Skip to content

RFC playing around req alloc

Pavel Begunkov io-uring 1 version In review

Get series with b4
b4 am [email protected]

Versions newest first

  1. v1 RFC Latest17 patches 9 repliesIn review

Patches in v1 No applies recorded yet

#Subject and filesSizeRepliesApplied
1/17io_uring: replace force_nonblock with flagsfs/io_uring.c+96 -820 replies
2/17io_uring: make op handlers always take issue flagsfs/io_uring.c+13 -120 replies
3/17io_uring: don't propagate io_comp_statefs/io_uring.c+67 -792 replies
4/17io_uring: don't keep submit_state on stackfs/io_uring.c+46 -440 replies
5/17io_uring: remove ctx from comp_statefs/io_uring.c+9 -110 replies
6/17io_uring: don't reinit submit state every timefs/io_uring.c+10 -50 replies
7/17io_uring: replace list with array for compl batchfs/io_uring.c+11 -240 replies
8/17io_uring: submit-completion free batchingfs/io_uring.c+29 -200 replies
9/17io_uring: remove fallback_reqfs/io_uring.c+2 -360 replies
10/17io_uring: count ctx refs separately from reqsfs/io_uring.c+5 -10 replies
11/17io_uring: persistent req cachefs/io_uring.c+13 -80 replies
12/17io_uring: feed reqs back into alloc cachefs/io_uring.c+12 -70 replies
13/17io_uring: use persistent request cachefs/io_uring.c+30 -231 reply
14/17io_uring: provide FIFO ordering for task_workfs/io-wq.h fs/io_uring.c include/linux/io_uring.h+108 -160 replies
15/17io_uring: enable req cache for task_work itemsfs/io_uring.c+21 -10 replies
16/17io_uring: take comp_state from ctxfs/io_uring.c+18 -190 replies
17/17io_uring: defer flushing cached reqsfs/io_uring.c+19 -101 reply

Reviews trailers per patch

No Reviewed-by, Acked-by or Tested-by trailers for v1 yet.

Read the v1 thread

Cover letter [PATCH RFC 00/17] playing around req alloc

Unfolding previous ideas on persistent req caches. 4-7 including slashed ~20% of overhead for nops benchmark, haven't done benchmarking personally for this yet, but according to perf should be ~30-40% in total. That's for IOPOLL + inline completion cases, obviously w/o async/IRQ completions.

Jens,
1. 11/17 removes deallocations on end of submit_sqes. Looks you forgot or just didn't do that.

2. lists are slow and not great cache-wise, that why at I want at least a combined approach from 12/17.

3. Instead of lists in "use persistent request cache" I had in mind a slightly different way: to grow the req alloc cache to 32-128 (or hint from the userspace), batch-alloc by 8 as before, and recycle _all_ reqs right into it. If overflows, do kfree().
It should give probabilistically high hit rate, amortising out most of allocations. Pros: it doesn't grow ~infinitely as lists can. Cons: there are always counter examples. But as I don't have numbers to back it, I took your implementation. Maybe, we'll reconsider later.

I'll revise tomorrow on a fresh head + do some performance testing, and is leaving it RFC until then.

Jens Axboe (3):
io_uring: use persistent request cache
io_uring: provide FIFO ordering for task_work
io_uring: enable req cache for task_work items

Pavel Begunkov (14):
io_uring: replace force_nonblock with flags
io_uring: make op handlers always take issue flags
io_uring: don't propagate io_comp_state
io_uring: don't keep submit_state on stack
io_uring: remove ctx from comp_state
io_uring: don't reinit submit state every time
io_uring: replace list with array for compl batch
io_uring: submit-completion free batching
io_uring: remove fallback_req
io_uring: count ctx refs separately from reqs
io_uring: persistent req cache
io_uring: feed reqs back into alloc cache
io_uring: take comp_state from ctx
io_uring: defer flushing cached reqs

 fs/io-wq.h               |   9 -
 fs/io_uring.c            | 716 ++++++++++++++++++++++-----------------
 include/linux/io_uring.h |  14 +
 3 files changed, 425 insertions(+), 314 deletions(-)
Signature · 1 line
2.24.0