[qemu.git] / block / nbd-client.c

/*
 * QEMU Block driver for  NBD
 *
 * Copyright (C) 2008 Bull S.A.S.
 *     Author: Laurent Vivier <[email protected]>
 *
 * Some parts:
 *    Copyright (C) 2007 Anthony Liguori <[email protected]>
 *
 * Permission is hereby granted, free of charge, to any person obtaining a copy
 * of this software and associated documentation files (the "Software"), to deal
 * in the Software without restriction, including without limitation the rights
 * to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
 * copies of the Software, and to permit persons to whom the Software is
 * furnished to do so, subject to the following conditions:
 *
 * The above copyright notice and this permission notice shall be included in
 * all copies or substantial portions of the Software.
 *
 * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
 * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
 * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
 * THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
 * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
 * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN
 * THE SOFTWARE.
 */

#include "nbd-client.h"
#include "qemu/sockets.h"

#define HANDLE_TO_INDEX(bs, handle) ((handle) ^ ((uint64_t)(intptr_t)bs))
#define INDEX_TO_HANDLE(bs, index)  ((index)  ^ ((uint64_t)(intptr_t)bs))

static void nbd_recv_coroutines_enter_all(NbdClientSession *s)
{
    int i;

    for (i = 0; i < MAX_NBD_REQUESTS; i++) {
        if (s->recv_coroutine[i]) {
            qemu_coroutine_enter(s->recv_coroutine[i], NULL);
        }
    }
}

static void nbd_teardown_connection(BlockDriverState *bs)
{
    NbdClientSession *client = nbd_get_client_session(bs);

    /* finish any pending coroutines */
    shutdown(client->sock, 2);
    nbd_recv_coroutines_enter_all(client);

    nbd_client_detach_aio_context(bs);
    closesocket(client->sock);
    client->sock = -1;
}

static void nbd_reply_ready(void *opaque)
{
    BlockDriverState *bs = opaque;
    NbdClientSession *s = nbd_get_client_session(bs);
    uint64_t i;
    int ret;

    if (s->reply.handle == 0) {
        /* No reply already in flight.  Fetch a header.  It is possible
         * that another thread has done the same thing in parallel, so
         * the socket is not readable anymore.
         */
        ret = nbd_receive_reply(s->sock, &s->reply);
        if (ret == -EAGAIN) {
            return;
        }
        if (ret < 0) {
            s->reply.handle = 0;
            goto fail;
        }
    }

    /* There's no need for a mutex on the receive side, because the
     * handler acts as a synchronization point and ensures that only
     * one coroutine is called until the reply finishes.  */
    i = HANDLE_TO_INDEX(s, s->reply.handle);
    if (i >= MAX_NBD_REQUESTS) {
        goto fail;
    }

    if (s->recv_coroutine[i]) {
        qemu_coroutine_enter(s->recv_coroutine[i], NULL);
        return;
    }

fail:
    nbd_teardown_connection(bs);
}

static void nbd_restart_write(void *opaque)
{
    BlockDriverState *bs = opaque;

    qemu_coroutine_enter(nbd_get_client_session(bs)->send_coroutine, NULL);
}

static int nbd_co_send_request(BlockDriverState *bs,
                               struct nbd_request *request,
                               QEMUIOVector *qiov, int offset)
{
    NbdClientSession *s = nbd_get_client_session(bs);
    AioContext *aio_context;
    int rc, ret;

    qemu_co_mutex_lock(&s->send_mutex);
    s->send_coroutine = qemu_coroutine_self();
    aio_context = bdrv_get_aio_context(bs);
    aio_set_fd_handler(aio_context, s->sock,
                       nbd_reply_ready, nbd_restart_write, bs);
    if (qiov) {
        if (!s->is_unix) {
            socket_set_cork(s->sock, 1);
        }
        rc = nbd_send_request(s->sock, request);
        if (rc >= 0) {
            ret = qemu_co_sendv(s->sock, qiov->iov, qiov->niov,
                                offset, request->len);
            if (ret != request->len) {
                rc = -EIO;
            }
        }
        if (!s->is_unix) {
            socket_set_cork(s->sock, 0);
        }
    } else {
        rc = nbd_send_request(s->sock, request);
    }
    aio_set_fd_handler(aio_context, s->sock, nbd_reply_ready, NULL, bs);
    s->send_coroutine = NULL;
    qemu_co_mutex_unlock(&s->send_mutex);
    return rc;
}

static void nbd_co_receive_reply(NbdClientSession *s,
    struct nbd_request *request, struct nbd_reply *reply,
    QEMUIOVector *qiov, int offset)
{
    int ret;

