Software interrupts are most often used to handle high-priority
requests while the process is doing low-priority
time-consuming tasks.
For example, suppose an application is copying a large file
across a network. The copy may take a long time (up to
several hours). During this time, the application wants to
see and respond to high-priority commands written to its
command message file (for example, requests for the number of
records copied so far, or that the copy stop immediately).
To accomplish this, the application posts a "software
interrupt" FREAD against its command message file.
The FREAD just signals to MPE/iX that this application
wants to know about any new commands written to this message
file. No data is transferred at the time the FREAD is
called. Software interrupt FREADs never wait; they
return to the application immediately.
The application then starts to perform the copy. Unlike
NOWAIT I/O, the application does not have to poll the command
message file repeatedly to see if data has been written there.
MPE/iX signals the application (with a software interrupt) that
data has been written to the command message
file and that the application should complete the I/O.
In our example, if a high-priority command is written to the
message file, MPE/iX immediately causes a software interrupt.
The part of the application performing the copy is stopped,
and MPE/iX
forces the execution of the application's interrupt
handling procedure. There the application completes the
read by calling IOWAIT or IODONTWAIT, and
processes the command. When the interrupt handling procedure
completes, the part of the application performing the copy
resumes automatically at the statement where it left off.
