[mirror_qemu.git] / block / io_uring.c

/*
 * Linux io_uring support.
 *
 * Copyright (C) 2009 IBM, Corp.
 * Copyright (C) 2009 Red Hat, Inc.
 * Copyright (C) 2019 Aarushi Mehta
 *
 * This work is licensed under the terms of the GNU GPL, version 2 or later.
 * See the COPYING file in the top-level directory.
 */
#include "qemu/osdep.h"
#include <liburing.h>
#include "block/aio.h"
#include "qemu/queue.h"
#include "block/block.h"
#include "block/raw-aio.h"
#include "qemu/coroutine.h"
#include "qapi/error.h"
#include "trace.h"

/* io_uring ring size */
#define MAX_ENTRIES 128

typedef struct LuringAIOCB {
    Coroutine *co;
    struct io_uring_sqe sqeq;
    ssize_t ret;
    QEMUIOVector *qiov;
    bool is_read;
    QSIMPLEQ_ENTRY(LuringAIOCB) next;

    /*
     * Buffered reads may require resubmission, see
     * luring_resubmit_short_read().
     */
    int total_read;
    QEMUIOVector resubmit_qiov;
} LuringAIOCB;

typedef struct LuringQueue {
    int plugged;
    unsigned int in_queue;
    unsigned int in_flight;
    bool blocked;
    QSIMPLEQ_HEAD(, LuringAIOCB) submit_queue;
} LuringQueue;

typedef struct LuringState {
    AioContext *aio_context;

    struct io_uring ring;

    /* io queue for submit at batch.  Protected by AioContext lock. */
    LuringQueue io_q;

    /* I/O completion processing.  Only runs in I/O thread.  */
    QEMUBH *completion_bh;
} LuringState;

/**
 * luring_resubmit:
 *
 * Resubmit a request by appending it to submit_queue.  The caller must ensure
 * that ioq_submit() is called later so that submit_queue requests are started.
 */
static void luring_resubmit(LuringState *s, LuringAIOCB *luringcb)
{
    QSIMPLEQ_INSERT_TAIL(&s->io_q.submit_queue, luringcb, next);
    s->io_q.in_queue++;
}

/**
 * luring_resubmit_short_read:
 *
 * Short reads are rare but may occur. The remaining read request needs to be
 * resubmitted.
 */
static void luring_resubmit_short_read(LuringState *s, LuringAIOCB *luringcb,
                                       int nread)
{
    QEMUIOVector *resubmit_qiov;
    size_t remaining;

    trace_luring_resubmit_short_read(s, luringcb, nread);

    /* Update read position */
    luringcb->total_read += nread;
    remaining = luringcb->qiov->size - luringcb->total_read;

    /* Shorten qiov */
    resubmit_qiov = &luringcb->resubmit_qiov;
    if (resubmit_qiov->iov == NULL) {
        qemu_iovec_init(resubmit_qiov, luringcb->qiov->niov);
    } else {
        qemu_iovec_reset(resubmit_qiov);
    }
    qemu_iovec_concat(resubmit_qiov, luringcb->qiov, luringcb->total_read,
                      remaining);

    /* Update sqe */
    luringcb->sqeq.off += nread;
    luringcb->sqeq.addr = (__u64)(uintptr_t)luringcb->resubmit_qiov.iov;
    luringcb->sqeq.len = luringcb->resubmit_qiov.niov;

    luring_resubmit(s, luringcb);
}

/**
 * luring_process_completions:
 * @s: AIO state
 *
 * Fetches completed I/O requests, consumes cqes and invokes their callbacks
 * The function is somewhat tricky because it supports nested event loops, for
 * example when a request callback invokes aio_poll().
 *
 * Function schedules BH completion so it  can be called again in a nested
 * event loop.  When there are no events left  to complete the BH is being
 * canceled.
 *
 */
static void luring_process_completions(LuringState *s)
{
    struct io_uring_cqe *cqes;
    int total_bytes;
    /*
     * Request completion callbacks can run the nested event loop.
     * Schedule ourselves so the nested event loop will "see" remaining
     * completed requests and process them.  Without this, completion
     * callbacks that wait for other requests using a nested event loop
     * would hang forever.
     *
     * This workaround is needed because io_uring uses poll_wait, which
     * is woken up when new events are added to the uring, thus polling on
     * the same uring fd will block unless more events are received.
     *
     * Other leaf block drivers (drivers that access the data themselves)
     * are networking based, so they poll sockets for data and run the
     * correct coroutine.
     */
    qemu_bh_schedule(s->completion_bh);

