CDA 4102/CDA 5155: Fall 2023
Hello, dear friend, you can consult us at any time if you have any questions, add WeChat: daixieit
CDA 4102/CDA 5155: Fall 2023
You will create a simple RISC-V simulator which will perform the following two tasks. Please develop your project in one (C, C++, Java or Python) source file to avoid the stress of combining multiple files before submission and making sure it still works correctly.
• Load a specified RISC-V text file1 and generate the assembly code equivalent to the input file (disassembler). Please see the sample input file and disassembly output in the project assignment.
• Generate the instruction-by-instruction simulation of the RISC-V code (simulator). It should also produce/print the contents of registers and data memories after execution of each instruction. Please see the sample simulation output file in the project assignment.
You do not have to implement any exception or interrupt handling for this project. We will use only valid testcases that will not create any exceptions. Please go through this document first, and then view the sample input/output files in the project assignment, before you start implementing the project.
Instructions
For reference, please use the RISC-V Instruction Set Architecture (riscv-ISA.pdf in the course website) to see the format for each instruction and pay attention to the following changes. For example, we introduced a break instruction, modified the opcode format, etc. In other words, you should exactly follow the details from riscv-ISA.pdf except the changes outlined in this document.
Your disassembler & simulator need to support the three categories of instructions shown in Figure 1.
|
Category-1 |
Category-2 |
Category-3 |
Category-4 |
|
beq, bne, blt, sw |
add, sub, and, or |
addi, andi, ori, sll, sra, lw |
jal, break |
Figure 1: Three categories of instructions
The format of Category-1 instructions is described in Figure 2. If the instruction belongs to Category-1, the right two bits (least significant bits) are always “00” preceded by 4 bits Opcode. Note that instead of using 7 bits opcode in RISC-V, we use 5 bits opcode as described in Figure 3. The remaining 25 bits of the instruction binary is exactly the same as the original RISC-V instruction set for that specific instruction.
|
Same as RISC-V instruction (25 bits) |
Opcode (5 bits) |
00 |
Figure 2: Format of Instructions in Category-1
Please pay attention to the exact description of instruction formats and its interpretation in MIPS instruction set manual. For example, in case of jal instruction, the 21-bit offset is shifted left by one bit (padded with 0 at LSB side), sign extended to form 32 bits, and then added to the address of the jal instruction to form the target address. Similarly, for beq, bne and blt instructions, the 12-bit offset is formed by concatenating bits in [31:25] with bits in [11:7] (see S-type format in risc-ISA.pdf), and then the 12-bit offset is shifted left by one bit, sign extended to form 32-bits, and added to the address of the current instruction to form the target address. The 12-bit offset Please note that we do not consider delay slot for this project. In other words, an instruction following the branch instruction should be treated as a regular instruction (see sample_simulation.txt).
|
Instruction |
Opcode |
|
beq |
00000 |
|
bne |
00001 |
|
blt |
00010 |
|
sw |
00011 |
Figure 3: Opcode for Category-1 instructions
If the instruction belongs to Category-2 which has the form “dest ← src1 op src2”, the rightmost two bits (least significant bits) are always “01” as shown in Figure 4. Then the following 5 bits serve as opcode as listed in Figure 5.
|
Same as RISC-V instruction (25 bits) |
Opcode (5 bits) |
01 |
Figure 4: Format of Category-2 instructions where both sources are registers
|
Instruction |
Opcode |
|
add |
00000 |
|
sub |
00001 |
|
and |
00010 |
|
or |
00011 |
Figure 5: Opcode for Category-2 instructions
If the instruction belongs to Category-3 which has the form “dest ← src1 op immediate_value”, the rightmost two bits (least significant bits) are always “ 10” . Then 5 bits for opcode as indicated in Figure 6. The instruction format is shown in Figure 7.
|
Instruction |
Opcode |
|
addi |
00000 |
|
andi |
00001 |
|
ori |
00010 |
|
sll |
00011 |
|
sra |
00100 |
|
lw |
00101 |
Figure 6: Opcode for Category-3 instructions
|
Same as RISC-V instruction (25 bits) |
Opcode (5 bits) |
10 |
Figure 7: Format of Category-3 instructions with source2 as immediate value
If the instruction belongs to Category-4, the rightmost two bits (least significant bits) are always “ 11” . Then 5 bits for opcode as indicated in Figure 8. The instruction format is shown in Figure 9.