    /* Wait until we're woken up by the read handler.  TODO: perhaps
     * peek at the next reply and avoid yielding if it's ours?  */
    qemu_coroutine_yield();
    *reply = s->reply;
    if (reply->handle != request->handle) {
        reply->error = EIO;
    } else {
        if (qiov && reply->error == 0) {
            ret = qemu_co_recvv(s->sock, qiov->iov, qiov->niov,
                                offset, request->len);
            if (ret != request->len) {
                reply->error = EIO;
            }
        }

        /* Tell the read handler to read another header.  */
        s->reply.handle = 0;
    }
}

static void nbd_coroutine_start(NbdClientSession *s,
   struct nbd_request *request)
{
    int i;

    /* Poor man semaphore.  The free_sema is locked when no other request
     * can be accepted, and unlocked after receiving one reply.  */
    if (s->in_flight >= MAX_NBD_REQUESTS - 1) {
        qemu_co_mutex_lock(&s->free_sema);
        assert(s->in_flight < MAX_NBD_REQUESTS);
    }
    s->in_flight++;

    for (i = 0; i < MAX_NBD_REQUESTS; i++) {
        if (s->recv_coroutine[i] == NULL) {
            s->recv_coroutine[i] = qemu_coroutine_self();
            break;
        }
    }

    assert(i < MAX_NBD_REQUESTS);
    request->handle = INDEX_TO_HANDLE(s, i);
}

static void nbd_coroutine_end(NbdClientSession *s,
    struct nbd_request *request)
{
    int i = HANDLE_TO_INDEX(s, request->handle);
    s->recv_coroutine[i] = NULL;
    if (s->in_flight-- == MAX_NBD_REQUESTS) {
        qemu_co_mutex_unlock(&s->free_sema);
    }
}

static int nbd_co_readv_1(BlockDriverState *bs, int64_t sector_num,
                          int nb_sectors, QEMUIOVector *qiov,
                          int offset)
{
    NbdClientSession *client = nbd_get_client_session(bs);
    struct nbd_request request = { .type = NBD_CMD_READ };
    struct nbd_reply reply;
    ssize_t ret;

    request.from = sector_num * 512;
    request.len = nb_sectors * 512;

    nbd_coroutine_start(client, &request);
    ret = nbd_co_send_request(bs, &request, NULL, 0);
    if (ret < 0) {
        reply.error = -ret;
    } else {
        nbd_co_receive_reply(client, &request, &reply, qiov, offset);
    }
    nbd_coroutine_end(client, &request);
    return -reply.error;

}

static int nbd_co_writev_1(BlockDriverState *bs, int64_t sector_num,
                           int nb_sectors, QEMUIOVector *qiov,
                           int offset)
{
    NbdClientSession *client = nbd_get_client_session(bs);
    struct nbd_request request = { .type = NBD_CMD_WRITE };
    struct nbd_reply reply;
    ssize_t ret;

    if (!bdrv_enable_write_cache(bs) &&
        (client->nbdflags & NBD_FLAG_SEND_FUA)) {
        request.type |= NBD_CMD_FLAG_FUA;
    }

    request.from = sector_num * 512;
    request.len = nb_sectors * 512;

    nbd_coroutine_start(client, &request);
    ret = nbd_co_send_request(bs, &request, qiov, offset);
    if (ret < 0) {
        reply.error = -ret;
    } else {
        nbd_co_receive_reply(client, &request, &reply, NULL, 0);
    }
    nbd_coroutine_end(client, &request);
    return -reply.error;
}

/* qemu-nbd has a limit of slightly less than 1M per request.  Try to
 * remain aligned to 4K. */
#define NBD_MAX_SECTORS 2040

int nbd_client_co_readv(BlockDriverState *bs, int64_t sector_num,
                        int nb_sectors, QEMUIOVector *qiov)
{
    int offset = 0;
    int ret;
    while (nb_sectors > NBD_MAX_SECTORS) {
        ret = nbd_co_readv_1(bs, sector_num, NBD_MAX_SECTORS, qiov, offset);
        if (ret < 0) {
            return ret;
        }
        offset += NBD_MAX_SECTORS * 512;
        sector_num += NBD_MAX_SECTORS;
        nb_sectors -= NBD_MAX_SECTORS;
    }
    return nbd_co_readv_1(bs, sector_num, nb_sectors, qiov, offset);
}

