Align file accesses with cache=off (Kevin Wolf, Laurent Vivier)

git-svn-id: svn://svn.savannah.nongnu.org/qemu/trunk@4367 c046a42c-6fe2-441c-8c8c-71466251a162

1 files changed, 113 insertions, 6 deletions

diff --git a/block-raw-posix.c b/block-raw-posix.c
index 6b0009e564..0ea365f1c8 100644
--- a/block-raw-posix.c
+++ b/block-raw-posix.c
@@ -77,10 +77,10 @@
 typedef struct BDRVRawState {
     int fd;
     int type;
+    int open_flags;
     unsigned int lseek_err_cnt;
 #if defined(__linux__)
     /* linux floppy specific */
-    int fd_open_flags;
     int64_t fd_open_time;
     int64_t fd_error_time;
     int fd_got_error;
@@ -111,6 +111,7 @@ static int raw_open(BlockDriverState *bs, const char *filename, int flags)
         open_flags |= O_DIRECT;
 #endif
 
+    s->open_flags = open_flags;
     s->type = FTYPE_FILE;
 
     fd = open(filename, open_flags, 0644);
@@ -141,7 +142,14 @@ static int raw_open(BlockDriverState *bs, const char *filename, int flags)
 #endif
 */
 
-static int raw_pread(BlockDriverState *bs, int64_t offset,
+/* 
+ * offset and count are in bytes, but must be multiples of 512 for files 
+ * opened with O_DIRECT. buf must be aligned to 512 bytes then.
+ *
+ * This function may be called without alignment if the caller ensures
+ * that O_DIRECT is not in effect.
+ */
+static int raw_pread_aligned(BlockDriverState *bs, int64_t offset,
                      uint8_t *buf, int count)
 {
     BDRVRawState *s = bs->opaque;
@@ -194,7 +202,14 @@ label__raw_read__success:
     return ret;
 }
 
-static int raw_pwrite(BlockDriverState *bs, int64_t offset,
+/* 
+ * offset and count are in bytes, but must be multiples of 512 for files 
+ * opened with O_DIRECT. buf must be aligned to 512 bytes then.
+ *
+ * This function may be called without alignment if the caller ensures
+ * that O_DIRECT is not in effect.
+ */
+static int raw_pwrite_aligned(BlockDriverState *bs, int64_t offset,
                       const uint8_t *buf, int count)
 {
     BDRVRawState *s = bs->opaque;
@@ -230,6 +245,67 @@ label__raw_write__success:
     return ret;
 }
 
+
+#ifdef O_DIRECT
+/* 
+ * offset and count are in bytes and possibly not aligned. For files opened 
+ * with O_DIRECT, necessary alignments are ensured before calling 
+ * raw_pread_aligned to do the actual read.
+ */
+static int raw_pread(BlockDriverState *bs, int64_t offset,
+                     uint8_t *buf, int count)
+{
+    BDRVRawState *s = bs->opaque;
+
+    if (unlikely((s->open_flags & O_DIRECT) &&
+            (offset % 512 || count % 512 || (uintptr_t) buf % 512))) {
+
+        int ret;
+
+        // Temporarily disable O_DIRECT for unaligned access
+        fcntl(s->fd, F_SETFL, s->open_flags & ~O_DIRECT);
+        ret = raw_pread_aligned(bs, offset, buf, count);
+        fcntl(s->fd, F_SETFL, s->open_flags);
+
+        return ret;
+
+    } else {
+        return raw_pread_aligned(bs, offset, buf, count);
+    }
+}
+
+/* 
+ * offset and count are in bytes and possibly not aligned. For files opened 
+ * with O_DIRECT, necessary alignments are ensured before calling 
+ * raw_pwrite_aligned to do the actual write.
+ */
+static int raw_pwrite(BlockDriverState *bs, int64_t offset,
+                      const uint8_t *buf, int count)
+{
+    BDRVRawState *s = bs->opaque;
+
+    if (unlikely((s->open_flags & O_DIRECT) &&
+            (offset % 512 || count % 512 || (uintptr_t) buf % 512))) {
+
+        int ret;
+
+        // Temporarily disable O_DIRECT for unaligned access
+        fcntl(s->fd, F_SETFL, s->open_flags & ~O_DIRECT);
+        ret = raw_pwrite_aligned(bs, offset, buf, count);
+        fcntl(s->fd, F_SETFL, s->open_flags);
+
+        return ret;
+    } else {
+        return raw_pwrite_aligned(bs, offset, buf, count);
+    }
+}
+
+#else
+#define raw_pread raw_pread_aligned
+#define raw_pwrite raw_pwrite_aligned
+#endif
+
+
 /***********************************************************/
 /* Unix AIO using POSIX AIO */
 
