libev/ev_iouring.c

/*
 * libev linux io_uring fd activity backend
 *
 * Copyright (c) 2019 Marc Alexander Lehmann <libev@schmorp.de>
 * All rights reserved.
 *
 * Redistribution and use in source and binary forms, with or without modifica-
 * tion, are permitted provided that the following conditions are met:
 *
 *   1.  Redistributions of source code must retain the above copyright notice,
 *       this list of conditions and the following disclaimer.
 *
 *   2.  Redistributions in binary form must reproduce the above copyright
 *       notice, this list of conditions and the following disclaimer in the
 *       documentation and/or other materials provided with the distribution.
 *
 * THIS SOFTWARE IS PROVIDED BY THE AUTHOR ``AS IS'' AND ANY EXPRESS OR IMPLIED
 * WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF MER-
 * CHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
 * EVENT SHALL THE AUTHOR BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPE-
 * CIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO,
 * PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS;
 * OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY,
 * WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTH-
 * ERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED
 * OF THE POSSIBILITY OF SUCH DAMAGE.
 *
 * Alternatively, the contents of this file may be used under the terms of
 * the GNU General Public License ("GPL") version 2 or any later version,
 * in which case the provisions of the GPL are applicable instead of
 * the above. If you wish to allow the use of your version of this file
 * only under the terms of the GPL and not to allow others to use your
 * version of this file under the BSD license, indicate your decision
 * by deleting the provisions above and replace them with the notice
 * and other provisions required by the GPL. If you do not delete the
 * provisions above, a recipient may use your version of this file under
 * either the BSD or the GPL.
 */

/*
 * general notes about linux io_uring:
 *
 * a) it's the best interface I have seen so far. on linux.
 * b) best is not necessarily very good.
 * c) it's better than the aio mess, doesn't suffer from the fork problems
 *    of linux aio or epoll and so on and so on. and you could do event stuff
 *    without any syscalls. what's not to like?
 * d) ok, it's vastly more complex, but that's ok, really.
 * e) why 3 mmaps instead of one? one would be more space-efficient,
 *    and I can't see what benefit three would have (other than being
 *    somehow resizable/relocatable, but that's apparently not possible).
 * f) hmm, it's practiclaly undebuggable (gdb can't access the memory, and
      the bizarre way structure offsets are commuinicated makes it hard to
 *    just print the ring buffer heads, even *iff* the memory were visible
 *    in gdb. but then, that's also ok, really.
 * g) well, you cannot specify a timeout when waiting for events. no,
 *    seriously, the interface doesn't support a timeout. never seen _that_
 *    before. sure, you can use a timerfd, but that's another syscall
 *    you could have avoided. overall, this bizarre omission smells
 *    like a µ-optimisation by the io_uring author for his personal
 *    applications, to the detriment of everybody else who just wants
 *    an event loop. but, umm, ok, if that's all, it could be worse.
 * h) there is a hardcoded limit of 4096 outstanding events. okay,
 *    at least there is no arbitrary low system-wide limit...
 * i) unlike linux aio, you *can* register more then the limit
 *    of fd events, and the kernel will "gracefully" signal an
 *    overflow, after which you could destroy and recreate the kernel
 *    state, a bit bigger, or fall back to e.g. poll. thats not
 *    totally insane, but kind of questions the point a high
 *    performance I/O framework when it doesn't really work
 *    under stress.
 * j) but, oh my! is has exactly the same bugs as the linux aio backend,
 *    where some undocumented poll combinations just fail.
 *    so we need epoll AGAIN as a fallback. AGAIN! epoll!! and of course,
 *    this is completely undocumented, have I mantioned this already?
 * k) overall, the *API* itself is, I dare to say, not a total trainwreck.
 *    the big isuess with it are the bugs requiring epoll, which might
 *    or might not get fixed (do I hold my breath?).
 */

#include <sys/timerfd.h>
#include <sys/mman.h>
#include <poll.h>

#define IOURING_INIT_ENTRIES 32

/*****************************************************************************/
/* syscall wrapdadoop - this section has the raw api/abi definitions */

#include <linux/fs.h>
#include <linux/types.h>

/* mostly directly taken from the kernel or documentation */

struct io_uring_sqe
{
  __u8 opcode;
  __u8 flags;
  __u16 ioprio;
  __s32 fd;
  __u64 off;
  __u64 addr;
  __u32 len;
  union {
    __kernel_rwf_t rw_flags;
    __u32 fsync_flags;
    __u16 poll_events;
    __u32 sync_range_flags;
    __u32 msg_flags;
  };
  __u64 user_data;
  union {
    __u16 buf_index;
    __u64 __pad2[3];
  };
};

struct io_uring_cqe
{
  __u64 user_data;
  __s32 res;
  __u32 flags;
};

struct io_sqring_offsets
{
  __u32 head;
  __u32 tail;
  __u32 ring_mask;
  __u32 ring_entries;
  __u32 flags;
  __u32 dropped;
  __u32 array;
  __u32 resv1;
  __u64 resv2;
};

struct io_cqring_offsets
{
  __u32 head;
  __u32 tail;
  __u32 ring_mask;
  __u32 ring_entries;
  __u32 overflow;
  __u32 cqes;
  __u64 resv[2];
};

struct io_uring_params
{
  __u32 sq_entries;
  __u32 cq_entries;
  __u32 flags;
  __u32 sq_thread_cpu;
  __u32 sq_thread_idle;
  __u32 resv[5];
  struct io_sqring_offsets sq_off;
  struct io_cqring_offsets cq_off;
};

