I'VE MADE MY SHARE OF ERRORS, SOME OF WHICH ARE DETECTED ONLY AFTER PUBLICATION. HERE IS A LIST OF SOME THAT I AM AWARE OF; PLEASE LET ME KNOW IF YOU DETECT OTHERS.
Thanks to Gennady Shmonin for pointing this out...
In the proof of Theorem 11, should be
Thanks to Marko Bertogna for pointing this out...
Equation 3 is incorrect. The second term in the min -- -- implicitly assumes that the job missing its deadline executes for time units, whereas it actually executes for strictly less than . Hence this second term should be -- ; for task systems with integer parameters, can be taken to be equal to one.
(A similar modification needs to be made for the definition of -- Equation 5 in the paper)
Thanks to Marko Bertogna for pointing this out...
Claim 5 in this paper states:
This statement is incorrect (although the error is not relevant to the remainder of the paper). The correct statement of this claim is as follows:
The corrected claim appears in the journalization of this paper (Sanjoy Baruah, Vincenzo Bonifaci, Alberto Marchetti-Spaccamela, and Sebastian Stiller. Improved multiprocessor global schedulability analysis. Real-Time Systems 46 (1), 2010.)
-
Sanjoy Baruah. Preemptive uniprocessor scheduling of non-cyclic GMF task systems. Proceedings of the IEEE International Conference on Embedded and Real-Time Computing Systems and Applications (RTCSA), Macau, China. August 2010. IEEE Computer Society Press.
Figure 1 in this paper incorrectly claims that the non-cyclic GMF task model is a generalization of the "traditional" GMF model. This is incorrect; a corrected version of this figure appears in (Sanjoy Baruah. The non-cyclic recurring real-time task model. Proceedings of the IEEE Real-Time Systems Symposium (RTSS), San Diego, CA. December 2010. IEEE Computer Society Press.)
-
Sanjoy Baruah. An improved global EDF schedulability test for uniform multiprocessors. Proceedings of the IEEE Real-Time Technology and Applications Symposium (RTAS), Stockholm, Sweden. April 2010. IEEE Computer Society Press.
Thanks to Haitao Zhu for pointing this out....
In the appendix, it is claimed (bullet item 4):
For a deadline miss to occur, it is necessary that
; hence
is a sufficient schedulability condition.
This is incorrect; the correct statement is
For a deadline miss to occur, it is necessary that
; hence
is a sufficient schedulability condition.
-
Sanjoy Baruah. Task assignment on two unrelated types of processors. Proceedings of the 17th International Conference on Real-Time and Network Systems (RTNS), Nantes, France. October, 2011.
Thanks to Gurulingesh Raravi for informing me of some significant errors in this paper. A corrected (and far more general) version of the results appear in (Andreas Wiese, Vincenzo Bonifaci, and Sanjoy Baruah. Partitioned EDF scheduling on a few types of unrelated multiprocessors. This is available off my home-page.)