int nbd_client_co_writev(BlockDriverState *bs, int64_t sector_num,
                         int nb_sectors, QEMUIOVector *qiov)
{
    int offset = 0;
    int ret;
    while (nb_sectors > NBD_MAX_SECTORS) {
        ret = nbd_co_writev_1(bs, sector_num, NBD_MAX_SECTORS, qiov, offset);
        if (ret < 0) {
            return ret;
        }
        offset += NBD_MAX_SECTORS * 512;
        sector_num += NBD_MAX_SECTORS;
        nb_sectors -= NBD_MAX_SECTORS;
    }
    return nbd_co_writev_1(bs, sector_num, nb_sectors, qiov, offset);
}

int nbd_client_co_flush(BlockDriverState *bs)
{
    NbdClientSession *client = nbd_get_client_session(bs);
    struct nbd_request request = { .type = NBD_CMD_FLUSH };
    struct nbd_reply reply;
    ssize_t ret;

    if (!(client->nbdflags & NBD_FLAG_SEND_FLUSH)) {
        return 0;
    }

    if (client->nbdflags & NBD_FLAG_SEND_FUA) {
        request.type |= NBD_CMD_FLAG_FUA;
    }

    request.from = 0;
    request.len = 0;

    nbd_coroutine_start(client, &request);
    ret = nbd_co_send_request(bs, &request, NULL, 0);
    if (ret < 0) {
        reply.error = -ret;
    } else {
        nbd_co_receive_reply(client, &request, &reply, NULL, 0);
    }
    nbd_coroutine_end(client, &request);
    return -reply.error;
}

int nbd_client_co_discard(BlockDriverState *bs, int64_t sector_num,
                          int nb_sectors)
{
    NbdClientSession *client = nbd_get_client_session(bs);
    struct nbd_request request = { .type = NBD_CMD_TRIM };
    struct nbd_reply reply;
    ssize_t ret;

    if (!(client->nbdflags & NBD_FLAG_SEND_TRIM)) {
        return 0;
    }
    request.from = sector_num * 512;
    request.len = nb_sectors * 512;

    nbd_coroutine_start(client, &request);
    ret = nbd_co_send_request(bs, &request, NULL, 0);
    if (ret < 0) {
        reply.error = -ret;
    } else {
        nbd_co_receive_reply(client, &request, &reply, NULL, 0);
    }
    nbd_coroutine_end(client, &request);
    return -reply.error;

}

void nbd_client_detach_aio_context(BlockDriverState *bs)
{
    aio_set_fd_handler(bdrv_get_aio_context(bs),
                       nbd_get_client_session(bs)->sock, NULL, NULL, NULL);
}

void nbd_client_attach_aio_context(BlockDriverState *bs,
                                   AioContext *new_context)
{
    aio_set_fd_handler(new_context, nbd_get_client_session(bs)->sock,
                       nbd_reply_ready, NULL, bs);
}

void nbd_client_close(BlockDriverState *bs)
{
    NbdClientSession *client = nbd_get_client_session(bs);
    struct nbd_request request = {
        .type = NBD_CMD_DISC,
        .from = 0,
        .len = 0
    };

    if (client->sock == -1) {
        return;
    }

    nbd_send_request(client->sock, &request);

    nbd_teardown_connection(bs);
}

int nbd_client_init(BlockDriverState *bs, int sock, const char *export,
                    Error **errp)
{
    NbdClientSession *client = nbd_get_client_session(bs);
    int ret;

    /* NBD handshake */
    logout("session init %s\n", export);
    qemu_set_block(sock);
    ret = nbd_receive_negotiate(sock, export,
                                &client->nbdflags, &client->size,
                                &client->blocksize, errp);
    if (ret < 0) {
        logout("Failed to negotiate with the NBD server\n");
        closesocket(sock);
        return ret;
    }

    qemu_co_mutex_init(&client->send_mutex);
    qemu_co_mutex_init(&client->free_sema);
    client->sock = sock;