#define IORING_OP_POLL_ADD    6
#define IORING_OP_POLL_REMOVE 7

#define IORING_ENTER_GETEVENTS 0x01

#define IORING_OFF_SQ_RING 0x00000000ULL
#define IORING_OFF_CQ_RING 0x08000000ULL
#define IORING_OFF_SQES    0x10000000ULL

inline_size
int
evsys_io_uring_setup (unsigned entries, struct io_uring_params *params)
{
  return ev_syscall2 (SYS_io_uring_setup, entries, params);
}

inline_size
int
evsys_io_uring_enter (int fd, unsigned to_submit, unsigned min_complete, unsigned flags, const sigset_t *sig, size_t sigsz)
{
  return ev_syscall6 (SYS_io_uring_enter, fd, to_submit, min_complete, flags, sig, sigsz);
}

/*****************************************************************************/
/* actual backed implementation */

/* we hope that volatile will make the compiler access this variables only once */
#define EV_SQ_VAR(name) *(volatile unsigned *)((char *)iouring_sq_ring + iouring_sq_ ## name)
#define EV_CQ_VAR(name) *(volatile unsigned *)((char *)iouring_cq_ring + iouring_cq_ ## name)

/* the index array */
#define EV_SQ_ARRAY     ((unsigned *)((char *)iouring_sq_ring + iouring_sq_array))

/* the submit/completion queue entries */
#define EV_SQES         ((struct io_uring_sqe *)         iouring_sqes)
#define EV_CQES         ((struct io_uring_cqe *)((char *)iouring_cq_ring + iouring_cq_cqes))

static
struct io_uring_sqe *
iouring_sqe_get (EV_P)
{
  unsigned tail = EV_SQ_VAR (tail);

  if (tail + 1 - EV_SQ_VAR (head) > EV_SQ_VAR (ring_entries))
    {
      /* queue full, flush */
      evsys_io_uring_enter (iouring_fd, iouring_to_submit, 0, 0, 0, 0);
      iouring_to_submit = 0;
    }

  assert (("libev: io_uring queue full after flush", tail + 1 - EV_SQ_VAR (head) <= EV_SQ_VAR (ring_entries)));

  return EV_SQES + (tail & EV_SQ_VAR (ring_mask));
}

inline_size
struct io_uring_sqe *
iouring_sqe_submit (EV_P_ struct io_uring_sqe *sqe)
{
  unsigned idx = sqe - EV_SQES;

  EV_SQ_ARRAY [idx] = idx;
  ECB_MEMORY_FENCE_RELEASE;
  ++EV_SQ_VAR (tail);
  /*ECB_MEMORY_FENCE_RELEASE; /* for the time being we assume this is not needed */
  ++iouring_to_submit;
}

/*****************************************************************************/

/* when the timerfd expires we simply note the fact,
 * as the purpose of the timerfd is to wake us up, nothing else.
 * the next iteration should re-set it.
 */
static void
iouring_tfd_cb (EV_P_ struct ev_io *w, int revents)
{
  iouring_tfd_to = EV_TSTAMP_HUGE;
}

static void
iouring_epoll_cb (EV_P_ struct ev_io *w, int revents)
{
  epoll_poll (EV_A_ 0);
}

/* called for full and partial cleanup */
ecb_cold
static int
iouring_internal_destroy (EV_P)
{
  close (iouring_tfd);
  close (iouring_fd);

  if (iouring_sq_ring != MAP_FAILED) munmap (iouring_sq_ring, iouring_sq_ring_size);
  if (iouring_cq_ring != MAP_FAILED) munmap (iouring_cq_ring, iouring_cq_ring_size);
  if (iouring_sqes    != MAP_FAILED) munmap (iouring_sqes   , iouring_sqes_size   );

  if (ev_is_active (&iouring_epoll_w)) ev_ref (EV_A); ev_io_stop (EV_A_ &iouring_epoll_w);
  if (ev_is_active (&iouring_tfd_w  )) ev_ref (EV_A); ev_io_stop (EV_A_ &iouring_tfd_w  );
}

ecb_cold
static int
iouring_internal_init (EV_P)
{
  struct io_uring_params params = { 0 };

  iouring_to_submit = 0;

  iouring_tfd     = -1;
  iouring_sq_ring = MAP_FAILED;
  iouring_cq_ring = MAP_FAILED;
  iouring_sqes    = MAP_FAILED;

  for (;;)
    {
      iouring_fd = evsys_io_uring_setup (iouring_entries, &params);

      if (iouring_fd >= 0)
        break; /* yippie */

      if (errno != EINVAL)
        return -1; /* we failed */

      /* EINVAL: lots of possible reasons, but maybe
       * it is because we hit the unqueryable hardcoded size limit
       */

      /* we hit the limit already, give up */
      if (iouring_max_entries)
        return -1;

      /* first time we hit EINVAL? assume we hit the limit, so go back and retry */
      iouring_entries >>= 1;
      iouring_max_entries = iouring_entries;
    }

  iouring_sq_ring_size = params.sq_off.array + params.sq_entries * sizeof (unsigned);
  iouring_cq_ring_size = params.cq_off.cqes  + params.cq_entries * sizeof (struct io_uring_cqe);
  iouring_sqes_size    =                       params.sq_entries * sizeof (struct io_uring_sqe);

  iouring_sq_ring = mmap (0, iouring_sq_ring_size, PROT_READ | PROT_WRITE,
                          MAP_SHARED | MAP_POPULATE, iouring_fd, IORING_OFF_SQ_RING);
  iouring_cq_ring = mmap (0, iouring_cq_ring_size, PROT_READ | PROT_WRITE,
                          MAP_SHARED | MAP_POPULATE, iouring_fd, IORING_OFF_CQ_RING);
  iouring_sqes    = mmap (0, iouring_sqes_size, PROT_READ | PROT_WRITE,
                          MAP_SHARED | MAP_POPULATE, iouring_fd, IORING_OFF_SQES);

  if (iouring_sq_ring == MAP_FAILED || iouring_cq_ring == MAP_FAILED || iouring_sqes == MAP_FAILED)
    return -1;