@@ -402,10 +478,26 @@ static BlockDriverAIOCB *raw_aio_read(BlockDriverState *bs,
         BlockDriverCompletionFunc *cb, void *opaque)
 {
     RawAIOCB *acb;
+    BDRVRawState *s = bs->opaque;
+
+    /* 
+     * If O_DIRECT is used and the buffer is not aligned fall back
+     * to synchronous IO.
+     */
+    if (unlikely((s->open_flags & O_DIRECT) && ((uintptr_t) buf % 512))) {
+        int ret;
+
+        acb = qemu_aio_get(bs, cb, opaque);
+        ret = raw_pread(bs, 512 * sector_num, buf, 512 * nb_sectors);
+        acb->common.cb(acb->common.opaque, ret);
+        qemu_aio_release(acb);
+        return &acb->common;
+    }
 
     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;
@@ -418,6 +510,21 @@ static BlockDriverAIOCB *raw_aio_write(BlockDriverState *bs,
         BlockDriverCompletionFunc *cb, void *opaque)
 {
     RawAIOCB *acb;
+    BDRVRawState *s = bs->opaque;
+
+    /* 
+     * If O_DIRECT is used and the buffer is not aligned fall back
+     * to synchronous IO.
+     */
+    if (unlikely((s->open_flags & O_DIRECT) && ((uintptr_t) buf % 512))) {
+        int ret;
+
+        acb = qemu_aio_get(bs, cb, opaque);
+        ret = raw_pwrite(bs, 512 * sector_num, buf, 512 * nb_sectors);
+        acb->common.cb(acb->common.opaque, ret);
+        qemu_aio_release(acb);
+        return &acb->common;
+    }
 
     acb = raw_aio_setup(bs, sector_num, (uint8_t*)buf, nb_sectors, cb, opaque);
     if (!acb)
@@ -679,7 +786,7 @@ static int hdev_open(BlockDriverState *bs, const char *filename, int flags)
         s->type = FTYPE_CD;
     } else if (strstart(filename, "/dev/fd", NULL)) {
         s->type = FTYPE_FD;
-        s->fd_open_flags = open_flags;
+        s->open_flags = open_flags;
         /* open will not fail even if no floppy is inserted */
         open_flags |= O_NONBLOCK;
     } else if (strstart(filename, "/dev/sg", NULL)) {
@@ -734,7 +841,7 @@ static int fd_open(BlockDriverState *bs)
 #endif
             return -EIO;
         }
-        s->fd = open(bs->filename, s->fd_open_flags);
+        s->fd = open(bs->filename, s->open_flags);
         if (s->fd < 0) {
             s->fd_error_time = qemu_get_clock(rt_clock);
             s->fd_got_error = 1;
@@ -831,7 +938,7 @@ static int raw_eject(BlockDriverState *bs, int eject_flag)
                 close(s->fd);
                 s->fd = -1;
             }
-            fd = open(bs->filename, s->fd_open_flags | O_NONBLOCK);
+            fd = open(bs->filename, s->open_flags | O_NONBLOCK);
             if (fd >= 0) {
                 if (ioctl(fd, FDEJECT, 0) < 0)
                     perror("FDEJECT");