block-raw.c

qemu性能直逼VMware的仿真器QEMU 的模擬速度約為實機的 25％；約為 Bochs 的 60 倍。Plex86、User-Mode-Linux、VMware 和 Virtual PC 則比

/*
 * Block driver for RAW files
 * 
 * Copyright (c) 2006 Fabrice Bellard
 * 
 * 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 "vl.h"
#include "block_int.h"
#include <assert.h>
#ifndef _WIN32
#include <aio.h>

#ifndef QEMU_TOOL
#include "exec-all.h"
#endif

#ifdef CONFIG_COCOA
#include <paths.h>
#include <sys/param.h>
#include <IOKit/IOKitLib.h>
#include <IOKit/IOBSD.h>
#include <IOKit/storage/IOMediaBSDClient.h>
#include <IOKit/storage/IOMedia.h>
#include <IOKit/storage/IOCDMedia.h>
//#include <IOKit/storage/IOCDTypes.h>
#include <CoreFoundation/CoreFoundation.h>
#endif

#ifdef __sun__
#define _POSIX_PTHREAD_SEMANTICS 1
#include <signal.h>
#include <sys/dkio.h>
#endif
#ifdef __linux__
#include <sys/ioctl.h>
#include <linux/cdrom.h>
#include <linux/fd.h>
#endif
#ifdef __FreeBSD__
#include <sys/disk.h>
#endif

//#define DEBUG_FLOPPY

#define FTYPE_FILE   0
#define FTYPE_CD     1
#define FTYPE_FD     2

/* if the FD is not accessed during that time (in ms), we try to
   reopen it to see if the disk has been changed */
#define FD_OPEN_TIMEOUT 1000

typedef struct BDRVRawState {
    int fd;
    int type;
#if defined(__linux__)
    /* linux floppy specific */
    int fd_open_flags;
    int64_t fd_open_time;
    int64_t fd_error_time;
    int fd_got_error;
    int fd_media_changed;
#endif
} BDRVRawState;

static int fd_open(BlockDriverState *bs);

static int raw_open(BlockDriverState *bs, const char *filename, int flags)
{
    BDRVRawState *s = bs->opaque;
    int fd, open_flags, ret;

    open_flags = O_BINARY;
    if ((flags & BDRV_O_ACCESS) == O_RDWR) {
        open_flags |= O_RDWR;
    } else {
        open_flags |= O_RDONLY;
        bs->read_only = 1;
    }
    if (flags & BDRV_O_CREAT)
        open_flags |= O_CREAT | O_TRUNC;

    s->type = FTYPE_FILE;

    fd = open(filename, open_flags, 0644);
    if (fd < 0) {
        ret = -errno;
        if (ret == -EROFS)
            ret = -EACCES;
        return ret;
    }
    s->fd = fd;
    return 0;
}

/* XXX: use host sector size if necessary with:
#ifdef DIOCGSECTORSIZE
        {
            unsigned int sectorsize = 512;
            if (!ioctl(fd, DIOCGSECTORSIZE, &sectorsize) &&
                sectorsize > bufsize)
                bufsize = sectorsize;
        }
#endif
#ifdef CONFIG_COCOA
        u_int32_t   blockSize = 512;
        if ( !ioctl( fd, DKIOCGETBLOCKSIZE, &blockSize ) && blockSize > bufsize) {
            bufsize = blockSize;
        }
#endif
*/

static int raw_pread(BlockDriverState *bs, int64_t offset, 
                     uint8_t *buf, int count)
{
    BDRVRawState *s = bs->opaque;
    int ret;
    
    ret = fd_open(bs);
    if (ret < 0)
        return ret;

    lseek(s->fd, offset, SEEK_SET);
    ret = read(s->fd, buf, count);
    return ret;
}

static int raw_pwrite(BlockDriverState *bs, int64_t offset, 
                      const uint8_t *buf, int count)
{
    BDRVRawState *s = bs->opaque;
    int ret;
    
    ret = fd_open(bs);
    if (ret < 0)
        return ret;

    lseek(s->fd, offset, SEEK_SET);
    ret = write(s->fd, buf, count);
    return ret;
}

