[qemu.git] / posix-aio-compat.c

/*
 * QEMU posix-aio emulation
 *
 * Copyright IBM, Corp. 2008
 *
 * Authors:
 *  Anthony Liguori   <aliguori@us.ibm.com>
 *
 * This work is licensed under the terms of the GNU GPL, version 2.  See
 * the COPYING file in the top-level directory.
 *
 */

#include <sys/ioctl.h>
#include <pthread.h>
#include <unistd.h>
#include <errno.h>
#include <time.h>
#include <string.h>
#include <stdlib.h>
#include <stdio.h>
#include "osdep.h"
#include "qemu-common.h"

#include "posix-aio-compat.h"

static pthread_mutex_t lock = PTHREAD_MUTEX_INITIALIZER;
static pthread_cond_t cond = PTHREAD_COND_INITIALIZER;
static pthread_t thread_id;
static pthread_attr_t attr;
static int max_threads = 64;
static int cur_threads = 0;
static int idle_threads = 0;
static TAILQ_HEAD(, qemu_paiocb) request_list;

#ifdef HAVE_PREADV
static int preadv_present = 1;
#else
static int preadv_present = 0;
#endif

static void die2(int err, const char *what)
{
    fprintf(stderr, "%s failed: %s\n", what, strerror(err));
    abort();
}

static void die(const char *what)
{
    die2(errno, what);
}

static void mutex_lock(pthread_mutex_t *mutex)
{
    int ret = pthread_mutex_lock(mutex);
    if (ret) die2(ret, "pthread_mutex_lock");
}

static void mutex_unlock(pthread_mutex_t *mutex)
{
    int ret = pthread_mutex_unlock(mutex);
    if (ret) die2(ret, "pthread_mutex_unlock");
}

static int cond_timedwait(pthread_cond_t *cond, pthread_mutex_t *mutex,
                           struct timespec *ts)
{
    int ret = pthread_cond_timedwait(cond, mutex, ts);
    if (ret && ret != ETIMEDOUT) die2(ret, "pthread_cond_timedwait");
    return ret;
}

static void cond_signal(pthread_cond_t *cond)
{
    int ret = pthread_cond_signal(cond);
    if (ret) die2(ret, "pthread_cond_signal");
}

static void thread_create(pthread_t *thread, pthread_attr_t *attr,
                          void *(*start_routine)(void*), void *arg)
{
    int ret = pthread_create(thread, attr, start_routine, arg);
    if (ret) die2(ret, "pthread_create");
}

static size_t handle_aiocb_ioctl(struct qemu_paiocb *aiocb)
{
	int ret;

	ret = ioctl(aiocb->aio_fildes, aiocb->aio_ioctl_cmd, aiocb->aio_ioctl_buf);
	if (ret == -1)
		return -errno;
	return ret;
}

#ifdef HAVE_PREADV

static ssize_t
qemu_preadv(int fd, const struct iovec *iov, int nr_iov, off_t offset)
{
    return preadv(fd, iov, nr_iov, offset);
}

static ssize_t
qemu_pwritev(int fd, const struct iovec *iov, int nr_iov, off_t offset)
{
    return pwritev(fd, iov, nr_iov, offset);
}

#else

static ssize_t
qemu_preadv(int fd, const struct iovec *iov, int nr_iov, off_t offset)
{
    return -ENOSYS;
}

static ssize_t
qemu_pwritev(int fd, const struct iovec *iov, int nr_iov, off_t offset)
{
    return -ENOSYS;
}

