Programming Interrupts for Data Acquisition on 80x86-Based Computers
T. Hayles and D. Potter
---------------------------------------------------------------------------
Table of Contents
* Introduction
* What are Interrupts?
* Interrupts for Data Acquisition
* Interrupts on the IBM PC
* Interrupt Programming
* Example of Interrupt Programing: IRQ_ESP.ASM
* Conclusion
---------------------------------------------------------------------------
Introduction
This application note explains how to use interrupt programming on
computers based on the 80x86 family of microprocessors. This family
includes the 8086, 8088, 80286, 80386, and 80486 microprocessors that are
used in PC/XT/AT, EISA, and IBM PS/2 computers. This application note
describes interrupts, explains how interrupts work and how they are
implemented on 80x86-based computers, and discusses programming guidelines
when designing an interrupt-driven application. Finally, this application
note contains example programming code to demonstrate how to program
interrupts for a data acquisition application.
What are Interrupts?
Interrupts are very important to the operation of any computer. Interrupts
give the processor the ability to respond quickly to its peripherals (such
as the keyboard and the hard disk) and to the outside world in general.
Without interrupts, a processor would be unable to service more than one
task efficiently and reliably. The importance of the interrupt is
illustrated when comparing an interrupt to a doorbell. If your door did not
have a doorbell, you would have to periodically go to the door to see if
anyone happened to be there at that time. Of course, that would be very
inefficient. With a doorbell, you only need to go to the door when the
doorbell rings, and you are then confident that someone is there waiting.
Likewise, it is not efficient for the processor to continually check
whether any of its peripherals require attention at a given time. An
interrupt is a doorbell to the processor to signal that some device needs
service.
During normal execution of a program, instructions are read from memory and
executed sequentially by the processor. The processor uses a special
register called the instruction pointer to keep track of the next
instruction to be executed. A set of general-purpose registers are used for
manipulation and temporary storage of any data used by the program.
An interrupt is a special input to the processor. When the processor is
signalled that an interrupt has occurred, the processor finishes the
instruction currently being executed and saves the instruction pointer and
a status word to the stack. The stack is a special block of memory used to
keep track of information during function calls and interrupts. The
processor uses a special register called the stack pointer to keep track of
the location of the top of the stack, which is where the last item was
added to the stack.
After the processor saves this information to the stack, the processor
branches to a special routine called an interrupt service routine or
interrupt handler. If needed, this routine can determine the source of the
interrupt and then service the device that requested the interrupt. This
process may involve reading a character from the keyboard and updating the
monitor, or moving data from an adapter board into computer memory. After
handling the interrupt, the last instruction executed in the interrupt
service routine is an interrupt return instruction. This instruction causes
the processor to restore the instruction pointer and status register with
the values that were saved to the stack at the time of the interrupt and to
resume execution.
Interrupts for Data Acquisition
Many data acquisition applications can benefit greatly from the interrupt
capability of the processor. Interrupts can be used for immediate
communication between the data acquisition hardware and the computer. They
can be used to time or synchronize data acquisition with external events or
to respond promptly to external events and alarm conditions. However,
interrupts are not always the best approach in data acquisition
applications. You must consider the advantages and disadvantages of various
data acquisition methods before deciding to use interrupt-driven
acquisition.
Data acquisition involves the monitoring of one or several sources of
physical data at a regular rate, which is often determined by a clock
signal. Commonly, the data originates from analog-to-digital converters
(ADCs), counter/timer devices, or switches. In general, there are three
approaches to acquiring data from an external device or synchronizing
communication between devices. These three approaches are described as
follows:
Polling - This method involves periodically reading the status of the
device to determine whether the device needs attention.
Direct Memory Access (DMA) - A dedicated processor, the DMA controller,
transparently transfers data from the device to computer memory, or vice
versa.
Interrupts - The device is configured to interrupt the processor whenever
the device requires attention.
