block: explicit I/O accounting

Decouple the I/O accounting from bdrv_aio_readv/writev/flush and
make the hardware models call directly into the accounting helpers.

This means:
 - we do not count internal requests from image formats in addition
   to guest originating I/O
 - we do not double count I/O ops if the device model handles it
   chunk wise
 - we only account I/O once it actuall is done
 - can extent I/O accounting to synchronous or coroutine I/O easily
 - implement I/O latency tracking easily (see the next patch)

I've conveted the existing device model callers to the new model,
device models that are using synchronous I/O and weren't accounted
before haven't been updated yet.  Also scsi hasn't been converted
to the end-to-end accounting as I want to defer that after the pending
scsi layer overhaul.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>

block.c

@@ -1942,13 +1942,13 @@ static QObject* bdrv_info_stats_bs(BlockDriverState *bs)
                              "'wr_highest_offset': %" PRId64 ","
                              "'flush_operations': %" PRId64
                              "} }",
-                             bs->rd_bytes,
+                             bs->nr_bytes[BDRV_ACCT_READ],
-                             bs->wr_bytes,
+                             bs->nr_bytes[BDRV_ACCT_WRITE],
-                             bs->rd_ops,
+                             bs->nr_ops[BDRV_ACCT_READ],
-                             bs->wr_ops,
+                             bs->nr_ops[BDRV_ACCT_WRITE],
                              bs->wr_highest_sector *
                              (uint64_t)BDRV_SECTOR_SIZE,
-                             bs->flush_ops);
+                             bs->nr_ops[BDRV_ACCT_FLUSH]);
     dict  = qobject_to_qdict(res);
 
     if (*bs->device_name) {
@@ -2262,7 +2262,6 @@ char *bdrv_snapshot_dump(char *buf, int buf_size, QEMUSnapshotInfo *sn)
     return buf;
 }
 
-
 /**************************************************************/
 /* async I/Os */
 
@@ -2271,7 +2270,6 @@ BlockDriverAIOCB *bdrv_aio_readv(BlockDriverState *bs, int64_t sector_num,
                                  BlockDriverCompletionFunc *cb, void *opaque)
 {
     BlockDriver *drv = bs->drv;
-    BlockDriverAIOCB *ret;
 
     trace_bdrv_aio_readv(bs, sector_num, nb_sectors, opaque);
 
@@ -2280,16 +2278,8 @@ BlockDriverAIOCB *bdrv_aio_readv(BlockDriverState *bs, int64_t sector_num,
     if (bdrv_check_request(bs, sector_num, nb_sectors))
         return NULL;
 
-    ret = drv->bdrv_aio_readv(bs, sector_num, qiov, nb_sectors,
+    return drv->bdrv_aio_readv(bs, sector_num, qiov, nb_sectors,
                                cb, opaque);
-
-    if (ret) {
-        /* Update stats even though technically transfer has not happened. */
-        bs->rd_bytes += (unsigned) nb_sectors * BDRV_SECTOR_SIZE;
-        bs->rd_ops++;
-    }
-
-    return ret;
 }
 
 typedef struct BlockCompleteData {
@@ -2356,9 +2346,6 @@ BlockDriverAIOCB *bdrv_aio_writev(BlockDriverState *bs, int64_t sector_num,
                                cb, opaque);
 
     if (ret) {
-        /* Update stats even though technically transfer has not happened. */
-        bs->wr_bytes += (unsigned) nb_sectors * BDRV_SECTOR_SIZE;
-        bs->wr_ops ++;
         if (bs->wr_highest_sector < sector_num + nb_sectors - 1) {
             bs->wr_highest_sector = sector_num + nb_sectors - 1;
         }
@@ -2612,8 +2599,6 @@ BlockDriverAIOCB *bdrv_aio_flush(BlockDriverState *bs,
 
     trace_bdrv_aio_flush(bs, opaque);
 