    /* Now that we're connected, set the socket to be non-blocking and
     * kick the reply mechanism.  */
    qemu_set_nonblock(sock);
    nbd_client_attach_aio_context(bs, bdrv_get_aio_context(bs));

    logout("Established connection with NBD server\n");
    return 0;
}
Commit	Line	Data
2302c1ca MAL	1	/*
	2	* QEMU Block driver for NBD
	3	*
	4	* Copyright (C) 2008 Bull S.A.S.
	5	* Author: Laurent Vivier <[email protected]>
	6	*
	7	* Some parts:
	8	* Copyright (C) 2007 Anthony Liguori <[email protected]>
	9	*
	10	* Permission is hereby granted, free of charge, to any person obtaining a copy
	11	* of this software and associated documentation files (the "Software"), to deal
	12	* in the Software without restriction, including without limitation the rights
	13	* to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
	14	* copies of the Software, and to permit persons to whom the Software is
	15	* furnished to do so, subject to the following conditions:
	16	*
	17	* The above copyright notice and this permission notice shall be included in
	18	* all copies or substantial portions of the Software.
	19	*
	20	* THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
	21	* IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
	22	* FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
	23	* THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
	24	* LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
	25	* OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN
	26	* THE SOFTWARE.
	27	*/
	28
	29	#include "nbd-client.h"
	30	#include "qemu/sockets.h"
	31
	32	#define HANDLE_TO_INDEX(bs, handle) ((handle) ^ ((uint64_t)(intptr_t)bs))
	33	#define INDEX_TO_HANDLE(bs, index) ((index) ^ ((uint64_t)(intptr_t)bs))
	34
69152c09 MAL	35	static void nbd_recv_coroutines_enter_all(NbdClientSession *s)
	36	{
	37	int i;
	38
	39	for (i = 0; i < MAX_NBD_REQUESTS; i++) {
	40	if (s->recv_coroutine[i]) {
	41	qemu_coroutine_enter(s->recv_coroutine[i], NULL);
	42	}
	43	}
	44	}
	45
f53a829b	46	static void nbd_teardown_connection(BlockDriverState *bs)
4a41a2d6	47	{
f53a829b HR	48	NbdClientSession *client = nbd_get_client_session(bs);
f53a829b HR	49
4a41a2d6 SH	50	/* finish any pending coroutines */
	51	shutdown(client->sock, 2);
	52	nbd_recv_coroutines_enter_all(client);
	53
f53a829b	54	nbd_client_detach_aio_context(bs);
4a41a2d6 SH	55	closesocket(client->sock);
	56	client->sock = -1;
	57	}
	58
2302c1ca MAL	59	static void nbd_reply_ready(void *opaque)
2302c1ca MAL	60	{
f53a829b HR	61	BlockDriverState *bs = opaque;
f53a829b HR	62	NbdClientSession *s = nbd_get_client_session(bs);
2302c1ca MAL	63	uint64_t i;
	64	int ret;
	65
	66	if (s->reply.handle == 0) {
	67	/* No reply already in flight. Fetch a header. It is possible
	68	* that another thread has done the same thing in parallel, so
	69	* the socket is not readable anymore.
	70	*/
	71	ret = nbd_receive_reply(s->sock, &s->reply);
	72	if (ret == -EAGAIN) {
	73	return;
	74	}
	75	if (ret < 0) {
	76	s->reply.handle = 0;
	77	goto fail;
	78	}
	79	}
	80
	81	/* There's no need for a mutex on the receive side, because the
	82	* handler acts as a synchronization point and ensures that only
	83	* one coroutine is called until the reply finishes. */
	84	i = HANDLE_TO_INDEX(s, s->reply.handle);
	85	if (i >= MAX_NBD_REQUESTS) {
	86	goto fail;
	87	}
	88
	89	if (s->recv_coroutine[i]) {
