Operating Systems 2014F: Assignment 5

Please submit the answers to the following questions via CULearn by midnight on Thursday, October 16, 2014. There 10 points in 8 questions.

Submit your answers as a single text file named "<username>-comp3000-assign5.txt" (where username is your MyCarletonOne username). The first four lines of this file should be "COMP 3000 Assignment 5", your name, student number, and the date of submission. You may wish to format your answers in Markdown to improve their appearance.

No other formats will be accepted. Submitting in another format will likely result in your assignment not being graded and you receiving no marks for this assignment. In particular do not submit a zip file, MS Word, or OpenOffice file as your answers document!

Don't forget to include what outside resources you used to complete each of your answers, including other students, man pages, and web resources. You do not need to list help from the instructor, TA, or information found in the textbook.

Questions

1. [1] Some x86 processors have support for PAE (physical address extensions), which allows for the use of more than 4 GiB of memory with a 32-bit processor. On a system with PAE, what is the size of pointers used in regular C programs? Why?
2. [1] On a system that supports both paging and segmentation (such as the x86 in 32-bit protected mode), do segment base registers hold physical or virtual addresses? Explain why briefly.
3. [1] Do the stack and data segments on the x86 (as referred to by %esp and %ebp) ever overlap? Explain briefly.
4. [1] If we have a system where virtual address 0x52D2C3A3 mapped to physical address 0x13A103A3, what is the largest page size that could be used for this mapping? Why?
5. [1] What is the xchg x86 instruction do? Why is it important for creating concurrent programs?
6. [1] Peterson's algorithm does not reliably work on modern machines. Explain why briefly by making reference to the memory hierarchy of modern systems (e.g., relationship between registers, caches, and main memory/RAM).
7. [1] Spinlocks cause a program to loop infinitely until a resource becomes available. High performance multi-threaded code often uses spinlocks rather than locks that yield the processor when waiting for the lock to become available. When are spinlocks preferable? Why?
8. [3] Create dotprod_lockfree.c by modifying dotprod_mutex.c (From this LLNL tutorial) so that it does not do any locking using mutexes or any other explicit locking mechanism. In other words, the program should still be multithreaded; the threads should be able to run without any race conditions corrupting the computation but without grabbing and releasing any locks (explicitly). Explain why your modifications are correct. (You may want to refer to the serial version to help you understand what the program is supposed to do.)