- 25 Jan, 2021 40 commits
-
-
Paolo Valente authored
Some BFQ mechanisms make their decisions on a bfq_queue basing also on whether the bfq_queue is I/O bound. In this respect, the current logic for evaluating whether a bfq_queue is I/O bound is rather rough. This commits replaces this logic with a more effective one. The new logic measures the percentage of time during which a bfq_queue is active, and marks the bfq_queue as I/O bound if the latter if this percentage is above a fixed threshold. Tested-by:
Jan Kara <jack@suse.cz> Signed-off-by:
Paolo Valente <paolo.valente@linaro.org> Signed-off-by:
Jens Axboe <axboe@kernel.dk>
-
Christoph Hellwig authored
When an already remapped bio is resubmitted (e.g. by blk_queue_split), bio_check_eod will compare the remapped bi_sector against the size of the partition, leading to spurious I/O failures. Skip the EOD check in this case. Fixes: 309dca30 ("block: store a block_device pointer in struct bio") Reported-by:
Christoph Hellwig <hch@lst.de> Signed-off-by:
-
Pavel Begunkov authored
The block layer spends quite a while in blkdev_direct_IO() to copy and initialise bio's bvec. However, if we've already got a bvec in the input iterator it might be reused in some cases, i.e. when new ITER_BVEC_FLAG_FIXED flag is set. Simple tests show considerable performance boost, and it also reduces memory footprint. Suggested-by:
Matthew Wilcox <willy@infradead.org> Reviewed-by:
Pavel Begunkov <asml.silence@gmail.com> Reviewed-by:
Ming Lei <ming.lei@redhat.com> Signed-off-by:
-
Pavel Begunkov authored
Add a helper function calculating the number of bvec segments we need to allocate to construct a bio. It doesn't change anything functionally, but will be used to not duplicate special cases in the future. Reviewed-by:
-
Pavel Begunkov authored
iov_iter_advance() is heavily used, but implemented through generic means. For bvecs there is a specifically crafted function for that, so use bvec_iter_advance() instead, it's faster and slimmer. Reviewed-by:
-
Christoph Hellwig authored
This saves one memory allocation, and ensures the bvecs aren't freed before the AIO completion. This will allow the lower level code to be optimized so that it can avoid allocating another bvec array. Signed-off-by:
-
Pavel Begunkov authored
Direct IO does not operate on the current working set of pages managed by the kernel, so it should not be accounted as memory stall to PSI infrastructure. The block layer and iomap direct IO use bio_iov_iter_get_pages() to build bios, and they are the only users of it, so to avoid PSI tracking for them clear out BIO_WORKINGSET flag. Do same for dio_bio_submit() because fs/direct_io constructs bios by hand directly calling bio_add_page(). Reported-by:
Christoph Hellwig <hch@infradead.org> Suggested-by:
Johannes Weiner <hannes@cmpxchg.org> Reviewed-by:
-
Pavel Begunkov authored
zero-length bvec segments are allowed in general, but not handled by bio and down the block layer so filtered out. This inconsistency may be confusing and prevent from optimisations. As zero-length segments are useless and places that were generating them are patched, declare them not allowed. Reviewed-by:
-
Pavel Begunkov authored
iter_file_splice_write() may spawn bvec segments with zero-length. In preparation for prohibiting them, filter out by hand at splice level. Reviewed-by:
-
Guoqing Jiang authored
We can remove 'q' from blk_execute_rq as well after the previous change in blk_execute_rq_nowait. And more importantly it never really was needed to start with given that we can trivial derive it from struct request. Cc: linux-scsi@vger.kernel.org Cc: virtualization@lists.linux-foundation.org Cc: linux-ide@vger.kernel.org Cc: linux-mmc@vger.kernel.org Cc: linux-nvme@lists.infradead.org Cc: linux-nfs@vger.kernel.org Acked-by: Ulf Hansson <ulf.hansson@linaro.org> # for mmc Signed-off-by:
Guoqing Jiang <guoqing.jiang@cloud.ionos.com> Signed-off-by:
-
Guoqing Jiang authored
The 'q' is not used since commit a1ce35fa ("block: remove dead elevator code"), also update the comment of the function. And more importantly it never really was needed to start with given that we can trivial derive it from struct request. Cc: target-devel@vger.kernel.org Cc: linux-scsi@vger.kernel.org Cc: virtualization@lists.linux-foundation.org Cc: linux-ide@vger.kernel.org Cc: linux-mmc@vger.kernel.org Cc: linux-nvme@lists.infradead.org Cc: linux-nfs@vger.kernel.org Signed-off-by:
-
Pan Bian authored
Free the request rq before returning error code. Fixes: 972248e9 ("scsi: bsg-lib: handle bidi requests without block layer help") Signed-off-by:
Pan Bian <bianpan2016@163.com> Reviewed-by:
-
Ming Lei authored
