[mirror_qemu.git] / hw / virtio-blk.c

/*
 * Virtio Block Device
 *
 * Copyright IBM, Corp. 2007
 *
 * 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 <qemu-common.h>
#include <sysemu.h>
#include "virtio-blk.h"
#include "block_int.h"
#ifdef __linux__
# include <scsi/sg.h>
#endif

typedef struct VirtIOBlock
{
    VirtIODevice vdev;
    BlockDriverState *bs;
    VirtQueue *vq;
    void *rq;
    char serial_str[BLOCK_SERIAL_STRLEN + 1];
    QEMUBH *bh;
} VirtIOBlock;

static VirtIOBlock *to_virtio_blk(VirtIODevice *vdev)
{
    return (VirtIOBlock *)vdev;
}

/* store identify data in little endian format
 */
static inline void put_le16(uint16_t *p, unsigned int v)
{
    *p = cpu_to_le16(v);
}

/* copy to *dst from *src, nul pad dst tail as needed to len bytes
 */
static inline void padstr(char *dst, const char *src, int len)
{
    while (len--)
        *dst++ = *src ? *src++ : '\0';
}

/* setup simulated identify data as appropriate for virtio block device
 *
 * ref: AT Attachment 8 - ATA/ATAPI Command Set (ATA8-ACS)
 */
static inline void virtio_identify_template(struct virtio_blk_config *bc)
{
    uint16_t *p = &bc->identify[0];
    uint64_t lba_sectors = bc->capacity;

    memset(p, 0, sizeof(bc->identify));
    put_le16(p + 0, 0x0);                            /* ATA device */
    padstr((char *)(p + 23), QEMU_VERSION, 8);       /* firmware revision */
    padstr((char *)(p + 27), "QEMU VIRT_BLK", 40);   /* model# */
    put_le16(p + 47, 0x80ff);                        /* max xfer 255 sectors */
    put_le16(p + 49, 0x0b00);                        /* support IORDY/LBA/DMA */
    put_le16(p + 59, 0x1ff);                         /* cur xfer 255 sectors */
    put_le16(p + 80, 0x1f0);                         /* support ATA8/7/6/5/4 */
    put_le16(p + 81, 0x16);
    put_le16(p + 82, 0x400);
    put_le16(p + 83, 0x400);
    put_le16(p + 100, lba_sectors);
    put_le16(p + 101, lba_sectors >> 16);
    put_le16(p + 102, lba_sectors >> 32);
    put_le16(p + 103, lba_sectors >> 48);
}

typedef struct VirtIOBlockReq
{
    VirtIOBlock *dev;
    VirtQueueElement elem;
    struct virtio_blk_inhdr *in;
    struct virtio_blk_outhdr *out;
    struct virtio_scsi_inhdr *scsi;
    QEMUIOVector qiov;
    struct VirtIOBlockReq *next;
} VirtIOBlockReq;

static void virtio_blk_req_complete(VirtIOBlockReq *req, int status)
{
    VirtIOBlock *s = req->dev;

    req->in->status = status;
    virtqueue_push(s->vq, &req->elem, req->qiov.size + sizeof(*req->in));
    virtio_notify(&s->vdev, s->vq);

    qemu_free(req);
}

static int virtio_blk_handle_write_error(VirtIOBlockReq *req, int error)
{
    BlockInterfaceErrorAction action = drive_get_onerror(req->dev->bs);
    VirtIOBlock *s = req->dev;

    if (action == BLOCK_ERR_IGNORE)
        return 0;

    if ((error == ENOSPC && action == BLOCK_ERR_STOP_ENOSPC)
            || action == BLOCK_ERR_STOP_ANY) {
        req->next = s->rq;
        s->rq = req;
        vm_stop(0);
    } else {
        virtio_blk_req_complete(req, VIRTIO_BLK_S_IOERR);
    }

    return 1;
}

static void virtio_blk_rw_complete(void *opaque, int ret)
{
    VirtIOBlockReq *req = opaque;

    if (ret && (req->out->type & VIRTIO_BLK_T_OUT)) {
        if (virtio_blk_handle_write_error(req, -ret))
            return;
    }

    virtio_blk_req_complete(req, VIRTIO_BLK_S_OK);
}

static VirtIOBlockReq *virtio_blk_alloc_request(VirtIOBlock *s)
{
    VirtIOBlockReq *req = qemu_mallocz(sizeof(*req));
    req->dev = s;
    return req;
}

static VirtIOBlockReq *virtio_blk_get_request(VirtIOBlock *s)
{
    VirtIOBlockReq *req = virtio_blk_alloc_request(s);

    if (req != NULL) {
        if (!virtqueue_pop(s->vq, &req->elem)) {
            qemu_free(req);
            return NULL;
        }
    }

    return req;
}

#ifdef __linux__
static void virtio_blk_handle_scsi(VirtIOBlockReq *req)
{
    struct sg_io_hdr hdr;
    int ret, size = 0;
    int status;
    int i;

