Errata Real-Time Architectures 2007/2008

This document maintains a list of errors, typos, and other kind of mistakes that have been identified in the book, slides, or articles for the real-time architectures course (2IN25 and 2IN60) during the academic year 2008/2009. The errata that have been marked with a “*” gave rise to points to students. 1 Book 1.1 Section 4.3.1 “Optimality” There is a typo in “Case 1” on page 80. Rather than “C1 < T2 – FT2, the relation should be “C1 ≤ T2 – FT2. This is immediately clear from the top-line of the text on p. 81. Note that • the relation in “Case 1” on p. 82 is correct (and the text of both cases are identical, apart from the typo on p. 80). • the relations in Fig. 4.5 (case (a)) and Fig. 4.6 reflect the actual schedules illustrated by the figures rather than the assumptions in the cases. 1.2 Section 4.4.2 “An Example”* The reason for not being able to guarantee the schedulability of the task set under RMS in this example is wrong. A better, i.e. less pessimistic, premise would be U > LL(2) and HB(τ1, τ2) > 2. Still, both tests are sufficient rather than exact tests. Using an exact test for RMS reveals that the task-set is indeed unschedulable under RMS. Motivation: U > ln 2 is the (lower) limit of the LL-bound as the number of tasks tends to infinity and should not be used to determine the schedulability of a task set consisting of a finite number of tasks. Counter Example: Whereas the following task set with τ1 (C1=3, T1=5) and τ2 (C2=1, T2=7) is schedulable using RMS since U (=0.74) < LL(2), it would appear not to be if the U > ln 2 condition is used. 1.3 Section 5.6 “Sporadic Server” The description of the SS algorithm is wrong; see also RTA.Exercises-5. A “proper” description of the SS algorithm can be found in Section 7.3 “Sporadic Servers” in [1].