#endif

/*
 * Check if we need to copy the data in the aiocb into a new
 * properly aligned buffer.
 */
static int aiocb_needs_copy(struct qemu_paiocb *aiocb)
{
    if (aiocb->aio_flags & QEMU_AIO_SECTOR_ALIGNED) {
        int i;

        for (i = 0; i < aiocb->aio_niov; i++)
            if ((uintptr_t) aiocb->aio_iov[i].iov_base % 512)
                return 1;
    }

    return 0;
}

static size_t handle_aiocb_rw_vector(struct qemu_paiocb *aiocb)
{
    size_t offset = 0;
    ssize_t len;

    do {
        if (aiocb->aio_type == QEMU_PAIO_WRITE)
            len = qemu_pwritev(aiocb->aio_fildes,
                               aiocb->aio_iov,
                               aiocb->aio_niov,
                               aiocb->aio_offset + offset);
         else
            len = qemu_preadv(aiocb->aio_fildes,
                              aiocb->aio_iov,
                              aiocb->aio_niov,
                              aiocb->aio_offset + offset);
    } while (len == -1 && errno == EINTR);

    if (len == -1)
        return -errno;
    return len;
}

static size_t handle_aiocb_rw_linear(struct qemu_paiocb *aiocb, char *buf)
{
    size_t offset = 0;
    size_t len;

    while (offset < aiocb->aio_nbytes) {
         if (aiocb->aio_type == QEMU_PAIO_WRITE)
             len = pwrite(aiocb->aio_fildes,
                          (const char *)buf + offset,
                          aiocb->aio_nbytes - offset,
                          aiocb->aio_offset + offset);
         else
             len = pread(aiocb->aio_fildes,
                         buf + offset,
                         aiocb->aio_nbytes - offset,
                         aiocb->aio_offset + offset);

         if (len == -1 && errno == EINTR)
             continue;
         else if (len == -1) {
             offset = -errno;
             break;
         } else if (len == 0)
             break;

         offset += len;
    }

    return offset;
}

static size_t handle_aiocb_rw(struct qemu_paiocb *aiocb)
{
    size_t nbytes;
    char *buf;

    if (!aiocb_needs_copy(aiocb)) {
        /*
         * If there is just a single buffer, and it is properly aligned
         * we can just use plain pread/pwrite without any problems.
         */
        if (aiocb->aio_niov == 1)
             return handle_aiocb_rw_linear(aiocb, aiocb->aio_iov->iov_base);

        /*
         * We have more than one iovec, and all are properly aligned.
         *
         * Try preadv/pwritev first and fall back to linearizing the
         * buffer if it's not supported.
         */
	if (preadv_present) {
            nbytes = handle_aiocb_rw_vector(aiocb);
            if (nbytes == aiocb->aio_nbytes)
	        return nbytes;
            if (nbytes < 0 && nbytes != -ENOSYS)
                return nbytes;
            preadv_present = 0;
        }

        /*
         * XXX(hch): short read/write.  no easy way to handle the reminder
         * using these interfaces.  For now retry using plain
         * pread/pwrite?
         */
    }

    /*
     * Ok, we have to do it the hard way, copy all segments into
     * a single aligned buffer.
     */
    buf = qemu_memalign(512, aiocb->aio_nbytes);
    if (aiocb->aio_type == QEMU_PAIO_WRITE) {
        char *p = buf;
        int i;

        for (i = 0; i < aiocb->aio_niov; ++i) {
            memcpy(p, aiocb->aio_iov[i].iov_base, aiocb->aio_iov[i].iov_len);
            p += aiocb->aio_iov[i].iov_len;
        }
    }

    nbytes = handle_aiocb_rw_linear(aiocb, buf);
    if (aiocb->aio_type != QEMU_PAIO_WRITE) {
        char *p = buf;
        size_t count = aiocb->aio_nbytes, copy;
        int i;

        for (i = 0; i < aiocb->aio_niov && count; ++i) {
            copy = count;
            if (copy > aiocb->aio_iov[i].iov_len)
                copy = aiocb->aio_iov[i].iov_len;
            memcpy(aiocb->aio_iov[i].iov_base, p, copy);
            p     += copy;
            count -= copy;
        }
    }
    qemu_vfree(buf);

    return nbytes;
}

static void *aio_thread(void *unused)
{
    pid_t pid;
    sigset_t set;

    pid = getpid();