For some applications, polling a device can be a very effective way to
service a data acquisition operation. The polling method is simple, easy to
understand, and easy to debug. If the processor can be dedicated to the
single task of monitoring and servicing the device, high acquisition rates
can be realized. However, dedication to a single task is often not
possible. Other devices may also require timely monitoring, or the
processor may be required to do other work, such as handling previously
acquired data. DMA or interrupts would then be a better approach.
DMA uses a DMA controller to move data from one location to another, making
fast data transfers of large blocks of data possible. DMA is fast because
dedicated hardware transfers the data and requires only one or two
read/write bus cycles per transfer. In addition, the processor is free for
other activities (such as processing acquired data) because the DMA
controller proceeds with the data acquisition in the background. However,
the processor may need to access the data on a point-by-point basis as it
is being acquired. DMA is not well-suited for this type of application, and
interrupts or polling may be preferable. Furthermore, the number of DMA
channels in a PC is limited, and a dedicated DMA channel is required for
each unique data acquisition source.
With interrupts, the processor can respond quickly to an event (such as a
clock signal) to transfer data or to synchronize different events. Unlike
DMA, an interrupt-driven system has the advantage of being able to process
the data on a point-by-point basis, which can be particularly important in
control applications. Consider a typical control application synchronized
with a clock signal, for example. The application may require that, with
every pulse of a clock, the processor read the result of an A/D conversion,
read the state of several digital lines, use these results to compute a new
control value, and write this value to a digital-to-analog converter (DAC).
DMA would not be useful for such an application, and a polling routine
would not directly provide synchronization with the clock signal.
Interrupts, however, make it easier to synchronize the input and output
with the clock signal and for the processor to perform other general
processing tasks in the foreground, if needed.
One disadvantage of interrupts, however, is that they require processor
time, and there can be a significant amount of software overhead time
associated with the interrupt service routine. When designing a data
acquisition system, you must estimate the amount of this overhead time,
which consists of the time required by the processor to vector to the
proper interrupt service routine, known as interrupt latency, and the time
necessary to perform the task required within the service routine. If the
overhead is too large for the required acquisition rate, then DMA or a very
efficient polling routine must be used. In applications that are time
critical, the interrupt service routine must be written to be as fast as
possible. As a result, interrupt service routines for data acquisition
applications are often written in assembly language.
Many driver software packages use both interrupt and DMA methods to handle
data acquisition servicing. LabDriver® and LabWindows®, for example, all
include sophisticated, high-level routines for data acquisition that take
advantage of the interrupt and DMA capabilities of the hardware.
Application developers should seriously consider using such a driver
package, which spares the developer the complexity and details of
programming the low-level hardware of the computer and the adapter board.
Using a driver package can significantly decrease the amount of time
required to develop an application.
Interrupts on the IBM PC
The 80x86 Family Register Set
The 80x86 family of microprocessors use 14 separate 16-bit registers that
can be grouped into four categories: general-purpose registers, segment
registers, offset registers, and the flags register. The registers, their
categories, and their use are listed in Table 1.
Table 1. Registers of the 8086 Family of Microprocessors
---------------------------------------------------------------------------
Register Category Use
AX General-purpose -
BX General-purpose -
CX General-purpose -
DX General-purpose -
CS Segment Code segment
DS Segment Data segment
ES Segment Extra segment
SS Segment Stack segment
SP Offset Stack pointer
BP Offset Base pointer
SI Offset Source index
DI Offset Destination index
IP Offset Instruction pointer
Flags Flags Status flags
The general-purpose registers are used by the processor for temporary
storage of data. To facilitate use of 8-bit and 16-bit values, each of
these 16-bit registers can also be addressed as a pair of 8-bit registers.
AL, AH, BL, BH, and so on are used to address the lower (L) or higher (H)
bytes.
The segment registers are used to store the base address of 64 kilobyte
memory segments. The absolute or complete address of a memory location is
obtained by combining the segment registers with an offset value stored in
one of the offset registers. A segment register identifies a distinct 64
kilobyte segment. A particular memory location within the segment is
indicated by a 16-bit offset register. For example, the memory location
01F0:0009 refers to byte 9 within segment 01F0.
