Skip to main content Link Menu Expand (external link) Document Search Copy Copied
CS 162 Project 0

Debug

The faulting instruction you observed in GDB should match the one you found in #3. Now that you have determined the faulting instruction, understood the purpose of the instruction, and walked through how the kernel initializes a user process, you are in a position to modify the kernel so that do-nothing runs correctly.

  1. (Autograder test 1) Modify the Pintos kernel so that do-nothing no longer crashes. Your change should be in the Pintos kernel, not the userspace program (do-nothing.c) or libraries in proj-pregame/src/lib. This should not involve extensive changes to the Pintos source code. Our staff solution solves this with a single-line change to process.c. After making this change, the do-nothing test should pass but all others will likely fail. Note: It is okay if your change seems like a hack. You will implement a better fix in the user programs project.

  2. (Autograder test 2) Now that you understand how to fix do-nothing, take a look at the do-not-much test. This test differs from do-nothing in that it returns argc instead of simply 162. For now, you can assume that do-not-much does not take in command-line arguments. (How does this relate to argc?) Implement a change that passes do-not-much. Again, this solution can be a hack. Hint: Refer back to your answer in Problem 5.

  3. (Autograder test 3) It is possible that your fix also works for the stack-align-v test, but there are solutions for do-nothing and do-not-much that do not. Take a look at the stack-align-v test. It behaves similarly to do-not-much, but it should return the value of esp % 16. Write down what this program should return (Hint: this can be found in stack-align-v.ck) as well as why this is the case. You may wish to review stack alignment from Discussion 0.

  4. Modify your fix if necessary so that all three of do-nothing, do-not-much, and stack-align-v pass. Please copy your solution to passing stack-align-v and justify it with explicit references to steps of the calling convention and corresponding offset values. Remember to run the autograder at this point if you have not already.

  5. Re-run GDB as before on do-nothing. Execute the loadusersymbols command, set a breakpoint at _start, and continue, to skip directly to the beginning of userspace execution. Using the disassemble and stepi commands, execute the do-nothing program instruction by instruction until you reach the int $0x30 instruction in proj-pregame/src/lib/user/syscall.c. At this point, print the top two words at the top of the stack by examining memory (Hint: x/2xw $esp) and copy the output.

  6. The int $0x30 instruction switches to kernel mode and pushes an interrupt stack frame onto the kernel stack for this process. Continue stepping through instruction-by-instruction until you reach syscall_handler. What are the values of args[0] and args[1], and how do they relate to your answer to the previous question?

Now, you can continue stepping through Pintos. Having completed running do-nothing, Pintos will proceed to shut down because we provided the -q option on the kernel command line. You can step through this in GDB if you’re curious how Pintos shuts down.

Congratulations! You’ve walked through Pintos starting up, running a user program to completion, and shutting down, in GDB. Hopefully this guided exercise helped you get acquainted with Pintos. Be sure to push your code to GitHub with the small change you made in order to make the do-nothing and stack-align-v tests pass. Check that the autograder gives you a full score on the coding portion.

Note: Your written answers will be graded by the readers, not by the autograder.