  iouring_sq_head         = params.sq_off.head;
  iouring_sq_tail         = params.sq_off.tail;
  iouring_sq_ring_mask    = params.sq_off.ring_mask;
  iouring_sq_ring_entries = params.sq_off.ring_entries;
  iouring_sq_flags        = params.sq_off.flags;
  iouring_sq_dropped      = params.sq_off.dropped;
  iouring_sq_array        = params.sq_off.array;

  iouring_cq_head         = params.cq_off.head;
  iouring_cq_tail         = params.cq_off.tail;
  iouring_cq_ring_mask    = params.cq_off.ring_mask;
  iouring_cq_ring_entries = params.cq_off.ring_entries;
  iouring_cq_overflow     = params.cq_off.overflow;
  iouring_cq_cqes         = params.cq_off.cqes;

  iouring_tfd = timerfd_create (CLOCK_MONOTONIC, TFD_CLOEXEC);

  if (iouring_tfd < 0)
    return iouring_tfd;

  iouring_tfd_to = EV_TSTAMP_HUGE;

  return 0;
}

ecb_cold
static void
iouring_fork (EV_P)
{
  iouring_internal_destroy (EV_A);

  while (iouring_internal_init (EV_A) < 0)
    ev_syserr ("(libev) io_uring_setup");

  /* forking epoll should also effectively unregister all fds from the backend */
  epoll_fork (EV_A);
  /* epoll_fork already did this. hopefully */
  /*fd_rearm_all (EV_A);*/

  ev_io_stop  (EV_A_ &iouring_epoll_w);
  ev_io_set   (EV_A_ &iouring_epoll_w, backend_fd, EV_READ);
  ev_io_start (EV_A_ &iouring_epoll_w);

  ev_io_stop  (EV_A_ &iouring_tfd_w);
  ev_io_set   (EV_A_ &iouring_tfd_w, iouring_tfd, EV_READ);
  ev_io_start (EV_A_ &iouring_tfd_w);
}

/*****************************************************************************/

static void
iouring_modify (EV_P_ int fd, int oev, int nev)
{
  if (ecb_expect_false (anfds [fd].eflags))
    {
      /* we handed this fd over to epoll, so undo this first */
      /* we do it manually because the optimisations on epoll_modify won't do us any good */
      epoll_ctl (iouring_fd, EPOLL_CTL_DEL, fd, 0);
      anfds [fd].eflags = 0;
      oev = 0;
    }

  if (oev)
    {
      /* we assume the sqe's are all "properly" initialised */
      struct io_uring_sqe *sqe = iouring_sqe_get (EV_A);
      sqe->opcode    = IORING_OP_POLL_REMOVE;
      sqe->fd        = fd;
      sqe->user_data = -1;
      iouring_sqe_submit (EV_A_ sqe);

      /* increment generation counter to avoid handling old events */
      ++anfds [fd].egen;
    }

  if (nev)
    {
      struct io_uring_sqe *sqe = iouring_sqe_get (EV_A);
      sqe->opcode      = IORING_OP_POLL_ADD;
      sqe->fd          = fd;
      sqe->user_data   = (uint32_t)fd | ((__u64)(uint32_t)anfds [fd].egen << 32);
      sqe->poll_events =
        (nev & EV_READ ? POLLIN : 0)
        | (nev & EV_WRITE ? POLLOUT : 0);
      iouring_sqe_submit (EV_A_ sqe);
    }
}

inline_size
void
iouring_tfd_update (EV_P_ ev_tstamp timeout)
{
  ev_tstamp tfd_to = mn_now + timeout;

  /* we assume there will be many iterations per timer change, so
   * we only re-set the timerfd when we have to because its expiry
   * is too late.
   */
  if (ecb_expect_false (tfd_to < iouring_tfd_to))
    {
       struct itimerspec its;

       iouring_tfd_to = tfd_to;
       EV_TS_SET (its.it_interval, 0.);
       EV_TS_SET (its.it_value, tfd_to);

       if (timerfd_settime (iouring_tfd, TFD_TIMER_ABSTIME, &its, 0) < 0)
         assert (("libev: iouring timerfd_settime failed", 0));
    }
}

inline_size
void
iouring_process_cqe (EV_P_ struct io_uring_cqe *cqe)
{
  int      fd  = cqe->user_data & 0xffffffffU;
  uint32_t gen = cqe->user_data >> 32;
  int      res = cqe->res;

  /* ignore fd removal events, if there are any. TODO: verify */
  if (cqe->user_data == (__u64)-1)
    abort ();//D

  assert (("libev: io_uring fd must be in-bounds", fd >= 0 && fd < anfdmax));

  /* documentation lies, of course. the result value is NOT like
   * normal syscalls, but like linux raw syscalls, i.e. negative
   * error numbers. fortunate, as otherwise there would be no way
   * to get error codes at all. still, why not document this?
   */

  /* ignore event if generation doesn't match */
  /* this should actually be very rare */
  if (ecb_expect_false (gen != (uint32_t)anfds [fd].egen))
    return;

  if (ecb_expect_false (res < 0))
    {
      if (res == -EINVAL)
        {
          /* we assume this error code means the fd/poll combination is buggy
           * and fall back to epoll.
           * this error code might also indicate a bug, but the kernel doesn't
           * distinguish between those two conditions, so... sigh...
           */

          epoll_modify (EV_A_ fd, 0, anfds [fd].events);
        }
      else if (res == -EBADF)
        {
          assert (("libev: event loop rejected bad fd", res != -EBADF));
          fd_kill (EV_A_ fd);
        }
      else
        {
          errno = -res;
          ev_syserr ("(libev) IORING_OP_POLL_ADD");
        }

      return;
    }

  /* feed events, we do not expect or handle POLLNVAL */
  fd_event (
    EV_A_
    fd,
    (res & (POLLOUT | POLLERR | POLLHUP) ? EV_WRITE : 0)
    | (res & (POLLIN | POLLERR | POLLHUP) ? EV_READ : 0)
  );

