- 19 Oct, 2021 40 commits
-
-
Pavel Begunkov authored
We pass iovec** into __io_import_iovec(), which should keep it, initialise and modify accordingly. It's expensive, return it directly from __io_import_iovec encoding errors with ERR_PTR if needed. io_import_iovec keeps the old interface, but it's inline and so everything is optimised nicely. Signed-off-by:
Pavel Begunkov <asml.silence@gmail.com> Link: https://lore.kernel.org/r/6230e9769982f03a8f86fa58df24666088c44d3e.1634314022.git.asml.silence@gmail.comSigned-off-by:
Jens Axboe <axboe@kernel.dk>
-
Pavel Begunkov authored
Delay loading req->rw.{addr,len} in io_import_iovec until it's really needed, so removing extra loads for the fixed path, which doesn't use them. Signed-off-by: -
Pavel Begunkov authored
Don't decide about locking based on io_wq_current_is_worker(), it's not consistent with all other code and is expensive, use issue_flags. Signed-off-by:
-
Pavel Begunkov authored
Don't load req->ctx in advance, it takes an extra register and the field stays valid even after opcode handlers. It also optimises out req->ctx load in io_iopoll_req_issued() once it's inlined. Signed-off-by:
-
Pavel Begunkov authored
Combine force_nonblock branches (which is already optimised by compiler), flip branches so the most hot/common path is the first, e.g. as with non on-stack iov setup, and add extra likely/unlikely attributions for errror paths. Signed-off-by:
-
Pavel Begunkov authored
Make io_import_iovec taking struct io_rw_state instead of an iter pointer. First it takes care of initialising iovec pointer, which can be forgotten. Even more, we can not init it if not needed, e.g. in case of IORING_OP_READ_FIXED or IORING_OP_READ. Also hide saving iter_state inside of it by splitting out an inline function of it to avoid extra ifs. Signed-off-by:
-
Pavel Begunkov authored
First, change IO_URING_F_NONBLOCK to take sign bit of the int, so checking for it can be turned into test + sign-based-jump, makes the binary smaller and may be faster. Then, instead of passing need_lock boolean into io_import_iovec() just give it issue_flags, which is already stored somewhere. Saves some space on stack, a couple of test + cmov operations and other conversions. note: we still leave force_nonblock = issue_flags & IO_URING_F_NONBLOCK variable, but it's optimised out by the compiler into testing issue_flags directly. Signed-off-by:
-
Pavel Begunkov authored
Currently io_read() and io_write() keep separate pointers to an iter and to struct iov_iter_state, which is not great for register spilling and requires more on-stack copies. They are both either on-stack or in req->async_data at the same time, so use struct io_rw_state and keep a pointer only to it, so having all the state with just one pointer. Signed-off-by:
-
Pavel Begunkov authored
Add a new struct io_rw_state storing all iov related bits: fast iov, iterator and iterator state. Not much changes here, simply convert struct io_async_rw to use it. Signed-off-by:
-
Pavel Begunkov authored
Don't override req->result in io_complete_rw_iopoll() when it's already of the same value, we have an if just above it, so move the assignment there. Also, add one simle unlikely() in __io_complete_rw_common(). Signed-off-by:
-
Pavel Begunkov authored
Rearrange io_read return handling so first we expect it completing successfully and only then checking for errors, which is a colder path. Signed-off-by:
-
Pavel Begunkov authored
Some of the functions keep issue_flags as int, change those to unsigned. Signed-off-by:
-
Pavel Begunkov authored
Apparently, percpu_ref_put/get() are expensive enough if done per request, get them in a batch and cache on the submission side to avoid getting it over and over again. Also, if we're completing under uring_lock, return refs back into the cache instead of perfcpu_ref_put(). Pretty similar to how we do tctx->cached_refs accounting, but fall back to normal putting when we already changed a rsrc node by the time of free. Signed-off-by:
-
Pavel Begunkov authored
io_req_set_rsrc_node() reloads loads req->ctx, however it's already in registers in all use cases, so better to pass it as a parameter. Signed-off-by:
-
Pavel Begunkov authored
[ 158.514382] WARNING: CPU: 5 PID: 15251 at fs/io_uring.c:1141 io_free_batch_list+0x269/0x360 [ 158.514426] RIP: 0010:io_free_batch_list+0x269/0x360 [ 158.514437] Call Trace: [ 158.514440] __io_submit_flush_completions+0xde/0x180 [ 158.514444] tctx_task_work+0x14a/0x220 [ 158.514447] task_work_run+0x64/0xa0 [ 158.514448] __do_sys_io_uring_enter+0x7c/0x970 [ 158.514450] __x64_sys_io_uring_enter+0x22/0x30 [ 158.514451] do_syscall_64+0x43/0x90 [ 158.514453] entry_SYSCALL_64_after_hwframe+0x44/0xae We should not touch request internals including req->comp_list.next after putting our ref if it's not final, e.g. we can start freeing requests from the free cache. Fixed: 62ca9cb93e7f8 ("io_uring: optimise io_free_batch_list()") Signed-off-by: -
Pavel Begunkov authored
task_work_add() takes care of waking up the thread, remove useless wake_up_process(). Signed-off-by:
-
Pavel Begunkov authored
Looking at the assembly, the compiler decided to reload req->opcode in io_op_defs[opcode].needs_file instead of one it had in a register, so store it in a temp variable so it can be optimised out. Also move the personality block later, it's better for spilling/etc. as it only depends on @sqe, which we're keeping anyway. By the way, zero req->opcode if it over IORING_OP_LAST, not a problem, at the moment but is safer. Signed-off-by:
-
Pavel Begunkov authored
struct io_submit_state's ->free_list and ->link are hotter and smaller than ->plug, place them first. Signed-off-by:
-
Pavel Begunkov authored
