HOLD must not be active during the time from the leading edge of RESET until 34 CLKs following the trailing edge of RESET unless the 80C286 is in the Halt condition. To ensure that the 80C286 remains in the Halt condition until the processor Reset operation is complete, no interrupts should occur after the execution of HLT until 34 CLKs after the trailing edge of the RESET pulse. LOCK The CPU asserts an active lock signal during InterruptAcknowledge cycles, the XCHG instruction, and during some descriptor accesses. Lock is also asserted when the LOCK prefix is used. The LOCK prefix may be used with the following ASM-286 assembly instructions; MOVS, INS and OUTS. For bus cycles other than Interrupt-Acknowledge cycles, Lock will be active for the first and subsequent cycles of a series of cycles to be locked. Lock will not be shown active during the last cycle to be locked. For the next-to-last cycle, Lock will become inactive at the end of the first TC regardless of the number of wait states inserted. For Interrupt-Acknowledge cycles, Lock will be active for each cycle, and will become inactive at the end of the first TC for each cycle regardless of the number of waitstates inserted. Instruction Fetching The 80C286 Bus Unit (BU) will fetch instructions ahead of the current instruction being executed. This activity is called prefetching. It occurs when the local bus would otherwise be idle and obeys the following rules: A prefetch bus operation starts when at least two bytes of the 6-byte prefetch queue are empty. The prefetcher normally performs word prefetches independent of the byte alignment of the code segment base in physical memory. The prefetcher will perform only a byte code fetch operation for control transfers to an instruction beginning on a numerically odd physical address. Prefetching stops whenever a control transfer or HLT instruction is decoded by the lU and placed into the instruction queue. In real address mode, the prefetcher may fetch up to 6 bytes beyond the last control transfer or HLT instruction in a code segment. In protected mode, the prefetcher will never cause a segment overrun exception. The prefetcher stops at the last physical memory word of the code segment. Exception 13 will occur if the program attempts to execute beyond the last full instruction in the code segment. If the last byte of a code segment appears on an even physical memory address, the prefetcher will read the next physical byte of memory (perform a word code fetch). The value of this byte is ignored and any attempt to execute it causes exception 13.
31
80C286
MEMORY CYCLE N - 1 TS φ1 CLK φ2 φ1 TC φ2 φ1 TS φ2 φ1 MEMORY CYCLE N TC φ2 φ1 TC φ2
PROC CLK
A23 - A0
VALID ADDR
VALID ADDR
VALID ADDR
S0 • S1
SRDY
READY
(SEE NOTE 8)
(SEE NOTE 9)
ARDY (SEE NOTE 10)
NOTES: 8. SRDYEN is active low. 9. If SRDYEN is high, the state of SRDY will not effect READY. 10. ARDYEN is active low. FIGURE 24. SYNCHRONOUS AND ASYNCHRONOUS READY
32
80C286
READ CYCLE TI φ2 CLK φ1 TS φ2 φ1 TC φ2 φ1 TS φ2 φ1 WRITE CYCLE TC φ2 φ1 TI φ2
A23 - A0
VALID ADDR
VALID ADDR
S0 • S1
D15 - D0 VALID READ DATA MRDC
VALID WRITE DATA
MWTC
DEN
DT/R
FIGURE 25. BACK TO BACK READ-WRITE CYCLE
33
80C286
WRITE CYCLE TI φ2 CLK φ1 TS φ2 φ1 TC φ2 φ1 TS φ2
READ CYCLE TC φ1 φ2 φ1 TI φ2
A23 - A0
VALID
VALID
S0 • S1 VALID READ DATA D15 - D0 VALID WRITE DATA
MRDC
MWTC
DEN
DT/R
FIGURE 26. BACK TO BACK WRITE-READ CYCLE
WRITE CYCLE N-1 TI φ2 CLK φ1 TS φ2 φ1 TC φ2 φ1 TS φ2 φ1 WRITE CYCLE N TC φ2 φ1 TI φ2
A23 - A0
VALID ADDR N-1
VALID ADDR N
S0 • S1
D15 - D0
VALID DATA N-1
VALID DATA N
MWTC
DEN
DT/R
(HIGH)
FIGURE 27. BACK TO BACK WRITE-WRITE CYCLE
Processor Extension Transfers
34
80C286
The processor extension interface uses I/O port addresses 00F8(H), and 00FC(H) which are part of the I/O port address range reserved by Intersil. An ESC instruction with Machine Status Word bits EM = 0 and TS = 0 will perform I/O bus operations to one or more of these I/O port addresses independent of the value of lOPL and CPL. ESC instructions with memory references enable the CPU to accept PEREQ inputs for processor extension operand transfers. The CPU will determine the operand starting address and read/write status of the instruction. For each operand transfer, two or three bus operations are performed, one word transfer with I/O port address 00FA(H) and one or two bus operations with memory. Three bus operations are required for each word operand aligned on an odd byte address. Interrupt Acknowledge Sequence Figure 29 illustrates an interrupt acknowledge sequence performed by the 80C286 in response to an INTR input. An interrupt acknowledge sequence consists of two INTA bus operations. The first allows a master 82C59A Programmable Interrupt Controller (PlC) to determine which if any of its slaves should return the interrupt vector. An eight bit vector is read on D0-D7 of the 80C286 during the second INTA bus operation to select an interrupt handler routine from the interrupt table. The Master Cascade Enable (MCE) signal of the 82C288 is used to enable the cascade address drivers during INTA bus operations (See Figure 29) onto the local address bus for distribution to slave interrupt controllers via the system address bus. The 80C286 emits the LOCK signal (active LOW) during TS of the first INTA bus operation. A local bus “hold” request will not be honored until the end of the second INTA bus operation. Three idle processor clocks are provided by the 80C286 between INTA bus operations to allow for the minimum INTA to INTA time and CAS (cascade address) out delay of the 82C59A. The second INTA bus operation must always have at least one extra TC state added via logic controlling READY. A23-A0 are in three-state OFF until after the first TC state of the second INTA bus operation. This prevents bus contention between the cascade address drivers and CPU address drivers. The extra TC state allows time for the 80C286 to resume driving the address lines for subsequent bus operations.