The instruction pointer (IP) holds the offset address of the next
instruction to be executed by the processor. The absolute address of the
instruction is determined by combining the IP offset register with the Code
Segment (CS) Register. The IP Register is automatically updated each time
the processor loads an instruction.
Another offset register is the Stack Pointer (SP) Register, which, when
combined with the Stack Segment Register (SS), points to the current top of
the stack. The stack is a last-in first-out (LIFO) memory structure. As
items or data are added, the stack gets larger. When an item is removed,
the first item to come off the stack is the last item added to the stack.
When data is added to or removed from the stack, the value of the stack
pointer is appropriately updated.
The Flags Register, as the name implies, stores the state of nine CPU
status and control flags. Six of the bits are status flags, which are
automatically set or cleared after the execution of every instruction. The
three other bits are used for control purposes. These bits include the
direction flag, the trap flag, and the interrupt flag. The interrupt flag
(IF) bit is used to enable or disable hardware interrupts.
The 80x86 Family Interrupts
The first 1024 bytes of memory are reserved for the interrupt vector table.
This table holds up to 256 vectors, each of which is a 4-byte pointer to a
specific interrupt service routine that is executed when the corresponding
interrupt is processed. The design of the 80x86 family requires certain
interrupt vectors to be used for specific functions. While many of these
vectors are reserved for software interrupts, a limited number of vectors
are reserved for hardware interrupts.
When a hardware interrupt occurs, the processor first responds by pushing
the contents of the Flags, CS, and IP Registers onto the stack and
disabling any further hardware interrupts by clearing the IF bit in the
Flags Register. The processor then looks to the system bus for an 8-bit
interrupt number, multiplies the number by 4 (the size of each vector in
bytes), and uses the result as an offset into the interrupt vector table.
The vector address pointed to by the result is then loaded into the CS and
IP Registers, and the processor resumes operation at this new location
pointed to by the instruction pointer. After processing the interrupt, the
processor restores the Flags, CS, and IP Registers with the values saved to
the stack when the interrupt occurred. Because the IF bit is cleared after
the Flags Register is saved to the stack, restoring the Flags Register also
resets the IF bit, which enables hardware interrupts.
Hardware Interrupts and the Intel 8259A Programmable Interrupt Controller
Hardware interrupts are sent to the processor through an Intel 8259A
programmable interrupt controller, which provides the system with eight
prioritized hardware interrupts. When the 8259A receives an interrupt
request, typically from a peripheral device, the controller drives high its
output (INT), which is connected to the processor's interrupt input pin
(INTR). The INTR pin is used by the processor to signal the occurrence of a
maskable interrupt. If the IF bit in the Flags Register is set, the
processor sends an acknowledge signal to the 8259A controller.
After receiving the interrupt acknowledge signal (INTA) from the processor,
the interrupt controller places the appropriate interrupt vector on the
system bus. The eight hardware interrupts (IRQ0 through IRQ7) on the 8259A
are mapped by the controller to interrupt vectors 8 through 15 (0Fh). The
interrupts are prioritized, IRQ0 having the highest priority and IRQ7
having the lowest priority. The PC design assigns interrupts IRQ0 through
IRQ7 for particular peripherals. Table 2 lists these assignments. IRQ0 and
IRQ1 are reserved for the system timer and the keyboard, respectively,
while IRQ2 is reserved for cascading additional interrupt controllers. IRQ3
through IRQ7 are assigned to various peripherals but are connected to the
I/O bus (see Figure 1). If these peripherals are not present, interrupts
IRQ3 through IRQ7 can be used for other purposes.
Table 2. Interrupt Assignments on the IBM PC/XT Computer
Interrupt Vector Index Action
IRQ0 08h Timer tick (18.2 times/sec)
IRQ1 09h Keyboard input
IRQ2 0Ah Reserved for cascading interrupts
IRQ3 0Bh COM2
IRQ4 0Ch COM1
IRQ5 0Dh Fixed disk controller
IRQ6 0Eh Floppy disk controller
IRQ7 0Fh Printer controller
Figure 1. Block Diagram of 8259A Interface to CPU and PC/XT Bus
The PC AT computer is designed with a second interrupt controller for eight
additional IRQ levels. The INT line of the second controller, the slave, is
cascaded to the IRQ2 line of the first controller, the master, as
illustrated in Figure 2. The eight interrupts (IRQ8 through IRQ15) from the
second controller are mapped to interrupt vectors 112 (70h) through 119
(77h). Because these additional interrupts are effectively connected to the
IRQ2 line of the master 8259A controller, they take priority over IRQ3
through IRQ7 of the master 8259A.