	90	qemu_coroutine_enter(s->recv_coroutine[i], NULL);
	91	return;
	92	}
	93
	94	fail:
f53a829b	95	nbd_teardown_connection(bs);
2302c1ca MAL	96	}
	97
	98	static void nbd_restart_write(void *opaque)
	99	{
f53a829b	100	BlockDriverState *bs = opaque;
2302c1ca	101
f53a829b	102	qemu_coroutine_enter(nbd_get_client_session(bs)->send_coroutine, NULL);
2302c1ca MAL	103	}
2302c1ca MAL	104
f53a829b HR	105	static int nbd_co_send_request(BlockDriverState *bs,
	106	struct nbd_request *request,
	107	QEMUIOVector *qiov, int offset)
2302c1ca	108	{
f53a829b	109	NbdClientSession *s = nbd_get_client_session(bs);
69447cd8	110	AioContext *aio_context;
2302c1ca MAL	111	int rc, ret;
	112
	113	qemu_co_mutex_lock(&s->send_mutex);
	114	s->send_coroutine = qemu_coroutine_self();
f53a829b	115	aio_context = bdrv_get_aio_context(bs);
69447cd8	116	aio_set_fd_handler(aio_context, s->sock,
f53a829b	117	nbd_reply_ready, nbd_restart_write, bs);
2302c1ca MAL	118	if (qiov) {
	119	if (!s->is_unix) {
	120	socket_set_cork(s->sock, 1);
	121	}
	122	rc = nbd_send_request(s->sock, request);
	123	if (rc >= 0) {
	124	ret = qemu_co_sendv(s->sock, qiov->iov, qiov->niov,
	125	offset, request->len);
	126	if (ret != request->len) {
	127	rc = -EIO;
	128	}
	129	}
	130	if (!s->is_unix) {
	131	socket_set_cork(s->sock, 0);
	132	}
	133	} else {
	134	rc = nbd_send_request(s->sock, request);
	135	}
f53a829b	136	aio_set_fd_handler(aio_context, s->sock, nbd_reply_ready, NULL, bs);
2302c1ca MAL	137	s->send_coroutine = NULL;
	138	qemu_co_mutex_unlock(&s->send_mutex);
	139	return rc;
	140	}
	141
	142	static void nbd_co_receive_reply(NbdClientSession *s,
	143	struct nbd_request request, struct nbd_reply reply,
	144	QEMUIOVector *qiov, int offset)
	145	{
	146	int ret;
	147
	148	/* Wait until we're woken up by the read handler. TODO: perhaps
	149	* peek at the next reply and avoid yielding if it's ours? */
	150	qemu_coroutine_yield();
	151	*reply = s->reply;
	152	if (reply->handle != request->handle) {
	153	reply->error = EIO;
	154	} else {
	155	if (qiov && reply->error == 0) {
	156	ret = qemu_co_recvv(s->sock, qiov->iov, qiov->niov,
	157	offset, request->len);
	158	if (ret != request->len) {
	159	reply->error = EIO;
	160	}
	161	}
	162
	163	/* Tell the read handler to read another header. */
	164	s->reply.handle = 0;
	165	}
	166	}
	167
	168	static void nbd_coroutine_start(NbdClientSession *s,
	169	struct nbd_request *request)
	170	{
	171	int i;
	172
	173	/* Poor man semaphore. The free_sema is locked when no other request
	174	* can be accepted, and unlocked after receiving one reply. */
	175	if (s->in_flight >= MAX_NBD_REQUESTS - 1) {
	176	qemu_co_mutex_lock(&s->free_sema);
	177	assert(s->in_flight < MAX_NBD_REQUESTS);
	178	}
	179	s->in_flight++;
	180
	181	for (i = 0; i < MAX_NBD_REQUESTS; i++) {
	182	if (s->recv_coroutine[i] == NULL) {
	183	s->recv_coroutine[i] = qemu_coroutine_self();
	184	break;
	185	}
	186	}
	187
	188	assert(i < MAX_NBD_REQUESTS);
	189	request->handle = INDEX_TO_HANDLE(s, i);
	190	}
	191
	192	static void nbd_coroutine_end(NbdClientSession *s,
	193	struct nbd_request *request)
	194	{
	195	int i = HANDLE_TO_INDEX(s, request->handle);
	196	s->recv_coroutine[i] = NULL;
	197	if (s->in_flight-- == MAX_NBD_REQUESTS) {
	198	qemu_co_mutex_unlock(&s->free_sema);
	199	}
	200	}
201
f53a829b	202	static int nbd_co_readv_1(BlockDriverState *bs, int64_t sector_num,