    /* block all signals */
    if (sigfillset(&set)) die("sigfillset");
    if (sigprocmask(SIG_BLOCK, &set, NULL)) die("sigprocmask");

    while (1) {
        struct qemu_paiocb *aiocb;
        size_t ret = 0;
        qemu_timeval tv;
        struct timespec ts;

        qemu_gettimeofday(&tv);
        ts.tv_sec = tv.tv_sec + 10;
        ts.tv_nsec = 0;

        mutex_lock(&lock);

        while (TAILQ_EMPTY(&request_list) &&
               !(ret == ETIMEDOUT)) {
            ret = cond_timedwait(&cond, &lock, &ts);
        }

        if (TAILQ_EMPTY(&request_list))
            break;

        aiocb = TAILQ_FIRST(&request_list);
        TAILQ_REMOVE(&request_list, aiocb, node);
        aiocb->active = 1;
        idle_threads--;
        mutex_unlock(&lock);

        switch (aiocb->aio_type) {
        case QEMU_PAIO_READ:
        case QEMU_PAIO_WRITE:
		ret = handle_aiocb_rw(aiocb);
		break;
        case QEMU_PAIO_IOCTL:
		ret = handle_aiocb_ioctl(aiocb);
		break;
	default:
		fprintf(stderr, "invalid aio request (0x%x)\n", aiocb->aio_type);
		ret = -EINVAL;
		break;
	}

        mutex_lock(&lock);
        aiocb->ret = ret;
        idle_threads++;
        mutex_unlock(&lock);

        if (kill(pid, aiocb->ev_signo)) die("kill failed");
    }

    idle_threads--;
    cur_threads--;
    mutex_unlock(&lock);

    return NULL;
}

static void spawn_thread(void)
{
    cur_threads++;
    idle_threads++;
    thread_create(&thread_id, &attr, aio_thread, NULL);
}

int qemu_paio_init(struct qemu_paioinit *aioinit)
{
    int ret;

    ret = pthread_attr_init(&attr);
    if (ret) die2(ret, "pthread_attr_init");

    ret = pthread_attr_setdetachstate(&attr, PTHREAD_CREATE_DETACHED);
    if (ret) die2(ret, "pthread_attr_setdetachstate");

    TAILQ_INIT(&request_list);

    return 0;
}

static int qemu_paio_submit(struct qemu_paiocb *aiocb, int type)
{
    aiocb->aio_type = type;
    aiocb->ret = -EINPROGRESS;
    aiocb->active = 0;
    mutex_lock(&lock);
    if (idle_threads == 0 && cur_threads < max_threads)
        spawn_thread();
    TAILQ_INSERT_TAIL(&request_list, aiocb, node);
    mutex_unlock(&lock);
    cond_signal(&cond);

    return 0;
}

int qemu_paio_read(struct qemu_paiocb *aiocb)
{
    return qemu_paio_submit(aiocb, QEMU_PAIO_READ);
}

int qemu_paio_write(struct qemu_paiocb *aiocb)
{
    return qemu_paio_submit(aiocb, QEMU_PAIO_WRITE);
}

int qemu_paio_ioctl(struct qemu_paiocb *aiocb)
{
    return qemu_paio_submit(aiocb, QEMU_PAIO_IOCTL);
}

ssize_t qemu_paio_return(struct qemu_paiocb *aiocb)
{
    ssize_t ret;

    mutex_lock(&lock);
    ret = aiocb->ret;
    mutex_unlock(&lock);

    return ret;
}

int qemu_paio_error(struct qemu_paiocb *aiocb)
{
    ssize_t ret = qemu_paio_return(aiocb);

    if (ret < 0)
        ret = -ret;
    else
        ret = 0;

    return ret;
}

int qemu_paio_cancel(int fd, struct qemu_paiocb *aiocb)
{
    int ret;

    mutex_lock(&lock);
    if (!aiocb->active) {
        TAILQ_REMOVE(&request_list, aiocb, node);
        aiocb->ret = -ECANCELED;
        ret = QEMU_PAIO_CANCELED;
    } else if (aiocb->ret == -EINPROGRESS)
        ret = QEMU_PAIO_NOTCANCELED;
    else
        ret = QEMU_PAIO_ALLDONE;
    mutex_unlock(&lock);