-    bs->flush_ops++;
-
     if (bs->open_flags & BDRV_O_NO_FLUSH) {
         return bdrv_aio_noop_em(bs, cb, opaque);
     }
@@ -3168,6 +3153,25 @@ int bdrv_in_use(BlockDriverState *bs)
     return bs->in_use;
 }
 
+void
+bdrv_acct_start(BlockDriverState *bs, BlockAcctCookie *cookie, int64_t bytes,
+        enum BlockAcctType type)
+{
+    assert(type < BDRV_MAX_IOTYPE);
+
+    cookie->bytes = bytes;
+    cookie->type = type;
+}
+
+void
+bdrv_acct_done(BlockDriverState *bs, BlockAcctCookie *cookie)
+{
+    assert(cookie->type < BDRV_MAX_IOTYPE);
+
+    bs->nr_bytes[cookie->type] += cookie->bytes;
+    bs->nr_ops[cookie->type]++;
+}
+
 int bdrv_img_create(const char *filename, const char *fmt,
                     const char *base_filename, const char *base_fmt,
                     char *options, uint64_t img_size, int flags)

block.h

@@ -255,6 +255,22 @@ int64_t bdrv_get_dirty_count(BlockDriverState *bs);
 void bdrv_set_in_use(BlockDriverState *bs, int in_use);
 int bdrv_in_use(BlockDriverState *bs);
 
+enum BlockAcctType {
+    BDRV_ACCT_READ,
+    BDRV_ACCT_WRITE,
+    BDRV_ACCT_FLUSH,
+    BDRV_MAX_IOTYPE,
+};
+
+typedef struct BlockAcctCookie {
+    int64_t bytes;
+    enum BlockAcctType type;
+} BlockAcctCookie;
+
+void bdrv_acct_start(BlockDriverState *bs, BlockAcctCookie *cookie,
+        int64_t bytes, enum BlockAcctType type);
+void bdrv_acct_done(BlockDriverState *bs, BlockAcctCookie *cookie);
+
 typedef enum {
     BLKDBG_L1_UPDATE,
 
@@ -307,3 +323,4 @@ typedef enum {
 void bdrv_debug_event(BlockDriverState *bs, BlkDebugEvent event);
 
 #endif
+

block_int.h

@@ -184,11 +184,8 @@ struct BlockDriverState {
     void *sync_aiocb;
 
     /* I/O stats (display with "info blockstats"). */
-    uint64_t rd_bytes;
+    uint64_t nr_bytes[BDRV_MAX_IOTYPE];
-    uint64_t wr_bytes;
+    uint64_t nr_ops[BDRV_MAX_IOTYPE];
-    uint64_t rd_ops;
-    uint64_t wr_ops;
-    uint64_t flush_ops;
     uint64_t wr_highest_sector;
 
     /* Whether the disk can expand beyond total_sectors */

hw/ide/ahci.c

@@ -710,6 +710,7 @@ static void ncq_cb(void *opaque, int ret)
     DPRINTF(ncq_tfs->drive->port_no, "NCQ transfer tag %d finished\n",
             ncq_tfs->tag);
 
+    bdrv_acct_done(ncq_tfs->drive->port.ifs[0].bs, &ncq_tfs->acct);
     qemu_sglist_destroy(&ncq_tfs->sglist);
     ncq_tfs->used = 0;
 }
@@ -756,6 +757,10 @@ static void process_ncq_command(AHCIState *s, int port, uint8_t *cmd_fis,
             ncq_tfs->is_read = 1;
 