Table 3. Additional Interrupt Assignments on IBM PC AT Computers
Interrupt Vector Index Action
IRQ8 70h Real-time clock
IRQ9 71h Redirect cascade
IRQ10 72h Reserved
IRQ11 73h Reserved
IRQ12 74h Auxiliary device
IRQ13 75h Math coprocessor exception
IRQ14 76h Fixed disk controller
IRQ15 77h Reserved
Figure 2. Cascaded Configuration of the Intel 8259A as Used in PC AT
Computers
A device, such as a plug-in adapter board, issues an interrupt by causing a
high-to-low transition on one of the IRQ lines. With most adapter boards
you can choose which IRQ line to use, usually by positioning a jumper. The
board is then programmed to assert an interrupt signal under certain
conditions, such as the completion of an A/D conversion or the pulse of a
counter/timer output. Many adapter boards are able to cause interrupts
under more than one condition. Therefore, the interrupt service routine
must determine the cause of the interrupt, usually by reading a status
register on the adapter board.
The Micro Channel bus, used in many of the IBM PS/2 computers, and the EISA
bus also use the cascaded Intel 8259A configuration shown in Figure 2 for
prioritized interrupts. The previous discussion applies to these computers
as well. However, there is one significant difference concerning how the
interrupt line works. While interrupts on PC/XT/AT computers are triggered
by a high-to-low transition, or edge, of the interrupt line, the Micro
Channel bus defines an interrupt signal as asserted by the state of the
interrupt line. With this configuration, referred to as level mode, an
interrupt is considered to be asserted, or active, as long as the interrupt
line is held in the high state. With the EISA bus, on the other hand, both
level mode and edge-triggered interrupts can be used.
Interrupt Programming
There are three basic operations to programming an interrupt-driven
application: installing the interrupt vector, which points to the interrupt
service routine, into the interrupt vector table; handling the interrupt
with an interrupt service routine; and removing the interrupt vector. These
operations are described as follows.
Installing the Interrupt Vector
Installing an interrupt vector consists of putting the correct address of
the interrupt service routine, or interrupt vector, in the interrupt vector
table, and enabling interrupts. First, you determine the proper location in
the interrupt vector table to install the handler address. As mentioned
previously, interrupts IRQ0 through IRQ7 map to interrupt level 8 through
15, and interrupts IRQ8 through IRQ15 map to interrupt levels 112 (70h)
through 119 (77h). For example, the address of a handler that is to service
an IRQ0 interrupt must be installed at location or index 8 in the vector
table. DOS has two functions, accessible through the DOS function call INT
21h, for setting and retrieving interrupt vectors. Function 25h is used to
set an interrupt vector. When function 25h is called, the AL Register holds
the interrupt table index, and the DS and DX Registers point to the address
of the interrupt service routine. Function 35h is used to get, or return
the value of, an interrupt vector from the table. This function should be
called before the interrupt is set so the original vector can be saved and
then restored when removing the interrupt.
For the interrupt controller to recognize an interrupt, the appropriate
interrupt must be enabled. Enabling interrupts is done via the Interrupt
Mask Register of the interrupt controller. This register contains bits that
mask each of the interrupt request lines. If the bit is cleared or set,
then the corresponding interrupt is enabled or disabled, respectively.
Whenever the interrupt controller is to be programmed, interrupts to the
processor must be disabled. Interrupts are disabled with the assembly
language call cli and re-enabled with the function sti. These two
instructions manipulate the interrupt flag (IF) bit in the Flags Register.
Interrupts should be also be disabled during any manipulation of the
interrupt vector table.