    return ret;
}
Commit	Line	Data
3c529d93 AL	1	/*
	2	* QEMU posix-aio emulation
	3	*
	4	* Copyright IBM, Corp. 2008
	5	*
	6	* Authors:
	7	* Anthony Liguori <aliguori@us.ibm.com>
	8	*
	9	* This work is licensed under the terms of the GNU GPL, version 2. See
	10	* the COPYING file in the top-level directory.
	11	*
	12	*/
	13
221f715d	14	#include <sys/ioctl.h>
3c529d93 AL	15	#include <pthread.h>
	16	#include <unistd.h>
	17	#include <errno.h>
30525aff	18	#include <time.h>
8653c015	19	#include <string.h>
	20	#include <stdlib.h>
	21	#include <stdio.h>
3c529d93	22	#include "osdep.h"
f141eafe	23	#include "qemu-common.h"
3c529d93 AL	24
	25	#include "posix-aio-compat.h"
	26
	27	static pthread_mutex_t lock = PTHREAD_MUTEX_INITIALIZER;
	28	static pthread_cond_t cond = PTHREAD_COND_INITIALIZER;
	29	static pthread_t thread_id;
a8227a5a	30	static pthread_attr_t attr;
3c529d93 AL	31	static int max_threads = 64;
	32	static int cur_threads = 0;
	33	static int idle_threads = 0;
	34	static TAILQ_HEAD(, qemu_paiocb) request_list;
	35
ceb42de8 AL	36	#ifdef HAVE_PREADV
	37	static int preadv_present = 1;
	38	#else
	39	static int preadv_present = 0;
	40	#endif
	41
8653c015	42	static void die2(int err, const char *what)
	43	{
	44	fprintf(stderr, "%s failed: %s\n", what, strerror(err));
	45	abort();
	46	}
	47
	48	static void die(const char *what)
	49	{
	50	die2(errno, what);
	51	}
	52
	53	static void mutex_lock(pthread_mutex_t *mutex)
	54	{
	55	int ret = pthread_mutex_lock(mutex);
	56	if (ret) die2(ret, "pthread_mutex_lock");
	57	}
	58
	59	static void mutex_unlock(pthread_mutex_t *mutex)
	60	{
	61	int ret = pthread_mutex_unlock(mutex);
	62	if (ret) die2(ret, "pthread_mutex_unlock");
	63	}
	64
	65	static int cond_timedwait(pthread_cond_t cond, pthread_mutex_t mutex,
	66	struct timespec *ts)
	67	{
	68	int ret = pthread_cond_timedwait(cond, mutex, ts);
	69	if (ret && ret != ETIMEDOUT) die2(ret, "pthread_cond_timedwait");
	70	return ret;
	71	}
	72
5d47e372	73	static void cond_signal(pthread_cond_t *cond)
8653c015	74	{
5d47e372	75	int ret = pthread_cond_signal(cond);
5d47e372	76	if (ret) die2(ret, "pthread_cond_signal");
8653c015	77	}
	78
	79	static void thread_create(pthread_t thread, pthread_attr_t attr,
	80	void (start_routine)(void), void arg)
	81	{
	82	int ret = pthread_create(thread, attr, start_routine, arg);
	83	if (ret) die2(ret, "pthread_create");
	84	}
	85
f141eafe AL	86	static size_t handle_aiocb_ioctl(struct qemu_paiocb *aiocb)
	87	{
	88	int ret;
	89
	90	ret = ioctl(aiocb->aio_fildes, aiocb->aio_ioctl_cmd, aiocb->aio_ioctl_buf);
	91	if (ret == -1)
	92	return -errno;
	93	return ret;
	94	}
	95
ceb42de8 AL	96	#ifdef HAVE_PREADV
	97
	98	static ssize_t
	99	qemu_preadv(int fd, const struct iovec *iov, int nr_iov, off_t offset)
	100	{
	101	return preadv(fd, iov, nr_iov, offset);
	102	}
	103