             DPRINTF(port, "tag %d aio read %ld\n", ncq_tfs->tag, ncq_tfs->lba);
+
+            bdrv_acct_start(ncq_tfs->drive->port.ifs[0].bs, &ncq_tfs->acct,
+                            (ncq_tfs->sector_count-1) * BDRV_SECTOR_SIZE,
+                            BDRV_ACCT_READ);
             ncq_tfs->aiocb = dma_bdrv_read(ncq_tfs->drive->port.ifs[0].bs,
                                            &ncq_tfs->sglist, ncq_tfs->lba,
                                            ncq_cb, ncq_tfs);
@@ -766,6 +771,10 @@ static void process_ncq_command(AHCIState *s, int port, uint8_t *cmd_fis,
             ncq_tfs->is_read = 0;
 
             DPRINTF(port, "tag %d aio write %ld\n", ncq_tfs->tag, ncq_tfs->lba);
+
+            bdrv_acct_start(ncq_tfs->drive->port.ifs[0].bs, &ncq_tfs->acct,
+                            (ncq_tfs->sector_count-1) * BDRV_SECTOR_SIZE,
+                            BDRV_ACCT_WRITE);
             ncq_tfs->aiocb = dma_bdrv_write(ncq_tfs->drive->port.ifs[0].bs,
                                             &ncq_tfs->sglist, ncq_tfs->lba,
                                             ncq_cb, ncq_tfs);

hw/ide/ahci.h

@@ -258,6 +258,7 @@ typedef struct NCQTransferState {
     AHCIDevice *drive;
     BlockDriverAIOCB *aiocb;
     QEMUSGList sglist;
+    BlockAcctCookie acct;
     int is_read;
     uint16_t sector_count;
     uint64_t lba;

hw/ide/atapi.c

@@ -104,17 +104,20 @@ static void cd_data_to_raw(uint8_t *buf, int lba)
     memset(buf, 0, 288);
 }
 
-static int cd_read_sector(BlockDriverState *bs, int lba, uint8_t *buf,
+static int cd_read_sector(IDEState *s, int lba, uint8_t *buf, int sector_size)
-                           int sector_size)
 {
     int ret;
 
     switch(sector_size) {
     case 2048:
-        ret = bdrv_read(bs, (int64_t)lba << 2, buf, 4);
+        bdrv_acct_start(s->bs, &s->acct, 4 * BDRV_SECTOR_SIZE, BDRV_ACCT_READ);
+        ret = bdrv_read(s->bs, (int64_t)lba << 2, buf, 4);
+        bdrv_acct_done(s->bs, &s->acct);
         break;
     case 2352:
-        ret = bdrv_read(bs, (int64_t)lba << 2, buf + 16, 4);
+        bdrv_acct_start(s->bs, &s->acct, 4 * BDRV_SECTOR_SIZE, BDRV_ACCT_READ);
+        ret = bdrv_read(s->bs, (int64_t)lba << 2, buf + 16, 4);
+        bdrv_acct_done(s->bs, &s->acct);
         if (ret < 0)
             return ret;
         cd_data_to_raw(buf, lba);
@@ -181,7 +184,7 @@ void ide_atapi_cmd_reply_end(IDEState *s)
     } else {
         /* see if a new sector must be read */
         if (s->lba != -1 && s->io_buffer_index >= s->cd_sector_size) {
-            ret = cd_read_sector(s->bs, s->lba, s->io_buffer, s->cd_sector_size);
+            ret = cd_read_sector(s, s->lba, s->io_buffer, s->cd_sector_size);
             if (ret < 0) {
                 ide_transfer_stop(s);
                 ide_atapi_io_error(s, ret);
@@ -250,6 +253,7 @@ static void ide_atapi_cmd_reply(IDEState *s, int size, int max_size)
     s->io_buffer_index = 0;
 
     if (s->atapi_dma) {
+        bdrv_acct_start(s->bs, &s->acct, size, BDRV_ACCT_READ);
         s->status = READY_STAT | SEEK_STAT | DRQ_STAT;
         s->bus->dma->ops->start_dma(s->bus->dma, s,
                                    ide_atapi_cmd_read_dma_cb);
@@ -322,10 +326,7 @@ static void ide_atapi_cmd_read_dma_cb(void *opaque, int ret)
         s->status = READY_STAT | SEEK_STAT;
         s->nsector = (s->nsector & ~7) | ATAPI_INT_REASON_IO | ATAPI_INT_REASON_CD;
         ide_set_irq(s->bus);
-    eot:
+        goto eot;
-        s->bus->dma->ops->add_status(s->bus->dma, BM_STATUS_INT);
-        ide_set_inactive(s);
-        return;
     }
 
