MCA Interrupt Procedures

Sources:
SA23-2647-00 RS/6000 Hardware Technical Reference - Micro Channel Architecture, 1st Ed., 1990
The Micro Channel Architecture Handbook (Chet Heath, PBUH!)
   You need to borrow it from the Internet Archive. Unfortunately, it is locked up tighter than a drum, so it is easier doing screen caps then recognizing it... YMMV, LFO
US5187781 Shared hardware interrupt circuit for personal computers


Interrupt Procedures

This section describes how all devices use interrupts. The procedure for sharing interrupts uses a request for interrupt services that is detected by the level of the interrupt request signal (level sensitive). These procedures involve the interaction between the hardware and an interrupt service routine.

To initiate an interrupt request, a device drives its -IRQ(N) active (N represents the assigned interrupt level for the device) and holds the signal active until it is reset by the interrupt service routine.

Each device provides an interrupt-pending bit within its address space. This bit is set by the device when it has an interrupt request pending and is reset by the interrupt handler when the interrupt request is serviced.

Note: The drivers for the interrupt signal must be disabled when the device is disabled.


Figure 1-60. Typical Adapter Interrupt Sharing Implementation

An example of the sequence of the hardware and interrupt service routine interaction is shown in the following.

    Hardware Operation

  1. An interrupt condition causes the hardware to drive -IRQ(N) active and sets an interrupt-pending latch, which can be read by software.

  2. Software Operation

  3. A master begins executing code at the beginning of the appropriate chain of interrupt handlers.

  4. The interrupt handler reads the interrupt-pending latch of the first device in the chain. If the latch is not set, the next device in the chain is tested. When the reporting device is detected, the handler executes the service routine.

  5. The interrupt service routine operates the device hardware.

  6. Hardware Operation

  7. The device hardware resets the interrupt-pending latch and the interrupt request because of interrupt-service-routine actions.

  8. Software Operation

  9. The interrupt service routine completes the interrupt processing.

  10. Hardware Operation

  11. If another interrupt is pending (-IRQ(N) driven active by another device), the sequence starts again at 1.


Edge Triggered Interrupts

"The PC AT uses an edge-triggered interrupt request signal. That is, where an adapter requests an interrupt so that its device may be serviced, it does so, e.g, by first driving the interrupt request line to a low level and then to a high level. The transition between the low level and the high level is received by the system as an interrupt request signal.

Where the computer is being used for multi-tasking and or by multi-users, it may be desirable to add more devices than there are hardware interrupt levels. This can create system problems. For example, if two tasks were to "talk" to the same printer at the same instant and time, the result would be incomprehensible. Still another task might require a special printer or plotter for output. Most interrupt-supporting adapters were designed to first drive the interrupt request line to a low level and then to a high level to provide a low-to-high edge signal which is interpreted, under control of software drivers, by the microprocessor and by the interrupt controllers as a request for servicing, i.e. a hardware interrupt request signal.

A problem arose where two such devices were connected to the same level interrupt and the devices requested interrupts at the same time or where one interrupt was being handled when a second request was generated. The occurrence of these conditions could cause lost signals and even physical damage to the I/O device adapter.

The PC AT could be used to provide sharing of an interrupt level between multiple devices, however, this required a change in design of many device adapters which were to share an interrupt request level and created compatibility problems with earlier designed cards. Adapter cards previously designed for non-interrupt level sharing were designed to hold the interrupt request line at a low level and then to drive the line high to generate a request signal.

In contrast the shared-interrupt card design, to operate in the PC AT, the PC XT or the PC called for the interrupt request line to float high, through pull-up resistors on each adapter and typically to be driven by open-collector drivers. Each adapter on the line may request an interrupt by momentarily pulsing the line to a low level. The high-to-low transition arms the Interrupt Controller; the low-to-high transition generates the interrupt. It was accordingly not possible in most cases to use the adapters not designed for shared interrupt use in a shared environment. It has been found that the use of more than one adapter card on an interrupt request level, whether specifically designed for interrupt sharing or not, may cause physical damage or loss of signal where more than one of the adapters is active at the same time."