Some high-level languages include function calls that operate in the same
manner as functions 25h and 35h. These functions provide a convenient way
to install interrupt vectors without having to resort to assembly language
programming. Table 4 lists these functions for some popular languages.
Consult the user manual for the particular language you are using for more
information concerning these or similar functions.
Table 4. Some Available Functions to Set and Get Interrupt Vectors
Language To Set Interrupt Vector To Get Interrupt Vector
DOS Function INT 25h INT 35h
Microsoft C 5.0 _dos_setvect _dos_getvect
Turbo C setvect getvect
Turbo Pascal 4.0 SetIntVec GetIntVec
Interrupt Handling
In an interrupt handler or service routine you must first save any
registers that the interrupt service routine will modify so the registers
can be restored before returning from the service routine. The interrupt
handler may then need to address the device to verify that the interrupt
originated from the device. Many devices are capable of causing interrupts
for a number of different reasons, so reading a Status Register may be
necessary to determine the exact cause of the interrupt.
Having verified the source and cause of the interrupt, the interrupt
service routine then does the particular task for that application. For
example, in a typical data acquisition application, the routine reads a
value from the adapter board, stores the value into memory, and increments
a counter. After handling the interrupt, the routine must restore the
processor registers that were saved at the beginning of the routine.
Finally, a non-specific end of interrupt (EOI) is sent to the interrupt
controller. If the EOI is not sent, no further interrupts of equal or lower
priority can occur.
Note: In systems with cascaded interrupt controllers (non-XT computers),
EOI instructions must be sent to both the master and the slave 8259A
interrupt controllers.
Again, interrupts to the processor must be disabled while writing the EOI
instructions to the interrupt controllers.
When writing an interrupt service routine, do not use DOS functions unless
special precautions are taken. DOS is not re-entrant. If the processor is
interrupted during a DOS call, and the service routine attempts to call
another DOS function, the computer may lock up. To prevent this problem,
you may use an undocumented INT 21h function, function 34h, to determine if
a DOS function is in progress. The 34h function returns a pointer in the ES
Register and the BX Register that points to a DOS busy flag (InDOS flag).
When an INT 21h function starts, the flag is incremented by one; when the
function ends, the flag is decremented by one. If the flag is zero, then no
DOS functions are currently being executed.
Note: The INT 21h function is undocumented and not officially supported by
IBM and Microsoft. Therefore, the function may not be available in all
versions of DOS.
Removing Interrupts
Finally, the application must remove the interrupt and restore the vector
table to its original state. First, disable the interrupt level using the
Interrupt Mask Register. Then, using the data stored during the interrupt
installation, the original interrupt vector is restored using function 25h
or a similar high-level function call.
Example of Interrupt Programing: IRQ_ESP.ASM
This section contains a listing of the file IRQ_ESP.ASM. This file is
assembly language code that installs and removes interrupt vectors and
programs the PC/XT/AT interrupt controller. Also included is an example of
an interrupt service routine. The listing consists of the three following
functions:
install_ISR: This function performs the steps necessary to install an
interrupt service routine and to enable the interrupt on the interrupt
controller. The IRQ number and the address of the interrupt service routine
are passed as parameters.
remove_ISR: This function uses the data saved by install_ISR to restore the
original interrupt vector and disable the interrupt in the interrupt
controller.
IRQ_Handler: This function contains the shell for an example interrupt
service routine. The example reads data from an adapter board, stores the
data into an array, and increments a counter. In addition, the function
sends the EOI to the interrupt controller.
Also included in the listing is an example of a C program that would use
these routines.
;******************************** IRQESP.ASM ********************************
; Engineering Software Package-Interrupt Controller Programming
; Rev A.0
; Copyright 1991 National Instruments Corporation.
; All rights reserved.
;
; This file contains code that installs and removes interrupt vectors
; and programs a PC/XT/AT interrupt controller. Included is an example of
; a user-written interrupt service routine. This code was written for
; the Microsoft Macro Assembler.