    /*
     * We require at least one output segment each for the virtio_blk_outhdr
     * and the SCSI command block.
     *
     * We also at least require the virtio_blk_inhdr, the virtio_scsi_inhdr
     * and the sense buffer pointer in the input segments.
     */
    if (req->elem.out_num < 2 || req->elem.in_num < 3) {
        virtio_blk_req_complete(req, VIRTIO_BLK_S_IOERR);
        return;
    }

    /*
     * No support for bidirection commands yet.
     */
    if (req->elem.out_num > 2 && req->elem.in_num > 3) {
        virtio_blk_req_complete(req, VIRTIO_BLK_S_UNSUPP);
        return;
    }

    /*
     * The scsi inhdr is placed in the second-to-last input segment, just
     * before the regular inhdr.
     */
    req->scsi = (void *)req->elem.in_sg[req->elem.in_num - 2].iov_base;
    size = sizeof(*req->in) + sizeof(*req->scsi);

    memset(&hdr, 0, sizeof(struct sg_io_hdr));
    hdr.interface_id = 'S';
    hdr.cmd_len = req->elem.out_sg[1].iov_len;
    hdr.cmdp = req->elem.out_sg[1].iov_base;
    hdr.dxfer_len = 0;

    if (req->elem.out_num > 2) {
        /*
         * If there are more than the minimally required 2 output segments
         * there is write payload starting from the third iovec.
         */
        hdr.dxfer_direction = SG_DXFER_TO_DEV;
        hdr.iovec_count = req->elem.out_num - 2;

        for (i = 0; i < hdr.iovec_count; i++)
            hdr.dxfer_len += req->elem.out_sg[i + 2].iov_len;

        hdr.dxferp = req->elem.out_sg + 2;

    } else if (req->elem.in_num > 3) {
        /*
         * If we have more than 3 input segments the guest wants to actually
         * read data.
         */
        hdr.dxfer_direction = SG_DXFER_FROM_DEV;
        hdr.iovec_count = req->elem.in_num - 3;
        for (i = 0; i < hdr.iovec_count; i++)
            hdr.dxfer_len += req->elem.in_sg[i].iov_len;

        hdr.dxferp = req->elem.in_sg;
        size += hdr.dxfer_len;
    } else {
        /*
         * Some SCSI commands don't actually transfer any data.
         */
        hdr.dxfer_direction = SG_DXFER_NONE;
    }

    hdr.sbp = req->elem.in_sg[req->elem.in_num - 3].iov_base;
    hdr.mx_sb_len = req->elem.in_sg[req->elem.in_num - 3].iov_len;
    size += hdr.mx_sb_len;

    ret = bdrv_ioctl(req->dev->bs, SG_IO, &hdr);
    if (ret) {
        status = VIRTIO_BLK_S_UNSUPP;
        hdr.status = ret;
        hdr.resid = hdr.dxfer_len;
    } else if (hdr.status) {
        status = VIRTIO_BLK_S_IOERR;
    } else {
        status = VIRTIO_BLK_S_OK;
    }

    req->scsi->errors = hdr.status;
    req->scsi->residual = hdr.resid;
    req->scsi->sense_len = hdr.sb_len_wr;
    req->scsi->data_len = hdr.dxfer_len;

    virtio_blk_req_complete(req, status);
}
#else
static void virtio_blk_handle_scsi(VirtIOBlockReq *req)
{
    virtio_blk_req_complete(req, VIRTIO_BLK_S_UNSUPP);
}
#endif /* __linux__ */

static void virtio_blk_handle_write(VirtIOBlockReq *req)
{
    BlockDriverAIOCB *acb;

    acb = bdrv_aio_writev(req->dev->bs, req->out->sector, &req->qiov,
                          req->qiov.size / 512, virtio_blk_rw_complete, req);
    if (!acb) {
        virtio_blk_req_complete(req, VIRTIO_BLK_S_IOERR);
    }
}

static void virtio_blk_handle_read(VirtIOBlockReq *req)
{
    BlockDriverAIOCB *acb;

