Hardware Reference
In-Depth Information
Lateness
The difference between the finishing time of a task and its deadline (
L
=
f
d
). Note that a negative lateness means that a task has completed before its
deadline.
−
Laxity
The maximum delay that a job can experience after its activation and still
complete within its deadline. At the arrival time, the laxity is equal to the relative
deadline minus the computation time (
D
−
C
). It is also called
slack time
.
Lifetime
The maximum time that can be represented inside the kernel.
Load
Computation time demanded by a task set in an interval, divided by the length
of the interval.
Mailbox
A communication buffer characterized by a message queue shared between
two or more jobs.
Message
A set of data, organized in a predetermined format for exchanging infor-
mation among tasks.
Mutual Exclusion
A kernel mechanism that serializes the execution of concurrent
tasks on critical sections of code.
Non-Preemptive Scheduling
A form of scheduling in which jobs, once started, can
continuously execute on the processor without interruption.
Optimal algorithm
A scheduling algorithm that minimizes some cost function de-
fined over the task set.
Overhead
The time required by the processor to manage all internal mechanisms
of the operating system, such as queuing jobs and messages, updating kernel data
structures, performing context switches, activating interrupt handlers, and so on.
Overload
Exceptional load condition on the processor, such that the computation
time demanded by the tasks in a certain interval exceeds the available processor
time in the same interval.
Overrun
Situation in which the computational demand of a job in a given interval
of time exceeds its expected value. It can occur because a task executes more
than expected (execution overrun) or it is activated earlier than expected (activation
overrun).
Search WWH ::
Custom Search