2020/9/28 CS 241 — Fall 2020 — Assignments 3 https://www.student.cs.uwaterloo.ca/~cs241/a3/ 1/6 CS 241 — Fall 2020 — Assignment 3 Assignments for CS 241 ← Assignment 2 Assignment 3 Assignment 4 → Friday, September 25th at 5:00 pm Friday, October 9th at 5:00 pm Friday, October 23rd at 5:00 pm P1 • P2 • P3 • P4 • P5 • P6 • P7 • P8 • P9 • P10 • Comprehensive Test In Assignment 3, you will incrementally write an assembler for MIPS assembly language (CS241 dialect). Reminder: For this and future assignments, be sure to run the command source /u/cs241/setup to gain access to the CS 241 tools. What Is An Assembler? So far in the course, you have used various tools like cs241.wordasm, cs241.binasm, mips.twoints and mips.array. The tools cs241.wordasm and cs241.binasm are MIPS assemblers. They convert ASCII text commands into MIPS machine language that can be executed by the MIPS emulators, which are mips.twoints and mips.array. This is important to understand: an assembler does not execute code. If the user writes something like div $0, $0, this will obviously cause an error if it is executed. However, the assembler's job is just to translate this instruction into machine language. When the emulator executes it, the error will happen, but you are not writing an emulator, just an assembler. The cs241.wordasm assembler you used on Assignment 1 was a primitive assembler that only supports the .word directive and nothing else. The cs241.binasm assembler you used on Assignment 2 supports all features of MIPS assembly language. Your task on Assignment 3 is essentially to write your own version of cs241.binasm. This means that, for valid assembly programs, you can check that your output is correct by comparing it to cs241.binasm using the diff command: ./my-assembler < input.asm > my-output.mips cs241.binasm < input.asm > expected-output.mips diff my-output.mips expected-output.mips If the files are the same, diff will produce no output, and this means your output is correct. Otherwise it will probably say something like Binary files my-output.mips and expected- output.mips differ. For invalid assembly programs, the only requirement is that your assembler produce an error message containing the string ERROR to standard error. The error message does not need to 2020/9/28 CS 241 — Fall 2020 — Assignments 3 https://www.student.cs.uwaterloo.ca/~cs241/a3/ 2/6 match the one produced by cs241.binasm. Marmoset Notes For each question, Marmoset has public and release tests, and often has secret tests as well. When you release test a submission, Marmoset will only report on public and release tests. No information about secret tests will be visible until after the assignment deadline. If you are using C++ for this assignment, Marmoset will run your programs with Valgrind, a tool that checks for memory-related errors. If Valgrind detects an error in your program, you will fail the Marmoset tests. A guide for debugging Valgrind errors is available. Marmoset is not yet available for this assignment. An announcement will be made on Piazza when the Marmoset tests are available. The mark weightings for the problems will be made available at the same time as the Marmoset tests. Writing An Assembler In the following problems, you will implement an assembler for progressively larger subsets of MIPS assembly language. Be sure to read the MIPS Assembly Language Specification. It should answer most questions you have about what is considered valid MIPS assembly language syntax. The problems may be done in either Racket or C++14. See language-specific notes for each option at the end of this document. We have provided a scanner (also called a tokenizer) for MIPS assembly language for each available language option (see language-specific notes). You should use this scanner as a starting point for your assembler. Each problem in requires you to submit a program that reads from standard input and writes to standard output as well as standard error. The input and output specifications are identical regardless of which language you choose. The only difference is that you must submit the appropriate .rkt or .cc file depending on your choice of language. For each problem, we ask you to implement support for additional instructions or features. You may submit the same assembler for all the problems. We encourage you to submit to Marmoset early. As soon as you implement support for the instructions specified by a problem, submit the current version of your assembler to Marmoset. That way, if you do not complete all of the problems before the deadline, you will still get credit for those that you did complete. Hint: Depending on the design decisions you make in your solutions to Problems 1 and 2, you may have to restructure your code to get a working solution to Problem 3. You may have to do further restructuring for Problem 4 and onwards. Therefore, you may want to read and 2020/9/28 CS 241 — Fall 2020 — Assignments 3 https://www.student.cs.uwaterloo.ca/~cs241/a3/ 3/6 understand all the problems before beginning Problem 1. However, if you find this overwhelming, you may find it easier to just focus on the problems in order. The decision is yours. Problem 1 (filename: asm.rkt or asm.cc) Begin by writing an assembler that correctly translates input containing no labels and no instructions other than .word. You may assume that the input to your assembler contains no labels and no instructions other than .word. Your assembler should never crash or leak memory, even if the input is not a valid assembly language program; it should produce an error message and return gracefully if the input is invalid. If the input is not a valid MIPS assembly language program, your assembler should print a message containing the word ERROR in all capitals to standard error and stop. It is good practice, but not a requirement, to embed ERROR within a meaningful error message. If the input contains a correct MIPS assembly language program, your assembler should output the equivalent MIPS machine language to standard