    while (io_uring_peek_cqe(&s->ring, &cqes) == 0) {
        LuringAIOCB *luringcb;
        int ret;

        if (!cqes) {
            break;
        }

        luringcb = io_uring_cqe_get_data(cqes);
        ret = cqes->res;
        io_uring_cqe_seen(&s->ring, cqes);
        cqes = NULL;

        /* Change counters one-by-one because we can be nested. */
        s->io_q.in_flight--;
        trace_luring_process_completion(s, luringcb, ret);

        /* total_read is non-zero only for resubmitted read requests */
        total_bytes = ret + luringcb->total_read;

        if (ret < 0) {
            /*
             * Only writev/readv/fsync requests on regular files or host block
             * devices are submitted. Therefore -EAGAIN is not expected but it's
             * known to happen sometimes with Linux SCSI. Submit again and hope
             * the request completes successfully.
             *
             * For more information, see:
             * https://lore.kernel.org/io-uring/20210727165811.284510-3-axboe@kernel.dk/T/#u
             *
             * If the code is changed to submit other types of requests in the
             * future, then this workaround may need to be extended to deal with
             * genuine -EAGAIN results that should not be resubmitted
             * immediately.
             */
            if (ret == -EINTR || ret == -EAGAIN) {
                luring_resubmit(s, luringcb);
                continue;
            }
        } else if (!luringcb->qiov) {
            goto end;
        } else if (total_bytes == luringcb->qiov->size) {
            ret = 0;
        /* Only read/write */
        } else {
            /* Short Read/Write */
            if (luringcb->is_read) {
                if (ret > 0) {
                    luring_resubmit_short_read(s, luringcb, ret);
                    continue;
                } else {
                    /* Pad with zeroes */
                    qemu_iovec_memset(luringcb->qiov, total_bytes, 0,
                                      luringcb->qiov->size - total_bytes);
                    ret = 0;
                }
            } else {
                ret = -ENOSPC;
            }
        }
end:
        luringcb->ret = ret;
        qemu_iovec_destroy(&luringcb->resubmit_qiov);

        /*
         * If the coroutine is already entered it must be in ioq_submit()
         * and will notice luringcb->ret has been filled in when it
         * eventually runs later. Coroutines cannot be entered recursively
         * so avoid doing that!
         */
        if (!qemu_coroutine_entered(luringcb->co)) {
            aio_co_wake(luringcb->co);
        }
    }
    qemu_bh_cancel(s->completion_bh);
}

static int ioq_submit(LuringState *s)
{
    int ret = 0;
    LuringAIOCB *luringcb, *luringcb_next;

    while (s->io_q.in_queue > 0) {
        /*
         * Try to fetch sqes from the ring for requests waiting in
         * the overflow queue
         */
        QSIMPLEQ_FOREACH_SAFE(luringcb, &s->io_q.submit_queue, next,
                              luringcb_next) {
            struct io_uring_sqe *sqes = io_uring_get_sqe(&s->ring);
            if (!sqes) {
                break;
            }
            /* Prep sqe for submission */
            *sqes = luringcb->sqeq;
            QSIMPLEQ_REMOVE_HEAD(&s->io_q.submit_queue, next);
        }
        ret = io_uring_submit(&s->ring);
        trace_luring_io_uring_submit(s, ret);
        /* Prevent infinite loop if submission is refused */
        if (ret <= 0) {
            if (ret == -EAGAIN || ret == -EINTR) {
                continue;
            }
            break;
        }
        s->io_q.in_flight += ret;
        s->io_q.in_queue  -= ret;
    }
    s->io_q.blocked = (s->io_q.in_queue > 0);

