xen: Don't peek behind the BlockDriverState abstraction

[qemu.git] / hw / xen_disk.c

diff --git a/hw/xen_disk.c b/hw/xen_disk.c
index 192e81746f23df6eab5ee789e0c5e2bc0b159ca4..fb68ed9bbf15f3d8f75c6f57fc9591937a410aee 100644 (file)
--- a/hw/xen_disk.c
+++ b/hw/xen_disk.c
@@ -14,6 +14,9 @@
  *
  *  You should have received a copy of the GNU General Public License along
  *  with this program; if not, see <http://www.gnu.org/licenses/>.
+ *
+ *  Contributions after 2012-01-13 are licensed under the terms of the
+ *  GNU GPL, version 2 or (at your option) any later version.
  */
 
 #include <stdio.h>
@@ -37,7 +40,6 @@
 #include <xen/io/xenbus.h>
 
 #include "hw.h"
-#include "block_int.h"
 #include "qemu-char.h"
 #include "xen_blkif.h"
 #include "xen_backend.h"
@@ -45,7 +47,6 @@
 
 /* ------------------------------------------------------------- */
 
-static int syncwrite    = 0;
 static int batch_maps   = 0;
 
 static int max_requests = 32;
@@ -64,6 +65,7 @@ struct ioreq {
     QEMUIOVector        v;
     int                 presync;
     int                 postsync;
+    uint8_t             mapped;
 
     /* grant mapping */
     uint32_t            domids[BLKIF_MAX_SEGMENTS_PER_REQUEST];
@@ -151,7 +153,7 @@ static void ioreq_finish(struct ioreq *ioreq)
     blkdev->requests_finished++;
 }
 
-static void ioreq_release(struct ioreq *ioreq)
+static void ioreq_release(struct ioreq *ioreq, bool finish)
 {
     struct XenBlkDev *blkdev = ioreq->blkdev;
 
@@ -159,7 +161,11 @@ static void ioreq_release(struct ioreq *ioreq)
     memset(ioreq, 0, sizeof(*ioreq));
     ioreq->blkdev = blkdev;
     QLIST_INSERT_HEAD(&blkdev->freelist, ioreq, list);
-    blkdev->requests_finished--;
+    if (finish) {
+        blkdev->requests_finished--;
+    } else {
+        blkdev->requests_inflight--;
+    }
 }
 
 /*
@@ -186,15 +192,10 @@ static int ioreq_parse(struct ioreq *ioreq)
             ioreq->presync = 1;
             return 0;
         }
-        if (!syncwrite) {
-            ioreq->presync = ioreq->postsync = 1;
-        }
+        ioreq->presync = ioreq->postsync = 1;
         /* fall through */
     case BLKIF_OP_WRITE:
         ioreq->prot = PROT_READ; /* from memory */
-        if (syncwrite) {
-            ioreq->postsync = 1;
-        }
         break;
     default:
         xen_be_printf(&blkdev->xendev, 0, "error: unknown operation (%d)\n",
@@ -245,7 +246,7 @@ static void ioreq_unmap(struct ioreq *ioreq)
     XenGnttab gnt = ioreq->blkdev->xendev.gnttabdev;
     int i;
 
-    if (ioreq->v.niov == 0) {
+    if (ioreq->v.niov == 0 || ioreq->mapped == 0) {
         return;
     }
     if (batch_maps) {
@@ -271,6 +272,7 @@ static void ioreq_unmap(struct ioreq *ioreq)
             ioreq->page[i] = NULL;
         }
     }
+    ioreq->mapped = 0;
 }
 
 static int ioreq_map(struct ioreq *ioreq)
@@ -278,7 +280,7 @@ static int ioreq_map(struct ioreq *ioreq)
     XenGnttab gnt = ioreq->blkdev->xendev.gnttabdev;
     int i;
 
-    if (ioreq->v.niov == 0) {
+    if (ioreq->v.niov == 0 || ioreq->mapped == 1) {
         return 0;
     }
     if (batch_maps) {
@@ -310,9 +312,12 @@ static int ioreq_map(struct ioreq *ioreq)
             ioreq->blkdev->cnt_map++;
         }
     }
+    ioreq->mapped = 1;
     return 0;
 }
 
+static int ioreq_runio_qemu_aio(struct ioreq *ioreq);
+
 static void qemu_aio_complete(void *opaque, int ret)
 {
     struct ioreq *ioreq = opaque;
@@ -324,9 +329,20 @@ static void qemu_aio_complete(void *opaque, int ret)
     }
 
     ioreq->aio_inflight--;
+    if (ioreq->presync) {
+        ioreq->presync = 0;
+        ioreq_runio_qemu_aio(ioreq);
+        return;
+    }
     if (ioreq->aio_inflight > 0) {
         return;
     }
+    if (ioreq->postsync) {
+        ioreq->postsync = 0;
+        ioreq->aio_inflight++;
+        bdrv_aio_flush(ioreq->blkdev->bs, qemu_aio_complete, ioreq);
+        return;
+    }
 