    acb = bdrv_aio_readv(req->dev->bs, req->out->sector, &req->qiov,
                         req->qiov.size / 512, virtio_blk_rw_complete, req);
    if (!acb) {
        virtio_blk_req_complete(req, VIRTIO_BLK_S_IOERR);
    }
}

static void virtio_blk_handle_output(VirtIODevice *vdev, VirtQueue *vq)
{
    VirtIOBlock *s = to_virtio_blk(vdev);
    VirtIOBlockReq *req;

    while ((req = virtio_blk_get_request(s))) {
        if (req->elem.out_num < 1 || req->elem.in_num < 1) {
            fprintf(stderr, "virtio-blk missing headers\n");
            exit(1);
        }

        if (req->elem.out_sg[0].iov_len < sizeof(*req->out) ||
            req->elem.in_sg[req->elem.in_num - 1].iov_len < sizeof(*req->in)) {
            fprintf(stderr, "virtio-blk header not in correct element\n");
            exit(1);
        }

        req->out = (void *)req->elem.out_sg[0].iov_base;
        req->in = (void *)req->elem.in_sg[req->elem.in_num - 1].iov_base;

        if (req->out->type & VIRTIO_BLK_T_SCSI_CMD) {
            virtio_blk_handle_scsi(req);
        } else if (req->out->type & VIRTIO_BLK_T_OUT) {
            qemu_iovec_init_external(&req->qiov, &req->elem.out_sg[1],
                                     req->elem.out_num - 1);
            virtio_blk_handle_write(req);
        } else {
            qemu_iovec_init_external(&req->qiov, &req->elem.in_sg[0],
                                     req->elem.in_num - 1);
            virtio_blk_handle_read(req);
        }
    }
    /*
     * FIXME: Want to check for completions before returning to guest mode,
     * so cached reads and writes are reported as quickly as possible. But
     * that should be done in the generic block layer.
     */
}

static void virtio_blk_dma_restart_bh(void *opaque)
{
    VirtIOBlock *s = opaque;
    VirtIOBlockReq *req = s->rq;

    qemu_bh_delete(s->bh);
    s->bh = NULL;

    s->rq = NULL;

    while (req) {
        virtio_blk_handle_write(req);
        req = req->next;
    }
}

static void virtio_blk_dma_restart_cb(void *opaque, int running, int reason)
{
    VirtIOBlock *s = opaque;

    if (!running)
        return;

    if (!s->bh) {
        s->bh = qemu_bh_new(virtio_blk_dma_restart_bh, s);
        qemu_bh_schedule(s->bh);
    }
}

static void virtio_blk_reset(VirtIODevice *vdev)
{
    /*
     * This should cancel pending requests, but can't do nicely until there
     * are per-device request lists.
     */
    qemu_aio_flush();
}

/* coalesce internal state, copy to pci i/o region 0
 */
static void virtio_blk_update_config(VirtIODevice *vdev, uint8_t *config)
{
    VirtIOBlock *s = to_virtio_blk(vdev);
    struct virtio_blk_config blkcfg;
    uint64_t capacity;
    int cylinders, heads, secs;

    bdrv_get_geometry(s->bs, &capacity);
    bdrv_get_geometry_hint(s->bs, &cylinders, &heads, &secs);
    memset(&blkcfg, 0, sizeof(blkcfg));
    stq_raw(&blkcfg.capacity, capacity);
    stl_raw(&blkcfg.seg_max, 128 - 2);
    stw_raw(&blkcfg.cylinders, cylinders);
    blkcfg.heads = heads;
    blkcfg.sectors = secs;
    blkcfg.size_max = 0;
    virtio_identify_template(&blkcfg);
    memcpy(&blkcfg.identify[VIRTIO_BLK_ID_SN], s->serial_str,
        VIRTIO_BLK_ID_SN_BYTES);
    memcpy(config, &blkcfg, sizeof(blkcfg));
}

static uint32_t virtio_blk_get_features(VirtIODevice *vdev)
{
    VirtIOBlock *s = to_virtio_blk(vdev);
    uint32_t features = 0;

    features |= (1 << VIRTIO_BLK_F_SEG_MAX);
    features |= (1 << VIRTIO_BLK_F_GEOMETRY);
#ifdef __linux__
    features |= (1 << VIRTIO_BLK_F_SCSI);
#endif
    if (strcmp(s->serial_str, "0"))
        features |= 1 << VIRTIO_BLK_F_IDENTIFY;