35
80C286
BUS HOLD ACKNOWLEDGE TC φ2 φ1 φ2 φ1 TI φ2 φ1 TH φ2
BUS CYCLE TYPE φ1 CLK
BUS HOLD ACKNOWLEDGE TH φ2 φ1 TH φ2 φ1 TH φ2 φ1 TS φ2 φ1 TC φ2
WRITE CYCLE TC φ1
(SEE NOTE 15) HOLD HLDA (SEE NOTE 11) 80C286 S1 • S0 A23 - A0 M/IO, COD/INTA BHE, LOCK (SEE NOTE 12) VALID (SEE NOTE 13) VALID (SEE NOTE 11) (SEE NOTE 14) (SEE NOTE 16)
D15 - D0 SRDY + SRDYEN
VALID
80C284
NOT READY NOT READY (SEE NOTE 17) ARDY + ARDYEN NOT READY NOT READY CMDLY DELAY ENABLE MWTC (SEE NOTE 17) READY
80C288
VOH DT/R DEN
ALE TS - STATUS CYCLE TC - COMMAND CYCLE
NOTES: 11. Status lines are held at a high impedance logic one by the 80C286 during a HOLD state. 12. Address, M/IO and COD/lNTA may start floating during any TC depending on when internal 80C286 bus arbiter decides to release bus to external HOLD. The float starts in φ2 of TC. 13. BHE and LOCK may start floating after the end of any TC depending on when internal 80C286 bus arbiter decides to release bus to external HOLD. The float starts in φ1 of TC. 14. The minimum HOLD to HLDA time is shown. Maximum is one TH longer. 15. The earliest HOLD time is shown. It will always allow a subsequent memory cycle if pending is shown. 16. The minimum HOLD to HLDA time is shown. Maximum is a function of the instruction, type of bus cycle and other machine state (i.e., Interrupts, Waits, Lock, etc.). 17. Asynchronous ready allows termination of the cycle. Synchronous ready does not signal ready in this example. Synchronous ready state is ignored after ready is signaled via the asynchronous input. FIGURE 28. MULTIBUS WRITE TERMINATED BY ASYNCHRONOUS READY WITH BUS HOLD
36
80C286
Local Bus Usage Priorities The 80C286 local bus is shared among several internal units and external HOLD requests. In case of simultaneous requests, their relative priorities are: (Highest) Any transfers which assert LOCK either explicitly (via the LOCK instruction prefix) or implicitly (i.e. some segment descriptor accesses, an interrupt acknowledge sequence, or an XCHG with memory). The second of the two byte bus operations required for an odd aligned word operand. The second or third cycle of a processor extension data transfer. Local bus request via HOLD input. Processor extension data operand transfer via PEREQ input. Data transfer performed by EU as part of an instruction. (Lowest) An instruction prefetch request from BU. The EU will inhibit prefetching two processor clocks in advance of any data transfers to minimize waiting by the EU for a prefetch to finish.
System Configurations
The versatile bus structure of the 80C286 micro-system, with a full complement of support chips, allows flexible configuration of a wide range of systems. The basic configuration, shown in Figure 30, is similar to an 80C86 maximum mode system. It includes the CPU plus an 82C59A interrupt controller, 82C284 clock generator, and the 82C288 Bus Controller. The 80C86 latches (82C82 and 82C83H) and transceivers (82C86H and 82C87H) may be used in an 80C286 microsystem. As indicated by the dashed lines in Figure 30, the ability to add processor extensions is an integral feature of 80C286 based microsystems. The processor extension interface allows external hardware to perform special functions and transfer data concurrent with CPU execution of other instructions. Full system integrity is maintained because the 80C286 supervises all data transfers and instruction execution for the processor extension. An 80C286 system which includes the 80287 numeric processor extension (NPX) uses this interface. The 80C286/80287 system has all the instructions and data types of an 80C86 or 80C88 with 8087 numeric processor extension. The 80287 NPX can perform numeric calculations and data transfers concurrently with CPU program execution. Numerics code and data have the same integrity as all other information protected by the 80C286 protection mechanism. The 80C286 can overlap chip select decoding and address propagation during the data transfer for the previous bus operation. This information is latched into the 82C82/83H's by ALE during the middle of a TS cycle. The latched chip select and address information remains stable during the bus operation while the next cycle's address is being decoded and propagated into the system. Decode logic can be implemented with a high speed PROM or PAL. The optional decode logic shown in Figure 30 takes advantage of the overlap between address and data of the 80C286 bus cycle to generate advanced memory and I/O select signals. This minimizes system performance degradation caused by address propagation and decode delays. In addition to selecting memory and I/O, the advanced selects may be used with configurations supporting local and system buses to enable the appropriate bus interface for each bus cycle. The COD/lNTA and M/IO signals are applied to the decode logic to distinguish between interrupt, I/O, code, and data bus cycles. By adding the 82289 bus arbiter chip the 80C286 provides a Multibus system bus interface as shown in Figure 31. The ALE output of the 82C288 for the Multibus bus is connected to its CMDLY input to delay the start of commands one system CLK as required to meet Multibus address and write data setup times. This arrangement will add at least one extra TC state to each bus operation which uses the Multibus. A second 82C288 bus controller and additional latches and transceivers could be added to the local bus of Figure 31. This configuration allows the 80C286 to support an on-board bus for local memory and peripherals, and the Multibus for system bus interfacing.
Halt or Shutdown Cycles The 80C286 externally indicates halt or shutdown conditions as a bus operation. These conditions occur due to a HLT instruction or multiple protection exceptions while attempting to execute one instruction. A halt or shutdown bus operation is signalled when S1, S0 and COD/lNTA are LOW and M/IO is HIGH. A1 HIGH indicates halt, and A1 LOW indicates shutdown. The 82C288 bus controller does not issue ALE, nor is READY required to terminate a halt or shutdown bus operation. During halt or shutdown, the 80C286 may service PEREQ or HOLD requests. A processor extension segment overrun during shutdown will inhibit further service of PEREQ. Either NMl or RESET will force the 80C286 out of either halt or shutdown. An INTR, if interrupts are enabled, or a processor extension segment overrun exception will also force the 80C286 out of halt.
