When a queue associated with a process remains empty, there are cases
where throughput gets boosted if the device is idled to await the
arrival of a new I/O request for that queue. Currently, BFQ assumes
that one of these cases is when the device has no internal queueing
(regardless of the properties of the I/O being served). Unfortunately,
this condition has proved to be too general. So, this commit refines it
as "the device has no internal queueing and is rotational".
This refinement provides a significant throughput boost with random
I/O, on flash-based storage without internal queueing. For example, on
a HiKey board, throughput increases by up to 125%, growing, e.g., from
6.9MB/s to 15.6MB/s with two or three random readers in parallel.
This commit also refactors the code related to device idling, for the
following reason. Finding the change that provides the above large
improvement has been slightly more difficult than it had to be,
because the logic that decides whether to idle the device is still
scattered across three functions. Almost all of the logic is in the
function bfq_bfqq_may_idle, but (1) part of the decision is made in
bfq_update_idle_window, and (2) the function bfq_bfqq_must_idle may
switch off idling regardless of the output of bfq_bfqq_may_idle. In
addition, both bfq_update_idle_window and bfq_bfqq_must_idle make
their decisions as a function of parameters that are used, for similar
purposes, also in bfq_bfqq_may_idle. This commit addresses this issue
by moving all the logic into bfq_bfqq_may_idle.
Signed-off-by: Paolo Valente <paolo.valente@linaro.org>
Signed-off-by: Luca Miccio <lucmiccio@gmail.com>
Signed-off-by: djb77 <dwayne.bakewell@gmail.com>
A set of processes may happen to perform interleaved reads, i.e.,requests
whose union would give rise to a sequential read pattern. There are two
typical cases: in the first case, processes read fixed-size chunks of
data at a fixed distance from each other, while in the second case processes
may read variable-size chunks at variable distances. The latter case occurs
for example with QEMU, which splits the I/O generated by the guest into
multiple chunks, and lets these chunks be served by a pool of cooperating
processes, iteratively assigning the next chunk of I/O to the first
available process. CFQ uses actual queue merging for the first type of
rocesses, whereas it uses preemption to get a sequential read pattern out
of the read requests performed by the second type of processes. In the end
it uses two different mechanisms to achieve the same goal: boosting the
throughput with interleaved I/O.
This patch introduces Early Queue Merge (EQM), a unified mechanism to get a
sequential read pattern with both types of processes. The main idea is
checking newly arrived requests against the next request of the active queue
both in case of actual request insert and in case of request merge. By doing
so, both the types of processes can be handled by just merging their queues.
EQM is then simpler and more compact than the pair of mechanisms used in
CFQ.
Finally, EQM also preserves the typical low-latency properties of BFQ, by
properly restoring the weight-raising state of a queue when it gets back to
a non-merged state.
Signed-off-by: Mauro Andreolini <mauro.andreolini@unimore.it>
Signed-off-by: Arianna Avanzini <avanzini@google.com>
Signed-off-by: Paolo Valente <paolo.valente@unimore.it>
Signed-off-by: Linus Walleij <linus.walleij@linaro.org>
Signed-off-by: Luca Grifo <lg@linux.com>
Signed-off-by: djb77 <dwayne.bakewell@gmail.com>
The general structure is borrowed from CFQ, as much of the code for
handling I/O contexts. Over time, several useful features have been
ported from CFQ as well (details in the changelog in README.BFQ). A
(bfq_)queue is associated to each task doing I/O on a device, and each
time a scheduling decision has to be made a queue is selected and served
until it expires.
- Slices are given in the service domain: tasks are assigned
budgets, measured in number of sectors. Once got the disk, a task
must however consume its assigned budget within a configurable
maximum time (by default, the maximum possible value of the
budgets is automatically computed to comply with this timeout).
This allows the desired latency vs "throughput boosting" tradeoff
to be set.
- Budgets are scheduled according to a variant of WF2Q+, implemented
using an augmented rb-tree to take eligibility into account while
preserving an O(log N) overall complexity.
- A low-latency tunable is provided; if enabled, both interactive
and soft real-time applications are guaranteed a very low latency.
- Latency guarantees are preserved also in the presence of NCQ.
- Also with flash-based devices, a high throughput is achieved
while still preserving latency guarantees.
- BFQ features Early Queue Merge (EQM), a sort of fusion of the
cooperating-queue-merging and the preemption mechanisms present
in CFQ. EQM is in fact a unified mechanism that tries to get a
sequential read pattern, and hence a high throughput, with any
set of processes performing interleaved I/O over a contiguous
sequence of sectors.
- BFQ supports full hierarchical scheduling, exporting a cgroups
interface. Since each node has a full scheduler, each group can
be assigned its own weight.
- If the cgroups interface is not used, only I/O priorities can be
assigned to processes, with ioprio values mapped to weights
with the relation weight = IOPRIO_BE_NR - ioprio.
- ioprio classes are served in strict priority order, i.e., lower
priority queues are not served as long as there are higher
priority queues. Among queues in the same class the bandwidth is
distributed in proportion to the weight of each queue. A very
thin extra bandwidth is however guaranteed to the Idle class, to
prevent it from starving.
Signed-off-by: Paolo Valente <paolo.valente@unimore.it>
Signed-off-by: Arianna Avanzini <avanzini@google.com>
Signed-off-by: Luca Grifo <lg@linux.com>
Signed-off-by: djb77 <dwayne.bakewell@gmail.com>