    if (s->io_q.in_flight) {
        /*
         * We can try to complete something just right away if there are
         * still requests in-flight.
         */
        luring_process_completions(s);
    }
    return ret;
}

static void luring_process_completions_and_submit(LuringState *s)
{
    aio_context_acquire(s->aio_context);
    luring_process_completions(s);

    if (!s->io_q.plugged && s->io_q.in_queue > 0) {
        ioq_submit(s);
    }
    aio_context_release(s->aio_context);
}

static void qemu_luring_completion_bh(void *opaque)
{
    LuringState *s = opaque;
    luring_process_completions_and_submit(s);
}

static void qemu_luring_completion_cb(void *opaque)
{
    LuringState *s = opaque;
    luring_process_completions_and_submit(s);
}

static bool qemu_luring_poll_cb(void *opaque)
{
    LuringState *s = opaque;

    return io_uring_cq_ready(&s->ring);
}

static void qemu_luring_poll_ready(void *opaque)
{
    LuringState *s = opaque;

    luring_process_completions_and_submit(s);
}

static void ioq_init(LuringQueue *io_q)
{
    QSIMPLEQ_INIT(&io_q->submit_queue);
    io_q->plugged = 0;
    io_q->in_queue = 0;
    io_q->in_flight = 0;
    io_q->blocked = false;
}

void luring_io_plug(BlockDriverState *bs, LuringState *s)
{
    trace_luring_io_plug(s);
    s->io_q.plugged++;
}

void luring_io_unplug(BlockDriverState *bs, LuringState *s)
{
    assert(s->io_q.plugged);
    trace_luring_io_unplug(s, s->io_q.blocked, s->io_q.plugged,
                           s->io_q.in_queue, s->io_q.in_flight);
    if (--s->io_q.plugged == 0 &&
        !s->io_q.blocked && s->io_q.in_queue > 0) {
        ioq_submit(s);
    }
}

/**
 * luring_do_submit:
 * @fd: file descriptor for I/O
 * @luringcb: AIO control block
 * @s: AIO state
 * @offset: offset for request
 * @type: type of request
 *
 * Fetches sqes from ring, adds to pending queue and preps them
 *
 */
static int luring_do_submit(int fd, LuringAIOCB *luringcb, LuringState *s,
                            uint64_t offset, int type)
{
    int ret;
    struct io_uring_sqe *sqes = &luringcb->sqeq;

    switch (type) {
    case QEMU_AIO_WRITE:
        io_uring_prep_writev(sqes, fd, luringcb->qiov->iov,
                             luringcb->qiov->niov, offset);
        break;
    case QEMU_AIO_READ:
        io_uring_prep_readv(sqes, fd, luringcb->qiov->iov,
                            luringcb->qiov->niov, offset);
        break;
    case QEMU_AIO_FLUSH:
        io_uring_prep_fsync(sqes, fd, IORING_FSYNC_DATASYNC);
        break;
    default:
        fprintf(stderr, "%s: invalid AIO request type, aborting 0x%x.\n",
                        __func__, type);
        abort();
    }
    io_uring_sqe_set_data(sqes, luringcb);

    QSIMPLEQ_INSERT_TAIL(&s->io_q.submit_queue, luringcb, next);
    s->io_q.in_queue++;
    trace_luring_do_submit(s, s->io_q.blocked, s->io_q.plugged,
                           s->io_q.in_queue, s->io_q.in_flight);
    if (!s->io_q.blocked &&
        (!s->io_q.plugged ||
         s->io_q.in_flight + s->io_q.in_queue >= MAX_ENTRIES)) {
        ret = ioq_submit(s);
        trace_luring_do_submit_done(s, ret);
        return ret;
    }
    return 0;
}