37
80C286
INTA CYCLE 1 BUS CYCLE TYPE CLK TC φ1 φ2 φ1 TS φ2 φ1 TC φ2 φ1 TC φ2 φ1 TI φ2 φ1 TI φ2 φ1 TI φ2 φ1 TS φ2 φ1 INTA CYCLE 2 TC φ2 φ1 TC φ2 φ1 TS φ2
S1 • S0
M/IO, COD/INTA
LOCK 80C286
(SEE NOTE 21)
(SEE NOTE 22) A23 - A0 DON’T CARE
(SEE NOTE 22)
BHE
DON’T CARE
D15 - D0
PREVIOUS WRITE CYCLE (SEE NOTE 19)
(SEE NOTE 18) VECTOR
(SEE NOTE 20)
READY NOT READY READY NOT READY READY
INTA
MCE
82C288
ALE
DT/R
DEN
NOTES: 18. Data is ignored. 19. First INTA cycle should have at least one wait state inserted to meet 82C59A minimum INTA pulse width. 20. Second INTA cycle must have at least one wait state inserted since the CPA will not drive A23-A0, BHE, and LOCK until after the first TC state. The CPU imposed one/clock delay prevents has contention between cascade address buffer being disabled by MCE ↓ and address outputs. 21. Without the wait state, the 80C286 address will not be valid for a memory cycle started immediately after the second INTA cycle. The 82C59A also requires one wait state for minimum INTA pulse width. 22. LOCK is active for the first INTA cycle to prevent the 82289 from releasing the bus between INTA cycles in a multi-master system. LOCK is also active for the second INTA cycle. 23. A23-A0 exits three-state OFF during φ2 of the second TC in the INTA cycle. FIGURE 29. INTERRUPT ACKNOWLEDGE SEQUENCE
38
VCC AEN MB CMDLY X2 S0 S1 READY CLK DECODE (OPTIONAL) M/IO MEMORY READ MEMORY WRITE I/O READ I/O WRITE INTERRUPT ACKNOWLEDGE
82C288 BUS CONTROLLER
VCC
RESET
X1 S0 RES S1 READY PCLK CLK EFI F/C
MRDC MWTC IORC IOWC INTA ALE MCE DEN DT/R
ADVANCED MEMORY AND I/O CHIP SELECTS
SYNC READY ENABLE ASYNC READY ENABLE RESET STB OE 82C284 CLOCK GENERATOR
SHDY RESET SRDYEN ARDY ARDYEN
ADDRESS BUS
80C286
FIGURE 30. BASIC 80C286 SYSTEM CONFIGURATION
39
PROCESSOR EXTENSION (OPTIONAL)
82C82 OR 82C83H LATCH
CAS0-2
A0 CHIP SELECT
M/IO LOCK CLK COD/INTA READY S1 A23 - A0 S0 NMI BHE HOLD HLDA ERROR INTR BUSY PEACK PEREQ 80C286 CPU D15 - D0
CS INT INTA WR RD SP/EN D0 - D7 82C59A INTERRUPT CONTROLLER
IR0 - IR7
OE 82C86H OR 82C87H TRANSCEIVER T
DATA BUS
VCC VCC MULTIBUS BUS ARBITRATION
SYSB/RESB BCLK INIT RESET CBRQ BREQ ALWAYS BPRO CBQLCK BPRN S0 BUSY S1 CBRQ READY LOCK CLK AEN M/IO 82289 BUS ARBITER
VCC AEN MEMORY READ MEMORY WRITE I/O READ I/O WRITE INTERRUPT ACKNOWLEDGE
X2 RES PCLK EFI F/C S0 S1 READY CLK
X1
RESET
MRDC MWTC IORC CMDLY IOWC INTA ALE S0 MCE S1 DEN READY CLK 82C288 DT/R BUS CONTROLLER M/IO
80C286
FIGURE 31. MULTIBUS SYSTEM BUS INTERFACE
40
SRDY RESET SRDYEN ARDY ARDYEN 82C284 CLOCK GENERATOR RESET M/IO LOCK CLK COD/INTA READY A23 - A0 S1 S0 BHE NMI HOLD HLDA ERROR INTR BUSY PEACK PEREQ 80C286 CPU D15 - D0 PROCESSOR EXTENSION (OPTIONAL)
SYNC READY ENABLE ASYNC READY ENABLE
STB OE 82C83H LATCH
ADDRESS BUS
CAS0-2 INT INTA WR RD SP/EN D0 - D7
A0 CS
CHIP SELECT
IR0 - IR7 82C59A INTERRUPT CONTROLLER
OE 82C87H TRANSCEIVER T
DATA BUS
80C286
/
Absolute Maximum Ratings
Supply Voltage . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ±8.0V Input, Output or I/O Voltage Applied. . . . . GND -1.0V to VCC +1.0V Storage Temperature Range . . . . . . . . . . . . . . . . . -65oC to +150oC Junction Temperature, PGA . . . . . . . . . . . . . . . . . . . . . . . . . +175oC PLCC . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . +150oC Lead Temperature (Soldering, 10s) . . . . . . . . . . . . . . . . . . . +300oC (PLCC - Lead Tips Only)
Thermal Information
Thermal Resistance (Typical) θJA (oC/W) θJC (oC/W) PDIP Package . . . . . . . . . . . . . . . . . . . 35 6 CERDIP Package . . . . . . . . . . . . . . . . 33 9 Maximum Package Power Dissipation PGA Package . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1.22W PLCC Package . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.2W Gate Count . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .22,500
CAUTION: Stresses above those listed in “Absolute Maximum Ratings” may cause permanent damage to the device. This is a stress only rating and operation of the device at these or any other conditions above those indicated in the operational sections of this specification is not implied.
Operating Conditions
Operating Voltage Range 80C286-10, -12 . . . . . . . . . . . . . . . . . . . . . . . . . . . +4.5V to +5.5V 80C286-16, -20, -25 . . . . . . . . . . . . . . . . . . . . . +4.75V to +5.25V Operating Temperature Range I80C286-10, -12, -16, -20 . . . . . . . . . . . . . . . . . . -40oC to +85oC C80C286-12, -16, -20, -25. . . . . . . . . . . . . . . . . . . . 0oC to +70oC
DC Electrical Specifications VCC = +5V ± 10%, TA = 0oC to +70oC (C80C286-12), VCC = +5V ± 5%, TA = 0oC to +70oC
(C80C286-16, -20, -25), VCC = +5V ± 10%, TA = -40oC to +85oC (I80C286-10, -12), VCC = +5V ± 5%, TA = -40oC to +85oC (I80C286-16, -20) MIN -0.5 2.0 -0.5 3.6 3.0 VCC -0.4 -10 -30 38 -50 -10 MAX 0.8 VCC +0.5 1.0 VCC +0.5 0.4 10 -500 200 -400 10 185 220 260 310 410 5 UNITS V V V V V V µA µA µA µA µA mA mA mA mA mA mA IOL = 2.0mA IOH = -2.0mA, IOH = -100µA VIN = GND or VCC Pins 29, 31, 57, 59, 61, 63-64 VIN = GND (See Note 28) VIN = 1.0V (See Note 24) VIN = 3.0V (See Note 25) VO = GND or VCC Pins 1, 7-8, 10-28, 32-34 80C286-10 (See Note 27) 80C286-12 (See Note 27) 80C286-16 (See Note 27) 80C286-20 (See Note 27) 80C286-25 (See Note 27) (See Note 26) TEST CONDITIONS
SYMBOL VIL VIH VILC VIHC VOL VOH II ISH IBHL IBHH IO ICCOP
PARAMETER Input LOW Voltage Input HIGH Voltage CLK Input LOW Voltage CLK Input HIGH Voltage Output LOW Voltage Output HIGH Voltage Input Leakage Current Input Sustaining Current on BUSY and ERROR Pins Input Sustaining Current LOW Input Sustaining Current HIGH Output Leakage Current Active Power Supply Current
ICCSB
Standby Power Supply Current
-
Capacitance TA = +25oC, All Measurements Referenced to Device GND