Add extra wq flushing for fallback_work, that's not necessary but safer if invariants of io_fallback_req_func() change. Signed-off-by:
-
Pavel Begunkov authored
Plugging is only needed with requests that also need a file, so hide plugging under a ->needs_file check. Also, place ->needs_file and ->plug bits into the same byte of io_op_defs, it may matter for compilers, e.g. only with the change a tested one decided to optimise two memory testb into a more with two register testb. Signed-off-by:
-
Pavel Begunkov authored
CQE result is a 32-bit integer, so the functions generating CQEs are better to accept not long but ints. Convert io_cqring_fill_event() and other helpers. Signed-off-by:
-
Pavel Begunkov authored
Inline io_poll_complete(), it's simple and doesn't have any particular purpose. Signed-off-by:
-
Pavel Begunkov authored
There is only a single user of io_req_needs_clean() inline it. Signed-off-by:
-
Pavel Begunkov authored
We keep struct io_completion only as a temporal storage of cflags, Place it in io_kiocb, it's cleaner, removes extra bits and even might be used for future optimisations. Signed-off-by:
-
Pavel Begunkov authored
->async_data is a slow path, so it won't matter much if we do the clean up inside io_clean_op(). Moreover, in many cases it's allocated together with setting one or more of IO_REQ_CLEAN_FLAGS flags, so it'd go through io_clean_op() anyway. Control ->async_data allocation with a new flag REQ_F_ASYNC_DATA, so we can do all the maintainence under io_req_needs_clean() fast check. Signed-off-by:
-
Pavel Begunkov authored
Delay reading the next node in io_free_batch_list(), allows the compiler to load the value a bit later improving register spilling in some cases. With gcc 11.1 it helped to move @task_refs variable from the stack to a register and optimises out a couple of per request instructions. Signed-off-by:
-
Pavel Begunkov authored
Attribute cold functions so compilers can optimise them for size. It shrinks the binary by 2.5-3% text data bss dec hex filename 90670 14002 8 104680 198e8 ./fs/io_uring.o 88053 14002 8 102063 18eaf ./fs/io_uring.o Signed-off-by:
-
Pavel Begunkov authored
Currenlty, we allocate one ctx reference per request at submission time and put them at free. It's batched and not so expensive but it still bloats the kernel, adds 2 function calls for rcu and adds some overhead for request counting in io_free_batch_list(). Always keep one reference with a request, even when it's freed and in io_uring request caches. There is extra work at ring exit / quiesce paths, which now need to put all cached requests. io_ring_exit_work() is already looping, so it's not a problem. Add hybrid-busy waiting to io_ctx_quiesce() as well for now. Signed-off-by:
-
Pavel Begunkov authored
->cq_wait and ->poll_wait and waken up in the same manner, use a single waitqueue for both of them. CQ waiters are queued exclusively, so wake up should first go over all pollers and that's what we need. Signed-off-by:
-
Pavel Begunkov authored
io_cqring_ev_posted() doesn't need to wake SQPOLL, it's either done by userspace or with task_work, but no action is required on request completion. Rip off bits waking it up in io_cqring_ev_posted(). Signed-off-by:
-
Pavel Begunkov authored
The invariant of io_wq_work_list is that it's empty IFF ->first is NULL, so no need to initially set ->last. With now having more users of the list it may play a role, i.e. used in each tw iteration and on every completion flushing. Signed-off-by:
-
Pavel Begunkov authored
Even after fully inlining io_alloc_req() my compiler does a NULL check in the path of successful allocation, no hacks like an empty dereference help it. Restructure io_alloc_req() by splitting out refilling part, so the compiler generate a slightly better binary. Signed-off-by:
-
Pavel Begunkov authored
io_req_complete_state() is inlined and used in lots of places, so we want to keep it concise. Move adding a request into a completion batch list from io_req_complete_state() into the consumer, i.e. __io_queue_sqe(). before vs after text data bss dec hex filename 91894 14002 8 105904 19db0 ./fs/io_uring.o 91046 14002 8 105056 19a60 ./fs/io_uring.o Signed-off-by:
-
Pavel Begunkov authored
Add two extra unlikely() in io_submit_sqes() and one around io_req_needs_clean() to help the compiler to avoid extra jumps in hot paths. Signed-off-by:
-
Pavel Begunkov authored
We want ->comp_list in the second cacheline, which is hotter comparing to the 3rd. Swap the field with ->link, which is not as hot and controlled by flags and so not accessed unless there is a link. By the way add a couple of comments for io_kiocb fields. Signed-off-by:
-
Pavel Begunkov authored
For some reason non-off IORING_OP_TIMEOUT always fails links, it's pretty inconvenient and unnecessary limits chaining after it to hard linking, which is far from ideal, e.g. doesn't pair well with timeout cancellation. Add a flag forcing it to not fail links on -ETIME. Signed-off-by:
-
Pavel Begunkov authored
Hiding a pointer to a struct io_buffer in rw.addr is error prone. We have some place in io_kiocb, so keep kbuf's in a separate field without aliasing and risks of it being misused. Signed-off-by:
-
Pavel Begunkov authored
Move io_req_prep() call inside of io_init_req(), it simplifies a bit error handling for callers. Signed-off-by:
-
Pavel Begunkov authored
Never return from io_drain_req() but punt to tw if we've got there but it's a false positive and we shouldn't actually drain. Signed-off-by:
-
Pavel Begunkov authored
Add a helper io_init_req_drain for initialising requests with IOSQE_DRAIN set. Also move bits from preambule of io_drain_req() in there, because we already modify all the bits needed inside the helper. Signed-off-by:
-