    return features;
}

static void virtio_blk_save(QEMUFile *f, void *opaque)
{
    VirtIOBlock *s = opaque;
    VirtIOBlockReq *req = s->rq;

    virtio_save(&s->vdev, f);
    
    while (req) {
        qemu_put_sbyte(f, 1);
        qemu_put_buffer(f, (unsigned char*)&req->elem, sizeof(req->elem));
        req = req->next;
    }
    qemu_put_sbyte(f, 0);
}

static int virtio_blk_load(QEMUFile *f, void *opaque, int version_id)
{
    VirtIOBlock *s = opaque;

    if (version_id != 2)
        return -EINVAL;

    virtio_load(&s->vdev, f);
    while (qemu_get_sbyte(f)) {
        VirtIOBlockReq *req = virtio_blk_alloc_request(s);
        qemu_get_buffer(f, (unsigned char*)&req->elem, sizeof(req->elem));
        req->next = s->rq;
        s->rq = req->next;
    }

    return 0;
}

VirtIODevice *virtio_blk_init(DeviceState *dev, DriveInfo *dinfo)
{
    VirtIOBlock *s;
    int cylinders, heads, secs;
    static int virtio_blk_id;
    char *ps;

    s = (VirtIOBlock *)virtio_common_init("virtio-blk", VIRTIO_ID_BLOCK,
                                          sizeof(struct virtio_blk_config),
                                          sizeof(VirtIOBlock));

    s->vdev.get_config = virtio_blk_update_config;
    s->vdev.get_features = virtio_blk_get_features;
    s->vdev.reset = virtio_blk_reset;
    s->bs = dinfo->bdrv;
    s->rq = NULL;
    if (strlen(ps = (char *)drive_get_serial(s->bs)))
        strncpy(s->serial_str, ps, sizeof(s->serial_str));
    else
        snprintf(s->serial_str, sizeof(s->serial_str), "0");
    s->bs->private = dev;
    bdrv_guess_geometry(s->bs, &cylinders, &heads, &secs);
    bdrv_set_geometry_hint(s->bs, cylinders, heads, secs);

    s->vq = virtio_add_queue(&s->vdev, 128, virtio_blk_handle_output);

    qemu_add_vm_change_state_handler(virtio_blk_dma_restart_cb, s);
    register_savevm("virtio-blk", virtio_blk_id++, 2,
                    virtio_blk_save, virtio_blk_load, s);