  /* io_uring is oneshot, so we need to re-arm the fd next iteration */
  /* this also means we usually have to do at least one syscall per iteration */
  anfds [fd].events = 0;
  fd_change (EV_A_ fd, EV_ANFD_REIFY);
}

/* called when the event queue overflows */
ecb_cold
static void
iouring_overflow (EV_P)
{
  /* we have two options, resize the queue (by tearing down
   * everything and recreating it, or living with it
   * and polling.
   * we implement this by resizing tghe queue, and, if that fails,
   * we just recreate the state on every failure, which
   * kind of is a very inefficient poll.
   * one danger is, due to the bios toward lower fds,
   * we will only really get events for those, so
   * maybe we need a poll() fallback, after all.
   */
  /*EV_CQ_VAR (overflow) = 0;*/ /* need to do this if we keep the state and poll manually */

  fd_rearm_all (EV_A);

  /* we double the size until we hit the hard-to-probe maximum */
  if (!iouring_max_entries)
    {
      iouring_entries <<= 1;
      iouring_fork (EV_A);
    }
  else
    {
      /* we hit the kernel limit, we should fall back to something else.
       * we can either poll() a few times and hope for the best,
       * poll always, or switch to epoll.
       * since we use epoll anyways, go epoll.
       */

      iouring_internal_destroy (EV_A);

      /* this should make it so that on return, we don'T call any uring functions */
      iouring_to_submit = 0;

      for (;;)
        {
          backend = epoll_init (EV_A_ 0);

          if (backend)
            break;

          ev_syserr ("(libev) iouring switch to epoll");
        }
    }
}

/* handle any events in the completion queue, return true if there were any */
static int
iouring_handle_cq (EV_P)
{
  unsigned head, tail, mask;
  
  head = EV_CQ_VAR (head);
  ECB_MEMORY_FENCE_ACQUIRE;
  tail = EV_CQ_VAR (tail);

  if (head == tail)
    return 0;

  /* it can only overflow if we have events, yes, yes? */
  if (ecb_expect_false (EV_CQ_VAR (overflow)))
    {
      iouring_overflow (EV_A);
      return 1;
    }

  mask = EV_CQ_VAR (ring_mask);

  do
    iouring_process_cqe (EV_A_ &EV_CQES [head++ & mask]);
  while (head != tail);

  EV_CQ_VAR (head) = head;
  ECB_MEMORY_FENCE_RELEASE;

  return 1;
}

static void
iouring_poll (EV_P_ ev_tstamp timeout)
{
  /* if we have events, no need for extra syscalls, but we might have to queue events */
  if (iouring_handle_cq (EV_A))
    timeout = EV_TS_CONST (0.);
  else
    /* no events, so maybe wait for some */
    iouring_tfd_update (EV_A_ timeout);

  /* only enter the kernel if we have somethign to submit, or we need to wait */
  if (timeout || iouring_to_submit)
    {
      int res;

      EV_RELEASE_CB;

      res = evsys_io_uring_enter (iouring_fd, iouring_to_submit, 1,
                                  timeout > EV_TS_CONST (0.) ? IORING_ENTER_GETEVENTS : 0, 0, 0);
      iouring_to_submit = 0;

      EV_ACQUIRE_CB;

      if (ecb_expect_false (res < 0))
        if (errno == EINTR)
          /* ignore */;
        else
          ev_syserr ("(libev) iouring setup");
      else
        iouring_handle_cq (EV_A);
    }
}

inline_size
int
iouring_init (EV_P_ int flags)
{
  if (!epoll_init (EV_A_ 0))
    return 0;

  ev_io_init  (&iouring_epoll_w, iouring_epoll_cb, backend_fd, EV_READ);
  ev_set_priority (&iouring_epoll_w, EV_MAXPRI);

  ev_io_init  (&iouring_tfd_w, iouring_tfd_cb, iouring_tfd, EV_READ);
  ev_set_priority (&iouring_tfd_w, EV_MAXPRI);

  iouring_entries     = IOURING_INIT_ENTRIES;
  iouring_max_entries = 0;

  if (iouring_internal_init (EV_A) < 0)
    {
      iouring_internal_destroy (EV_A);
      return 0;
    }

  ev_io_start (EV_A_ &iouring_epoll_w);
  ev_unref (EV_A); /* watcher should not keep loop alive */

  ev_io_start (EV_A_ &iouring_tfd_w);
  ev_unref (EV_A); /* watcher should not keep loop alive */

  backend_modify = iouring_modify;
  backend_poll   = iouring_poll;

  return EVBACKEND_IOURING;
}

inline_size
void
iouring_destroy (EV_P)
{
  iouring_internal_destroy (EV_A);
  epoll_destroy (EV_A);
}

