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>

6 files changed, 86 insertions, 21 deletions

diff --git a/hw/ide/ahci.c b/hw/ide/ahci.c
index 29521babf7..f4fa1545bd 100644
--- a/hw/ide/ahci.c
+++ b/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);
diff --git a/hw/ide/ahci.h b/hw/ide/ahci.h
index e456193b2b..832539c23c 100644
--- a/hw/ide/ahci.h
+++ b/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;
diff --git a/hw/ide/atapi.c b/hw/ide/atapi.c
index fe2fb0b806..c552320122 100644
--- a/hw/ide/atapi.c
+++ b/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,
-                           int sector_size)
+static int cd_read_sector(IDEState *s, int lba, uint8_t *buf, 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:
-        s->bus->dma->ops->add_status(s->bus->dma, BM_STATUS_INT);
-        ide_set_inactive(s);
-        return;
+        goto eot;
     }
 
     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,
diff --git a/hw/ide/core.c b/hw/ide/core.c
index d145b19b0c..40abc1edd2 100644
--- a/hw/ide/core.c
+++ b/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))
@@ -610,7 +613,10 @@ handle_rw_error:
     return;
 
 eot:
-   ide_set_inactive(s);
+    if (s->dma_cmd == IDE_DMA_READ || s->dma_cmd == IDE_DMA_WRITE) {
+        bdrv_acct_done(s->bs, &s->acct);
+    }
+    ide_set_inactive(s);
 }
 
 static void ide_sector_start_dma(IDEState *s, enum ide_dma_cmd dma_cmd)
@@ -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);
diff --git a/hw/ide/internal.h b/hw/ide/internal.h
index 02e805f070..7f5ef8de1d 100644
--- a/hw/ide/internal.h
+++ b/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;
diff --git a/hw/ide/macio.c b/hw/ide/macio.c
index 44fb3fef60..fdf5d75082 100644
--- a/hw/ide/macio.c
+++ b/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);
-        return;
+        goto done;
     }
 
     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);
-        return;
+        goto done;
     }
 
     /* 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);
-        io->dma_end(opaque);
-        return;
+        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);
-        return;
+        goto done;
     }
 
     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);
-	return;
+        goto done;
     }
 
     /* 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);
 }