int coroutine_fn luring_co_submit(BlockDriverState *bs, LuringState *s, int fd,
                                  uint64_t offset, QEMUIOVector *qiov, int type)
{
    int ret;
    LuringAIOCB luringcb = {
        .co         = qemu_coroutine_self(),
        .ret        = -EINPROGRESS,
        .qiov       = qiov,
        .is_read    = (type == QEMU_AIO_READ),
    };
    trace_luring_co_submit(bs, s, &luringcb, fd, offset, qiov ? qiov->size : 0,
                           type);
    ret = luring_do_submit(fd, &luringcb, s, offset, type);

    if (ret < 0) {
        return ret;
    }

    if (luringcb.ret == -EINPROGRESS) {
        qemu_coroutine_yield();
    }
    return luringcb.ret;
}

void luring_detach_aio_context(LuringState *s, AioContext *old_context)
{
    aio_set_fd_handler(old_context, s->ring.ring_fd, false,
                       NULL, NULL, NULL, NULL, s);
    qemu_bh_delete(s->completion_bh);
    s->aio_context = NULL;
}

void luring_attach_aio_context(LuringState *s, AioContext *new_context)
{
    s->aio_context = new_context;
    s->completion_bh = aio_bh_new(new_context, qemu_luring_completion_bh, s);
    aio_set_fd_handler(s->aio_context, s->ring.ring_fd, false,
                       qemu_luring_completion_cb, NULL,
                       qemu_luring_poll_cb, qemu_luring_poll_ready, s);
}

LuringState *luring_init(Error **errp)
{
    int rc;
    LuringState *s = g_new0(LuringState, 1);
    struct io_uring *ring = &s->ring;

    trace_luring_init_state(s, sizeof(*s));

    rc = io_uring_queue_init(MAX_ENTRIES, ring, 0);
    if (rc < 0) {
        error_setg_errno(errp, errno, "failed to init linux io_uring ring");
        g_free(s);
        return NULL;
    }

    ioq_init(&s->io_q);
    return s;