Revision:	1.5
Committed:	Tue Nov 19 13:09:00 2019 UTC (4 years, 10 months ago) by root
Content type:	text/plain
Branch:	MAIN
CVS Tags:	EV-rel-4_28, EV-rel-4_29, EV-rel-4_30
Changes since 1.4:	+0 -3 lines
Log Message:	* empty log message *
#	Content
1	/*
2	* libev linux io_uring fd activity backend
3	*
4	* Copyright (c) 2019 Marc Alexander Lehmann <libev@schmorp.de>
5	* All rights reserved.
6	*
7	* Redistribution and use in source and binary forms, with or without modifica-
8	* tion, are permitted provided that the following conditions are met:
9	*
10	* 1. Redistributions of source code must retain the above copyright notice,
11	* this list of conditions and the following disclaimer.
12	*
13	* 2. Redistributions in binary form must reproduce the above copyright
14	* notice, this list of conditions and the following disclaimer in the
15	* documentation and/or other materials provided with the distribution.
16	*
17	* THIS SOFTWARE IS PROVIDED BY THE AUTHOR ``AS IS'' AND ANY EXPRESS OR IMPLIED
18	* WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF MER-
19	* CHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO
20	* EVENT SHALL THE AUTHOR BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPE-
21	* CIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO,
22	* PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS;
23	* OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY,
24	* WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTH-
25	* ERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED
26	* OF THE POSSIBILITY OF SUCH DAMAGE.
27	*
28	* Alternatively, the contents of this file may be used under the terms of
29	* the GNU General Public License ("GPL") version 2 or any later version,
30	* in which case the provisions of the GPL are applicable instead of
31	* the above. If you wish to allow the use of your version of this file
32	* only under the terms of the GPL and not to allow others to use your
33	* version of this file under the BSD license, indicate your decision
34	* by deleting the provisions above and replace them with the notice
35	* and other provisions required by the GPL. If you do not delete the
36	* provisions above, a recipient may use your version of this file under
37	* either the BSD or the GPL.
38	*/
39
40	/*
41	* general notes about linux io_uring:
42	*
43	* a) it's the best interface I have seen so far. on linux.
44	* b) best is not necessarily very good.
45	* c) it's better than the aio mess, doesn't suffer from the fork problems
46	* of linux aio or epoll and so on and so on. and you could do event stuff
47	* without any syscalls. what's not to like?
48	* d) ok, it's vastly more complex, but that's ok, really.
49	* e) why 3 mmaps instead of one? one would be more space-efficient,
50	* and I can't see what benefit three would have (other than being
51	* somehow resizable/relocatable, but that's apparently not possible).
52	* f) hmm, it's practiclaly undebuggable (gdb can't access the memory, and
53	the bizarre way structure offsets are commuinicated makes it hard to
54	* just print the ring buffer heads, even iff the memory were visible
55	* in gdb. but then, that's also ok, really.
56	* g) well, you cannot specify a timeout when waiting for events. no,
57	* seriously, the interface doesn't support a timeout. never seen _that_
58	* before. sure, you can use a timerfd, but that's another syscall
59	* you could have avoided. overall, this bizarre omission smells
60	* like a µ-optimisation by the io_uring author for his personal
61	* applications, to the detriment of everybody else who just wants
62	* an event loop. but, umm, ok, if that's all, it could be worse.
63	* h) there is a hardcoded limit of 4096 outstanding events. okay,
64	* at least there is no arbitrary low system-wide limit...
65	* i) unlike linux aio, you can register more then the limit
66	* of fd events, and the kernel will "gracefully" signal an
67	* overflow, after which you could destroy and recreate the kernel
68	* state, a bit bigger, or fall back to e.g. poll. thats not
69	* totally insane, but kind of questions the point a high
70	* performance I/O framework when it doesn't really work
71	* under stress.
72	* j) but, oh my! is has exactly the same bugs as the linux aio backend,
73	* where some undocumented poll combinations just fail.
74	* so we need epoll AGAIN as a fallback. AGAIN! epoll!! and of course,
75	* this is completely undocumented, have I mantioned this already?
76	* k) overall, the API itself is, I dare to say, not a total trainwreck.
77	* the big isuess with it are the bugs requiring epoll, which might
78	* or might not get fixed (do I hold my breath?).
79	*/
80
81	#include <sys/timerfd.h>
82	#include <sys/mman.h>
83	#include <poll.h>
84
85	#define IOURING_INIT_ENTRIES 32
86
87	/*****************************************************************************/
88	/* syscall wrapdadoop - this section has the raw api/abi definitions */
89
90	#include <linux/fs.h>
91	#include <linux/types.h>
92
93	/* mostly directly taken from the kernel or documentation */
94
95	struct io_uring_sqe
96	{