SYMBOL CCLK CIN CI/O NOTES: 24. IBHL should be measured after lowering VIN to GND and then raising to 1.0V on the following pins: 36-51, 66, 67. 25. IBHH should be measured after raising VIN to VCC and then lowering to 3.0V on the following pins: 4-6, 36-51, 66-68. 26. ICCSB tested with the clock stopped in phase two of the processor clock cycle. VIN = VCC or GND, VCC = VCC (Max), outputs unloaded. 27. ICCOP measured at 10MHz for the 80C286-10, 12.5MHz for the 80C286-12, 16MHz for the 80C286-16, 20MHz for the 80C286-20, and 25MHz for the 80C286-25. VIN = 2.4V or 0.4V, VCC = VCC (Max), outputs unloaded. 28. ISH should be measured after raising VIN to VCC and then lowering to GND on pins 53 and 54. PARAMETER CLK Input Capacitance Other Input Capacitance I/O Capacitance TYP 10 10 10 UNITS pF pF pF TEST CONDITIONS FREQ = 1MHz
41
80C286
AC Electrical Specifications VCC = +5V ±10%, TA = 0oC to +70oC (C80C286-12), TA = -40oC to +85oC (I80C286-10, -12)
VCC = +5V ±5%, TA = 0oC to +70oC (C80C286-16), TA = -40oC to +85oC (I80C286-16) AC Timings are Referenced to 0.8V and 2.0V Points of the Signals as Illustrated in Data Sheet Waveforms, Unless Otherwise Specified 10MHz SYMBOL PARAMETER MIN MAX 12.5MHz MIN MAX 16MHz MIN MAX UNIT TEST CONDITION
TIMING REQUIREMENTS 1 2 3 17 18 4 5 6 7 8 9 10 11 20 System Clock (CLK) Period System Clock (CLK) LOW Time System Clock (CLK) HIGH Time System Clock (CLK) RISE Time System Clock (CLK) FALL Time Asynchronous Inputs SETUP Time Asynchronous Inputs HOLD Time RESET SETUP Time RESET HOLD Time Read Data SETUP Time Read Data HOLD Time READY SETUP Time READY HOLD Time Input RISE/FALL Times 50 12 16 20 20 19 0 8 4 26 25 8 8 10 40 11 13 15 15 10 0 5 4 20 20 8 8 8 31 7 11 5 5 10 0 5 3 12 5 5 5 6 ns ns ns ns ns ns ns ns ns ns ns ns ns ns 0.8V to 2.0V At 1.0V At 3.6V 1.0V to 3.6V 3.6V to 1.0V (Note 29) (Note 29)
TIMING RESPONSES 12A 12B 13 14 15 16 19 NOTES: 29. Asynchronous inputs are INTR, NMl, HOLD, PEREQ, ERROR, and BUSY. This specification is given only for testing purposes, to assure recognition at a specific CLK edge. 30. Delay from 1.0V on the CLK to 0.8V or 2.0V. 31. Output load: CL = 100pF. 32. Delay measured from address either reaching 0.8V or 2.0V (valid) to status going active reaching 0.8V or status going inactive reaching 2.0V. 33. Delay from 1.0V on the CLK to Float (no current drive) condition. 34. Delay from 1.0V on the CLK to 0.8V for min. (HOLD time) and to 2.0V for max. (inactive delay). 35. Delay from 1.0V on the CLK to 2.0V for min. (HOLD time) and to 0.8V for max. (active delay). 36. Delay from 1.0V on the CLK to 2.0V. Status/PEACK Active Delay Status/PEACK Inactive Delay Address Valid Delay Write Data Valid Delay Address/Status/Data Float Delay HLDA Valid Delay Address Valid to Status SETUP Time 1 1 1 0 0 0 27 22 30 35 40 47 47 1 1 1 0 0 0 22 21 24 32 31 32 25 1 1 1 0 0 0 16 18 20 27 28 29 25 ns ns ns ns ns ns ns 1, (Notes 31, 35) 1, (Notes 31, 34) 1, (Notes 30, 31) 1, (Notes 30, 31) 2, (Note 33) 1, (Notes 31, 36) 1, (Notes 31, 32)
AC Test Conditions
TEST CONDITION 1 2 IL (CONSTANT CURRENT SOURCE) |2.0mA| -6mA (VOH to Float) 8mA (VOL to Float) CL 100pF 100pF
42
80C286
AC Electrical Specifications VCC = +5V ±5%, TA = 0oC to +70oC (C80C286-20, -25), TA = -40oC to +85oC (l80C286-20)
AC Timings are Referenced to the 1.5V Point of the Signals as Illustrated in Data Sheet Waveforms, Unless Otherwise Specified 20MHz SYMBOL TIMING REQUIREMENTS 1 2 3 17 18 4 5 6 7 8 9 10 11 20 System Clock (CLK) Period System Clock (CLK) LOW Time System Clock (CLK) HIGH Time System Clock (CLK) RISE Time System Clock (CLK) FALL Time Asynchronous Inputs SETUP Time Asynchronous Inputs HOLD Time RESET SETUP Time RESET HOLD Time Read Data SETUP Time Read Data HOLD Time READY SETUP Time READY HOLD Time Input RISE/FALL Times 25 6 9 4 4 10 0 3 2 10 3 4 4 6 20 5 7 4 4 10 0 3 2 9 3 4 4 6 ns ns ns ns ns ns ns ns ns ns ns ns ns ns 0.8V to 2.0V At 1.0V At 3.6V 1.0V to 3.6V 3.6V to 1.0V (Note 37) (Note 37) PARAMETER MIN MAX 25MHz MIN MAX UNIT TEST CONDITION
TIMING RESPONSES 12A 12B 13 14 15 16 19 NOTES: 37. Asynchronous inputs are INTR, NMl, HOLD, PEREQ, ERROR, and BUSY. This specification is given only for testing purposes, to assure recognition at a specific CLK edge. 38. Delay from 1.0V on the CLK to 1.5V. 39. Output load: CL = 100pF. 40. Delay measured from address reaching 1.5V to status reaching 1.5V. 41. Delay from 1.0V on the CLK to Float (no current drive) condition. 42. Delay from 1.0V on the CLK to 1.5V. Status/PEACK Active Delay Status/PEACK Inactive Delay Address Valid Delay Write Data Valid Delay Address/Status/Data Float Delay HLDA Valid Delay Address Valid to Status SETUP Time 1 1 1 0 0 0 9 15 16 23 27 25 20 1 1 1 0 0 0 12 12 13 20 24 24 19 ns ns ns ns ns ns ns 1, (Notes 39, 42) 1, (Notes 39, 42) 1, (Notes 38, 39) 1, (Notes 38, 39) 2, (Note 41) 1, (Notes 38, 39) 1, (Notes 39, 40)
AC Test Conditions
TEST CONDITION 1 2 IL (CONSTANT CURRENT SOURCE) |2.0mA| -6mA (VOH to Float) 8mA (VOL to Float) CL 100pF 100pF
43
80C286 AC Specifications
(Continued) C80C86-12, -16 I80C286-10, -12, -16 AC DRIVE AND MEASURE POINTS - CLK INPUT
4.0V
3.6V
3.6V
CLK INPUT 1.0V 0.45V 1.0V
4.0V 3.6V CLK INPUT 1.0V 0.45V 1.0V 3.6V
tSETUP
tHOLD
2.4V 2.0V OTHER DEVICE INPUT 0.8V 0.4V tDELAY (MAX) tDELAY (MAX) 2.0V DEVICE OUTPUT 0.8V 0.8V 2.0V
NOTE: For AC testing, input rise and fall times are driven at 1ns per volt. FIGURE 32.
44
80C286 AC Specifications
(Continued) C80C286-20, -25 I80C286-20 AC DRIVE AND MEASURE POINTS - CLK INPUT
4.0V 3.6V CLK INPUT 1.0V 0.45V 1.0V 3.6V
4.0V 3.6V CLK INPUT 1.0V 0.45V 1.0V 3.6V
tSETUP
tHOLD
2.4V 2.0V OTHER DEVICE INPUT 0.8V 0.4V tDELAY 0.8V 2.0V
DEVICE OUTPUT
1.5V
NOTE: Typical Output Rise/Fall Time is 6ns. For AC testing, input rise and fall times are driven at 1ns per volt. FIGURE 33.
45
80C286
AC Electrical Specifications 82C284 and 82C288 Timing Specifications are given for reference only and no guarantee is implied. 82C284 Timing
10MHz SYMBOL PARAMETER MIN MAX 12.5MHz MIN MAX 16MHz MIN MAX UNIT TEST CONDITION
TIMING REQUIREMENTS 11 12 13 14 SRDY/SRDYEN Setup Time SRDY/SRDYEN Hold Time ARDY/ARDYEN Setup Time ARDY/ARDYEN Hold Time 15 2 5 30 15 2 5 25 10 1 3 20 ns ns ns ns (Note 43) (Note 43)
TIMING RESPONSES 19 PCLK Delay 0 20 0 16 0 15 ns CL = 75pF, IOL = 5mA, IOH = 1mA
NOTE: 43. These times are given for testing purposes to ensure a predetermined action.