This bioset is just for allocating bio only from bio_next_split, and it needn't bvecs, so remove the flag. Cc: linux-bcache@vger.kernel.org Cc: Coly Li <colyli@suse.de> Reviewed-by:
Coly Li <colyli@suse.de> Reviewed-by:
Hannes Reinecke <hare@suse.de> Signed-off-by:
-
Ming Lei authored
bvec_alloc(), bvec_free() and bvec_nr_vecs() are only used inside block layer core functions, no need to declare them in public header. Reviewed-by:
-
Ming Lei authored
bvec_alloc() may allocate more bio vectors than requested, so set .bi_max_vecs as actual allocated vector number, instead of the requested number. This way can help fs build bigger bio because new bio often won't be allocated until the current one becomes full. Reviewed-by:
-
Ming Lei authored
The inline bvecs won't be used if user needn't bvecs by not passing BIOSET_NEED_BVECS, so don't allocate bvecs in this situation. Reviewed-by:
-
Ming Lei authored
q->bio_split is only used by bio_split() for fast cloning bio, and no need to allocate bvecs, so remove this flag. Reviewed-by:
-
Ming Lei authored
Managing bio slab cache via xarray by using slab cache size as xarray index, and storing 'struct bio_slab' instance into xarray. So code is simplified a lot, meantime it becomes more readable than before. Reviewed-by:
-
huhai authored
As we can see, returns parent_sched_may_change whether sd->next_in_service changes or not, so remove this judgment. Signed-off-by:
huhai <huhai@tj.kylinos.cn> Acked-by:
-
Jan Kara authored
Currently when non-mq aware IO scheduler (BFQ, mq-deadline) is used for a queue with multiple HW queues, the performance it rather bad. The problem is that these IO schedulers use queue-wide locking and their dispatch function does not respect the hctx it is passed in and returns any request it finds appropriate. Thus locality of request access is broken and dispatch from multiple CPUs just contends on IO scheduler locks. For these IO schedulers there's little point in dispatching from multiple CPUs. Instead dispatch always only from a single CPU to limit contention. Below is a comparison of dbench runs on XFS filesystem where the storage is a raid card with 64 HW queues and to it attached a single rotating disk. BFQ is used as IO scheduler: clients MQ SQ MQ-Patched Amean 1 39.12 (0.00%) 43.29 * -10.67%* 36.09 * 7.74%* Amean 2 128.58 (0.00%) 101.30 * 21.22%* 96.14 * 25.23%* Amean 4 577.42 (0.00%) 494.47 * 14.37%* 508.49 * 11.94%* Amean 8 610.95 (0.00%) 363.86 * 40.44%* 362.12 * 40.73%* Amean 16 391.78 (0.00%) 261.49 * 33.25%* 282.94 * 27.78%* Amean 32 324.64 (0.00%) 267.71 * 17.54%* 233.00 * 28.23%* Amean 64 295.04 (0.00%) 253.02 * 14.24%* 242.37 * 17.85%* Amean 512 10281.61 (0.00%) 10211.16 * 0.69%* 10447.53 * -1.61%* Numbers are times so lower is better. MQ is stock 5.10-rc6 kernel. SQ is the same kernel with megaraid_sas.host_tagset_enable=0 so that the card advertises just a single HW queue. MQ-Patched is a kernel with this patch applied. You can see multiple hardware queues heavily hurt performance in combination with BFQ. The patch restores the performance. Signed-off-by: -
Jan Kara authored
This reverts commit b445547e. Since both mq-deadline and BFQ completely ignore hctx they are passed to their dispatch function and dispatch whatever request they deem fit checking whether any request for a particular hctx is queued is just pointless since we'll very likely get a request from a different hctx anyway. In the following commit we'll deal with lock contention in these IO schedulers in presence of multiple HW queues in a different way. Signed-off-by:
-
Paolo Valente authored
This commits preserves I/O-dispatch plugging for a special symmetric case that may suddenly turn into asymmetric: the case where only one bfq_queue, say bfqq, is busy. In this case, not expiring bfqq does not cause any harm to any other queues in terms of service guarantees. In contrast, it avoids the following unlucky sequence of events: (1) bfqq is expired, (2) a new queue with a lower weight than bfqq becomes busy (or more queues), (3) the new queue is served until a new request arrives for bfqq, (4) when bfqq is finally served, there are so many requests of the new queue in the drive that the pending requests for bfqq take a lot of time to be served. In particular, event (2) may case even already dispatched requests of bfqq to be delayed, inside the drive. So, to avoid this series of events, the scenario is preventively declared as asymmetric also if bfqq is the only busy queues. By doing so, I/O-dispatch plugging is performed for bfqq. Tested-by:
-
Paolo Valente authored