97	__u8 opcode;
98	__u8 flags;
99	__u16 ioprio;
100	__s32 fd;
101	__u64 off;
102	__u64 addr;
103	__u32 len;
104	union {
105	__kernel_rwf_t rw_flags;
106	__u32 fsync_flags;
107	__u16 poll_events;
108	__u32 sync_range_flags;
109	__u32 msg_flags;
110	};
111	__u64 user_data;
112	union {
113	__u16 buf_index;
114	__u64 __pad2[3];
115	};
116	};
117
118	struct io_uring_cqe
119	{
120	__u64 user_data;
121	__s32 res;
122	__u32 flags;
123	};
124
125	struct io_sqring_offsets
126	{
127	__u32 head;
128	__u32 tail;
129	__u32 ring_mask;
130	__u32 ring_entries;
131	__u32 flags;
132	__u32 dropped;
133	__u32 array;
134	__u32 resv1;
135	__u64 resv2;
136	};
137
138	struct io_cqring_offsets
139	{
140	__u32 head;
141	__u32 tail;
142	__u32 ring_mask;
143	__u32 ring_entries;
144	__u32 overflow;
145	__u32 cqes;
146	__u64 resv[2];
147	};
148
149	struct io_uring_params
150	{
151	__u32 sq_entries;
152	__u32 cq_entries;
153	__u32 flags;
154	__u32 sq_thread_cpu;
155	__u32 sq_thread_idle;
156	__u32 resv[5];
157	struct io_sqring_offsets sq_off;
158	struct io_cqring_offsets cq_off;
159	};
160
161	#define IORING_OP_POLL_ADD 6
162	#define IORING_OP_POLL_REMOVE 7
163
164	#define IORING_ENTER_GETEVENTS 0x01
165
166	#define IORING_OFF_SQ_RING 0x00000000ULL
167	#define IORING_OFF_CQ_RING 0x08000000ULL
168	#define IORING_OFF_SQES 0x10000000ULL
169
170	inline_size
171	int
172	evsys_io_uring_setup (unsigned entries, struct io_uring_params *params)
173	{
174	return ev_syscall2 (SYS_io_uring_setup, entries, params);
175	}
176
177	inline_size
178	int
179	evsys_io_uring_enter (int fd, unsigned to_submit, unsigned min_complete, unsigned flags, const sigset_t *sig, size_t sigsz)
180	{
181	return ev_syscall6 (SYS_io_uring_enter, fd, to_submit, min_complete, flags, sig, sigsz);
182	}
183
184	/*****************************************************************************/
185	/* actual backed implementation */
186
187	/* we hope that volatile will make the compiler access this variables only once */
188	#define EV_SQ_VAR(name) (volatile unsigned )((char *)iouring_sq_ring + iouring_sq_ ## name)
189	#define EV_CQ_VAR(name) (volatile unsigned )((char *)iouring_cq_ring + iouring_cq_ ## name)
190
191	/* the index array */
192	#define EV_SQ_ARRAY ((unsigned )((char )iouring_sq_ring + iouring_sq_array))
193
194	/* the submit/completion queue entries */
195	#define EV_SQES ((struct io_uring_sqe *) iouring_sqes)
196	#define EV_CQES ((struct io_uring_cqe )((char )iouring_cq_ring + iouring_cq_cqes))
197
198	static
199	struct io_uring_sqe *
200	iouring_sqe_get (EV_P)
201	{
202	unsigned tail = EV_SQ_VAR (tail);
203
204	if (tail + 1 - EV_SQ_VAR (head) > EV_SQ_VAR (ring_entries))
205	{
206	/* queue full, flush */
207	evsys_io_uring_enter (iouring_fd, iouring_to_submit, 0, 0, 0, 0);
208	iouring_to_submit = 0;
209	}
210
211	assert (("libev: io_uring queue full after flush", tail + 1 - EV_SQ_VAR (head) <= EV_SQ_VAR (ring_entries)));
212
213	return EV_SQES + (tail & EV_SQ_VAR (ring_mask));
214	}
215
216	inline_size
217	struct io_uring_sqe *
218	iouring_sqe_submit (EV_P_ struct io_uring_sqe *sqe)
219	{
220	unsigned idx = sqe - EV_SQES;
221
222	EV_SQ_ARRAY [idx] = idx;
223	ECB_MEMORY_FENCE_RELEASE;
224	++EV_SQ_VAR (tail);
225	/ECB_MEMORY_FENCE_RELEASE; / for the time being we assume this is not needed */
226	++iouring_to_submit;
227	}
228
229	/*****************************************************************************/
230
231	/* when the timerfd expires we simply note the fact,
232	* as the purpose of the timerfd is to wake us up, nothing else.
233	* the next iteration should re-set it.
234	*/
235	static void
236	iouring_tfd_cb (EV_P_ struct ev_io *w, int revents)
237	{
238	iouring_tfd_to = EV_TSTAMP_HUGE;
239	}
240
241	static void
242	iouring_epoll_cb (EV_P_ struct ev_io *w, int revents)
243	{
244	epoll_poll (EV_A_ 0);
245	}
246
247	/* called for full and partial cleanup */
248	ecb_cold
249	static int
250	iouring_internal_destroy (EV_P)
251	{
252	close (iouring_tfd);
253	close (iouring_fd);
254
255	if (iouring_sq_ring != MAP_FAILED) munmap (iouring_sq_ring, iouring_sq_ring_size);
256	if (iouring_cq_ring != MAP_FAILED) munmap (iouring_cq_ring, iouring_cq_ring_size);
257	if (iouring_sqes != MAP_FAILED) munmap (iouring_sqes , iouring_sqes_size );
258
259	if (ev_is_active (&iouring_epoll_w)) ev_ref (EV_A); ev_io_stop (EV_A_ &iouring_epoll_w);
260	if (ev_is_active (&iouring_tfd_w )) ev_ref (EV_A); ev_io_stop (EV_A_ &iouring_tfd_w );
261	}
262
263	ecb_cold
264	static int
265	iouring_internal_init (EV_P)
266	{
267	struct io_uring_params params = { 0 };
268
269	iouring_to_submit = 0;
270
271	iouring_tfd = -1;
272	iouring_sq_ring = MAP_FAILED;
273	iouring_cq_ring = MAP_FAILED;
274	iouring_sqes = MAP_FAILED;
275
276	for (;;)
277	{
278	iouring_fd = evsys_io_uring_setup (iouring_entries, &params);
279
280	if (iouring_fd >= 0)
281	break; /* yippie */
282
283	if (errno != EINVAL)
284	return -1; /* we failed */
285
286	/* EINVAL: lots of possible reasons, but maybe
287	* it is because we hit the unqueryable hardcoded size limit
288	*/
289
290	/* we hit the limit already, give up */
291	if (iouring_max_entries)
