This preview has intentionally blurred sections. Sign up to view the full version.View Full Document
Unformatted text preview: CSE4/586 (Spring 2009): Homework 5 Due by Apr 23 Thursday, in class. 1. (20 points) In Lamport’s logical clock based mutual exclusion protocol (Chapter 7) why do we need to have the following check for a process i entering the critical section? For all other j , knownT [ j ] must be larger than the timestamp of req i . Answer: Note that, the other condition (call it condition1) for process i to enter CS is that: req i is at the head of reqQ i . Due to channel delays and concurrent execution, different processes may have differing views on the pending requests for CS. It is possible that there exists a process j such that req j is at the head of reqQ j , while req i is at the head of reqQ i . If condition1 had been the only condition for entering the CS, then both processes would be free to enter CS concurrently, violating the specifications for the mutual exclusion problem. The condition2, that states: For all other j , knownT [ j ] must be larger than the timestamp of req i , ensures that any process i waits long enough to guarantee that it learns about all the pending requests-for-CS in the system that are less than req i . (This is because the channels are FIFO, and if i gets a message from j with timestamp larger than req i , it is guaranteed to have received any req j with timestamp less than req i . In other words, this condition enables a correct implementation of total order on requests-for-CS across processes by forcing them to wait until they learn about all pending requests at other processes that are less than their...
View Full Document
This note was uploaded on 01/04/2010 for the course CSE 123 taught by Professor Qwerty during the Spring '09 term at École Normale Supérieure.
- Spring '09