82C288 Timing
10MHz SYMBOL PARAMETER MIN MAX 12.5MHz MIN MAX 16MHz MIN MAX UNIT TEST CONDITION
TIMING REQUIREMENTS 12 13 CMDLY Setup Time CMDLY Hold Time 15 1 15 1 10 0 ns ns
TIMING RESPONSES 16 17 19 20 21 22 23 24 29 30 NOTE: 44. These times are given for testing purposes to ensure a predetermined action. ALE Active Delay ALE Inactive Delay DT/R Read Active Delay DEN Read Active Delay DEN Read Inactive Delay DT/R Read Inactive Delay DEN Write Active Delay DEN Write Inactive Delay Command Active Delay from CLK Command Inactive Delay from CLK 1 3 5 3 3 3 16 19 23 21 23 24 23 23 21 20 1 3 5 3 3 3 16 19 23 21 21 18 23 23 21 20 1 5 5 3 3 3 12 15 18 16 14 14 17 15 15 15 ns ns ns ns ns ns ns ns ns ns CL = 300pF IOL = 32mA Max CL = 150pF IOL = 16mA Max IOL = 1mA Max
46
80C286 Waveforms
READ CYCLE WRITE CYCLE ILLUSTRATED WITH ZERO ILLUSTRATED WITH ONE WAIT STATES WAIT STATE TS TC TS 1 φ2 φ2 φ1 φ2 φ1 φ2 φ1 READ (TS OR TS) TC φ2 φ1 TC φ2 φ1
BUS CYCLE TYPE VOH CLK VOL S1 • S0 2 3
TI
12A
12B
19 13 80C286 A23 - A0 M/IO, COD, INTA BHE, LOCK VALID ADDRESS 13 VALID CONTROL 9 8 D15 - D0 13
19
VALID ADDRESS 13 VALID CONTROL
VALID IF TS
14 VALID WRITE DATA
15
VALID READ DATA 11 10 READY 12 11 SRDY + SRDYEN 19 82C284 13 ARDY + ARDYEN 19 PLCK 16 ALE 12 13 CMDLY 29 82C288 MWTC 29 MRDC 19 DT/R 20 21 DEN 22 23 24 30 (SEE NOTE 1) 30 13 12 12 13 17 19 20 14 10 11
FIGURE 34. MAJOR CYCLE TIMING NOTE: The modified timing is due to the CMDLY signal being active.
47
80C286 Waveforms
BUS CYCLE TYPE VCH CLK VCL PCLK (SEE NOTE 47) 4 INTR, NMI HOLD, PEREQ (SEE NOTE 45) 4 ERROR, BUSY (SEE NOTE 46) RESET 5 VCL 7 6 5 VCH CLK (SEE NOTE 47) φ1 19 19 RESET TX φ2 φ1 TX φ2 CLK VCL 7 (SEE NOTE 47) 6
(Continued)
VCH φ2 φ1 TX φ2 φ1
φ1
FIGURE 35. 80C286 ASYNCHRONOUS INPUT SIGNAL TIMING NOTES: 45. PCLK indicates which processor cycle phase will occur on the next CLK, PCLK may not indicate the correct phase until the first cycle is performed. 46. These inputs are asynchronous. The setup and hold times shown assure recognition for testing purposes.
FIGURE 36. 80C286 RESET INPUT TIMING AND SUBSEQUENT PROCESSOR CYCLE PHASE NOTE: 47. When RESET meets the setup time shown, the next CLK will start or repeat φ1 of a processor cycle.
48
80C286 Waveforms
(Continued)
BUS CYCLE TYPE VCH CLK VCL HLDA 16
TH φ1
TS OR TI φ2 φ1
φ2
φ1
TI
φ2
φ1
TH
φ2
16 (SEE NOTE 51) 12A (SEE NOTE 50)
S1 • S0 80C286 12B CLK IF TS 15 IF NPX TRANSFER
15
(SEE NOTE 50)
(SEE NOTE 48) BHE, LOCK A23 - A0, M/IO, COD/INTA 13 (SEE NOTE 52) VALID (SEE NOTE 49) 14 (SEE NOTE 53) D15 - D0 VALID IF WRITE 80C284 15 15
PCLK
FIGURE 37. EXITING AND ENTERING HOLD NOTES: 48. These signals may not be driven by the 80C286 during the time shown. The worst case in terms of latest float time is shown. 49. The data bus will be driven as shown if the cycle before TI in the diagram was a write TC. 50. The 80C286 puts its status pins in a high impedance logic one state during TH. 51. For HOLD request set up to HLDA, refer to Figure 29. 52. BHE and LOCK are driven at this time but will not become valid until TS. 53. The data bus will remain in a high impedance state if a read cycle is performed.
49
80C286 Waveforms
BUS CYCLE TYPE VCH CLK VCL S1 • S0 MEMORY ADDRESS IF PROC. EXT. TO MEMORY TRANSFER I/O PORT ADDRESS 00FA(H) IF MEMORY TO PROC. EXT. TRANSFER I/O READ IF PROC. EXT. TO MEMORY MEMORY READ IF MEMORY TO PROC. EXT. MEMORY WRITE IF PROC. EXT. TO MEMORY I/O WRITE IF MEMORY TO PROC. EXT. TI φ2 φ1 TS φ2 1 TC φ2 φ1 TS φ2 φ1 TC φ2 φ1 TI
(Continued)
A23 - A0 M/IO, COD INTA PEACK
12A
12B (SEE NOTE 54) (SEE NOTE 55) 4
I/O PORT ADDRESS 00FA(H) IF PROC. EXT. TO MEMORY TRANSFER MEMORY ADDRESS IF MEMORY TO PROC. EXT. TRANSFER
5
PEREQ
ASSUMING WORD-ALIGNED MEMORY OPERAND. IF ODD ALIGNED, 80C286 TRANSFERS TO/FROM MEMORY BYTE-AT-A-TIME WITH TWO MEMORY CYCLES.
FIGURE 38. 80C286 PEREQ/PEACK TIMING FOR ONE TRANSFER ONLY NOTES: 54. PEACK always goes active during the first bus operation of a processor extension data operand transfer sequence. The first bus operation will be either a memory read at operand address or I/O read at port address 00FA(H). 55. To prevent a second processor extension data operand transfer, the worst case maximum time (Shown above) is 3 x 1 - 12AMAX - 4 MIN. The actual configuration dependent, maximum time is: 3 x 1 - 12AMAX - 4 MIN + N x 2 x 1 . N is the number of extra TC states added to either the first or second bus operation of the processor extension data operand transfer sequence.