     s->io_buffer_index = 0;
@@ -343,9 +344,11 @@ static void ide_atapi_cmd_read_dma_cb(void *opaque, int ret)
 #ifdef DEBUG_AIO
     printf("aio_read_cd: lba=%u n=%d\n", s->lba, n);
 #endif
+
     s->bus->dma->iov.iov_base = (void *)(s->io_buffer + data_offset);
     s->bus->dma->iov.iov_len = n * 4 * 512;
     qemu_iovec_init_external(&s->bus->dma->qiov, &s->bus->dma->iov, 1);
+
     s->bus->dma->aiocb = bdrv_aio_readv(s->bs, (int64_t)s->lba << 2,
                                        &s->bus->dma->qiov, n * 4,
                                        ide_atapi_cmd_read_dma_cb, s);
@@ -355,6 +358,12 @@ static void ide_atapi_cmd_read_dma_cb(void *opaque, int ret)
                             ASC_MEDIUM_NOT_PRESENT);
         goto eot;
     }
+
+    return;
+eot:
+    bdrv_acct_done(s->bs, &s->acct);
+    s->bus->dma->ops->add_status(s->bus->dma, BM_STATUS_INT);
+    ide_set_inactive(s);
 }
 
 /* start a CD-CDROM read command with DMA */
@@ -368,6 +377,8 @@ static void ide_atapi_cmd_read_dma(IDEState *s, int lba, int nb_sectors,
     s->io_buffer_size = 0;
     s->cd_sector_size = sector_size;
 
+    bdrv_acct_start(s->bs, &s->acct, s->packet_transfer_size, BDRV_ACCT_READ);
+
     /* XXX: check if BUSY_STAT should be set */
     s->status = READY_STAT | SEEK_STAT | DRQ_STAT | BUSY_STAT;
     s->bus->dma->ops->start_dma(s->bus->dma, s,

hw/ide/core.c

@@ -473,7 +473,10 @@ void ide_sector_read(IDEState *s)
 #endif
         if (n > s->req_nb_sectors)
             n = s->req_nb_sectors;
+
+        bdrv_acct_start(s->bs, &s->acct, n * BDRV_SECTOR_SIZE, BDRV_ACCT_READ);
         ret = bdrv_read(s->bs, sector_num, s->io_buffer, n);
+        bdrv_acct_done(s->bs, &s->acct);
         if (ret != 0) {
             if (ide_handle_rw_error(s, -ret,
                 BM_STATUS_PIO_RETRY | BM_STATUS_RETRY_READ))
@@ -610,6 +613,9 @@ handle_rw_error:
     return;
 
 eot:
+    if (s->dma_cmd == IDE_DMA_READ || s->dma_cmd == IDE_DMA_WRITE) {
+        bdrv_acct_done(s->bs, &s->acct);
+    }
     ide_set_inactive(s);
 }
 
@@ -619,6 +625,20 @@ static void ide_sector_start_dma(IDEState *s, enum ide_dma_cmd dma_cmd)
     s->io_buffer_index = 0;
     s->io_buffer_size = 0;
     s->dma_cmd = dma_cmd;
+
+    switch (dma_cmd) {
+    case IDE_DMA_READ:
+        bdrv_acct_start(s->bs, &s->acct, s->nsector * BDRV_SECTOR_SIZE,
+                        BDRV_ACCT_READ);
+        break;
+    case IDE_DMA_WRITE:
+        bdrv_acct_start(s->bs, &s->acct, s->nsector * BDRV_SECTOR_SIZE,
+                        BDRV_ACCT_WRITE);
+        break;
+    default:
+        break;
+    }
+
     s->bus->dma->ops->start_dma(s->bus->dma, s, ide_dma_cb);
 }
 