/***********************************************************/
/* Unix AIO using POSIX AIO */

typedef struct RawAIOCB {
    BlockDriverAIOCB common;
    struct aiocb aiocb;
    struct RawAIOCB *next;
} RawAIOCB;

static int aio_sig_num = SIGUSR2;
static RawAIOCB *first_aio; /* AIO issued */
static int aio_initialized = 0;

static void aio_signal_handler(int signum)
{
#ifndef QEMU_TOOL
    CPUState *env = cpu_single_env;
    if (env) {
        /* stop the currently executing cpu because a timer occured */
        cpu_interrupt(env, CPU_INTERRUPT_EXIT);
#ifdef USE_KQEMU
        if (env->kqemu_enabled) {
            kqemu_cpu_interrupt(env);
        }
#endif
    }
#endif
}

void qemu_aio_init(void)
{
    struct sigaction act;

    aio_initialized = 1;
    
    sigfillset(&act.sa_mask);
    act.sa_flags = 0; /* do not restart syscalls to interrupt select() */
    act.sa_handler = aio_signal_handler;
    sigaction(aio_sig_num, &act, NULL);

#if defined(__GLIBC__) && defined(__linux__)
    {
        /* XXX: aio thread exit seems to hang on RedHat 9 and this init
           seems to fix the problem. */
        struct aioinit ai;
        memset(&ai, 0, sizeof(ai));
        ai.aio_threads = 1;
        ai.aio_num = 1;
        ai.aio_idle_time = 365 * 100000;
        aio_init(&ai);
    }
#endif
}

void qemu_aio_poll(void)
{
    RawAIOCB *acb, **pacb;
    int ret;

    for(;;) {
        pacb = &first_aio;
        for(;;) {
            acb = *pacb;
            if (!acb)
                goto the_end;
            ret = aio_error(&acb->aiocb);
            if (ret == ECANCELED) {
                /* remove the request */
                *pacb = acb->next;
                qemu_aio_release(acb);
            } else if (ret != EINPROGRESS) {
                /* end of aio */
                if (ret == 0) {
                    ret = aio_return(&acb->aiocb);
                    if (ret == acb->aiocb.aio_nbytes)
                        ret = 0;
                    else
                        ret = -EINVAL;
                } else {
                    ret = -ret;
                }
                /* remove the request */
                *pacb = acb->next;
                /* call the callback */
                acb->common.cb(acb->common.opaque, ret);
                qemu_aio_release(acb);
                break;
            } else {
                pacb = &acb->next;
            }
        }
    }
 the_end: ;
}

/* Wait for all IO requests to complete.  */
void qemu_aio_flush(void)
{
    qemu_aio_wait_start();
    qemu_aio_poll();
    while (first_aio) {
        qemu_aio_wait();
    }
    qemu_aio_wait_end();
}

/* wait until at least one AIO was handled */
static sigset_t wait_oset;

void qemu_aio_wait_start(void)
{
    sigset_t set;

    if (!aio_initialized)
        qemu_aio_init();
    sigemptyset(&set);
    sigaddset(&set, aio_sig_num);
    sigprocmask(SIG_BLOCK, &set, &wait_oset);
}

void qemu_aio_wait(void)
{
    sigset_t set;
    int nb_sigs;

#ifndef QEMU_TOOL
    if (qemu_bh_poll())
        return;
#endif
    sigemptyset(&set);
    sigaddset(&set, aio_sig_num);
    sigwait(&set, &nb_sigs);
    qemu_aio_poll();
}

void qemu_aio_wait_end(void)
{
    sigprocmask(SIG_SETMASK, &wait_oset, NULL);
}

static RawAIOCB *raw_aio_setup(BlockDriverState *bs,
        int64_t sector_num, uint8_t *buf, int nb_sectors,
        BlockDriverCompletionFunc *cb, void *opaque)
{
    BDRVRawState *s = bs->opaque;
    RawAIOCB *acb;

    if (fd_open(bs) < 0)
        return NULL;