"Referring now to the drawings there is shown in FIG. 1 two device adapter cards shown generally as 1a and 1b. Included on adapter 1a, 1b are edge-triggered interrupt request circuits shown generally as 3a and 3b. As described above the interrupt request signal is a momentary pulse generated by the interrupt request circuit, 3a, 3b. The interrupt request line 5 is first pulsed low to arm the Interrupt Controller 22, (see FIG. 3B), and then pulsed high to request the interrupt. The transition from low to high generated by circuits 3a and 3b and transmitted to the interrupt request line 5 is received by the microprocessor and interrupt controllers as a request for attention by an I/O adapter 1a, 1b. To create this transition of the interrupt request line 5 from low to high line 5 must first be driven low and then driven high. Circuits 3a and 3b provide this transition. Circuits 3a, 3b in this exemplary instance includes power supply lines 7a, 7b and ground lines 9a, 9b which are connected to a common source of power and ground on the system planar when the adapter cards are plugged into the I/O connectors or slots 22 (see FIG. 3C). Each circuit 3a, 3b incorporate a low transistor 11a, 11b; a high transistor, 13a, 13b, and a diode 15a, 15b. An interrupt request signal is generated as follows: The low transistor, 11a, 11b is turned on to drive the interrupt request line 5 to a low level by connecting it, through adapter interrupt request output line 17a, 17b to ground 9a, 9b. The signal is then generated by turning the low transistor 11a, 11b off and high transistor 13a, 13b on. This sequence provides the low to high transition, i.e. a rising edge signal, that tells the system that a device adapter is requesting attention through a hardware interrupt request.

Short Circuit Interrupt on PC vs. MCA Bus

"Conventional edge-triggered cards in PC/XT/AT designed computers use these bipolar driver circuits. Normally, they are not designed to share interrupts. Even if they did, the quick pulse that produces the voltage edge needed to signal the interrupt would not unduly stress the circuitry. It would likely be too brief to do damage. Run the same circuity in level-sensitive mode or repeatedly overlap the pulses, and the first shared interrupt could result in a circuit failure". Micro Channel Architecture Handbook, page 96.

So, IBM was faced with a problem if they used the older bipolar "Totem Pole" circuit. "When the outputs of two bipolar chips are connected together, they fight one another. When one chip is on and the other is off, the result is effectively a short circuit. The "on" chip pushes out all the current it can to try and force the "off" output into an "on" condition - something it cannot do. The "on" chip may pump more current than it is designed to, and the "off" chip is forced to absorb all the current to remain off. With time, the resulting overload can damage either or both chips." Micro Channel Architecture Handbook, pages 85-86.

Note: This illustration from "The Micro Channel Architecture Handbook" uses conventional current flow, as evidenced by the arrow head pointing down into ground. Actual electron flow is up from ground to the more positive end. Think of your car battery. Negative ground. Except the old stuff, or European cars... Also note that the emitter arrow head on the transistors points to the negative side (electrons flow into the arrow point).

"The problem with using this design in a shared interrupt environment can now be understood. As an example, assume that adapter la has driven interrupt request line 5 to a low level by connecting interrupt request line 5 to ground 9a through active transistor 11a at the same time that adapter 1b high transistor 13b is active. This would result in momentarily grounding the system power supply source through power supply line 7b, transistor 13b, diode 15b, and transistor 11a. This destructive passage of current could result not only in loss of the interrupt request signal from adapter 1b, but in damage to the physical components such as the transistors. Resistors could be placed between the adapter cards and the interrupt request line to limit the current, however, this solution would decrease the "noise" margin for the logic and aggravate noise problems in the system. These problems can be overcome by the system of the present invention."