	104	static ssize_t
	105	qemu_pwritev(int fd, const struct iovec *iov, int nr_iov, off_t offset)
	106	{
	107	return pwritev(fd, iov, nr_iov, offset);
	108	}
	109
	110	#else
	111
	112	static ssize_t
	113	qemu_preadv(int fd, const struct iovec *iov, int nr_iov, off_t offset)
	114	{
	115	return -ENOSYS;
	116	}
	117
	118	static ssize_t
	119	qemu_pwritev(int fd, const struct iovec *iov, int nr_iov, off_t offset)
	120	{
	121	return -ENOSYS;
	122	}
	123
	124	#endif
	125
f141eafe AL	126	/*
	127	* Check if we need to copy the data in the aiocb into a new
	128	* properly aligned buffer.
	129	*/
	130	static int aiocb_needs_copy(struct qemu_paiocb *aiocb)
	131	{
	132	if (aiocb->aio_flags & QEMU_AIO_SECTOR_ALIGNED) {
	133	int i;
	134
	135	for (i = 0; i < aiocb->aio_niov; i++)
	136	if ((uintptr_t) aiocb->aio_iov[i].iov_base % 512)
	137	return 1;
	138	}
	139
	140	return 0;
	141	}
	142
ceb42de8 AL	143	static size_t handle_aiocb_rw_vector(struct qemu_paiocb *aiocb)
	144	{
	145	size_t offset = 0;
	146	ssize_t len;
	147
	148	do {
	149	if (aiocb->aio_type == QEMU_PAIO_WRITE)
	150	len = qemu_pwritev(aiocb->aio_fildes,
	151	aiocb->aio_iov,
	152	aiocb->aio_niov,
	153	aiocb->aio_offset + offset);
	154	else
	155	len = qemu_preadv(aiocb->aio_fildes,
	156	aiocb->aio_iov,
	157	aiocb->aio_niov,
	158	aiocb->aio_offset + offset);
	159	} while (len == -1 && errno == EINTR);
	160
	161	if (len == -1)
	162	return -errno;
	163	return len;
	164	}
	165
f141eafe	166	static size_t handle_aiocb_rw_linear(struct qemu_paiocb aiocb, char buf)
221f715d AL	167	{
221f715d AL	168	size_t offset = 0;
f141eafe	169	size_t len;
221f715d AL	170
221f715d AL	171	while (offset < aiocb->aio_nbytes) {
f141eafe AL	172	if (aiocb->aio_type == QEMU_PAIO_WRITE)
	173	len = pwrite(aiocb->aio_fildes,
	174	(const char *)buf + offset,
	175	aiocb->aio_nbytes - offset,
	176	aiocb->aio_offset + offset);
	177	else
	178	len = pread(aiocb->aio_fildes,
	179	buf + offset,
221f715d AL	180	aiocb->aio_nbytes - offset,
221f715d AL	181	aiocb->aio_offset + offset);
221f715d	182
f141eafe AL	183	if (len == -1 && errno == EINTR)
	184	continue;
	185	else if (len == -1) {
	186	offset = -errno;
	187	break;
	188	} else if (len == 0)
	189	break;
	190
	191	offset += len;
221f715d AL	192	}
	193
	194	return offset;
	195	}
	196
f141eafe	197	static size_t handle_aiocb_rw(struct qemu_paiocb *aiocb)
221f715d	198	{
f141eafe AL	199	size_t nbytes;
	200	char *buf;
	201
ceb42de8	202	if (!aiocb_needs_copy(aiocb)) {
f141eafe AL	203	/*
	204	* If there is just a single buffer, and it is properly aligned
	205	* we can just use plain pread/pwrite without any problems.
	206	*/
ceb42de8 AL	207	if (aiocb->aio_niov == 1)
	208	return handle_aiocb_rw_linear(aiocb, aiocb->aio_iov->iov_base);
	209
	210	/*
	211	* We have more than one iovec, and all are properly aligned.
	212	*