292	return -1;
293
294	/* first time we hit EINVAL? assume we hit the limit, so go back and retry */
295	iouring_entries >>= 1;
296	iouring_max_entries = iouring_entries;
297	}
298
299	iouring_sq_ring_size = params.sq_off.array + params.sq_entries * sizeof (unsigned);
300	iouring_cq_ring_size = params.cq_off.cqes + params.cq_entries * sizeof (struct io_uring_cqe);
301	iouring_sqes_size = params.sq_entries * sizeof (struct io_uring_sqe);
302
303	iouring_sq_ring = mmap (0, iouring_sq_ring_size, PROT_READ \| PROT_WRITE,
304	MAP_SHARED \| MAP_POPULATE, iouring_fd, IORING_OFF_SQ_RING);
305	iouring_cq_ring = mmap (0, iouring_cq_ring_size, PROT_READ \| PROT_WRITE,
306	MAP_SHARED \| MAP_POPULATE, iouring_fd, IORING_OFF_CQ_RING);
307	iouring_sqes = mmap (0, iouring_sqes_size, PROT_READ \| PROT_WRITE,
308	MAP_SHARED \| MAP_POPULATE, iouring_fd, IORING_OFF_SQES);
309
310	if (iouring_sq_ring == MAP_FAILED \|\| iouring_cq_ring == MAP_FAILED \|\| iouring_sqes == MAP_FAILED)
311	return -1;
312
313	iouring_sq_head = params.sq_off.head;
314	iouring_sq_tail = params.sq_off.tail;
315	iouring_sq_ring_mask = params.sq_off.ring_mask;
316	iouring_sq_ring_entries = params.sq_off.ring_entries;
317	iouring_sq_flags = params.sq_off.flags;
318	iouring_sq_dropped = params.sq_off.dropped;
319	iouring_sq_array = params.sq_off.array;
320
321	iouring_cq_head = params.cq_off.head;
322	iouring_cq_tail = params.cq_off.tail;
323	iouring_cq_ring_mask = params.cq_off.ring_mask;
324	iouring_cq_ring_entries = params.cq_off.ring_entries;
325	iouring_cq_overflow = params.cq_off.overflow;
326	iouring_cq_cqes = params.cq_off.cqes;
327
328	iouring_tfd = timerfd_create (CLOCK_MONOTONIC, TFD_CLOEXEC);
329
330	if (iouring_tfd < 0)
331	return iouring_tfd;
332
333	iouring_tfd_to = EV_TSTAMP_HUGE;
334
335	return 0;
336	}
337
338	ecb_cold
339	static void
340	iouring_fork (EV_P)
341	{
342	iouring_internal_destroy (EV_A);
343
344	while (iouring_internal_init (EV_A) < 0)
345	ev_syserr ("(libev) io_uring_setup");
346
347	/* forking epoll should also effectively unregister all fds from the backend */
348	epoll_fork (EV_A);
349	/* epoll_fork already did this. hopefully */
350	/fd_rearm_all (EV_A);/
351
352	ev_io_stop (EV_A_ &iouring_epoll_w);
353	ev_io_set (EV_A_ &iouring_epoll_w, backend_fd, EV_READ);
354	ev_io_start (EV_A_ &iouring_epoll_w);
355
356	ev_io_stop (EV_A_ &iouring_tfd_w);
357	ev_io_set (EV_A_ &iouring_tfd_w, iouring_tfd, EV_READ);
358	ev_io_start (EV_A_ &iouring_tfd_w);
359	}
360
361	/*****************************************************************************/
362
363	static void
364	iouring_modify (EV_P_ int fd, int oev, int nev)
365	{
366	if (ecb_expect_false (anfds [fd].eflags))
367	{
368	/* we handed this fd over to epoll, so undo this first */
369	/* we do it manually because the optimisations on epoll_modify won't do us any good */
370	epoll_ctl (iouring_fd, EPOLL_CTL_DEL, fd, 0);
371	anfds [fd].eflags = 0;
372	oev = 0;
373	}
374
375	if (oev)
376	{
377	/* we assume the sqe's are all "properly" initialised */
378	struct io_uring_sqe *sqe = iouring_sqe_get (EV_A);
379	sqe->opcode = IORING_OP_POLL_REMOVE;
380	sqe->fd = fd;
381	sqe->user_data = -1;
382	iouring_sqe_submit (EV_A_ sqe);
383
384	/* increment generation counter to avoid handling old events */
385	++anfds [fd].egen;
386	}
387
388	if (nev)
389	{
390	struct io_uring_sqe *sqe = iouring_sqe_get (EV_A);
391	sqe->opcode = IORING_OP_POLL_ADD;
392	sqe->fd = fd;
393	sqe->user_data = (uint32_t)fd \| ((__u64)(uint32_t)anfds [fd].egen << 32);
394	sqe->poll_events =
395	(nev & EV_READ ? POLLIN : 0)
396	\| (nev & EV_WRITE ? POLLOUT : 0);
397	iouring_sqe_submit (EV_A_ sqe);
398	}
399	}
400
401	inline_size
402	void
403	iouring_tfd_update (EV_P_ ev_tstamp timeout)
404	{
405	ev_tstamp tfd_to = mn_now + timeout;
406
407	/* we assume there will be many iterations per timer change, so
408	* we only re-set the timerfd when we have to because its expiry
409	* is too late.
410	*/
411	if (ecb_expect_false (tfd_to < iouring_tfd_to))
412	{
413	struct itimerspec its;
414
415	iouring_tfd_to = tfd_to;
416	EV_TS_SET (its.it_interval, 0.);
417	EV_TS_SET (its.it_value, tfd_to);
418
419	if (timerfd_settime (iouring_tfd, TFD_TIMER_ABSTIME, &its, 0) < 0)
420	assert (("libev: iouring timerfd_settime failed", 0));
421	}
422	}
423
424	inline_size
425	void
426	iouring_process_cqe (EV_P_ struct io_uring_cqe *cqe)
427	{
428	int fd = cqe->user_data & 0xffffffffU;
429	uint32_t gen = cqe->user_data >> 32;
430	int res = cqe->res;
431
432	/* ignore fd removal events, if there are any. TODO: verify */
433	if (cqe->user_data == (__u64)-1)
434	abort ();//D
435
436	assert (("libev: io_uring fd must be in-bounds", fd >= 0 && fd < anfdmax));
437
438	/* documentation lies, of course. the result value is NOT like
439	* normal syscalls, but like linux raw syscalls, i.e. negative
440	* error numbers. fortunate, as otherwise there would be no way
441	* to get error codes at all. still, why not document this?
442	*/
443
444	/* ignore event if generation doesn't match */
445	/* this should actually be very rare */