}

void luring_cleanup(LuringState *s)
{
    io_uring_queue_exit(&s->ring);
    trace_luring_cleanup_state(s);
    g_free(s);
}
Commit	Line	Data
6663a0a3 AM	1	/*
	2	* Linux io_uring support.
	3	*
	4	* Copyright (C) 2009 IBM, Corp.
	5	* Copyright (C) 2009 Red Hat, Inc.
	6	* Copyright (C) 2019 Aarushi Mehta
	7	*
	8	* This work is licensed under the terms of the GNU GPL, version 2 or later.
	9	* See the COPYING file in the top-level directory.
	10	*/
	11	#include "qemu/osdep.h"
	12	#include <liburing.h>
6663a0a3 AM	13	#include "block/aio.h"
	14	#include "qemu/queue.h"
	15	#include "block/block.h"
	16	#include "block/raw-aio.h"
	17	#include "qemu/coroutine.h"
	18	#include "qapi/error.h"
d803f590	19	#include "trace.h"
6663a0a3 AM	20
	21	/* io_uring ring size */
	22	#define MAX_ENTRIES 128
	23
	24	typedef struct LuringAIOCB {
	25	Coroutine *co;
	26	struct io_uring_sqe sqeq;
	27	ssize_t ret;
	28	QEMUIOVector *qiov;
	29	bool is_read;
	30	QSIMPLEQ_ENTRY(LuringAIOCB) next;
	31
	32	/*
	33	* Buffered reads may require resubmission, see
	34	* luring_resubmit_short_read().
	35	*/
	36	int total_read;
	37	QEMUIOVector resubmit_qiov;
	38	} LuringAIOCB;
	39
	40	typedef struct LuringQueue {
	41	int plugged;
	42	unsigned int in_queue;
	43	unsigned int in_flight;
	44	bool blocked;
	45	QSIMPLEQ_HEAD(, LuringAIOCB) submit_queue;
	46	} LuringQueue;
	47
	48	typedef struct LuringState {
	49	AioContext *aio_context;
	50
	51	struct io_uring ring;
	52
	53	/* io queue for submit at batch. Protected by AioContext lock. */
	54	LuringQueue io_q;
	55
	56	/* I/O completion processing. Only runs in I/O thread. */
	57	QEMUBH *completion_bh;
	58	} LuringState;
	59
	60	/**
	61	* luring_resubmit:
	62	*
	63	* Resubmit a request by appending it to submit_queue. The caller must ensure
	64	* that ioq_submit() is called later so that submit_queue requests are started.
	65	*/
	66	static void luring_resubmit(LuringState s, LuringAIOCB luringcb)
	67	{
	68	QSIMPLEQ_INSERT_TAIL(&s->io_q.submit_queue, luringcb, next);
	69	s->io_q.in_queue++;
	70	}
	71
	72	/**
	73	* luring_resubmit_short_read:
	74	*
be6a166f SH	75	* Short reads are rare but may occur. The remaining read request needs to be
be6a166f SH	76	* resubmitted.
6663a0a3 AM	77	*/
	78	static void luring_resubmit_short_read(LuringState s, LuringAIOCB luringcb,
	79	int nread)
	80	{
	81	QEMUIOVector *resubmit_qiov;
	82	size_t remaining;
	83
d803f590 AM	84	trace_luring_resubmit_short_read(s, luringcb, nread);
d803f590 AM	85
6663a0a3	86	/* Update read position */
c06fc7ce	87	luringcb->total_read += nread;
6663a0a3 AM	88	remaining = luringcb->qiov->size - luringcb->total_read;
	89
	90	/* Shorten qiov */
	91	resubmit_qiov = &luringcb->resubmit_qiov;
	92	if (resubmit_qiov->iov == NULL) {
	93	qemu_iovec_init(resubmit_qiov, luringcb->qiov->niov);
	94	} else {
	95	qemu_iovec_reset(resubmit_qiov);
	96	}
	97	qemu_iovec_concat(resubmit_qiov, luringcb->qiov, luringcb->total_read,
	98	remaining);
	99
	100	/* Update sqe */
c06fc7ce	101	luringcb->sqeq.off += nread;
6663a0a3 AM	102	luringcb->sqeq.addr = (__u64)(uintptr_t)luringcb->resubmit_qiov.iov;
	103	luringcb->sqeq.len = luringcb->resubmit_qiov.niov;
	104
	105	luring_resubmit(s, luringcb);
	106	}
	107
	108	/**
	109	* luring_process_completions:
	110	* @s: AIO state
	111	*
	112	* Fetches completed I/O requests, consumes cqes and invokes their callbacks