|
Instruction |
Opcode |
|
jal |
00000 |
|
break |
11111 |
Figure 8: Opcode for Category-4 instructions
|
Same as RISC-V instruction (25 bits) |
Opcode (5 bits) |
11 |
Figure 9: Format of Category-4 instructions
All signed numbers should be interpreted using 2’s complement arithmetic. Note that the signed numbers can be in registers, data memories or inside an instruction (e.g., the immediate field is signed for addi). Most importantly, each location (register or data memory) can be treated differently based on the context. For example, an arithmetic instruction (e.g., add) will treat the content of a register as a signed number (in 2’s complement arithmetic), whereas a logical operation (e.g., and) will treat the same register content as an unsigned number (sequence of bits). Please go through riscv-ISA.pdf to understand how each instruction treats its operands (signed or unsigned). Assume that all unassigned register and data memory locations are 0.
Sample Input/output Files
Your program will be given a text input file (see sample.txt). This file will contain a sequence of 32-bit instruction words starting at address "256". The final instruction in the sequence of instructions is always break. There will be only one break instruction. Following the break instruction (immediately after break), there is a sequence of 32-bit 2's complement signed integers for the program data up to the end of the file. The newline character can be either “\n” (linux) or “\r\n” (windows). Your code should work for both cases. Please download the sample input/output files using “Save As” instead of using copy/paste of the content.
Your RISC-V simulator (with executable name as Vsim) should accept an input file (inputfilename.txt) in the following command format and produce two output files in the same directory: disassembly.txt (contains disassembled output) and simulation.txt (contains the simulation trace). Please hardcode the names of the output files. Please do not hardcode the input filename. It will be specified when running your program. For example, it can be “sample.txt” or “test.txt” .
Vsim inputfilename.txt
Correct handling of the sample input file (with possible different data values) will be used to determine 60% of the credit. The remaining 40% will be determined from other valid test cases that you will not have access prior to grading. It is recommended that you construct your own sample input files with which to further test your disassembler/simulator. It is okay to share your new testcases with other students in the class as long as it does not lead to similarity in the project source code.
The disassembler output file should contain 3 columns of data with each column separated by one tab character (‘\t’ or char(9)). See the sample disassembly file in the project1 assignment.
1. The text (e.g., 0’s and 1’s) string representing the 32-bit data word at that location.
2. The address (in decimal) of that location
3. The disassembled instruction.
Note, if you are displaying an instruction, the third column should contain every part of the instruction, with each argument separated by a comma and then a space (“, ”).
The simulation output file should have the following format.
20 hyphens and a new line
Cycle < cycleNumber >:< tab >< instr_Address >< tab >< instr_string >
< blank_line >
Registers
x00: < tab >< int(x0) >< tab >< int(x1) >...< tab >< int(x7) >
x08: < tab >< int(x8) >< tab >< int(x9) >...< tab >< int(x15) >
x16: < tab >< int(x16) >< tab >< int(x17) >...< tab >< int(x23) >
x24: < tab >< int(x24) >< tab >< int(x25) >...< tab >< int(x31) >
< blank_line >
Data
< firstDataAddress >: < tab >< display 8 data words as integers with tabs in between >
..... < continue until the last data word >
Display all integer values in decimal. Immediate values should be preceded by a “#” symbol. Note that some instructions take signed immediate values while others take unsigned immediate values. You will have to make sure you properly display a signed or unsigned value depending on the context.
Because we will be using “diff -w -B” to check your output versus the expected outputs, please follow the output formatting. Mismatches will be treated as wrong output and will lead to score penalty.
The project assignment contains the following sample programs/files to test your
disassembler/simulator.
• sample.txt : This is the input to your program.
• sample_disassembly.txt : This is what your program should produce as disassembled output.
• sample_simulation.txt : This is what your program should output as simulation trace.
2023-09-22