The following patches apply over linus's tree and this patchset https://lore.kernel.org/all/20211007214448.6282-1-michael.christie@xxxxxxxxxx/ which allows us to check the vhost owner thread's RLIMITs: It looks like that patchset has been ok'd by all the major parties and just needs a small cleanup to apply to Jens and Paul trees, so I wanted to post my threading patches based over it for review. The following patches allow us to support multiple vhost workers per device. I ended up just doing Stefan's original idea where userspace has the kernel create a worker and we pass back the pid. This has the benefit over the workqueue and userspace thread approach where we only have one'ish code path in the kernel during setup to detect old tools. The main IO paths and device/vq setup/teardown paths all use common code. I've also included a patch for qemu so you can get an idea of how it works. If we are ok with the kernel code then I'll break that up into a patchset and send to qemu-devel for review. Results: -------- fio jobs 1 2 4 8 12 16 ---------------------------------------------------------- 1 worker 84k 492k 510k - - - worker per vq 184k 380k 744k 1422k 2256k 2434k Notes: 0. This used a simple fio command: fio --filename=/dev/sdb --direct=1 --rw=randrw --bs=4k \ --ioengine=libaio --iodepth=128 --numjobs=$JOBS_ABOVE and I used a VM with 16 vCPUs and 16 virtqueues. 1. The patches were tested with emulate_pr=0 and these patches: https://lore.kernel.org/all/yq1tuhge4bg.fsf@xxxxxxxxxxxxxxxxxxxx/t/ which are in mkp's scsi branches for the next kernel. They fix the perf issues where IOPs dropped at 12 vqs/jobs. 2. Because we have a hard limit of 1024 cmds, if the num jobs * iodepth was greater than 1024, I would decrease iodepth. So 12 jobs used 85 cmds, and 16 used 64. 3. The perf issue above at 2 jobs is because when we only have 1 worker we execute more cmds per vhost_work due to all vqs funneling to one worker. This results in less context switches and better batching without having to tweak any settings. I'm working on patches to add back batching during lio completion and do polling on the submission side. We will still want the threading patches, because if we batch at the fio level plus use the vhost theading patches, we can see a big boost like below. So hopefully doing it at the kernel will allow apps to just work without having to be smart like fio. fio using io_uring and batching with the iodepth_batch* settings: fio jobs 1 2 4 8 12 16 ------------------------------------------------------------- 1 worker 494k 520k - - - - worker per vq 496k 878k 1542k 2436k 2304k 2590k V3: - fully convert vhost code to use vq based APIs instead of leaving it half per dev and half per vq. - rebase against kernel worker API. - Drop delayed worker creation. We always create the default worker at VHOST_SET_OWNER time. Userspace can create and bind workers after that. v2: - change loop that we take a refcount to the worker in - replaced pid == -1 with define. - fixed tabbing/spacing coding style issue - use hash instead of list to lookup workers. - I dropped the patch that added an ioctl cmd to get a vq's worker's pid. I saw we might do a generic netlink interface instead.