	113	* The function is somewhat tricky because it supports nested event loops, for
	114	* example when a request callback invokes aio_poll().
	115	*
	116	* Function schedules BH completion so it can be called again in a nested
	117	* event loop. When there are no events left to complete the BH is being
	118	* canceled.
	119	*
	120	*/
	121	static void luring_process_completions(LuringState *s)
	122	{
	123	struct io_uring_cqe *cqes;
	124	int total_bytes;
	125	/*
	126	* Request completion callbacks can run the nested event loop.
	127	* Schedule ourselves so the nested event loop will "see" remaining
	128	* completed requests and process them. Without this, completion
	129	* callbacks that wait for other requests using a nested event loop
	130	* would hang forever.
	131	*
	132	* This workaround is needed because io_uring uses poll_wait, which
	133	* is woken up when new events are added to the uring, thus polling on
	134	* the same uring fd will block unless more events are received.
	135	*
	136	* Other leaf block drivers (drivers that access the data themselves)
	137	* are networking based, so they poll sockets for data and run the
	138	* correct coroutine.
	139	*/
	140	qemu_bh_schedule(s->completion_bh);
	141
	142	while (io_uring_peek_cqe(&s->ring, &cqes) == 0) {
	143	LuringAIOCB *luringcb;
	144	int ret;
	145
	146	if (!cqes) {
	147	break;
	148	}
	149
	150	luringcb = io_uring_cqe_get_data(cqes);
	151	ret = cqes->res;
	152	io_uring_cqe_seen(&s->ring, cqes);
	153	cqes = NULL;
	154
	155	/* Change counters one-by-one because we can be nested. */
	156	s->io_q.in_flight--;
d803f590	157	trace_luring_process_completion(s, luringcb, ret);
6663a0a3 AM	158
	159	/* total_read is non-zero only for resubmitted read requests */
	160	total_bytes = ret + luringcb->total_read;
	161
	162	if (ret < 0) {
54caccb3 FE	163	/*
	164	* Only writev/readv/fsync requests on regular files or host block
	165	* devices are submitted. Therefore -EAGAIN is not expected but it's
	166	* known to happen sometimes with Linux SCSI. Submit again and hope
	167	* the request completes successfully.
	168	*
	169	* For more information, see:
	170	* https://lore.kernel.org/io-uring/20210727165811.284510-3-axboe@kernel.dk/T/#u
	171	*
	172	* If the code is changed to submit other types of requests in the
	173	* future, then this workaround may need to be extended to deal with
	174	* genuine -EAGAIN results that should not be resubmitted
	175	* immediately.
	176	*/
	177	if (ret == -EINTR \|\| ret == -EAGAIN) {
6663a0a3 AM	178	luring_resubmit(s, luringcb);
	179	continue;
	180	}
	181	} else if (!luringcb->qiov) {
	182	goto end;
	183	} else if (total_bytes == luringcb->qiov->size) {
	184	ret = 0;
	185	/* Only read/write */
	186	} else {
	187	/* Short Read/Write */
	188	if (luringcb->is_read) {
	189	if (ret > 0) {
	190	luring_resubmit_short_read(s, luringcb, ret);
	191	continue;
	192	} else {
	193	/* Pad with zeroes */
	194	qemu_iovec_memset(luringcb->qiov, total_bytes, 0,
	195	luringcb->qiov->size - total_bytes);
	196	ret = 0;
	197	}
	198	} else {
74e4a8a9	199	ret = -ENOSPC;
6663a0a3 AM	200	}
	201	}
	202	end:
	203	luringcb->ret = ret;
	204	qemu_iovec_destroy(&luringcb->resubmit_qiov);
	205
	206	/*
	207	* If the coroutine is already entered it must be in ioq_submit()
	208	* and will notice luringcb->ret has been filled in when it
