Linux advisory file lock
Linux exposes a file-locking primitive with the command flock.
It is advisory in that the kernel does not enforce exclusive file
access - it is entirely up to the processes to check the lock status
before doing any mutually-exclusive file I/O operation.
#!/bin/bash
F=/tmp/lockfile
exec 200<$F
flock -sn 200
RETVAL=$?
if [ $RETVAL -ne 1 ] ; then
echo "Failed to acquire lock; check conflict using 'lsof $F'"
exit $RETVAL
else
echo "sleeping with locked $F"
sleep 60
fi
exit 0
flock -s acquires shared-lock. Replace it with flock -x to acquire
exclusive lock.
Typically, you would write your code such that reader/consumer can proceed as long as it can acquire the shared-lock. But a writer/producer must acquire exclusive-lock before it can proceed so that it does not change the resource state while there are one or more readers.
Linux offers another lock primitive via syscall fcntl() which
does not have CLI interface. Debian/Ubuntu dpkg/apt uses this lock
for exclusive access. See apt/apt-pkg/contrib/fileutl.cc