    return &s->vdev;
}
Commit	Line	Data
6e02c38d AL	1	/*
	2	* Virtio Block Device
	3	*
	4	* Copyright IBM, Corp. 2007
	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
869a5c6d AL	14	#include <qemu-common.h>
869a5c6d AL	15	#include <sysemu.h>
6e02c38d AL	16	#include "virtio-blk.h"
6e02c38d AL	17	#include "block_int.h"
1063b8b1 CH	18	#ifdef __linux__
	19	# include <scsi/sg.h>
	20	#endif
6e02c38d AL	21
	22	typedef struct VirtIOBlock
	23	{
	24	VirtIODevice vdev;
	25	BlockDriverState *bs;
	26	VirtQueue *vq;
869a5c6d	27	void *rq;
bf011293	28	char serial_str[BLOCK_SERIAL_STRLEN + 1];
213189ab	29	QEMUBH *bh;
6e02c38d AL	30	} VirtIOBlock;
	31
	32	static VirtIOBlock to_virtio_blk(VirtIODevice vdev)
	33	{
	34	return (VirtIOBlock *)vdev;
	35	}
	36
bf011293	37	/* store identify data in little endian format
	38	*/
	39	static inline void put_le16(uint16_t *p, unsigned int v)
	40	{
	41	*p = cpu_to_le16(v);
	42	}
	43
	44	/* copy to dst from src, nul pad dst tail as needed to len bytes
	45	*/
	46	static inline void padstr(char dst, const char src, int len)
	47	{
	48	while (len--)
	49	dst++ = src ? *src++ : '\0';
	50	}
	51
	52	/* setup simulated identify data as appropriate for virtio block device
	53	*
	54	* ref: AT Attachment 8 - ATA/ATAPI Command Set (ATA8-ACS)
	55	*/
	56	static inline void virtio_identify_template(struct virtio_blk_config *bc)
	57	{
	58	uint16_t *p = &bc->identify[0];
	59	uint64_t lba_sectors = bc->capacity;
	60
	61	memset(p, 0, sizeof(bc->identify));
	62	put_le16(p + 0, 0x0); /* ATA device */
	63	padstr((char )(p + 23), QEMU_VERSION, 8); / firmware revision */
	64	padstr((char )(p + 27), "QEMU VIRT_BLK", 40); / model# */
	65	put_le16(p + 47, 0x80ff); /* max xfer 255 sectors */
	66	put_le16(p + 49, 0x0b00); /* support IORDY/LBA/DMA */
	67	put_le16(p + 59, 0x1ff); /* cur xfer 255 sectors */
	68	put_le16(p + 80, 0x1f0); /* support ATA8/7/6/5/4 */
	69	put_le16(p + 81, 0x16);
	70	put_le16(p + 82, 0x400);
	71	put_le16(p + 83, 0x400);
	72	put_le16(p + 100, lba_sectors);
	73	put_le16(p + 101, lba_sectors >> 16);
	74	put_le16(p + 102, lba_sectors >> 32);
	75	put_le16(p + 103, lba_sectors >> 48);
	76	}
	77
6e02c38d AL	78	typedef struct VirtIOBlockReq
	79	{
	80	VirtIOBlock *dev;
	81	VirtQueueElement elem;
	82	struct virtio_blk_inhdr *in;
	83	struct virtio_blk_outhdr *out;
1063b8b1	84	struct virtio_scsi_inhdr *scsi;
d28a1b6e	85	QEMUIOVector qiov;
869a5c6d	86	struct VirtIOBlockReq *next;
6e02c38d AL	87	} VirtIOBlockReq;
6e02c38d AL	88
869a5c6d AL	89	static void virtio_blk_req_complete(VirtIOBlockReq *req, int status)
	90	{
	91	VirtIOBlock *s = req->dev;
	92
	93	req->in->status = status;
d28a1b6e	94	virtqueue_push(s->vq, &req->elem, req->qiov.size + sizeof(*req->in));
869a5c6d AL	95	virtio_notify(&s->vdev, s->vq);
869a5c6d AL	96
869a5c6d AL	97	qemu_free(req);
	98	}
	99
	100	static int virtio_blk_handle_write_error(VirtIOBlockReq *req, int error)
	101	{
	102	BlockInterfaceErrorAction action = drive_get_onerror(req->dev->bs);
	103	VirtIOBlock *s = req->dev;
	104
	105	if (action == BLOCK_ERR_IGNORE)
	106	return 0;
	107
	108	if ((error == ENOSPC && action == BLOCK_ERR_STOP_ENOSPC)
	109	\|\| action == BLOCK_ERR_STOP_ANY) {
	110	req->next = s->rq;
	111	s->rq = req;
	112	vm_stop(0);
	113	} else {
	114	virtio_blk_req_complete(req, VIRTIO_BLK_S_IOERR);
	115	}
	116
	117	return 1;
	118	}
	119
6e02c38d AL	120	static void virtio_blk_rw_complete(void *opaque, int ret)
	121	{
	122	VirtIOBlockReq *req = opaque;
6e02c38d	123