	209	* eventually runs later. Coroutines cannot be entered recursively
	210	* so avoid doing that!
	211	*/
	212	if (!qemu_coroutine_entered(luringcb->co)) {
	213	aio_co_wake(luringcb->co);
	214	}
	215	}
	216	qemu_bh_cancel(s->completion_bh);
	217	}
	218
	219	static int ioq_submit(LuringState *s)
	220	{
	221	int ret = 0;
	222	LuringAIOCB luringcb, luringcb_next;
	223
	224	while (s->io_q.in_queue > 0) {
	225	/*
	226	* Try to fetch sqes from the ring for requests waiting in
	227	* the overflow queue
	228	*/
	229	QSIMPLEQ_FOREACH_SAFE(luringcb, &s->io_q.submit_queue, next,
	230	luringcb_next) {
	231	struct io_uring_sqe *sqes = io_uring_get_sqe(&s->ring);
	232	if (!sqes) {
	233	break;
	234	}
	235	/* Prep sqe for submission */
	236	*sqes = luringcb->sqeq;
	237	QSIMPLEQ_REMOVE_HEAD(&s->io_q.submit_queue, next);
	238	}
	239	ret = io_uring_submit(&s->ring);
d803f590	240	trace_luring_io_uring_submit(s, ret);
6663a0a3 AM	241	/* Prevent infinite loop if submission is refused */
6663a0a3 AM	242	if (ret <= 0) {
b4e44c99	243	if (ret == -EAGAIN \|\| ret == -EINTR) {
6663a0a3 AM	244	continue;
	245	}
	246	break;
	247	}
	248	s->io_q.in_flight += ret;
	249	s->io_q.in_queue -= ret;
	250	}
	251	s->io_q.blocked = (s->io_q.in_queue > 0);
	252
	253	if (s->io_q.in_flight) {
	254	/*
	255	* We can try to complete something just right away if there are
	256	* still requests in-flight.
	257	*/
	258	luring_process_completions(s);
	259	}
	260	return ret;
	261	}
	262
	263	static void luring_process_completions_and_submit(LuringState *s)
	264	{
	265	aio_context_acquire(s->aio_context);
	266	luring_process_completions(s);
	267
	268	if (!s->io_q.plugged && s->io_q.in_queue > 0) {
	269	ioq_submit(s);
	270	}
	271	aio_context_release(s->aio_context);
	272	}
	273
	274	static void qemu_luring_completion_bh(void *opaque)
	275	{
	276	LuringState *s = opaque;
	277	luring_process_completions_and_submit(s);
	278	}
	279
	280	static void qemu_luring_completion_cb(void *opaque)
	281	{
	282	LuringState *s = opaque;
	283	luring_process_completions_and_submit(s);
	284	}
	285
daffeb02 AM	286	static bool qemu_luring_poll_cb(void *opaque)
	287	{
	288	LuringState *s = opaque;
daffeb02	289
826cc324 SH	290	return io_uring_cq_ready(&s->ring);
	291	}
	292
	293	static void qemu_luring_poll_ready(void *opaque)
	294	{
	295	LuringState *s = opaque;
daffeb02	296
826cc324	297	luring_process_completions_and_submit(s);
daffeb02 AM	298	}
daffeb02 AM	299
6663a0a3 AM	300	static void ioq_init(LuringQueue *io_q)
	301	{
	302	QSIMPLEQ_INIT(&io_q->submit_queue);
	303	io_q->plugged = 0;
	304	io_q->in_queue = 0;
	305	io_q->in_flight = 0;
	306	io_q->blocked = false;
	307	}
	308
	309	void luring_io_plug(BlockDriverState bs, LuringState s)
	310	{
d803f590	311	trace_luring_io_plug(s);
6663a0a3 AM	312	s->io_q.plugged++;
	313	}
	314
	315	void luring_io_unplug(BlockDriverState bs, LuringState s)
	316	{
	317	assert(s->io_q.plugged);
d803f590 AM	318	trace_luring_io_unplug(s, s->io_q.blocked, s->io_q.plugged,
d803f590 AM	319	s->io_q.in_queue, s->io_q.in_flight);
6663a0a3 AM	320	if (--s->io_q.plugged == 0 &&
	321	!s->io_q.blocked && s->io_q.in_queue > 0) {
	322	ioq_submit(s);
	323	}
	324	}
	325
	326	/**
	327	* luring_do_submit:
	328	* @fd: file descriptor for I/O
	329	* @luringcb: AIO control block
	330	* @s: AIO state
	331	* @offset: offset for request
	332	* @type: type of request
	333	*