BFQ tags some bfq_queues as interactive or soft_rt if it deems that these bfq_queues contain the I/O of, respectively, interactive or soft real-time applications. BFQ privileges both these special types of bfq_queues over normal bfq_queues. To privilege a bfq_queue, BFQ mainly raises the weight of the bfq_queue. In particular, soft_rt bfq_queues get a higher weight than interactive bfq_queues. A bfq_queue may turn from interactive to soft_rt. And this leads to a tricky issue. Soft real-time applications usually start with an I/O-bound, interactive phase, in which they load themselves into main memory. BFQ correctly detects this phase, and keeps the bfq_queues associated with the application in interactive mode for a while. Problems arise when the I/O pattern of the application finally switches to soft real-time. One of the conditions for a bfq_queue to be deemed as soft_rt is that the bfq_queue does not consume too much bandwidth. But the bfq_queues associated with a soft real-time application consume as much bandwidth as they can in the loading phase of the application. So, after the application becomes truly soft real-time, a lot of time should pass before the average bandwidth consumed by its bfq_queues finally drops to a value acceptable for soft_rt bfq_queues. As a consequence, there might be a time gap during which the application is not privileged at all, because its bfq_queues are not interactive any longer, but cannot be deemed as soft_rt yet. To avoid this problem, BFQ pretends that an interactive bfq_queue consumes zero bandwidth, and allows an interactive bfq_queue to switch to soft_rt. Yet, this fake zero-bandwidth consumption easily causes the bfq_queue to often switch to soft_rt deceptively, during its loading phase. As in soft_rt mode, the bfq_queue gets its bandwidth correctly computed, and therefore soon switches back to interactive. Then it switches again to soft_rt, and so on. These spurious fluctuations usually cause losses of throughput, because they deceive BFQ's mechanisms for boosting throughput (injection, I/O-plugging avoidance, ...). This commit addresses this issue as follows: 1) It does compute actual bandwidth consumption also for interactive bfq_queues. This avoids the above false positives. 2) When a bfq_queue switches from interactive to normal mode, the consumed bandwidth is reset (forgotten). This allows the bfq_queue to enjoy soft_rt very quickly. In particular, two alternatives are possible in this switch: - the bfq_queue still has backlog, and therefore there is a budget already scheduled to serve the bfq_queue; in this case, the scheduling of the current budget of the bfq_queue is not hindered, because only the scheduling of the next budget will be affected by the weight drop. After that, if the bfq_queue is actually in a soft_rt phase, and becomes empty during the service of its current budget, which is the natural behavior of a soft_rt bfq_queue, then the bfq_queue will be considered as soft_rt when its next I/O arrives. If, in contrast, the bfq_queue remains constantly non-empty, then its next budget will be scheduled with a low weight, which is the natural treatment for an I/O-bound (non soft_rt) bfq_queue. - the bfq_queue is empty; in this case, the bfq_queue may be considered unjustly soft_rt when its new I/O arrives. Yet the problem is now much smaller than before, because it is unlikely that more than one spurious fluctuation occurs. Tested-by: -
Paolo Valente authored
BFQ heuristics try to detect interactive I/O, and raise the weight of the queues containing such an I/O. Yet, if also the user changes the weight of a queue (i.e., the user changes the ioprio of the process associated with that queue), then it is most likely better to prevent BFQ heuristics from silently changing the same weight. Tested-by:
-
Paolo Valente authored
Tests on slower machines showed current window to be way too small. This commit increases it. Tested-by:
-
Jia Cheng Hu authored
Since commit c5089591c3ba ("block, bfq: detect wakers and unconditionally inject their I/O"), when the in-service bfq_queue, say Q, is temporarily empty, BFQ checks whether there are I/O requests to inject (also) from the waker bfq_queue for Q. To this goal, the value pointed by bfqq->waker_bfqq->next_rq must be controlled. However, the current implementation mistakenly looks at bfqq->next_rq, which instead points to the next request of the currently served queue. This mistake evidently causes losses of throughput in scenarios with waker bfq_queues. This commit corrects this mistake. Fixes: c5089591c3ba ("block, bfq: detect wakers and unconditionally inject their I/O") Signed-off-by:Jia Cheng Hu <jia.jiachenghu@gmail.com> Signed-off-by:
-
Paolo Valente authored