d28a1b6e	124	if (ret && (req->out->type & VIRTIO_BLK_T_OUT)) {
869a5c6d AL	125	if (virtio_blk_handle_write_error(req, -ret))
869a5c6d AL	126	return;
6e02c38d AL	127	}
6e02c38d AL	128
869a5c6d AL	129	virtio_blk_req_complete(req, VIRTIO_BLK_S_OK);
869a5c6d AL	130	}
6e02c38d	131
869a5c6d AL	132	static VirtIOBlockReq virtio_blk_alloc_request(VirtIOBlock s)
	133	{
	134	VirtIOBlockReq req = qemu_mallocz(sizeof(req));
487414f1	135	req->dev = s;
869a5c6d	136	return req;
6e02c38d AL	137	}
	138
	139	static VirtIOBlockReq virtio_blk_get_request(VirtIOBlock s)
	140	{
869a5c6d	141	VirtIOBlockReq *req = virtio_blk_alloc_request(s);
6e02c38d	142
869a5c6d AL	143	if (req != NULL) {
	144	if (!virtqueue_pop(s->vq, &req->elem)) {
	145	qemu_free(req);
	146	return NULL;
	147	}
6e02c38d AL	148	}
	149
	150	return req;
	151	}
	152
1063b8b1 CH	153	#ifdef __linux__
	154	static void virtio_blk_handle_scsi(VirtIOBlockReq *req)
	155	{
	156	struct sg_io_hdr hdr;
	157	int ret, size = 0;
	158	int status;
	159	int i;
	160
	161	/*
	162	* We require at least one output segment each for the virtio_blk_outhdr
	163	* and the SCSI command block.
	164	*
	165	* We also at least require the virtio_blk_inhdr, the virtio_scsi_inhdr
	166	* and the sense buffer pointer in the input segments.
	167	*/
	168	if (req->elem.out_num < 2 \|\| req->elem.in_num < 3) {
	169	virtio_blk_req_complete(req, VIRTIO_BLK_S_IOERR);
	170	return;
	171	}
	172
	173	/*
	174	* No support for bidirection commands yet.
	175	*/
	176	if (req->elem.out_num > 2 && req->elem.in_num > 3) {
	177	virtio_blk_req_complete(req, VIRTIO_BLK_S_UNSUPP);
	178	return;
	179	}
	180
	181	/*
	182	* The scsi inhdr is placed in the second-to-last input segment, just
	183	* before the regular inhdr.
	184	*/
	185	req->scsi = (void *)req->elem.in_sg[req->elem.in_num - 2].iov_base;
	186	size = sizeof(req->in) + sizeof(req->scsi);
	187
	188	memset(&hdr, 0, sizeof(struct sg_io_hdr));
	189	hdr.interface_id = 'S';
	190	hdr.cmd_len = req->elem.out_sg[1].iov_len;
	191	hdr.cmdp = req->elem.out_sg[1].iov_base;
	192	hdr.dxfer_len = 0;
	193
	194	if (req->elem.out_num > 2) {
	195	/*
	196	* If there are more than the minimally required 2 output segments
	197	* there is write payload starting from the third iovec.
	198	*/
	199	hdr.dxfer_direction = SG_DXFER_TO_DEV;
	200	hdr.iovec_count = req->elem.out_num - 2;
	201
	202	for (i = 0; i < hdr.iovec_count; i++)
	203	hdr.dxfer_len += req->elem.out_sg[i + 2].iov_len;
	204
	205	hdr.dxferp = req->elem.out_sg + 2;
	206
	207	} else if (req->elem.in_num > 3) {
	208	/*
	209	* If we have more than 3 input segments the guest wants to actually
	210	* read data.
	211	*/
	212	hdr.dxfer_direction = SG_DXFER_FROM_DEV;
	213	hdr.iovec_count = req->elem.in_num - 3;
	214	for (i = 0; i < hdr.iovec_count; i++)
	215	hdr.dxfer_len += req->elem.in_sg[i].iov_len;
	216
217	hdr.dxferp = req->elem.in_sg;
218	size += hdr.dxfer_len;
219	} else {
220	/*
221	* Some SCSI commands don't actually transfer any data.
222	*/
223	hdr.dxfer_direction = SG_DXFER_NONE;
224	}
225
226	hdr.sbp = req->elem.in_sg[req->elem.in_num - 3].iov_base;
227	hdr.mx_sb_len = req->elem.in_sg[req->elem.in_num - 3].iov_len;
228	size += hdr.mx_sb_len;
229
230	ret = bdrv_ioctl(req->dev->bs, SG_IO, &hdr);
231	if (ret) {
232	status = VIRTIO_BLK_S_UNSUPP;
233	hdr.status = ret;
234	hdr.resid = hdr.dxfer_len;
235	} else if (hdr.status) {
236	status = VIRTIO_BLK_S_IOERR;
237	} else {
238	status = VIRTIO_BLK_S_OK;
239	}
240
241	req->scsi->errors = hdr.status;
242	req->scsi->residual = hdr.resid;