BUS CYCLE TYPE VCH CLK VCL RESET S1 • S0 PEACK A23 - A0 BHE 13 M/IO COD/INTA UNKNOWN 13 LOCK UNKNOWN 15 DATA 16 HLDA UNKNOWN (SEE NOTE 58) UNKNOWN 13 UNKNOWN 19 (SEE NOTE 56) AT LEAST 16 CLK PERIODS 7 (SEE NOTE 57) 12B 6
φ2
φ1
TX
φ2
φ1
TX
φ2
φ1
TX
φ2
φ1
TI
φ2
FIGURE 39. INITIAL 80C286 PIN STATE DURING RESET NOTES: 56. Setup time for RESET ↑ may be violated with the consideration that φ1 of the processor clock may begin one system CLK period later. 57. Setup and hold times for RESET ↓ must be met for proper operation, but RESET ↓ may occur during φ1 or φ2. 58. The data bus is only guaranteed to be in a high impedance state at the time shown.
50
80C286 Waveforms
BYTE 1
(Continued)
BYTE 2 BYTE 3 BYTE 4 BYTE 5 BYTE 6
7654321076543210 LOW DISP/DATA OPCODE d w MOD REG R/M HIGH DISP/DATA LOW DATA HIGH DATA
REGISTER OPERAND REGISTERS TO USE IN OFFSET CALCULATION REGISTER OPERAND/EXTENSION OF OPCODE REGISTER MODE/MEMORY MODE WITH DISPLACEMENT LENGTH WORD/BYTE OPERATION DIRECTION IS TO REGISTER DIRECTION IS FROM REGISTER OPERATION (INSTRUCTION) CODE
FIGURE 40A. SHORT OPCODE FORMAT EXAMPLE
BYTE 1
BYTE 2
BYTE 3
BYTE 4
BYTE 5
765432107654321076543210 LOW DISP MOD REG R/M HIGH DISP
LONG OPCODE
FIGURE 40B. LONG OPCODE FORMAT EXAMPLE FIGURE 40. 80C286 INSTRUCTION FORMAT EXAMPLES
80C286 Instruction Set Summary
Instruction Timing Notes The instruction clock counts listed below establish the maximum execution rate of the 80C286. With no delays in bus cycles, the actual clock count of an 80C286 program will average 5% more than the calculated clock count, due to instruction sequences which execute faster than they can be fetched from memory. To calculate elapsed times for instruction sequences, multiply the sum of all instruction clock counts, as listed in the table below, by the processor clock period. An 12.5MHz processor clock has a clock period of 80 nanoseconds and requires an 80C286 system clock (CLK input) of 25MHz. Instruction Clock Count Assumptions 1. The instruction has been perfected, decoded and is ready for execution. Control transfer instruction clock counts include all time required to fetch, decode, and prepare the next instruction for execution. 2. Bus cycles do not require wait states. 3. There are no processor extension data transfer or local bus HOLD requests. 4. No exceptions occur during instruction execution. Instruction Set Summary Notes Addressing displacements selected by the MOD field are not shown. If necessary they appear after the instruction fields shown. Above/below refers to unsigned value. Greater refers to more positive signed values. Less refers to less positive (more negative) signed values if d = 1, then “to” register; if d = 0 then “from” register if w = 1, then word instruction; if w = 0, then byte instruction if s = 0, then 16-bit immediate data form the operand if s = 1, then an immediate data byte is sign-extended to form the 16-bit operand x don’t care z used for string primitives for comparison with ZF FLAG If two clock counts are given, the smaller refers to a register operand and the larger refers to a memory operand * = add one clock if offset calculation requires summing 3 elements n = number of times repeated m = number of bytes of code in next instruction Level (L) - Lexical nesting level of the procedure The following comments describe possible exceptions, side effects and allowed usage for instructions in both operating modes of the 80C286.
51
80C286
Real Address Mode Only 1. This is a protected mode instruction. Attempted execution in real address mode will result in an undefined opcode exception (6). 2. A segment overrun exception (13) will occur if a word operand references at offset FFFF(H) is attempted. 3. This instruction may be executed in real address mode to initialize the CPU for protected mode. 4. The IOPL and NT fields will remain 0. 5. Processor extension segment overrun interrupt (9) will occur if the operand exceeds the segment limit. Either Mode 6. An exception may occur, depending on the value of the operand. 7. LOCK is automatically asserted regardless of the presence or absence of the LOCK instruction prefix. 8. LOCK does not remain active between all operand transfers. Protected Virtual Address Mode Only 9. A general protection exception (13) will occur if the memory operand cannot be used due to either a segment limit or access rights violation. If a stack segment limit is violated, a stack segment overrun exception (12) occurs. 10. For segment load operations, the CPL, RPL and DPL must agree with privilege rules to avoid an exception. The segment must be present to avoid a not-present exception (11). If the SS register is the destination and a 80C286 Instruction Set Summary FUNCTION FORMAT CLOCK COUNT REAL ADDRES S MODE DATA TRANSFER MOV = Move Register to Register/Mem- 1000100w mod ory r/m Register/Memory to Regis- 1000101w mod ter r/m reg reg data if w=1 2, 3 2, 3 (Note 59) (Note 59) 2, 5 2, 5 (Note 59) (Note 59) 2, 3 2, 3 (Note 59) (Note 59) 2 5 2 5 2 9 2 2 2 9 9 9 PROTECTED VIRTUAL ADDRESS MODE COMMENTS REAL ADDRES S MODE PROTECTED VIRTUAL ADDRESS MODE segment not-present violation occurs, a stack exception (12) occurs. 11. All segment descriptor accesses in the GDT or LDT made by this instruction will automatically assert LOCK to maintain descriptor integrity in multiprocessor systems. 12. JMP, CALL, INT, RET, IRET instructions referring to another code segment will cause a general protection exception (13) if any privilege rule is violated. 13. A general protection exception (13) occurs if CPL ≠ 0. 14. A general protection exception (13) occurs if CPL > IOPL. 15. The IF field of the flag word is not updated if CPL > IOPL. The IOPL field is updated only if CPL = 0. 16. Any violation of privilege rules as applied to the selector operand does not cause a protection exception; rather, the instruction does not return a result and the zero flag is cleared. 17. If the starting address of the memory operand violates a segment limit, or an invalid access is attempted, a general protection exception (13) will occur before the ESC instruction is executed. A stack segment overrun exception (12) will occur if the stack limit is violated by the operand’s starting address. If a segment limit is violated during an attempted data transfer then a processor extension segment overrun exception (9) occurs. 18. The destination of an INT, JMP, CALL, RET or IRET instruction must be in the defined limit of a code segment or a general protection exception (13) will occur.