	334	* Fetches sqes from ring, adds to pending queue and preps them
	335	*
	336	*/
	337	static int luring_do_submit(int fd, LuringAIOCB luringcb, LuringState s,
	338	uint64_t offset, int type)
	339	{
d803f590	340	int ret;
6663a0a3 AM	341	struct io_uring_sqe *sqes = &luringcb->sqeq;
	342
	343	switch (type) {
	344	case QEMU_AIO_WRITE:
	345	io_uring_prep_writev(sqes, fd, luringcb->qiov->iov,
	346	luringcb->qiov->niov, offset);
	347	break;
	348	case QEMU_AIO_READ:
	349	io_uring_prep_readv(sqes, fd, luringcb->qiov->iov,
	350	luringcb->qiov->niov, offset);
	351	break;
	352	case QEMU_AIO_FLUSH:
	353	io_uring_prep_fsync(sqes, fd, IORING_FSYNC_DATASYNC);
	354	break;
	355	default:
	356	fprintf(stderr, "%s: invalid AIO request type, aborting 0x%x.\n",
	357	__func__, type);
	358	abort();
	359	}
	360	io_uring_sqe_set_data(sqes, luringcb);
	361
	362	QSIMPLEQ_INSERT_TAIL(&s->io_q.submit_queue, luringcb, next);
	363	s->io_q.in_queue++;
d803f590 AM	364	trace_luring_do_submit(s, s->io_q.blocked, s->io_q.plugged,
d803f590 AM	365	s->io_q.in_queue, s->io_q.in_flight);
6663a0a3 AM	366	if (!s->io_q.blocked &&
	367	(!s->io_q.plugged \|\|
	368	s->io_q.in_flight + s->io_q.in_queue >= MAX_ENTRIES)) {
d803f590 AM	369	ret = ioq_submit(s);
	370	trace_luring_do_submit_done(s, ret);
	371	return ret;
6663a0a3 AM	372	}
	373	return 0;
	374	}
	375
	376	int coroutine_fn luring_co_submit(BlockDriverState bs, LuringState s, int fd,
	377	uint64_t offset, QEMUIOVector *qiov, int type)
	378	{
	379	int ret;
	380	LuringAIOCB luringcb = {
	381	.co = qemu_coroutine_self(),
	382	.ret = -EINPROGRESS,
	383	.qiov = qiov,
	384	.is_read = (type == QEMU_AIO_READ),
	385	};
d803f590 AM	386	trace_luring_co_submit(bs, s, &luringcb, fd, offset, qiov ? qiov->size : 0,
d803f590 AM	387	type);
6663a0a3	388	ret = luring_do_submit(fd, &luringcb, s, offset, type);
d803f590	389
6663a0a3 AM	390	if (ret < 0) {
	391	return ret;
	392	}
	393
	394	if (luringcb.ret == -EINPROGRESS) {
	395	qemu_coroutine_yield();
	396	}
	397	return luringcb.ret;
	398	}
	399
	400	void luring_detach_aio_context(LuringState s, AioContext old_context)
	401	{
826cc324 SH	402	aio_set_fd_handler(old_context, s->ring.ring_fd, false,
826cc324 SH	403	NULL, NULL, NULL, NULL, s);
6663a0a3 AM	404	qemu_bh_delete(s->completion_bh);
	405	s->aio_context = NULL;
	406	}
	407
	408	void luring_attach_aio_context(LuringState s, AioContext new_context)
	409	{
	410	s->aio_context = new_context;
	411	s->completion_bh = aio_bh_new(new_context, qemu_luring_completion_bh, s);
	412	aio_set_fd_handler(s->aio_context, s->ring.ring_fd, false,
826cc324 SH	413	qemu_luring_completion_cb, NULL,
826cc324 SH	414	qemu_luring_poll_cb, qemu_luring_poll_ready, s);
6663a0a3 AM	415	}
	416
	417	LuringState luring_init(Error *errp)
	418	{
	419	int rc;
	420	LuringState *s = g_new0(LuringState, 1);
	421	struct io_uring *ring = &s->ring;
	422
d803f590 AM	423	trace_luring_init_state(s, sizeof(*s));
d803f590 AM	424
6663a0a3 AM	425	rc = io_uring_queue_init(MAX_ENTRIES, ring, 0);
	426	if (rc < 0) {
	427	error_setg_errno(errp, errno, "failed to init linux io_uring ring");
	428	g_free(s);
	429	return NULL;
	430	}
	431
	432	ioq_init(&s->io_q);
e2848bc5	433	return s;
7845e731	434
6663a0a3 AM	435	}
	436
	437	void luring_cleanup(LuringState *s)
	438	{
	439	io_uring_queue_exit(&s->ring);
d803f590	440	trace_luring_cleanup_state(s);
bd89f936	441	g_free(s);
6663a0a3	442	}