;
;****************************************************************************
Public _install_ISR ; _install_ISR and _remove_ISR are
Public _remove_ISR ; PUBLIC so that they can be called
; from a C program.
Public _IRQ_Handler ; _IRQ_Handler is PUBLIC so that its
; address can be passed to
; _install_ISR by a C program.
; These external declarations enable the assembly language interrupt
; service routine to access data declared in the calling C program. Because
; the C compiler renames io_buffer as _io_buffer, these names are prepended
; with an underscore.
EXTRN _io_buffer:WORD
EXTRN _io_count:WORD
MASTER_IRQ_BASE EQU 20h ; I/O addresses of the master interrupt
MASTER_IRQ_MASK EQU 21h ; controller.
SLAVE_IRQ_BASE EQU 0A0h ; I/O addresses of the slave interrupt
SLAVE_IRQ_MASK EQU 0A1h ; controller.
NON_SPEC_EOI EQU 20h ; The bit pattern for a non-specific
; end of interrupt command.
IRQ_TEXT SEGMENT WORD PUBLIC 'CODE'
ASSUME CS:IRQ_TEXT
IRQ_num DB ? ; IRQ_num contains the IRQ number (0
; to 15).
INT_num DB ? ; INT_num contains the INT number or
; index into the interrupt vector
; table (8 to 15 or 112 to 119).
old_int_off DW ? ; old_int_off and old_int_seg contain
old_int_seg DW ? ; the OFFSET and SEGMENT of the
; original interrupt vector.
;***************************************************************************
; _install_ISR performs the steps necessary to install a user-
; defined interrupt service routine and to enable the interrupt. The
; name is prepended with an underscore so that _install_ISR can be called
; from a C program (the C compiler automatically prepends an underscore
; to every global name). The syntax for the call from a C program is as
; follows:
;
; install_ISR (IRQ_num, ISR_proc);
;
; Because _install_ISR is declared as a far procedure, _install_ISR must
; also be so declared in the calling C program or, equivalently, the C
; program must be compiled with the /AL flag.
;***************************************************************************
_install_ISR proc far
push bp ; Save the registers to be used and
mov bp, sp ; set up the base pointer to the stack
push ax ; frame containing the arguments to
push bx ; this procedure.
push cx
push dx
push ds
push es
mov ax, cs ; Set up ds to point to the segment
mov ds, ax ; containing the data items. In this
; case the data segment is also the
; CODE segment.
mov al, [bp+6] ; Grab the IRQ number from the stack.
mov IRQ_num, al ; Store the IRQ number for later use.
cmp al, 7 ; If the IRQ number is 0 to 7, you must
ja slave ; add 8 to it to index the right place
add al, 8 ; in the interrupt vector table. If the
jmp getvec ; IRQ number is 8 to 15, you must add
; 104 to index the right place. This
slave: add al, 104 ; index is also called the INT number.
getvec: mov INT_num, al ; Store the INT number for later use.
mov ah, 35h ; Get the previous ISR vector in es and
int 21h ; bx. al already contains the correct
mov old_int_off, bx ; index into the interrupt vector table.
mov old_int_seg, es ; The old vector is needed when removing
; the user-defined ISR vector.
cli ; Disable interrupts while installing
; a vector and changing the masks.
push ds ; Save the current value of ds.
mov al, INT_num ;
mov ah, 25h ; To install your own vector, ds and dx
mov dx, [bp + 10] ; must point to it when calling DOS
mov ds, dx ; function 25h and al must contain
mov dx, [bp + 8] ; the INT number, or index, into the
int 21h ; interrupt vector table.
pop ds ; Restore ds.
; For the interrupt controller to recognize your interrupt, the interrupt
; must be "unmasked". The MASTER_IRQ_MASK Register controls IRQ levels
; 0 to 7. Clearing bit x (for example, bit 5) enables IRQ level x (for
; example, IRQ 5). Setting a bit disables the corresponding IRQ level.
