Spring 2010: EE 382N-4   Unique: 16820

Advanced Embedded Systems Architecture

Lab Assignment #2: Due Mar 21, 2010

 

 

AIM: Develop an interrupt handler to measure the interrupt latency of Linux running on the TLL platform using the I/O ports on the FPGA.

 

Note: This is a group lab exercise and every group should demonstrate the working of the code on the board. A group can have a maximum of 2 members.

 

Tutorial: device_driver + FPGA. Go through this link first.

 

In this tutorial, an adder is implemented on the FPGA. The user application code interacts with the adder through the device driver.

 

 

Specification:

 

The functional diagram of the required setup is shown in the block diagram below.

 

 

 

1)    The user application is the software running on linux which in turn is running on the ARM processor.

a.    The application should send 10 pairs of numbers to the device driver, issue the `add` command and once addition is complete (for all 10 pairs), receive the results.

1.    However, this should be implemented in a while loop (free running – without user intervention).

2.    After the `add` command is sent to the device driver, the loop will wait for the interrupt from the FPGA and then issue the command to read the sum (#10).

b.    It will also host the interrupt signal (SIGIO) handler to measure the latency from the time the input data are sent to the FPGA to the time when the interrupt is received. You can use the C function gettimeofday() to note the time.

c.    List the average, minimum and maximum latency over 10 runs (run your program 10 times) as part of your report.

d.    Discuss your user application code, as part of your report.

 

2)    The device driver will be designed as a loadable module, thus saving kernel memory.

a.    The device driver file adder.c (from the demo) should be modified to host the interrupt functions.

b.    The interrupt functions will initialize the required handlers, request for the specific interrupt to be serviced etc.

c.    An example of the same setup for GPIO (as against FPGA) is given below. Relate the following programs to lecture 9 for a better understanding of the code.

1.    main.c is the user application

2.    mx21_gpio.c is the corresponding device driver

3.    The interrupt number for the GPIO pin used here is #107. The corresponding interrupt number for the FPGA is #240.

4.    The MX21 board documentation at /scratch/Documentation/MC9328MX21RM.pdf. Chapter 15 contains a note about the GPIOs. The configuration details of the GPIO registers are given in that chapter. This example configures the port PB10 as output and loopbacked to PB11 as input. You do not require these for the corresponding FPGA implementation.

d.    The above example can be used as a reference to modify the adder.c program.

e.    Configure the interrupt as rising edge sensitive.

f.     Notify the application code about the occurrence of the interrupt using the kill_fasync routine, similar to the one in mx21_gpio.c

g.    The `read` command to read the results should be implemented as a separate read function (refer copy_to_user function) and not part of the ioctl function. This will make the read process faster.

h.    Discuss the changes that you made to the example device driver (adder.c), as part of your report.

 

3)    The adder is implemented in the FPGA (top.v)

a.    Receives data from the user application through the device driver.

b.    FPGA is memory mapped starting from 0xD3000000 and 0xCC000000 of the ARM processor as stated in the docs.

c.    The top.v (from the demo) is the verilog file that gets burned into the FPGA. Modify it

1.    to receive, store and add the 10 pairs of numbers on the reception of an `add` command,

2.    to raise the interrupt signal when the addition is complete,

1.    MZ_CPLD_MISC14 is the signal corresponding to interrupt #240, in top.v

3.    to make the interrupt signal a pulse which will go back to zero state automatically after a certain delay,

4.    to have one of the DIP switches (or address 0xCC000000 has the value ‘1’ in it) on the board to act as the enable for the interrupt,

1.    Interrupts should not be generated when the DIP switch is off (or when 0xCC000000 has say ‘2’ written into it)

2.    DIP_SW in top.v is the signal corresponding to the switches.

3.    In top.v, 0xCC000000 – 0xCFFFFFFF is the bigger of the memory mapped regions corresponding to the FPGA, the other one is 0xD3000000 – 0xD3FFFFFF

5.    to toggle an LED whenever address 0xD3000004 is accessed indicating the data arrival from the ARM processor (optional), and

6.    to map the interrupt pulse to another LED on the board for observation (optional).

d.    Discuss the changes to top.v, as part of your report.