@@ -641,7 +661,10 @@ void ide_sector_write(IDEState *s)
     n = s->nsector;
     if (n > s->req_nb_sectors)
         n = s->req_nb_sectors;
+
+    bdrv_acct_start(s->bs, &s->acct, n * BDRV_SECTOR_SIZE, BDRV_ACCT_READ);
     ret = bdrv_write(s->bs, sector_num, s->io_buffer, n);
+    bdrv_acct_done(s->bs, &s->acct);
 
     if (ret != 0) {
         if (ide_handle_rw_error(s, -ret, BM_STATUS_PIO_RETRY))
@@ -685,6 +708,7 @@ static void ide_flush_cb(void *opaque, int ret)
         }
     }
 
+    bdrv_acct_done(s->bs, &s->acct);
     s->status = READY_STAT | SEEK_STAT;
     ide_set_irq(s->bus);
 }
@@ -698,6 +722,7 @@ void ide_flush_cache(IDEState *s)
         return;
     }
 
+    bdrv_acct_start(s->bs, &s->acct, 0, BDRV_ACCT_FLUSH);
     acb = bdrv_aio_flush(s->bs, ide_flush_cb, s);
     if (acb == NULL) {
         ide_flush_cb(s, -EIO);

hw/ide/internal.h

@@ -440,6 +440,7 @@ struct IDEState {
     int lba;
     int cd_sector_size;
     int atapi_dma; /* true if dma is requested for the packet cmd */
+    BlockAcctCookie acct;
     /* ATA DMA state */
     int io_buffer_size;
     QEMUSGList sg;

hw/ide/macio.c

@@ -52,8 +52,7 @@ static void pmac_ide_atapi_transfer_cb(void *opaque, int ret)
         m->aiocb = NULL;
         qemu_sglist_destroy(&s->sg);
         ide_atapi_io_error(s, ret);
-        io->dma_end(opaque);
+        goto done;
-        return;
     }
 
     if (s->io_buffer_size > 0) {
@@ -71,8 +70,7 @@ static void pmac_ide_atapi_transfer_cb(void *opaque, int ret)
         ide_atapi_cmd_ok(s);
 
     if (io->len == 0) {
-        io->dma_end(opaque);
+        goto done;
-        return;
     }
 
     /* launch next transfer */
@@ -92,9 +90,14 @@ static void pmac_ide_atapi_transfer_cb(void *opaque, int ret)
         /* Note: media not present is the most likely case */
         ide_atapi_cmd_error(s, SENSE_NOT_READY,
                             ASC_MEDIUM_NOT_PRESENT);
+        goto done;
+    }
+    return;
+
+done:
+    bdrv_acct_done(s->bs, &s->acct);
     io->dma_end(opaque);
     return;
-    }
 }
 
 static void pmac_ide_transfer_cb(void *opaque, int ret)
@@ -109,8 +112,7 @@ static void pmac_ide_transfer_cb(void *opaque, int ret)
         m->aiocb = NULL;
         qemu_sglist_destroy(&s->sg);
 	ide_dma_error(s);
-        io->dma_end(io);
+        goto done;
-        return;
     }
 
     sector_num = ide_get_sector(s);
@@ -130,10 +132,8 @@ static void pmac_ide_transfer_cb(void *opaque, int ret)
     }
 
     /* end of DMA ? */
-
     if (io->len == 0) {
-        io->dma_end(io);
+        goto done;
-	return;
     }
 