446	if (ecb_expect_false (gen != (uint32_t)anfds [fd].egen))
447	return;
448
449	if (ecb_expect_false (res < 0))
450	{
451	if (res == -EINVAL)
452	{
453	/* we assume this error code means the fd/poll combination is buggy
454	* and fall back to epoll.
455	* this error code might also indicate a bug, but the kernel doesn't
456	* distinguish between those two conditions, so... sigh...
457	*/
458
459	epoll_modify (EV_A_ fd, 0, anfds [fd].events);
460	}
461	else if (res == -EBADF)
462	{
463	assert (("libev: event loop rejected bad fd", res != -EBADF));
464	fd_kill (EV_A_ fd);
465	}
466	else
467	{
468	errno = -res;
469	ev_syserr ("(libev) IORING_OP_POLL_ADD");
470	}
471
472	return;
473	}
474
475	/* feed events, we do not expect or handle POLLNVAL */
476	fd_event (
477	EV_A_
478	fd,
479	(res & (POLLOUT \| POLLERR \| POLLHUP) ? EV_WRITE : 0)
480	\| (res & (POLLIN \| POLLERR \| POLLHUP) ? EV_READ : 0)
481	);
482
483	/* io_uring is oneshot, so we need to re-arm the fd next iteration */
484	/* this also means we usually have to do at least one syscall per iteration */
485	anfds [fd].events = 0;
486	fd_change (EV_A_ fd, EV_ANFD_REIFY);
487	}
488
489	/* called when the event queue overflows */
490	ecb_cold
491	static void
492	iouring_overflow (EV_P)
493	{
494	/* we have two options, resize the queue (by tearing down
495	* everything and recreating it, or living with it
496	* and polling.
497	* we implement this by resizing tghe queue, and, if that fails,
498	* we just recreate the state on every failure, which
499	* kind of is a very inefficient poll.
500	* one danger is, due to the bios toward lower fds,
501	* we will only really get events for those, so
502	* maybe we need a poll() fallback, after all.
503	*/
504	/EV_CQ_VAR (overflow) = 0;/ /* need to do this if we keep the state and poll manually */
505
506	fd_rearm_all (EV_A);
507
508	/* we double the size until we hit the hard-to-probe maximum */
509	if (!iouring_max_entries)
510	{
511	iouring_entries <<= 1;
512	iouring_fork (EV_A);
513	}
514	else
515	{
516	/* we hit the kernel limit, we should fall back to something else.
517	* we can either poll() a few times and hope for the best,
518	* poll always, or switch to epoll.
519	* since we use epoll anyways, go epoll.
520	*/
521
522	iouring_internal_destroy (EV_A);
523
524	/* this should make it so that on return, we don'T call any uring functions */
525	iouring_to_submit = 0;
526
527	for (;;)
528	{
529	backend = epoll_init (EV_A_ 0);
530
531	if (backend)
532	break;
533
534	ev_syserr ("(libev) iouring switch to epoll");
535	}
536	}
537	}
538
539	/* handle any events in the completion queue, return true if there were any */
540	static int
541	iouring_handle_cq (EV_P)
542	{
543	unsigned head, tail, mask;
544
545	head = EV_CQ_VAR (head);
546	ECB_MEMORY_FENCE_ACQUIRE;
547	tail = EV_CQ_VAR (tail);
548
549	if (head == tail)
550	return 0;
551
552	/* it can only overflow if we have events, yes, yes? */
553	if (ecb_expect_false (EV_CQ_VAR (overflow)))
554	{
555	iouring_overflow (EV_A);
556	return 1;
557	}
558
559	mask = EV_CQ_VAR (ring_mask);
560
561	do
562	iouring_process_cqe (EV_A_ &EV_CQES [head++ & mask]);
563	while (head != tail);
564
565	EV_CQ_VAR (head) = head;
566	ECB_MEMORY_FENCE_RELEASE;
567
568	return 1;
569	}
570
571	static void
572	iouring_poll (EV_P_ ev_tstamp timeout)
573	{
574	/* if we have events, no need for extra syscalls, but we might have to queue events */
575	if (iouring_handle_cq (EV_A))
576	timeout = EV_TS_CONST (0.);
577	else
578	/* no events, so maybe wait for some */
579	iouring_tfd_update (EV_A_ timeout);
580
581	/* only enter the kernel if we have somethign to submit, or we need to wait */
582	if (timeout \|\| iouring_to_submit)
583	{
584	int res;
585
586	EV_RELEASE_CB;
587
588	res = evsys_io_uring_enter (iouring_fd, iouring_to_submit, 1,
589	timeout > EV_TS_CONST (0.) ? IORING_ENTER_GETEVENTS : 0, 0, 0);
590	iouring_to_submit = 0;
591
592	EV_ACQUIRE_CB;
593
594	if (ecb_expect_false (res < 0))
595	if (errno == EINTR)
596	/* ignore */;
597	else
598	ev_syserr ("(libev) iouring setup");
599	else
600	iouring_handle_cq (EV_A);
601	}
602	}
603
604	inline_size
605	int
606	iouring_init (EV_P_ int flags)
607	{
608	if (!epoll_init (EV_A_ 0))
609	return 0;
610
611	ev_io_init (&iouring_epoll_w, iouring_epoll_cb, backend_fd, EV_READ);
612	ev_set_priority (&iouring_epoll_w, EV_MAXPRI);
613
614	ev_io_init (&iouring_tfd_w, iouring_tfd_cb, iouring_tfd, EV_READ);
615	ev_set_priority (&iouring_tfd_w, EV_MAXPRI);
616
617	iouring_entries = IOURING_INIT_ENTRIES;
618	iouring_max_entries = 0;
619
620	if (iouring_internal_init (EV_A) < 0)
621	{
622	iouring_internal_destroy (EV_A);
623	return 0;
624	}
625
626	ev_io_start (EV_A_ &iouring_epoll_w);
627	ev_unref (EV_A); /* watcher should not keep loop alive */
628
629	ev_io_start (EV_A_ &iouring_tfd_w);
630	ev_unref (EV_A); /* watcher should not keep loop alive */
631
632	backend_modify = iouring_modify;
633	backend_poll = iouring_poll;
634
635	return EVBACKEND_IOURING;
636	}
637
638	inline_size
639	void
640	iouring_destroy (EV_P)
641	{
642	iouring_internal_destroy (EV_A);
643	epoll_destroy (EV_A);
644	}
645