Immediate to Register/Mem- 1100011w mod 000 data ory r/m Immediate to Register Memory to Accumulator 1011w reg data 1010000w addr-low data if w = 1 addr-high
52
80C286
80C286 Instruction Set Summary (Continued) FUNCTION FORMAT CLOCK COUNT REAL ADDRES S MODE Accumulator to Memory 1010001w addr-low addr-high 3 PROTECTED VIRTUAL ADDRESS MODE 3 COMMENTS REAL ADDRES S MODE 2 2 2 PROTECTED VIRTUAL ADDRESS MODE 9 9, 10, 11 9
Register/Memory to Seg- 10001110 mod 0 reg ment Register r/m Segment Register to Regis- 10001100 mod 0 reg ter/Memory r/m PUSH = Push Memory Register Segment Register Immediate PUSHA = Push All POP = Pop Memory Register Segment Register POPA = Pop All XCHG = Exchange Register/Memory with Reg- 1000011w mod ister r/m Register with Accumulator IN = Input From Fixed Port Variable Port OUT = Output To Fixed Port Variable Port 1110011w port 1110111w 1110010w port 1110110w 10010 reg reg 10001111 mod r/m 01011 reg 000 111 reg (reg ≠ 01) 000 11111111 mod r/m 01010 reg 000 110 reg data if s = 0 110
2, 5 17, 19 (Note 59) (Note 59) 2, 3 2, 3 (Note 59) (Note 59)
5 5 (Note 59) (Note 59) 3 3 3 17 3 3 3 17
2 2 2 2 2
9 9 9 9 9
011010s0 data 01100000
5 5 (Note 59) (Note 59) 5 5 19 5 20 19
2 2 2 2
9 9 9, 10, 11 9
01100001
3, 5 3, 5 (Note 59) (Note 59) 3 3
2, 7
7, 9
5 5
5 5
14 14
3 3 5
3 3 5
14 14 9
XLAT = Translate Byte to AL 11010111
53
80C286
80C286 Instruction Set Summary (Continued) FUNCTION FORMAT CLOCK COUNT REAL ADDRES S MODE LEA = Load EA to Register 10001101 mod r/m LDS = Load Pointer to DS LES = Load Pointer to ES LAHF Load AH with Flags 11000101 mod r/m 11000100 mod r/m 10011111 reg reg (mod ≠ 11) reg (mod ≠ 1) PROTECTED VIRTUAL ADDRESS MODE COMMENTS REAL ADDRES S MODE PROTECTED VIRTUAL ADDRESS MODE
3 3 (Note 59) (Note 59) 7 21 (Note 59) (Note 59) 7 21 (Note 59) (Note 59) 2 2 3 5 2 2 3 5 2 2, 4 9 9, 15 2 2 9, 10, 11 9, 10, 11
SAHF = Store AH into Flags 10011110 PUSHF = Push Flags POPF = Pop Flags ARlTHMETlC ADD = Add Reg/Memory with Register 000000dw mod to r/m Either Immediate ter/Memory to Regis- 100000sw mod r/m 0000010w data reg 10011100 10011101
2, 7 2, 7 (Note 59) (Note 59) data if sw = 01 3, 7 3, 7 (Note 59) (Note 59) 3 3
2
9
000 data data if w = 1
2
9
Immediate to Accumulator ADC = Add with Carry
Reg/Memory with Register 000100dw mod to r/m Either Immediate ter/Memory to Regis- 100000sw mod r/m 0001010w data
reg
2, 7 2, 7 (Note 59) (Note 59) data if sw = 01 3, 7 3, 7 (Note 59) (Note 59) 3 3
2
9
010 data data if w = 1
2
9
Immediate to Accumulator INC = Increment Register/Memory Register SUB = Subtract
1111111w mod r/m 01000 reg
000
2, 7 2, 7 (Note 59) (Note 59) 2 2
2
9
Reg/Memory and Register 001010dw mod to r/m Either
reg
2, 7 2, 7 (Note 59) (Note 59)
2
9
54
80C286
80C286 Instruction Set Summary (Continued) FUNCTION FORMAT CLOCK COUNT REAL ADDRES S MODE Immediate from ter/Memory Regis- 100000sw mod r/m 101 data data if w = 1 data if sw = 01 PROTECTED VIRTUAL ADDRESS MODE COMMENTS REAL ADDRES S MODE 2 PROTECTED VIRTUAL ADDRESS MODE 9
3, 7 3, 7 (Note 59) (Note 59) 3 3
Immediate from Accumula- 0010110w data tor SBB = Subtract with Borrow Reg/Memory and Register 000110dw mod to r/m Either Immediate ter/Memory from Regis- 100000sw mod r/m reg
2, 7 2, 7 (Note 59) (Note 59) data if sw = 01 3, 7 3, 7 (Note 59) (Note 59) 3 3
2
9
011 data data if w = 1
2
9
Immediate from Accumula- 0001110w data tor DEC = Decrement Register/Memory Register CMP = Compare Register/Memory with Reg- 0011101w mod ister r/m Register with ter/Memory Immediate ter/Memory with Regis- 0011100w mod r/m Regis- 100000sw mod r/m reg reg 1111111w mod r/m 01001 reg 001
2, 7 2, 7 (Note 59) (Note 59) 2 2
2
9
2, 6 2, 6 (Note 59) (Note 59) 2, 7 2, 7 (Note 59) (Note 59) data if sw = 01 3, 6 3, 6 (Note 59) (Note 59) 3 2 3 3 3 3 3 7 (Note 59) 3 3 3 3
2 2 2
9 9 9
111 data data if w = 1 011
Immediate with Accumula- 0011110w data tor NEG = Change Sign 1111011w mod r/m
2
7
AAA = ASCII Adjust for Add 00110111 DAA = Decimal Adjust for 00100111 Add AAS = ASCII Adjust for 00111111 Subtract DAS = Decimal Adlust for 00101111 Subtract MUL = Multiply (Unsigned) 1111011w mod r/m 100
55
80C286
80C286 Instruction Set Summary (Continued) FUNCTION FORMAT CLOCK COUNT REAL ADDRES S MODE Register - Byte Register - Word Memory - Byte Memory - Word IMUL = Integer Multiply (Signed) Register - Byte Register - Word Memory - Byte Memory - Word IMUL = Interger Immediate 011010s1 mod Multiply (Signed) r/m DIV = Divide (Unsigned) Register - Byte Register - Word Memory - Byte Memory - Word IDIV = (Signed) Integer Divide 1111011w mod r/m 111 17 25 17 25 6 6 2, 6 2, 6 6 6 6, 9 6, 9 1111011w mod r/m reg data 110 14 22 14 22 6 6 2, 6 2, 6 6 6 6, 9 6, 9 1111011w mod r/m 101 13 21 13 21 2 2 2 9 9 9 13 21 PROTECTED VIRTUAL ADDRESS MODE 13 21 2 2 9 9 COMMENTS REAL ADDRES S MODE PROTECTED VIRTUAL ADDRESS MODE
16 16 (Note 59) (Note 59) 24 24 (Note 59) (Note 59)
16 16 (Note 59) (Note 59) 24 24 (Note 59) (Note 59) data if s = 21, 24 21, 24 0 (Note 59) (Note 59)
17 17 (Note 59) (Note 59) 25 25 (Note 59) (Note 59)
Register - Byte Register - Word Memory - Byte Memory - Word AAM = ASCII Adjust for 11010100 00001010 Multiply
20 20 (Note 59) (Note 59) 28 28 (Note 59) (Note 59) 16 16
56
80C286