     ioreq->status = ioreq->aio_errors ? BLKIF_RSP_ERROR : BLKIF_RSP_OKAY;
     ioreq_unmap(ioreq);
@@ -345,7 +361,8 @@ static int ioreq_runio_qemu_aio(struct ioreq *ioreq)
 
     ioreq->aio_inflight++;
     if (ioreq->presync) {
-        bdrv_flush(blkdev->bs); /* FIXME: aio_flush() ??? */
+        bdrv_aio_flush(ioreq->blkdev->bs, qemu_aio_complete, ioreq);
+        return 0;
     }
 
     switch (ioreq->req.operation) {
@@ -373,9 +390,6 @@ static int ioreq_runio_qemu_aio(struct ioreq *ioreq)
         goto err;
     }
 
-    if (ioreq->postsync) {
-        bdrv_flush(blkdev->bs); /* FIXME: aio_flush() ??? */
-    }
     qemu_aio_complete(ioreq, 0);
 
     return 0;
@@ -446,7 +460,7 @@ static void blk_send_response_all(struct XenBlkDev *blkdev)
     while (!QLIST_EMPTY(&blkdev->finished)) {
         ioreq = QLIST_FIRST(&blkdev->finished);
         send_notify += blk_send_response_one(ioreq);
-        ioreq_release(ioreq);
+        ioreq_release(ioreq, true);
     }
     if (send_notify) {
         xen_be_send_notify(&blkdev->xendev);
@@ -502,7 +516,7 @@ static void blk_handle_requests(struct XenBlkDev *blkdev)
             if (blk_send_response_one(ioreq)) {
                 xen_be_send_notify(&blkdev->xendev);
             }
-            ioreq_release(ioreq);
+            ioreq_release(ioreq, false);
             continue;
         }
 
@@ -522,6 +536,15 @@ static void blk_bh(void *opaque)
     blk_handle_requests(blkdev);
 }
 
+/*
+ * We need to account for the grant allocations requiring contiguous
+ * chunks; the worst case number would be
+ *     max_req * max_seg + (max_req - 1) * (max_seg - 1) + 1,
+ * but in order to keep things simple just use
+ *     2 * max_req * max_seg.
+ */
+#define MAX_GRANTS(max_req, max_seg) (2 * (max_req) * (max_seg))
+
 static void blk_alloc(struct XenDevice *xendev)
 {
     struct XenBlkDev *blkdev = container_of(xendev, struct XenBlkDev, xendev);
@@ -533,6 +556,11 @@ static void blk_alloc(struct XenDevice *xendev)
     if (xen_mode != XEN_EMULATE) {
         batch_maps = 1;
     }
+    if (xc_gnttab_set_max_grants(xendev->gnttabdev,
+            MAX_GRANTS(max_requests, BLKIF_MAX_SEGMENTS_PER_REQUEST)) < 0) {
+        xen_be_printf(xendev, 0, "xc_gnttab_set_max_grants failed: %s\n",
+                      strerror(errno));
+    }
 }
 
 static int blk_init(struct XenDevice *xendev)
@@ -581,10 +609,10 @@ static int blk_init(struct XenDevice *xendev)
     }
 
     /* read-only ? */
+    qflags = BDRV_O_NOCACHE | BDRV_O_CACHE_WB | BDRV_O_NATIVE_AIO;
     if (strcmp(blkdev->mode, "w") == 0) {
-        qflags = BDRV_O_RDWR;
+        qflags |= BDRV_O_RDWR;
     } else {
-        qflags = 0;
         info  |= VDISK_READONLY;
     }
 
@@ -621,7 +649,7 @@ static int blk_init(struct XenDevice *xendev)
     if (blkdev->file_size < 0) {
         xen_be_printf(&blkdev->xendev, 1, "bdrv_getlength: %d (%s) | drv %s\n",
                       (int)blkdev->file_size, strerror(-blkdev->file_size),
-                      blkdev->bs->drv ? blkdev->bs->drv->format_name : "-");
+                      bdrv_get_format_name(blkdev->bs) ?: "-");
         blkdev->file_size = 0;
     }
 
@@ -723,6 +751,7 @@ static void blk_disconnect(struct XenDevice *xendev)
         if (!blkdev->dinfo) {
             /* close/delete only if we created it ourself */
             bdrv_close(blkdev->bs);
+            bdrv_detach_dev(blkdev->bs, blkdev);
             bdrv_delete(blkdev->bs);
         }
         blkdev->bs = NULL;
@@ -741,6 +770,10 @@ static int blk_free(struct XenDevice *xendev)
     struct XenBlkDev *blkdev = container_of(xendev, struct XenBlkDev, xendev);
     struct ioreq *ioreq;
 
+    if (blkdev->bs || blkdev->sring) {
+        blk_disconnect(xendev);
+    }
+
     while (!QLIST_EMPTY(&blkdev->freelist)) {
         ioreq = QLIST_FIRST(&blkdev->freelist);
         QLIST_REMOVE(ioreq, list);