     /* launch next transfer */
@@ -163,6 +163,12 @@ static void pmac_ide_transfer_cb(void *opaque, int ret)
 
     if (!m->aiocb)
         pmac_ide_transfer_cb(io, -1);
+    return;
+done:
+    if (s->dma_cmd == IDE_DMA_READ || s->dma_cmd == IDE_DMA_WRITE) {
+        bdrv_acct_done(s->bs, &s->acct);
+    }
+    io->dma_end(io);
 }
 
 static void pmac_ide_transfer(DBDMA_io *io)
@@ -172,10 +178,22 @@ static void pmac_ide_transfer(DBDMA_io *io)
 
     s->io_buffer_size = 0;
     if (s->drive_kind == IDE_CD) {
+        bdrv_acct_start(s->bs, &s->acct, io->len, BDRV_ACCT_READ);
         pmac_ide_atapi_transfer_cb(io, 0);
         return;
     }
 
+    switch (s->dma_cmd) {
+    case IDE_DMA_READ:
+        bdrv_acct_start(s->bs, &s->acct, io->len, BDRV_ACCT_READ);
+        break;
+    case IDE_DMA_WRITE:
+        bdrv_acct_start(s->bs, &s->acct, io->len, BDRV_ACCT_WRITE);
+        break;
+    default:
+        break;
+    }
+
     pmac_ide_transfer_cb(io, 0);
 }
 

hw/scsi-disk.c

@@ -57,6 +57,7 @@ typedef struct SCSIDiskReq {
     struct iovec iov;
     QEMUIOVector qiov;
     uint32_t status;
+    BlockAcctCookie acct;
 } SCSIDiskReq;
 
 struct SCSIDiskState
@@ -107,10 +108,13 @@ static void scsi_cancel_io(SCSIRequest *req)
 static void scsi_read_complete(void * opaque, int ret)
 {
     SCSIDiskReq *r = (SCSIDiskReq *)opaque;
+    SCSIDiskState *s = DO_UPCAST(SCSIDiskState, qdev, r->req.dev);
     int n;
 
     r->req.aiocb = NULL;
 
+    bdrv_acct_done(s->bs, &r->acct);
+
     if (ret) {
         if (scsi_handle_rw_error(r, -ret, SCSI_REQ_STATUS_RETRY_READ)) {
             return;
@@ -161,6 +165,8 @@ static void scsi_read_data(SCSIRequest *req)
 
     r->iov.iov_len = n * 512;
     qemu_iovec_init_external(&r->qiov, &r->iov, 1);
+
+    bdrv_acct_start(s->bs, &r->acct, n * BDRV_SECTOR_SIZE, BDRV_ACCT_READ);
     r->req.aiocb = bdrv_aio_readv(s->bs, r->sector, &r->qiov, n,
                               scsi_read_complete, r);
     if (r->req.aiocb == NULL) {
@@ -207,11 +213,14 @@ static int scsi_handle_rw_error(SCSIDiskReq *r, int error, int type)
 static void scsi_write_complete(void * opaque, int ret)
 {
     SCSIDiskReq *r = (SCSIDiskReq *)opaque;
+    SCSIDiskState *s = DO_UPCAST(SCSIDiskState, qdev, r->req.dev);
     uint32_t len;
     uint32_t n;
 
     r->req.aiocb = NULL;
 