The value of the I/O plugging (idling) timeout is used also as the think-time threshold to decide whether a process has a short think time. In this respect, a good value of this timeout for rotational drives is un the order of several ms. Yet, this is often too long a time interval to be effective as a think-time threshold. This commit mitigates this problem (by a lot, according to tests), by halving the threshold. Tested-by:
-
Christoph Hellwig authored
Now that no fast path lookups in the partition table are left, there is no point in micro-optimizing the data structure for it. Just use a bog standard xarray. Signed-off-by:
Tejun Heo <tj@kernel.org> Signed-off-by:
-
Christoph Hellwig authored
There is good reason to iterate backwards when deleting all partitions in del_gendisk, just like we don't in blk_drop_partitions. Signed-off-by:
-
Christoph Hellwig authored
Add a helper to call kobject_uevent for the disk and all partitions, and unexport the disk_part_iter_* helpers that are now only used in the core block code. Signed-off-by:
-
Christoph Hellwig authored
Remove the reverse map from a sector to a partition for I/O accounting by simply using ->bi_bdev. Signed-off-by:
-
Christoph Hellwig authored
Rework the I/O accounting for bio based drivers to use ->bi_bdev. This means all drivers can now simply use bio_start_io_acct to start accounting, and it will take partitions into account automatically. To end I/O account either bio_end_io_acct can be used if the driver never remaps I/O to a different device, or bio_end_io_acct_remapped if the driver did remap the I/O. Signed-off-by:
-
Christoph Hellwig authored
There is no good reason to reassign ->bi_bdev when remapping the partition-relative block number to the device wide one, as all the information required by the drivers comes from the gendisk anyway. Keeping the original ->bi_bdev alive will allow to greatly simplify the partition-away I/O accounting. Signed-off-by:
-
Christoph Hellwig authored
Merge a few checks for whole devices vs partitions to streamline the sanity checks. Signed-off-by:
-
Christoph Hellwig authored
Replace the gendisk pointer in struct bio with a pointer to the newly improved struct block device. From that the gendisk can be trivially accessed with an extra indirection, but it also allows to directly look up all information related to partition remapping. Signed-off-by:
-
Christoph Hellwig authored
The block layer already checks for this conditions in bio_check_eod before calling the driver. Signed-off-by:
-
Christoph Hellwig authored
The block layer already checks for this conditions in bio_check_eod before calling the driver. Signed-off-by:
-
Christoph Hellwig authored
Unconditionally call set_disk_ro now that it only updates the hardware state. This allows to properly set up the Linux devices read-only when the controller turns a previously writable namespace read-only. Signed-off-by:
Keith Busch <kbusch@kernel.org> Reviewed-by:
Martin K. Petersen <martin.petersen@oracle.com> Reviewed-by:
-
Christoph Hellwig authored
Now that the hardware read-only state can't be changed by the BLKROSET ioctl, the code in this method is not required anymore. Signed-off-by:
Ilya Dryomov <idryomov@gmail.com> Reviewed-by:
-
Christoph Hellwig authored
Change the policy so that a BLKROSET on the whole device also affects partitions. To quote Martin K. Petersen: It's very common for database folks to twiddle the read-only state of block devices and partitions. I know that our users will find it very counter-intuitive that setting /dev/sda read-only won't prevent writes to /dev/sda1. The existing behavior is inconsistent in the sense that doing: # blockdev --setro /dev/sda # echo foo > /dev/sda1 permits writes. But: # blockdev --setro /dev/sda <something triggers revalidate> # echo foo > /dev/sda1 doesn't. And a subsequent: # blockdev --setrw /dev/sda # echo foo > /dev/sda1 doesn't work either since sda1's read-only policy has been inherited from the whole-disk device. You need to do: # blockdev --rereadpt after setting the whole-disk device rw to effectuate the same change on the partitions, otherwise they are stuck being read-only indefinitely. However, setting the read-only policy on a partition does *not* require the revalidate step. As a matter of fact, doing the revalidate will blow away the policy setting you just made. So the user needs to take different actions depending on whether they are trying to read-protect a whole-disk device or a partition. Despite using the same ioctl. That is really confusing. I have lost count how many times our customers have had data clobbered because of ambiguity of the existing whole-disk device policy. The current behavior violates the principle of least surprise by letting the user think they write protected the whole disk when they actually didn't. Suggested-by:
-