80C286 Instruction Set Summary (Continued) FUNCTION FORMAT CLOCK COUNT REAL ADDRES S MODE AAD = ASCII Adjust for 11010101 00001010 Divide CBW = Convert Byte to 10011000 Word CWD = Convert Word to 10011001 Double Word LOGIC Shift/Rotate Instructions Register/Memory by 1 Register/Memory by CL 1101000w mod r/m 1101001w mod r/m mod r/m TTT TTT TTT count TTT 000 001 010 011 100 101 111 AND = And Reg/Memory and Register 001000dw mod to r/m Either Immediate ter/Memory to Regis- 1000000w mod r/m 0010010w data reg 2, 7 2, 7 (Note 59) (Note 59) data if w = 3, 7 3, 7 1 (Note 59) (Note 59) 3 3 2 9 Instruction ROL ROR RCL RCR SHL/SAL SHR SAR 2, 7 2, 7 (Note 59) (Note 59) 5+n, 8+n 5+n, 8+n (Note 59) (Note 59) 5+n, 8+n 5+n, 8+n (Note 59) (Note 59) 2 2 2 9 9 9 14 2 2 PROTECTED VIRTUAL ADDRESS MODE 14 2 2 COMMENTS REAL ADDRES S MODE PROTECTED VIRTUAL ADDRESS MODE
Register/Memory by Count 1100000
100 data data if w = 1
2
9
Immediate to Accumulator
TEST = And Function to Flags, No Result Register/Memory and Reg- 1000010w mod ister r/m reg 2, 6 2, 6 (Note 59) (Note 59) 2 9
57
80C286
80C286 Instruction Set Summary (Continued) FUNCTION FORMAT CLOCK COUNT REAL ADDRES S MODE Immediate Data and Regis- 1111011w mod ter/Memory r/m Immediate Data and Accu- 1010100w data mulator OR = Or Reg/Memory and Register 000010dw mod to r/m Either Immediate ter/Memory to Regis- 1000000w mod r/m 0000110w data reg 2, 7 2, 7 (Note 59) (Note 59) data if w = 3, 7 3, 7 1 (Note 59) (Note 59) 3 3 2 9 000 data data if w = 1 PROTECTED VIRTUAL ADDRESS MODE COMMENTS REAL ADDRES S MODE 2 PROTECTED VIRTUAL ADDRESS MODE 9
data if w = 3, 6 3, 6 1 (Note 59) (Note 59) 3 3
001 data data if w = 1
2
9
Immediate to Accumulator XOR = Exclusive or
Reg/Memory and Register 001100dw mod to r/m Either Immediate ter/Memory to Regis- 1000000w mod r/m 0011010w data
reg
2, 7 2, 7 (Note 59) (Note 59) data if w = 3, 7 3, 7 1 (Note 59) (Note 59) 3 3
2
9
reg data data if w = 1 010
2
9
Immediate to Accumulator NOT = Invert ter/Memory
Regis- 1111011w mod r/m
2, 7 2, 7 (Note 59) (Note 59)
2
9
STRING MANIPULATION MOVS = Move Byte/Word CMPS = Byte/Word 1010010w 5 8 7 5 3 5 5 5 8 7 5 3 5 5 2 2 2 2 2 2 2 9 9 9 9 9 9, 14 9, 14
Compare 1010011w 1010111w
SCAS = Scan Byte/Word
LODS = Load Byte/Word to 1010110w AL/AX STOS = Store Byte/Word 1010101w from AL/A INS = Input Byte/Word from 0110110w DX Port OUTS = Output Byte/Word 0110111w to DX Port
58
80C286 Instruction Set Summary (Continued) FUNCTION REAL ADDRES S MODE REAL ADDRES S MODE Repeated by Count in CX MOVS = Move String CMPS = Compare String SCAS = Scan String LODS = Load String STOS = Store String INS = Input String OUTS = Output String CONTROL TRANSFER CALL = Call Direct Within Segment Register/Memory Within Segment Direct Intersegment Protected Mode Only (Direct Intersegment) Via Call Gate to Same Privilege Level Via Call Gate to Different Privilege Level, No Parameters Indirect 11111111 mod r/m 010 11101000 disp-low disp-high 7+m 7+m 7 + m, 7 + m, 11 + m 11 + m (Note 59) (Note 59) 13 + m Segment Selector 41 + m 82 + m 8, 11, 12, 18 8, 11, 12, 18 26 + m 2 2, 8 18 8, 9, 18 11110011 0110111w 11110011 0110110w 11110011 1010101w 4 + 3n 5 + 4n 5 + 4n 11110011 1010110w 5 + 4n 1111001z 1010111w 5 + 8n 5 + 8n 5 + 4n 4 + 3n 5 + 4n 5 + 4n 1111001z 1010011w 5 + 9n 5 + 9n 11110011 1010010w 5 + 4n 5 + 4n 2 2, 8 2, 8 2, 8 2, 8 2 2 9 8, 9 8, 9 8, 9 8, 9 9, 14 9, 14 PROTECTED VIRTUAL ADDRESS MODE PROTECTED VIRTUAL ADDRESS MODE FORMAT CLOCK COUNT COMMENTS
80C286
All Intersil U.S. products are manufactured, assembled and tested utilizing ISO9000 quality systems. Intersil Corporation’s quality certifications can be viewed at www.intersil.com/design/quality
Intersil products are sold by description only. Intersil Corporation reserves the right to make changes in circuit design, software and/or specifications at any time without notice. Accordingly, the reader is cautioned to verify that data sheets are current before placing orders. Information furnished by Intersil is believed to be accurate and reliable. However, no responsibility is assumed by Intersil or its subsidiaries for its use; nor for any infringements of patents or other rights of third parties which may result from its use. No license is granted by implication or otherwise under any patent or patent rights of Intersil or its subsidiaries.
For information regarding Intersil Corporation and its products, see www.intersil.com
59 10011010 Segment Offset
2
11, 12,18
80C286
80C286 Instruction Set Summary (Continued) FUNCTION FORMAT CLOCK COUNT REAL ADDRES S MODE Via Call Gate to Different Privilege Level, X Parameters Via TSS Via Task Gate Indirect Intersegment 11111111 mod r/m 011 mod ≠ 11 PROTECTED VIRTUAL ADDRESS MODE 86 + 4x + m COMMENTS REAL ADDRES S MODE PROTECTED VIRTUAL ADDRESS MODE 8, 11, 12, 18
177 + m 182 + m 16 + m 29 + m (Note 59) (Note 59) 2
8, 11, 12, 18 8, 11, 12, 18 8, 9, 11, 12, 18
Protected Mode Only (Indirect Intersegment) Via Call Gate to Same Privilege Level Via Call Gate to Different Privilege Level, No Parameters Via Call Gate to Different Privilege Level, X Parameters Via TSS Protected Mode Only (Indirect Intersegment) (Continued) Via Task Gate JMP = Unconditional Jump Short/Long Direct Within Segment Register/Memory Indirect Within Segment Direct Intersegment Protected Mode Only (Direct Intersegment) Via Call Gate to Same Privilege Level Via TSS Via Task Gate Indirect Intersegment 11111111 mod r/m 101 mod ≠ 11 11101011 disp-low 11101001 disp-low 11111111 mod r/m 100 disp-high 7+m 7+m 7+m 7+m 2 18 18 9, 18 185 + m (Note 59) 8, 9, 11, 12, 18 44 + m (Note 59) 83 + m (Note 59) 90 + 4x + m (Note 59) 180 + m (Note 59) 8, 9, 11, 12, 18 8, 9, 11, 12, 18 8, 9, 11, 12, 18 8, 9, 11, 12, 18
7 + m, 7 + m, 11 + m 11 + m (Note 59) (Note 59) 11 + m 23 + m
11101010 Segment Offset Segment Selector
11, 12, 18
38 + m 175 + m 180 + m 15 + m 26 + m (Note 59) (Note 59) 2
8, 11,12,18 8, 11,12,18 8, 11,12,18 8, 9, 11, 12, 18
Protected Mode Only (Indirect Intersegment)
60