	213	* Try preadv/pwritev first and fall back to linearizing the
	214	* buffer if it's not supported.
	215	*/
	216	if (preadv_present) {
	217	nbytes = handle_aiocb_rw_vector(aiocb);
	218	if (nbytes == aiocb->aio_nbytes)
	219	return nbytes;
	220	if (nbytes < 0 && nbytes != -ENOSYS)
	221	return nbytes;
	222	preadv_present = 0;
	223	}
	224
	225	/*
	226	* XXX(hch): short read/write. no easy way to handle the reminder
	227	* using these interfaces. For now retry using plain
	228	* pread/pwrite?
	229	*/
f141eafe	230	}
221f715d	231
f141eafe AL	232	/*
	233	* Ok, we have to do it the hard way, copy all segments into
	234	* a single aligned buffer.
	235	*/
	236	buf = qemu_memalign(512, aiocb->aio_nbytes);
	237	if (aiocb->aio_type == QEMU_PAIO_WRITE) {
	238	char *p = buf;
	239	int i;
	240
	241	for (i = 0; i < aiocb->aio_niov; ++i) {
	242	memcpy(p, aiocb->aio_iov[i].iov_base, aiocb->aio_iov[i].iov_len);
	243	p += aiocb->aio_iov[i].iov_len;
	244	}
	245	}
	246
	247	nbytes = handle_aiocb_rw_linear(aiocb, buf);
	248	if (aiocb->aio_type != QEMU_PAIO_WRITE) {
	249	char *p = buf;
	250	size_t count = aiocb->aio_nbytes, copy;
	251	int i;
	252
	253	for (i = 0; i < aiocb->aio_niov && count; ++i) {
	254	copy = count;
	255	if (copy > aiocb->aio_iov[i].iov_len)
	256	copy = aiocb->aio_iov[i].iov_len;
	257	memcpy(aiocb->aio_iov[i].iov_base, p, copy);
	258	p += copy;
	259	count -= copy;
	260	}
	261	}
	262	qemu_vfree(buf);
	263
	264	return nbytes;
221f715d AL	265	}
221f715d AL	266
3c529d93 AL	267	static void aio_thread(void unused)
3c529d93 AL	268	{
a8227a5a	269	pid_t pid;
3c529d93 AL	270	sigset_t set;
3c529d93 AL	271
a8227a5a	272	pid = getpid();
a8227a5a	273
3c529d93	274	/* block all signals */
8653c015	275	if (sigfillset(&set)) die("sigfillset");
8653c015	276	if (sigprocmask(SIG_BLOCK, &set, NULL)) die("sigprocmask");
3c529d93 AL	277
	278	while (1) {
	279	struct qemu_paiocb *aiocb;
221f715d	280	size_t ret = 0;
30525aff	281	qemu_timeval tv;
	282	struct timespec ts;
	283
	284	qemu_gettimeofday(&tv);
	285	ts.tv_sec = tv.tv_sec + 10;
	286	ts.tv_nsec = 0;
3c529d93	287
8653c015	288	mutex_lock(&lock);
3c529d93 AL	289
	290	while (TAILQ_EMPTY(&request_list) &&
	291	!(ret == ETIMEDOUT)) {
8653c015	292	ret = cond_timedwait(&cond, &lock, &ts);
3c529d93 AL	293	}
3c529d93 AL	294
514f7a27	295	if (TAILQ_EMPTY(&request_list))
3c529d93 AL	296	break;
	297
	298	aiocb = TAILQ_FIRST(&request_list);
	299	TAILQ_REMOVE(&request_list, aiocb, node);
3c529d93	300	aiocb->active = 1;
3c529d93	301	idle_threads--;
8653c015	302	mutex_unlock(&lock);
3c529d93	303
221f715d AL	304	switch (aiocb->aio_type) {
	305	case QEMU_PAIO_READ:
	306	case QEMU_PAIO_WRITE:
f141eafe	307	ret = handle_aiocb_rw(aiocb);
221f715d AL	308	break;
	309	case QEMU_PAIO_IOCTL:
	310	ret = handle_aiocb_ioctl(aiocb);
	311	break;
	312	default:
	313	fprintf(stderr, "invalid aio request (0x%x)\n", aiocb->aio_type);
	314	ret = -EINVAL;
	315	break;
	316	}
3c529d93	317
8653c015	318	mutex_lock(&lock);
221f715d	319	aiocb->ret = ret;
3c529d93	320	idle_threads++;
8653c015	321	mutex_unlock(&lock);
3c529d93	322
a8227a5a	323	if (kill(pid, aiocb->ev_signo)) die("kill failed");
3c529d93 AL	324	}
	325
	326	idle_threads--;
	327	cur_threads--;
8653c015	328	mutex_unlock(&lock);
3c529d93 AL	329
	330	return NULL;
	331	}
	332
8653c015	333	static void spawn_thread(void)
3c529d93	334	{
3c529d93 AL	335	cur_threads++;
3c529d93 AL	336	idle_threads++;
8653c015	337	thread_create(&thread_id, &attr, aio_thread, NULL);
3c529d93 AL	338	}
	339
	340	int qemu_paio_init(struct qemu_paioinit *aioinit)
	341	{
a8227a5a	342	int ret;
	343
	344	ret = pthread_attr_init(&attr);
	345	if (ret) die2(ret, "pthread_attr_init");
	346
	347	ret = pthread_attr_setdetachstate(&attr, PTHREAD_CREATE_DETACHED);
	348	if (ret) die2(ret, "pthread_attr_setdetachstate");
	349
3c529d93 AL	350	TAILQ_INIT(&request_list);
	351
	352	return 0;
	353	}
	354
221f715d	355	static int qemu_paio_submit(struct qemu_paiocb *aiocb, int type)
3c529d93	356	{
221f715d	357	aiocb->aio_type = type;
3c529d93 AL	358	aiocb->ret = -EINPROGRESS;
3c529d93 AL	359	aiocb->active = 0;
8653c015	360	mutex_lock(&lock);
3c529d93 AL	361	if (idle_threads == 0 && cur_threads < max_threads)
	362	spawn_thread();
	363	TAILQ_INSERT_TAIL(&request_list, aiocb, node);
8653c015	364	mutex_unlock(&lock);
5d47e372	365	cond_signal(&cond);
3c529d93 AL	366
	367	return 0;
	368	}
	369
	370	int qemu_paio_read(struct qemu_paiocb *aiocb)
	371	{
221f715d	372	return qemu_paio_submit(aiocb, QEMU_PAIO_READ);
3c529d93 AL	373	}
	374
	375	int qemu_paio_write(struct qemu_paiocb *aiocb)
	376	{
221f715d AL	377	return qemu_paio_submit(aiocb, QEMU_PAIO_WRITE);
	378	}
	379
	380	int qemu_paio_ioctl(struct qemu_paiocb *aiocb)
	381	{
	382	return qemu_paio_submit(aiocb, QEMU_PAIO_IOCTL);
3c529d93 AL	383	}
	384
	385	ssize_t qemu_paio_return(struct qemu_paiocb *aiocb)
	386	{
	387	ssize_t ret;
	388
8653c015	389	mutex_lock(&lock);
3c529d93	390	ret = aiocb->ret;
8653c015	391	mutex_unlock(&lock);
3c529d93 AL	392
	393	return ret;
	394	}
	395
	396	int qemu_paio_error(struct qemu_paiocb *aiocb)
	397	{
	398	ssize_t ret = qemu_paio_return(aiocb);
	399
	400	if (ret < 0)
	401	ret = -ret;
	402	else
	403	ret = 0;
	404
	405	return ret;
	406	}
	407
	408	int qemu_paio_cancel(int fd, struct qemu_paiocb *aiocb)
	409	{
	410	int ret;
	411
8653c015	412	mutex_lock(&lock);
3c529d93 AL	413	if (!aiocb->active) {
	414	TAILQ_REMOVE(&request_list, aiocb, node);
	415	aiocb->ret = -ECANCELED;
	416	ret = QEMU_PAIO_CANCELED;
	417	} else if (aiocb->ret == -EINPROGRESS)
	418	ret = QEMU_PAIO_NOTCANCELED;
	419	else
	420	ret = QEMU_PAIO_ALLDONE;
8653c015	421	mutex_unlock(&lock);
3c529d93 AL	422
	423	return ret;
	424	}