243	req->scsi->sense_len = hdr.sb_len_wr;
244	req->scsi->data_len = hdr.dxfer_len;
245
246	virtio_blk_req_complete(req, status);
247	}
248	#else
249	static void virtio_blk_handle_scsi(VirtIOBlockReq *req)
250	{
251	virtio_blk_req_complete(req, VIRTIO_BLK_S_UNSUPP);
252	}
253	#endif /* __linux__ */
254
d28a1b6e	255	static void virtio_blk_handle_write(VirtIOBlockReq *req)
869a5c6d	256	{
87b245db CH	257	BlockDriverAIOCB *acb;
	258
	259	acb = bdrv_aio_writev(req->dev->bs, req->out->sector, &req->qiov,
	260	req->qiov.size / 512, virtio_blk_rw_complete, req);
	261	if (!acb) {
	262	virtio_blk_req_complete(req, VIRTIO_BLK_S_IOERR);
	263	}
d28a1b6e	264	}
869a5c6d	265
d28a1b6e AL	266	static void virtio_blk_handle_read(VirtIOBlockReq *req)
d28a1b6e AL	267	{
87b245db CH	268	BlockDriverAIOCB *acb;
	269
	270	acb = bdrv_aio_readv(req->dev->bs, req->out->sector, &req->qiov,
	271	req->qiov.size / 512, virtio_blk_rw_complete, req);
	272	if (!acb) {
	273	virtio_blk_req_complete(req, VIRTIO_BLK_S_IOERR);
	274	}
869a5c6d AL	275	}
869a5c6d AL	276
6e02c38d AL	277	static void virtio_blk_handle_output(VirtIODevice vdev, VirtQueue vq)
	278	{
	279	VirtIOBlock *s = to_virtio_blk(vdev);
	280	VirtIOBlockReq *req;
	281
	282	while ((req = virtio_blk_get_request(s))) {
6e02c38d AL	283	if (req->elem.out_num < 1 \|\| req->elem.in_num < 1) {
	284	fprintf(stderr, "virtio-blk missing headers\n");
	285	exit(1);
	286	}
	287
	288	if (req->elem.out_sg[0].iov_len < sizeof(*req->out) \|\|
	289	req->elem.in_sg[req->elem.in_num - 1].iov_len < sizeof(*req->in)) {
	290	fprintf(stderr, "virtio-blk header not in correct element\n");
	291	exit(1);
	292	}
	293
	294	req->out = (void *)req->elem.out_sg[0].iov_base;
	295	req->in = (void *)req->elem.in_sg[req->elem.in_num - 1].iov_base;
	296
	297	if (req->out->type & VIRTIO_BLK_T_SCSI_CMD) {
1063b8b1	298	virtio_blk_handle_scsi(req);
6e02c38d	299	} else if (req->out->type & VIRTIO_BLK_T_OUT) {
d28a1b6e AL	300	qemu_iovec_init_external(&req->qiov, &req->elem.out_sg[1],
	301	req->elem.out_num - 1);
	302	virtio_blk_handle_write(req);
6e02c38d	303	} else {
d28a1b6e AL	304	qemu_iovec_init_external(&req->qiov, &req->elem.in_sg[0],
	305	req->elem.in_num - 1);
	306	virtio_blk_handle_read(req);
6e02c38d AL	307	}
	308	}
	309	/*
	310	* FIXME: Want to check for completions before returning to guest mode,
	311	* so cached reads and writes are reported as quickly as possible. But
	312	* that should be done in the generic block layer.
	313	*/
	314	}
	315
213189ab	316	static void virtio_blk_dma_restart_bh(void *opaque)
869a5c6d AL	317	{
	318	VirtIOBlock *s = opaque;
	319	VirtIOBlockReq *req = s->rq;
	320
213189ab MA	321	qemu_bh_delete(s->bh);
213189ab MA	322	s->bh = NULL;
869a5c6d AL	323
	324	s->rq = NULL;
	325
	326	while (req) {
	327	virtio_blk_handle_write(req);
	328	req = req->next;
	329	}
	330	}
	331
213189ab MA	332	static void virtio_blk_dma_restart_cb(void *opaque, int running, int reason)
	333	{
	334	VirtIOBlock *s = opaque;
	335
	336	if (!running)
	337	return;
	338
	339	if (!s->bh) {
	340	s->bh = qemu_bh_new(virtio_blk_dma_restart_bh, s);
	341	qemu_bh_schedule(s->bh);
	342	}
	343	}
	344
6e02c38d AL	345	static void virtio_blk_reset(VirtIODevice *vdev)
	346	{
	347	/*
	348	* This should cancel pending requests, but can't do nicely until there
	349	* are per-device request lists.
	350	*/
	351	qemu_aio_flush();
	352	}
	353
bf011293	354	/* coalesce internal state, copy to pci i/o region 0
bf011293	355	*/