; The SLAVE_IRQ_MASK Register controls IRQ levels 8 to 15. Clearing bit x (for
; example, bit 2) enables IRQ level x + 8 (for example, IRQ 10).
mov bx, 1 ; bx holds the new mask.
mov cl, IRQ_num ; The new mask contains all ones, except
shl bx, cl ; in the bit position corresponding to
not bx ; your IRQ level.
in al, MASTER_IRQ_MASK ; Read the current master mask.
and al, bl ; AND the mask with the low byte of the
out MASTER_IRQ_MASK, al ; new mask, and write out the result.
jmp $+2 ; This process provides a delay between
; reads and writes between these chips.
in al, SLAVE_IRQ_MASK ; Read the current slave mask.
and al, bh ; AND the mask with the high byte of the
out SLAVE_IRQ_MASK, al ; new mask, and write out the result.
pop es ; Restore all the registers.
pop ds
pop dx
pop cx
pop bx
pop ax
pop bp
sti ; Enable interrupts and
ret ; return.
_install_ISR endp
;****************************************************************************
; This is the shell of an interrupt handler.
;****************************************************************************
_IRQ_Handler proc far
push ax ; Save ax and ds.
push ds
; Code to handle the interrupt goes here - usually a register on the
; adapter board is read to make sure that is where the interrupt came
; from and, if the adapter can generate an interrupt for more than
; one reason, to determine the reason for the interrupt. Once
; verified that the interrupt indeed came from the correct adapter
; board, this routine might read a value from the board, store the
; value in an array, and increment a counter.
; Here is an example of such code:
push dx ; Save any other registers used
push di ; in this section.
mov ax, SEG _io_count ; Set up ds to point to the segment
mov ds, ax ; where _io_count is declared.
mov di, _io_count ; Load the sample count into di.
shl di, 1 ; Multiply the integer count by 2 to
; convert it to a byte count.
mov dx, 0236h ; Load an adapter address.
in ax, dx ; Read something from the adapter.
mov _io_buffer[di], ax ; Store the value; this assumes that
; _io_buffer and _io_count are in the
; same segment and that adding di to
; the beginning of the array does not
; cross over into the next segment.
inc word ptr _io_count ; Increment the sample count.
pop di ; Restore registers used in the
pop dx ; adapter handling code.
; Code to tell the adapter that the interrupt has been serviced goes
; here - usually the adapter contains a location that, when written
; to, causes the adapter to cease asserting the interrupt. This must
; be done before the next step (sending EOI to the interrupt control-
; ler) or the same interrupt is asserted again, and the computer may
; lock up (caught in an endless loop of asserting and servicing the
; same interrupt).
; Now send a non-specific end of interrupt to the interrupt con-
; troller. If the IRQ level is 8 to 15, send EOI to the slave
; first and then to the master. Otherwise, only send EOI to
; the master. It is not necessary to explicitly disable interrupts
; here because the CPU has already done so during the interrupt
; acknowledge cycle.
mov ax, cs ; Set ds to point to the segment
mov ds, ax ; where IRQ_num is declared.
mov al, NON_SPEC_EOI ; Move the EOI code into al.
cmp byte ptr IRQ_num, 8 ; If the IRQ level is less than 8,
jb Ack_Master ; send EOI to the master only.
out SLAVE_IRQ_BASE, al ; Otherwise, send EOI to the slave.
jmp $+2 ; Delay.
Ack_Master:
out MASTER_IRQ_BASE, al ; Send EOI to the master.
pop ds ; Restore remaining registers.
pop ax
iret ; Return with the IRET instruction.
_IRQ_Handler endp
;****************************************************************************
; _remove_ISR uses the data saved by _install_ISR to restore the
; original interrupt vector and to set the proper bit in the mask
; registers to disable the interrupt. _remove_ISR can be called from a
; C program by the following statement:
;
; remove_ISR();
;
; Remember, this is a far procedure like _install_ISR.
