Anti-starvation interrupt protocol
View Patent ↗An anti-starvation interrupt protocol for use in avoiding livelock in a multiprocessor computer system is provided. At least one processor is configured to include first and second control status registers (CSRs). The first CSR buffers information, such as interrupts, received by the processor, while the second CSR keeps track of the priority level of the interrupts. When an interrupt controller receives an interrupt, it issues a write transaction to the first CSR at the processor. If the first CSR has room to accept the write transaction, the processor returns an acknowledgement, whereas if the first CSR is already full, the processor returns a no acknowledgment. In response to a no acknowledgment, the interrupt controller increments an interrupt starvation counter, and checks to see whether the counter exceeds a threshold. If not, the interrupt controller waits a preset time and reposts the write transaction. If it does, the interrupt controller issues a write transaction having a higher priority to the second CSR. In response, the processor copies all of the pending interrupts from the first CSR into the memory subsystem, thereby freeing up the first CSR to accept additional write transactions.
1. A method for avoiding starvation of interrupts of a computer system, the computer system having at least one processor for servicing the interrupts, a memory subsystem, one or more input/output (I/O) devices configured to issue interrupts and at least one interrupt controller, the method comprising the steps of:
providing an interrupt queue register having a fixed number of entries for storing interrupts to be serviced by the at least one processor;
posting interrupts generated by the one or more I/O devices to the interrupt queue register;
determining whether the interrupt queue register is full and unable to store a given interrupt; and
if the interrupt queue register is full, one of waiting a preset time and reposting the given interrupt to the interrupt queue register, and draining the contents of the interrupt queue register into the memory subsystem so that the interrupt queue register can store the given interrupt.
2. The method of claim 1 further comprising the steps of:
if the interrupt queue register has a free entry for storing the given interrupt, sending an acknowledgement message (ACK) to the interrupt controller; and
if the interrupt queue register is full and cannot store the given interrupt, sending a non-acknowledgment message (NACK) to the interrupt controller.
3. The method of claim 2 further comprising the steps of:
providing an interrupt starvation counter that is operable by the interrupt controller;
in response to receiving an ACK from the processor, clearing the interrupt starvation counter; and
in response to receiving a NACK from the processor, adjusting the interrupt starvation counter.
4. The method of claim 3 wherein the step of adjusting comprises one of incrementing or decrementing the interrupt starvation counter.
5. The method of claim 3 further comprising the steps of:
determining whether the interrupt starvation counter exceeds a threshold; and
if the interrupt starvation counter does not exceed the threshold, performing the step of waiting the preset time and reposting the given interrupt to the interrupt queue register.
6. The method of claim 5 further comprising the steps of:
providing an interrupt request register having a plurality of bit positions that are settable to indicate a priority level of an interrupt posted to the processor;
if the interrupt starvation counter exceeds the threshold, setting the interrupt request register with a high priority indication; and
in response to the setting of the interrupt request register with the high priority indication, performing the step of draining the contents of the interrupt queue register into the memory subsystem.
7. The method of claim 6 further comprising the step of, in response to the posting of an interrupt to the interrupt queue register, setting the interrupt request register with a low priority indication.
8. The method of claim 7 wherein the interrupt request register includes a first bit segment the setting of which is indicative of a low priority interrupt and a second bit segment the setting of which is indicative of a high priority interrupt.
9. The method of claim 8 wherein each bit segment of the interrupt request register is associated with an interrupt line leading into the processor.
10. The method of claim 6 wherein the contents of the interrupt queue register drained into the memory subsystem are placed in a first-in-first-out (FIFO) queue.
11. The method of claim 10 further comprising the step of servicing the interrupts stored at the FIFO queue.
12. The method of claim 11 wherein the interrupts issued by the I/O devices are one of level sensitive interrupts (LSIs) and message signaled interrupts (MSIs).
13. The method of claim 1 wherein the interrupts issued by the I/O devices are one of level sensitive interrupts (LSIs) and message signaled interrupts (MSIs).
14. The method of claim 1 wherein the step of draining the contents of the interrupt queue register into the memory subsystem is performed by Privileged Architecture Library (PAL) code running at the processor.
15. The method of claim 1 wherein
the computer system includes a plurality of processors configured to issue interrupts, and
the interrupt controller receives interrupts from one or more selected processors of the computer system and posts such processor initiated interrupts to the interrupt queue of the at least one processor as with the I/O device interrupts.
16. A computer system comprising:
a plurality of input/output (I/O) devices configured to issue interrupts;
at least one processor for servicing the interrupts;
a memory subsystem;
an interrupt controller configured to receive the interrupts from the I/O devices and to post those interrupts to the at least one processor for servicing; and
an interrupt queue register for storing the interrupts posted to the at least one processor, wherein
upon receiving a newly posted interrupt, the at least one processor determines whether the interrupt queue register is already full and, if so, drains the contents of the interrupt queue register into the memory subsystem.
17. The computer system of claim 16 further comprising an interrupt starvation counter that is operable by the interrupt controller, wherein
if the interrupt queue register has a free entry for the new interrupt, sending an acknowledgement (ACK) message from the at least one processor to the interrupt controller, and
in response to the ACK, clearing the interrupt starvation counter.
18. The computer system of claim 17 further comprising
if the interrupt queue register does not have a free entry for the new interrupt, sending a non-acknowledgement (NACK) message from the at least one processor to the interrupt controller,
in response to the NACK, adjusting the interrupt starvation counter,
determining whether the interrupt starvation counter exceeds a threshold,
if the interrupt starvation counter does not exceed the threshold, waiting a predetermined time and reposting the new interrupt to the at least one processor.
19. The computer system of claim 18 further comprising an interrupt request register having a plurality of bit positions that are settable to indicate a priority level of an interrupt posted to the processor, wherein
if the interrupt starvation counter exceeds the threshold, setting the interrupt request register with a high priority indication; and
in response to the setting of the interrupt request register with the high priority indication, draining the contents of the interrupt queue register into the memory subsystem.
20. The computer system of claim 16 wherein
the computer system has a plurality of processors configured to issue interrupts, and
the interrupt controller posts the processor initiated interrupts to the at least one processor for servicing as with the I/O device interrupts.