6e02c38d AL	356	static void virtio_blk_update_config(VirtIODevice vdev, uint8_t config)
	357	{
	358	VirtIOBlock *s = to_virtio_blk(vdev);
	359	struct virtio_blk_config blkcfg;
	360	uint64_t capacity;
	361	int cylinders, heads, secs;
	362
	363	bdrv_get_geometry(s->bs, &capacity);
	364	bdrv_get_geometry_hint(s->bs, &cylinders, &heads, &secs);
5c5dafdc	365	memset(&blkcfg, 0, sizeof(blkcfg));
6e02c38d AL	366	stq_raw(&blkcfg.capacity, capacity);
	367	stl_raw(&blkcfg.seg_max, 128 - 2);
	368	stw_raw(&blkcfg.cylinders, cylinders);
	369	blkcfg.heads = heads;
	370	blkcfg.sectors = secs;
c7085da7	371	blkcfg.size_max = 0;
bf011293	372	virtio_identify_template(&blkcfg);
	373	memcpy(&blkcfg.identify[VIRTIO_BLK_ID_SN], s->serial_str,
	374	VIRTIO_BLK_ID_SN_BYTES);
6e02c38d AL	375	memcpy(config, &blkcfg, sizeof(blkcfg));
	376	}
	377
	378	static uint32_t virtio_blk_get_features(VirtIODevice *vdev)
	379	{
bf011293	380	VirtIOBlock *s = to_virtio_blk(vdev);
1063b8b1 CH	381	uint32_t features = 0;
	382
	383	features \|= (1 << VIRTIO_BLK_F_SEG_MAX);
	384	features \|= (1 << VIRTIO_BLK_F_GEOMETRY);
	385	#ifdef __linux__
	386	features \|= (1 << VIRTIO_BLK_F_SCSI);
	387	#endif
bf011293	388	if (strcmp(s->serial_str, "0"))
bf011293	389	features \|= 1 << VIRTIO_BLK_F_IDENTIFY;
1063b8b1 CH	390
1063b8b1 CH	391	return features;
6e02c38d AL	392	}
	393
	394	static void virtio_blk_save(QEMUFile f, void opaque)
	395	{
	396	VirtIOBlock *s = opaque;
869a5c6d AL	397	VirtIOBlockReq *req = s->rq;
869a5c6d AL	398
6e02c38d	399	virtio_save(&s->vdev, f);
869a5c6d AL	400
	401	while (req) {
	402	qemu_put_sbyte(f, 1);
	403	qemu_put_buffer(f, (unsigned char*)&req->elem, sizeof(req->elem));
	404	req = req->next;
	405	}
	406	qemu_put_sbyte(f, 0);
6e02c38d AL	407	}
	408
	409	static int virtio_blk_load(QEMUFile f, void opaque, int version_id)
	410	{
	411	VirtIOBlock *s = opaque;
	412
869a5c6d	413	if (version_id != 2)
6e02c38d AL	414	return -EINVAL;
	415
	416	virtio_load(&s->vdev, f);
869a5c6d AL	417	while (qemu_get_sbyte(f)) {
	418	VirtIOBlockReq *req = virtio_blk_alloc_request(s);
	419	qemu_get_buffer(f, (unsigned char*)&req->elem, sizeof(req->elem));
	420	req->next = s->rq;
	421	s->rq = req->next;
	422	}
6e02c38d AL	423
	424	return 0;
	425	}
	426
d176c495	427	VirtIODevice virtio_blk_init(DeviceState dev, DriveInfo *dinfo)
6e02c38d AL	428	{
	429	VirtIOBlock *s;
	430	int cylinders, heads, secs;
	431	static int virtio_blk_id;
bf011293	432	char *ps;
cf21e106	433
53c25cea PB	434	s = (VirtIOBlock *)virtio_common_init("virtio-blk", VIRTIO_ID_BLOCK,
	435	sizeof(struct virtio_blk_config),
	436	sizeof(VirtIOBlock));
6e02c38d AL	437
	438	s->vdev.get_config = virtio_blk_update_config;
	439	s->vdev.get_features = virtio_blk_get_features;
	440	s->vdev.reset = virtio_blk_reset;
d176c495	441	s->bs = dinfo->bdrv;
869a5c6d	442	s->rq = NULL;
d176c495	443	if (strlen(ps = (char *)drive_get_serial(s->bs)))
bf011293	444	strncpy(s->serial_str, ps, sizeof(s->serial_str));
	445	else
	446	snprintf(s->serial_str, sizeof(s->serial_str), "0");
d176c495	447	s->bs->private = dev;
6e02c38d AL	448	bdrv_guess_geometry(s->bs, &cylinders, &heads, &secs);
	449	bdrv_set_geometry_hint(s->bs, cylinders, heads, secs);
	450
	451	s->vq = virtio_add_queue(&s->vdev, 128, virtio_blk_handle_output);
	452
869a5c6d AL	453	qemu_add_vm_change_state_handler(virtio_blk_dma_restart_cb, s);
869a5c6d AL	454	register_savevm("virtio-blk", virtio_blk_id++, 2,
6e02c38d AL	455	virtio_blk_save, virtio_blk_load, s);
6e02c38d AL	456
53c25cea	457	return &s->vdev;
6e02c38d	458	}