+    bdrv_acct_done(s->bs, &r->acct);
+
     if (ret) {
         if (scsi_handle_rw_error(r, -ret, SCSI_REQ_STATUS_RETRY_WRITE)) {
             return;
@@ -252,6 +261,8 @@ static void scsi_write_data(SCSIRequest *req)
     n = r->iov.iov_len / 512;
     if (n) {
         qemu_iovec_init_external(&r->qiov, &r->iov, 1);
+
+        bdrv_acct_start(s->bs, &r->acct, n * BDRV_SECTOR_SIZE, BDRV_ACCT_WRITE);
         r->req.aiocb = bdrv_aio_writev(s->bs, r->sector, &r->qiov, n,
                                    scsi_write_complete, r);
         if (r->req.aiocb == NULL) {
@@ -854,13 +865,19 @@ static int scsi_disk_emulate_command(SCSIDiskReq *r, uint8_t *outbuf)
         buflen = 8;
         break;
     case SYNCHRONIZE_CACHE:
+    {
+        BlockAcctCookie acct;
+
+        bdrv_acct_start(s->bs, &acct, 0, BDRV_ACCT_FLUSH);
         ret = bdrv_flush(s->bs);
+        bdrv_acct_done(s->bs, &acct);
         if (ret < 0) {
             if (scsi_handle_rw_error(r, -ret, SCSI_REQ_STATUS_RETRY_FLUSH)) {
                 return -1;
             }
         }
         break;
+    }
     case GET_CONFIGURATION:
         memset(outbuf, 0, 8);
         /* ??? This should probably return much more information.  For now

hw/virtio-blk.c

@@ -47,6 +47,7 @@ typedef struct VirtIOBlockReq
     struct virtio_scsi_inhdr *scsi;
     QEMUIOVector qiov;
     struct VirtIOBlockReq *next;
+    BlockAcctCookie acct;
 } VirtIOBlockReq;
 
 static void virtio_blk_req_complete(VirtIOBlockReq *req, int status)
@@ -58,8 +59,6 @@ static void virtio_blk_req_complete(VirtIOBlockReq *req, int status)
     stb_p(&req->in->status, status);
     virtqueue_push(s->vq, &req->elem, req->qiov.size + sizeof(*req->in));
     virtio_notify(&s->vdev, s->vq);
-
-    g_free(req);
 }
 
 static int virtio_blk_handle_rw_error(VirtIOBlockReq *req, int error,
@@ -81,6 +80,8 @@ static int virtio_blk_handle_rw_error(VirtIOBlockReq *req, int error,
         vm_stop(VMSTOP_DISKFULL);
     } else {
         virtio_blk_req_complete(req, VIRTIO_BLK_S_IOERR);
+        bdrv_acct_done(s->bs, &req->acct);
+        g_free(req);
         bdrv_mon_event(s->bs, BDRV_ACTION_REPORT, is_read);
     }
 
@@ -100,6 +101,8 @@ static void virtio_blk_rw_complete(void *opaque, int ret)
     }
 
     virtio_blk_req_complete(req, VIRTIO_BLK_S_OK);
+    bdrv_acct_done(req->dev->bs, &req->acct);
+    g_free(req);
 }
 
 static void virtio_blk_flush_complete(void *opaque, int ret)
@@ -113,6 +116,8 @@ static void virtio_blk_flush_complete(void *opaque, int ret)
     }
 
     virtio_blk_req_complete(req, VIRTIO_BLK_S_OK);
+    bdrv_acct_done(req->dev->bs, &req->acct);
+    g_free(req);
 }
 
 static VirtIOBlockReq *virtio_blk_alloc_request(VirtIOBlock *s)
@@ -155,6 +160,7 @@ static void virtio_blk_handle_scsi(VirtIOBlockReq *req)
      */
     if (req->elem.out_num < 2 || req->elem.in_num < 3) {
         virtio_blk_req_complete(req, VIRTIO_BLK_S_IOERR);
+        g_free(req);
         return;
     }
 
@@ -163,6 +169,7 @@ static void virtio_blk_handle_scsi(VirtIOBlockReq *req)
      */
     if (req->elem.out_num > 2 && req->elem.in_num > 3) {
         virtio_blk_req_complete(req, VIRTIO_BLK_S_UNSUPP);
+        g_free(req);
         return;
     }
 
@@ -229,11 +236,13 @@ static void virtio_blk_handle_scsi(VirtIOBlockReq *req)
     stl_p(&req->scsi->data_len, hdr.dxfer_len);
 
     virtio_blk_req_complete(req, status);
+    g_free(req);
 }
 #else
 static void virtio_blk_handle_scsi(VirtIOBlockReq *req)
 {
     virtio_blk_req_complete(req, VIRTIO_BLK_S_UNSUPP);
+    g_free(req);
 }
 #endif /* __linux__ */
 