;***************************************************************************
_remove_ISR proc far
push ax ; Save the registers.
push bx
push cx
push dx
push ds
mov ax, cs ; Set up ds to point to the segment
mov ds, ax ; containing the data; in this case,
; ds is also the CODE segment.
mov bx, 1 ; bx holds the new mask.
mov cl, IRQ_num ; It contains all zeroes except
shl bx, cl ; in the bit position corresponding to
; your IRQ level.
cli ; Disable interrupts while changing
; masks and installing ISR vectors.
in al, MASTER_IRQ_MASK ; Read the current master mask.
or al, bl ; OR the current master mask with the
; low byte of the mask and
out MASTER_IRQ_MASK, al ; write the new mask out.
jmp $+2 ; Delay.
in al, SLAVE_IRQ_MASK ; Read the current slave mask.
or al, bh ; OR the current slave mask with the
; high byte of the mask and
out SLAVE_IRQ_MASK, al ; write the new mask out.
mov al, INT_num ; Restore the original vector by
mov dx, old_int_off ; installing the vector the same way
mov ds, old_int_seg ; you installed your ISR vector.
mov ah, 25h
int 21h
pop ds ; Restore the registers.
pop dx
pop cx
pop bx
pop ax
sti ; Enable interrupts.
ret
_remove_ISR endp
IRQ_TEXT ENDS
END
/***************************************************************************
* Here is an example of a C program that would use the routines in
* IRQESP.ASM.
****************************************************************************/<P>
extern void far install_ISR();
extern void far remove_ISR();
extern void far IRQ_Handler();
int io_buffer[1000];
int io_count;
main()
{
int i;
/* Initialize the sample counter. */
io_count = 0;
/* Install the interrupt service routine, IRQ_Handler(), for IRQ 5. */
install_ISR (5, IRQ_Handler);
/* Program your adapter board as necessary to generate the interrupts. */
/* When done (perhaps when io_count reaches 1000), disable your board's
interrupts. */
/* Re-install the previous interrupt service routine. */
remove_ISR ();
/* Look at the data. */
for (i=0; i< io_count; i++) printf("\n%d",io_buffer[i]);
}
Conclusion
Interrupts can be a powerful mechanism for servicing data acquisition and
control operations. With interrupts, the processor can respond quickly and
efficiently to data acquisition hardware. Besides simply transferring
acquired data, interrupts can be used to synchronize different events, or
process data as it is acquired. However, interrupts are not always the best
method for data acquisition. DMA is useful for moving blocks of data at
high rates with minimal processor intervention. Polling is the most
straightforward technique, but not very effective for multiple devices or
high throughput rates.
The PC design uses the Intel 8259A programmable interrupt controller to
provide several prioritized interrupts for the I/O bus. A data acquisition
board plugged into the I/O bus can interrupt the processor by asserting the
appropriate interrupt line. The processor then jumps to an interrupt
service routine located at a preprogrammed address. The example program,
IRQESP.ASM, included in this application note shows you how to program
interrupts.
Alternatively, National Instruments has developed sophisticated driver
software for PC/XT/AT, Micro Channel, and EISA data acquisition hardware.
This software includes high-level, easy-to-use functions that take
advantage of the interrupt and DMA capabilities of the data acquisition
hardware. Many applications developers who use driver-level software
instead of low-level hardware programming realize a substantial savings in
time and effort.
References
IBM Corporation. 1984 IBM Personal Computer AT Technical Reference.
IBM Corporation, Boca Raton, FL, 1984.
IBM Corporation. 1981 IBM Personal Computer Technical Reference.
IBM Corporation, Boca Raton, FL, 1981.
IBM Corporation. 1988 IBM Personal System/2 Hardware Interface Technical
Reference.
IBM Corporation, Boca Raton, FL, 1988.
Microsoft Corporation. 1984 Microsoft MS-DOS Operating System Programmer's
Reference Manual (PN 036-014-003). Microsoft Corporation, Bellevue, WA,
1984.
---------------------------------------------------------------------------
If you have further questions or comments feel free to contact us.
[Image] ..Back to Table of Contents
1 1 340022-01 © Copyright 1991 National Instruments Corporation. All rights
reserved. April 1991
---------------------------------------------------------------------------
[Image]
Home|Catalog|Newsletter|Support|Solutions|Worldwide|Education|Request Form
Comments
---------------------------------------------------------------------------
(c)1996 National Instruments