2302c1ca MAL	203	int nb_sectors, QEMUIOVector *qiov,
	204	int offset)
	205	{
f53a829b	206	NbdClientSession *client = nbd_get_client_session(bs);
b1b27b64	207	struct nbd_request request = { .type = NBD_CMD_READ };
2302c1ca MAL	208	struct nbd_reply reply;
	209	ssize_t ret;
	210
2302c1ca MAL	211	request.from = sector_num * 512;
	212	request.len = nb_sectors * 512;
	213
	214	nbd_coroutine_start(client, &request);
f53a829b	215	ret = nbd_co_send_request(bs, &request, NULL, 0);
2302c1ca MAL	216	if (ret < 0) {
	217	reply.error = -ret;
	218	} else {
	219	nbd_co_receive_reply(client, &request, &reply, qiov, offset);
	220	}
	221	nbd_coroutine_end(client, &request);
	222	return -reply.error;
	223
	224	}
	225
f53a829b	226	static int nbd_co_writev_1(BlockDriverState *bs, int64_t sector_num,
2302c1ca MAL	227	int nb_sectors, QEMUIOVector *qiov,
	228	int offset)
	229	{
f53a829b	230	NbdClientSession *client = nbd_get_client_session(bs);
b1b27b64	231	struct nbd_request request = { .type = NBD_CMD_WRITE };
2302c1ca MAL	232	struct nbd_reply reply;
	233	ssize_t ret;
	234
f53a829b	235	if (!bdrv_enable_write_cache(bs) &&
2302c1ca MAL	236	(client->nbdflags & NBD_FLAG_SEND_FUA)) {
	237	request.type \|= NBD_CMD_FLAG_FUA;
	238	}
	239
	240	request.from = sector_num * 512;
	241	request.len = nb_sectors * 512;
	242
	243	nbd_coroutine_start(client, &request);
f53a829b	244	ret = nbd_co_send_request(bs, &request, qiov, offset);
2302c1ca MAL	245	if (ret < 0) {
	246	reply.error = -ret;
	247	} else {
	248	nbd_co_receive_reply(client, &request, &reply, NULL, 0);
	249	}
	250	nbd_coroutine_end(client, &request);
	251	return -reply.error;
	252	}
	253
	254	/* qemu-nbd has a limit of slightly less than 1M per request. Try to
	255	* remain aligned to 4K. */
	256	#define NBD_MAX_SECTORS 2040
	257
f53a829b HR	258	int nbd_client_co_readv(BlockDriverState *bs, int64_t sector_num,
f53a829b HR	259	int nb_sectors, QEMUIOVector *qiov)
2302c1ca MAL	260	{
	261	int offset = 0;
	262	int ret;
	263	while (nb_sectors > NBD_MAX_SECTORS) {
f53a829b	264	ret = nbd_co_readv_1(bs, sector_num, NBD_MAX_SECTORS, qiov, offset);
2302c1ca MAL	265	if (ret < 0) {
	266	return ret;
	267	}
	268	offset += NBD_MAX_SECTORS * 512;
	269	sector_num += NBD_MAX_SECTORS;
	270	nb_sectors -= NBD_MAX_SECTORS;
	271	}
f53a829b	272	return nbd_co_readv_1(bs, sector_num, nb_sectors, qiov, offset);
2302c1ca MAL	273	}
2302c1ca MAL	274
f53a829b HR	275	int nbd_client_co_writev(BlockDriverState *bs, int64_t sector_num,
f53a829b HR	276	int nb_sectors, QEMUIOVector *qiov)
2302c1ca MAL	277	{
	278	int offset = 0;
	279	int ret;
	280	while (nb_sectors > NBD_MAX_SECTORS) {
f53a829b	281	ret = nbd_co_writev_1(bs, sector_num, NBD_MAX_SECTORS, qiov, offset);
2302c1ca MAL	282	if (ret < 0) {
	283	return ret;
	284	}
	285	offset += NBD_MAX_SECTORS * 512;
	286	sector_num += NBD_MAX_SECTORS;
	287	nb_sectors -= NBD_MAX_SECTORS;
	288	}
f53a829b	289	return nbd_co_writev_1(bs, sector_num, nb_sectors, qiov, offset);
2302c1ca MAL	290	}
2302c1ca MAL	291
f53a829b	292	int nbd_client_co_flush(BlockDriverState *bs)
2302c1ca	293	{
f53a829b	294	NbdClientSession *client = nbd_get_client_session(bs);
b1b27b64	295	struct nbd_request request = { .type = NBD_CMD_FLUSH };
2302c1ca MAL	296	struct nbd_reply reply;
	297	ssize_t ret;
	298
	299	if (!(client->nbdflags & NBD_FLAG_SEND_FLUSH)) {
	300	return 0;
	301	}
	302