@@ -266,6 +275,8 @@ static void virtio_blk_handle_flush(VirtIOBlockReq *req, MultiReqBuffer *mrb)
 {
     BlockDriverAIOCB *acb;
 
+    bdrv_acct_start(req->dev->bs, &req->acct, 0, BDRV_ACCT_FLUSH);
+
     /*
      * Make sure all outstanding writes are posted to the backing device.
      */
@@ -284,6 +295,8 @@ static void virtio_blk_handle_write(VirtIOBlockReq *req, MultiReqBuffer *mrb)
 
     sector = ldq_p(&req->out->sector);
 
+    bdrv_acct_start(req->dev->bs, &req->acct, req->qiov.size, BDRV_ACCT_WRITE);
+
     trace_virtio_blk_handle_write(req, sector, req->qiov.size / 512);
 
     if (sector & req->dev->sector_mask) {
@@ -317,6 +330,8 @@ static void virtio_blk_handle_read(VirtIOBlockReq *req)
 
     sector = ldq_p(&req->out->sector);
 
+    bdrv_acct_start(req->dev->bs, &req->acct, req->qiov.size, BDRV_ACCT_READ);
+
     if (sector & req->dev->sector_mask) {
         virtio_blk_rw_complete(req, -EIO);
         return;
@@ -370,6 +385,7 @@ static void virtio_blk_handle_request(VirtIOBlockReq *req,
                 s->serial ? s->serial : "",
                 MIN(req->elem.in_sg[0].iov_len, VIRTIO_BLK_ID_BYTES));
         virtio_blk_req_complete(req, VIRTIO_BLK_S_OK);
+        g_free(req);
     } else if (type & VIRTIO_BLK_T_OUT) {
         qemu_iovec_init_external(&req->qiov, &req->elem.out_sg[1],
                                  req->elem.out_num - 1);

hw/xen_disk.c

@@ -79,6 +79,7 @@ struct ioreq {
 
     struct XenBlkDev    *blkdev;
     QLIST_ENTRY(ioreq)   list;
+    BlockAcctCookie     acct;
 };
 
 struct XenBlkDev {
@@ -401,6 +402,7 @@ static void qemu_aio_complete(void *opaque, int ret)
     ioreq->status = ioreq->aio_errors ? BLKIF_RSP_ERROR : BLKIF_RSP_OKAY;
     ioreq_unmap(ioreq);
     ioreq_finish(ioreq);
+    bdrv_acct_done(ioreq->blkdev->bs, &ioreq->acct);
     qemu_bh_schedule(ioreq->blkdev->bh);
 }
 
@@ -419,6 +421,7 @@ static int ioreq_runio_qemu_aio(struct ioreq *ioreq)
 
     switch (ioreq->req.operation) {
     case BLKIF_OP_READ:
+        bdrv_acct_start(blkdev->bs, &ioreq->acct, ioreq->v.size, BDRV_ACCT_READ);
         ioreq->aio_inflight++;
         bdrv_aio_readv(blkdev->bs, ioreq->start / BLOCK_SIZE,
                        &ioreq->v, ioreq->v.size / BLOCK_SIZE,
@@ -429,6 +432,8 @@ static int ioreq_runio_qemu_aio(struct ioreq *ioreq)
         if (!ioreq->req.nr_segments) {
             break;
         }
+
+        bdrv_acct_start(blkdev->bs, &ioreq->acct, ioreq->v.size, BDRV_ACCT_WRITE);
         ioreq->aio_inflight++;
         bdrv_aio_writev(blkdev->bs, ioreq->start / BLOCK_SIZE,
                         &ioreq->v, ioreq->v.size / BLOCK_SIZE,