output. The error checking and output requirements above apply to this and all future problems on the assignment. Hint: there are relatively few ways in which an assembly language program can be valid (and all the valid forms are spelled out here), but many ways in which it can be invalid. You will find it much easier to write code that looks for valid input and rejects everything unexpected, rather than code that explicitly looks for all the different ways in which the input could be invalid. Problem 2 (filename: asm.rkt or asm.cc) Add support for label definitions to your assembler. In addition, if the input is a correct MIPS assembly program, your assembler should output a symbol table: a listing of the names and values of all defined labels to standard error. The list should be printed on several lines, one line for each label in the input. Each line should consist of the label (without the trailing colon), followed by a space, followed by the value of the label (in decimal). The labels may appear in the symbol table in any order. For example, the following input: begin: .word 2 middle: .word 0 .word 0 end: Should print the following to stderr (but possibly with the lines reordered): begin 0 middle 4 end 12 2020/9/28 CS 241 — Fall 2020 — Assignments 3 https://www.student.cs.uwaterloo.ca/~cs241/a3/ 4/6 In handling labels, you may use any data structure or data structures you choose, but be sure to take efficiency into account. Problem 3 (filename: asm.rkt or asm.cc) Modify your assembler to allow labels to be defined and also to be used as operands. Henceforth, you no longer need to output a symbol table as in Problem 2 (although you will not be penalized if you do so). Problem 4 (filename: asm.rkt or asm.cc) Modify your assembler to correctly handle jr and jalr instructions. Problem 5 (filename: asm.rkt or asm.cc) Modify your assembler to correctly handle add, sub, slt, and sltu instructions. Problem 6 (filename: asm.rkt or asm.cc) Modify your assembler to correctly handle beq and bne instructions with an integer or hex constant as the branch offset. Problem 7 (filename: asm.rkt or asm.cc) Modify your assembler to correctly handle beq and bne instructions with a label as the branch target operand. Problem 8 (filename: asm.rkt or asm.cc) Modify your assembler to correctly handle the lis, mflo, and mfhi instructions. Problem 9 (filename: asm.rkt or asm.cc) Modify your assembler to correctly handle the mult, multu, div, and divu instructions. Problem 10 (filename: asm.rkt or asm.cc) Modify your assembler to correctly handle the sw and lw instructions. Comprehensive Test (filename: asm.rkt or asm.cc) For this problem, you are not required to implement any new features. At this point, your assembler should correctly translate all valid MIPS assembly language programs, and write ERROR to standard error for any input that is not a valid MIPS assembly language program. 2020/9/28 CS 241 — Fall 2020 — Assignments 3 https://www.student.cs.uwaterloo.ca/~cs241/a3/ 5/6 When you are confident that your assembler is complete and correct, submit your assembler to the project A3Comprehensive on Marmoset. Marmoset will test it thoroughly with many large and complex programs, both valid and invalid. This comprehensive test involves large programs which may have as many as 100,000 lines. If your assembler is inefficient (runs in quadratic time or worse in terms of the input size) it will likely exceed Marmoset's time limit and fail some of the test cases. Disable debugging output from your assembler before submitting to this problem. If you produce too much debugging output and exceed Marmoset's file size limit, you will automatically fail the tests. After you submit to the comprehensive test, you may wish to re-submit your assembler to Problems 1 through 10 to ensure that these problems have the most up-to-date version of your code. Your Marmoset mark is based on your best on-time submission, so you will not be penalized for extra submissions to these problems. Click here to return to the top of Assignment 3. Language-Specific Details Racket The provided starter asm_rkt.zip has a function called scan that takes as input a string and returns a list of tokens. The Using Racket in CS 241 document contains hints and techniques for using Racket to write the assembler. See also the comments in the provided scanner. Run a Racket program using the command: racket asm.rkt Click here to return to the top of the page. C++ The provided starter asm_cpp.zip has a method called scan that takes as input a string and returns a vector of tokens. When submitting to Marmoset, if you have chosen C++, you will need to add all your files to a .zip file and submit that to Marmoset. The top level directory of your .zip file must contain the asm.cc file. For example, if your zip file contains a directory called a3 and the asm.cc file is stored under this directory, Marmoset will not be able to find the file. The STL Quick Reference for CS 241 document outlines the parts of the STL most likely to be of use in CS 241. 2020/9/28 CS 241 — Fall 2020 — Assignments 3 https://www.student.cs.uwaterloo.ca/~cs241/a3/ 6/6 You are strongly advised to check for memory-related errors by vetting your programs with Valgrind. To do this, run "valgrind program optionsAndArguments" instead of just "program optionsAndArguments". Marmoset will run your submissions with Valgrind as well, and will reject any submission that contains memory-related errors. Be aware that running Valgrind increases the execution time of your program by a factor of 5 to 20. See the following page on Debugging Valgrind Errors for a discussion of common Valgrind errors and how to resolve them. Compile a program in C++ using the command "g++-6 -g -std=c++14 -lm -Wl,--warn-common,-- fatal-warnings -o asm asm.cc scanner.cc". This command will create a file called asm containing the compiled code. Run the program using the command: ./asm Or use valgrind ./asm to run with Valgrind. Click here to return to the top of the page.
欢迎咨询51作业君