2302c1ca MAL	303	if (client->nbdflags & NBD_FLAG_SEND_FUA) {
	304	request.type \|= NBD_CMD_FLAG_FUA;
	305	}
	306
	307	request.from = 0;
	308	request.len = 0;
	309
	310	nbd_coroutine_start(client, &request);
f53a829b	311	ret = nbd_co_send_request(bs, &request, NULL, 0);
2302c1ca MAL	312	if (ret < 0) {
	313	reply.error = -ret;
	314	} else {
	315	nbd_co_receive_reply(client, &request, &reply, NULL, 0);
	316	}
	317	nbd_coroutine_end(client, &request);
	318	return -reply.error;
	319	}
	320
f53a829b HR	321	int nbd_client_co_discard(BlockDriverState *bs, int64_t sector_num,
f53a829b HR	322	int nb_sectors)
2302c1ca	323	{
f53a829b	324	NbdClientSession *client = nbd_get_client_session(bs);
b1b27b64	325	struct nbd_request request = { .type = NBD_CMD_TRIM };
2302c1ca MAL	326	struct nbd_reply reply;
	327	ssize_t ret;
	328
	329	if (!(client->nbdflags & NBD_FLAG_SEND_TRIM)) {
	330	return 0;
	331	}
2302c1ca MAL	332	request.from = sector_num * 512;
	333	request.len = nb_sectors * 512;
	334
	335	nbd_coroutine_start(client, &request);
f53a829b	336	ret = nbd_co_send_request(bs, &request, NULL, 0);
2302c1ca MAL	337	if (ret < 0) {
	338	reply.error = -ret;
	339	} else {
	340	nbd_co_receive_reply(client, &request, &reply, NULL, 0);
	341	}
	342	nbd_coroutine_end(client, &request);
	343	return -reply.error;
	344
	345	}
	346
f53a829b	347	void nbd_client_detach_aio_context(BlockDriverState *bs)
69447cd8	348	{
f53a829b HR	349	aio_set_fd_handler(bdrv_get_aio_context(bs),
f53a829b HR	350	nbd_get_client_session(bs)->sock, NULL, NULL, NULL);
69447cd8 SH	351	}
69447cd8 SH	352
f53a829b HR	353	void nbd_client_attach_aio_context(BlockDriverState *bs,
f53a829b HR	354	AioContext *new_context)
69447cd8	355	{
f53a829b HR	356	aio_set_fd_handler(new_context, nbd_get_client_session(bs)->sock,
f53a829b HR	357	nbd_reply_ready, NULL, bs);
69447cd8 SH	358	}
69447cd8 SH	359
f53a829b	360	void nbd_client_close(BlockDriverState *bs)
2302c1ca	361	{
f53a829b	362	NbdClientSession *client = nbd_get_client_session(bs);
b1b27b64 MAL	363	struct nbd_request request = {
	364	.type = NBD_CMD_DISC,
	365	.from = 0,
	366	.len = 0
	367	};
2302c1ca	368
4a41a2d6 SH	369	if (client->sock == -1) {
	370	return;
	371	}
	372
	373	nbd_send_request(client->sock, &request);
5ad283eb	374
f53a829b	375	nbd_teardown_connection(bs);
2302c1ca MAL	376	}
2302c1ca MAL	377
f53a829b HR	378	int nbd_client_init(BlockDriverState bs, int sock, const char export,
f53a829b HR	379	Error **errp)
2302c1ca	380	{
f53a829b	381	NbdClientSession *client = nbd_get_client_session(bs);
2302c1ca MAL	382	int ret;
	383
	384	/* NBD handshake */
e2bc625f	385	logout("session init %s\n", export);
e53a18e4	386	qemu_set_block(sock);
e2bc625f	387	ret = nbd_receive_negotiate(sock, export,
2302c1ca	388	&client->nbdflags, &client->size,
1ce52846	389	&client->blocksize, errp);
2302c1ca MAL	390	if (ret < 0) {
	391	logout("Failed to negotiate with the NBD server\n");
	392	closesocket(sock);
	393	return ret;
	394	}
	395
	396	qemu_co_mutex_init(&client->send_mutex);
	397	qemu_co_mutex_init(&client->free_sema);
2302c1ca MAL	398	client->sock = sock;
	399
	400	/* Now that we're connected, set the socket to be non-blocking and
	401	* kick the reply mechanism. */
	402	qemu_set_nonblock(sock);
f53a829b	403	nbd_client_attach_aio_context(bs, bdrv_get_aio_context(bs));
2302c1ca MAL	404
	405	logout("Established connection with NBD server\n");
	406	return 0;
	407	}