    acb = qemu_aio_get(bs, cb, opaque);
    if (!acb)
        return NULL;
    acb->aiocb.aio_fildes = s->fd;
    acb->aiocb.aio_sigevent.sigev_signo = aio_sig_num;
    acb->aiocb.aio_sigevent.sigev_notify = SIGEV_SIGNAL;
    acb->aiocb.aio_buf = buf;
    acb->aiocb.aio_nbytes = nb_sectors * 512;
    acb->aiocb.aio_offset = sector_num * 512;
    acb->next = first_aio;
    first_aio = acb;
    return acb;
}

static BlockDriverAIOCB *raw_aio_read(BlockDriverState *bs,
        int64_t sector_num, uint8_t *buf, int nb_sectors,
        BlockDriverCompletionFunc *cb, void *opaque)
{
    RawAIOCB *acb;

    acb = raw_aio_setup(bs, sector_num, buf, nb_sectors, cb, opaque);
    if (!acb)
        return NULL;
    if (aio_read(&acb->aiocb) < 0) {
        qemu_aio_release(acb);
        return NULL;
    } 
    return &acb->common;
}

static BlockDriverAIOCB *raw_aio_write(BlockDriverState *bs,
        int64_t sector_num, const uint8_t *buf, int nb_sectors,
        BlockDriverCompletionFunc *cb, void *opaque)
{
    RawAIOCB *acb;

    acb = raw_aio_setup(bs, sector_num, (uint8_t*)buf, nb_sectors, cb, opaque);
    if (!acb)
        return NULL;
    if (aio_write(&acb->aiocb) < 0) {
        qemu_aio_release(acb);
        return NULL;
    } 
    return &acb->common;
}

static void raw_aio_cancel(BlockDriverAIOCB *blockacb)
{
    int ret;
    RawAIOCB *acb = (RawAIOCB *)blockacb;
    RawAIOCB **pacb;

    ret = aio_cancel(acb->aiocb.aio_fildes, &acb->aiocb);
    if (ret == AIO_NOTCANCELED) {
        /* fail safe: if the aio could not be canceled, we wait for
           it */
        while (aio_error(&acb->aiocb) == EINPROGRESS);
    }

    /* remove the callback from the queue */
    pacb = &first_aio;
    for(;;) {
        if (*pacb == NULL) {
            break;
        } else if (*pacb == acb) {
            *pacb = acb->next;
            qemu_aio_release(acb);
            break;
        }
        pacb = &acb->next;
    }
}

static void raw_close(BlockDriverState *bs)
{
    BDRVRawState *s = bs->opaque;
    if (s->fd >= 0) {
        close(s->fd);
        s->fd = -1;
    }
}

static int raw_truncate(BlockDriverState *bs, int64_t offset)
{
    BDRVRawState *s = bs->opaque;
    if (s->type != FTYPE_FILE)
        return -ENOTSUP;
    if (ftruncate(s->fd, offset) < 0)
        return -errno;
    return 0;
}

static int64_t  raw_getlength(BlockDriverState *bs)
{
    BDRVRawState *s = bs->opaque;
    int fd = s->fd;
    int64_t size;
#ifdef _BSD
    struct stat sb;
#endif
#ifdef __sun__
    struct dk_minfo minfo;
    int rv;
#endif
    int ret;

    ret = fd_open(bs);
    if (ret < 0)
        return ret;

#ifdef _BSD
    if (!fstat(fd, &sb) && (S_IFCHR & sb.st_mode)) {
#ifdef DIOCGMEDIASIZE
	if (ioctl(fd, DIOCGMEDIASIZE, (off_t *)&size))
#endif
#ifdef CONFIG_COCOA
        size = LONG_LONG_MAX;
#else
        size = lseek(fd, 0LL, SEEK_END);
#endif
    } else
#endif
#ifdef __sun__
    /*
     * use the DKIOCGMEDIAINFO ioctl to read the size.
     */
    rv = ioctl ( fd, DKIOCGMEDIAINFO, &minfo );
    if ( rv != -1 ) {
        size = minfo.dki_lbsize * minfo.dki_capacity;
    } else /* there are reports that lseek on some devices
              fails, but irc discussion said that contingency
              on contingency was overkill */
#endif
    {
        size = lseek(fd, 0, SEEK_END);
    }
    return size;
}

static int raw_create(const char *filename, int64_t total_size,
                      const char *backing_file, int flags)
{