Level Sensitive Interrupts

"The present invention allows multiple adapters to be used on a single interrupt level without loss of an interrupt request signal or physical failure of a system element. The present invention utilizes diodes positioned between each adapter and the interrupt request line. Further, a pull-down resistor and a Schmitt trigger are incorporated in the interrupt request line in the preferred embodiment. The diodes prevent the short-circuiting of the adapters when two or more active adapters are in electrical contact with the hardware interrupt request line at the same time. The pull-down resistor and Schmitt trigger are used to convert the diode-modified interrupt signal to a signal which can be utilized by the microprocessor and interrupt controller system under control of software device drivers."

Referring now to FIG. 2, there is shown a modification to the circuitry shown in FIG. 1 in accordance with the present invention. The advantages of this invention may be obtained by incorporating diodes 19a, 19b between the adapter interrupt request output lines 17a, 17b and the interrupt request control line 5. Diodes 19a, 19b prohibit the passage of current from one adapter e.g. 1a to a second adapter 1b. It is well known that diodes can be designed to prohibit the passage of electrical current in one direction only, and to limit the amount of current passing. The preferred diode for the present concept would be a diode that had a minimal forward voltage drop characteristic. The introduction of these diodes requires other modifications to the system to allow adapter cards 1a, 1b to remain compatible with existing systems. A pull-down resistor and ground shown generally, as 23 are used to keep the interrupt request line 5 at a low level. A Schmitt trigger 21 is provided to detect a slow transition of the signal placed on interrupt request line 5 by adapter card circuit 3a, 3b. The output of the Schmitt Trigger 21 is connected to the input of an Interrupt Controller 20 (see FIG. 3B). This combination of components provide in response to the low-to-high signal generated by the adapters 1a, 1b over adapter output lines 17a, 17b a signal on the interrupt request line 6, that transitions from low to high and is suitable as an edge-trigger interrupt request signal.

In operation, if any of the drivers, circuits 3a and 3b are active high, the interrupt request line 5 will be high. If all drivers are low, or if all drivers are not active, the interrupt request line 5 will be low. The Schmitt trigger inverter 21 inverts this signal to provide a low signal to the processor 29 (see FIG. 3A) and interrupt request controllers 20 over interrupt request bus 6 (see FIG. 3 for a typical planar bus system layout). The following Table demonstrates the possible combination of operations

Adapter Output Interrupt Request Signal
1a 1b Original Inverted
0 0 0 1
0 1 1 0
1 0 1 0
1 1 1 0

The original signal is the logical "OR" of all active high requests usable for low-to-high edge detection processes. An inverted signal is the low active level request potentially useable for low level-sensitive, as contrasted to edge-triggered, detection for certain adapter combinations.

No invalid levels are produced and no potential overheating of the adapter circuits will result. The inverted output combination is similar to that produced by open collector drivers in a level sensitive interrupt system such as is used in the Family II models of personal computers listed above.

It should be pointed out that custom software drivers are required to be installed in the operating system of Family I computers for adapters 1a, 1b to be able to share interrupt levels in accordance with the present concept. Further, only adapters that can respond to these drivers and hold the interrupt request active until the activating adapter can be selected from the other possible adapters could be used. These drivers are written to hold the interrupt request low except when the interrupt is requested. On existing adapters, where an active edge-triggered interrupt request is not held low by the hardware, or where software drivers cannot hold the interrupt request line 5 low, these adapters could not share an interrupt level. Many adapters however meet these requirements for sharing, other adapters could be designed to work in non shared edge-triggered systems and also work in the present system.

The present concept provides a level sensitive system that can be operated in a manner similar to the level sensitive system exemplified by the Family II computers. The Schmitt trigger provides a level sensitive, as contrasted to edge-triggered, signal to the processor 24 and interrupt controllers 22."

Content created and/or collected by:
Louis Ohland, Peter Wendt, William Walsh, David Beem, Tatsuo Sunagawa, Jim Shorney, Tim Clarke, Kevin Bowling, Tomáš Slavotínek, and many others.

Ardent Tool of Capitalism - MAD Edition! is maintained by Tomáš Slavotínek.
Last update: 19 Nov 2021 - Changes & Credits | Legal Info & Contact