Hardware Reference
In-Depth Information
τ
k
τ
i
f
d
i
i
Figure 4.15
More accurate calculation of the interference on
τ
i
by higher priority tasks.
Note that this test is sufficient but not necessary for guaranteeing the schedulability of
the task set. This is due to the fact that
I
i
is calculated by assuming that each higher-
D
i
priority task
τ
j
times on
τ
i
. However, as shown in Figure 4.15,
the actual interference can be smaller than
I
i
, since
τ
i
may terminate earlier.
exactly interferes
T
j
To find a sufficient and necessary schedulability test for DM, the exact interleaving of
higher-priority tasks must be evaluated for each process. In general, this procedure is
quite costly since, for each task
τ
i
, it requires the construction of the schedule until
D
i
. Audsley et al. [ABRW92, ABR
+
93] proposed an efficient method for evalu-
ating the exact interference on periodic tasks and derived a sufficient and necessary
schedulability test for DM, called Response Time Analysis.
4.5.2
RESPONSE TIME ANALYSIS
According to the method proposed by Audsley at al., the longest response time
R
i
of a
periodic task
τ
i
is computed, at the critical instant, as the sum of its computation time
and the interference
I
i
of the higher priority tasks:
R
i
=
C
i
+
I
i
,
where
R
i
T
j
C
j
.
i−
1
I
i
=
j
=1
Hence,
R
i
T
j
C
j
.
i−
1
R
i
=
C
i
+
(4.17)
j
=1
No simple solution exists for this equation since
R
i
appears on both sides. Thus, the
worst-case response time of task
τ
i
is given by the smallest value of
R
i
that satisfies
Equation (4.17). Note, however, that only a subset of points in the interval [0
,D
i
] need
Search WWH ::
Custom Search