Deadline IO scheduler tunables==============================This little file attempts to document how the deadline io scheduler works.In particular, it will clarify the meaning of the exposed tunables that may beof interest to power users.Each io queue has a set of io scheduler tunables associated with it. Thesetunables control how the io scheduler works. You can find these entriesin:/sys/block/<device>/ioschedassuming that you have sysfs mounted on /sys. If you don't have sysfs mounted,you can do so by typing:# mount none /sys -t sysfs********************************************************************************read_expire	(in ms)-----------The goal of the deadline io scheduler is to attempt to guarentee a startservice time for a request. As we focus mainly on read latencies, this istunable. When a read request first enters the io scheduler, it is assigneda deadline that is the current time + the read_expire value in units ofmiliseconds.write_expire	(in ms)-----------Similar to read_expire mentioned above, but for writes.fifo_batch----------When a read request expires its deadline, we must move some requests fromthe sorted io scheduler list to the block device dispatch queue. fifo_batchcontrols how many requests we move, based on the cost of each request. Arequest is either qualified as a seek or a stream. The io scheduler knowsthe last request that was serviced by the drive (or will be serviced rightbefore this one). See seek_cost and stream_unit.write_starved	(number of dispatches)-------------When we have to move requests from the io scheduler queue to the blockdevice dispatch queue, we always give a preference to reads. However, wedon't want to starve writes indefinitely either. So writes_starved controlshow many times we give preference to reads over writes. When that has beendone writes_starved number of times, we dispatch some writes based on thesame criteria as reads.front_merges	(bool)------------Sometimes it happens that a request enters the io scheduler that is contigiouswith a request that is already on the queue. Either it fits in the back of thatrequest, or it fits at the front. That is called either a back merge candidateor a front merge candidate. Due to the way files are typically laid out,back merges are much more common than front merges. For some work loads, youmay even know that it is a waste of time to spend any time attempting tofront merge requests. Setting front_merges to 0 disables this functionality.Front merges may still occur due to the cached last_merge hint, but sincethat comes at basically 0 cost we leave that on. We simply disable therbtree front sector lookup when the io scheduler merge function is called.Nov 11 2002, Jens Axboe <axboe@suse.de>