[Qemu-devel] [PATCH v6 0/4] qemu-img: add preallocation=full

[Qemu-devel] [PATCH v6 0/4] qemu-img: add preallocation=full

[Hu Tao]

This series implements full image preallocation to create a non-sparse image
file at creation time, both for raw and qcow2 format. The purpose is to avoid
performance deterioration of the guest cause by sparse image.

v6:
  - add `Since 2.0' to PreallocMode 
  - apply total_size change to raw-win32.c as well

Hu Tao (4):
  qapi: introduce PreallocMode and a new PreallocMode full.
  raw, qcow2: don't convert file size to sector size
  raw-posix: Add full image preallocation option
  qcow2: Add full image preallocation option

 block/qcow2.c     | 91 ++++++++++++++++++++++++++++++++++++++++++++++++-------
 block/raw-posix.c | 45 +++++++++++++++++++++------
 block/raw-win32.c |  4 +--
 qapi-schema.json  | 14 +++++++++
 4 files changed, 132 insertions(+), 22 deletions(-)

-- 
1.8.0

[Qemu-devel] [PATCH v6 1/4] qapi: introduce PreallocMode and a new PreallocMode full.

[Hu Tao]

This patch prepares for the subsequent patches.

Reviewed-by: Fam Zheng <famz@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Signed-off-by: Hu Tao <hutao@cn.fujitsu.com>
---
 block/qcow2.c    |  8 ++++----
 qapi-schema.json | 14 ++++++++++++++
 2 files changed, 18 insertions(+), 4 deletions(-)

diff --git a/block/qcow2.c b/block/qcow2.c
index 0b4335c..d54d289 100644
--- a/block/qcow2.c
+++ b/block/qcow2.c
@@ -1454,7 +1454,7 @@ static int preallocate(BlockDriverState *bs)
 
 static int qcow2_create2(const char *filename, int64_t total_size,
                          const char *backing_file, const char *backing_format,
-                         int flags, size_t cluster_size, int prealloc,
+                         int flags, size_t cluster_size, PreallocMode prealloc,
                          QEMUOptionParameter *options, int version,
                          Error **errp)
 {
@@ -1624,7 +1624,7 @@ static int qcow2_create(const char *filename, QEMUOptionParameter *options,
     uint64_t sectors = 0;
     int flags = 0;
     size_t cluster_size = DEFAULT_CLUSTER_SIZE;
-    int prealloc = 0;
+    PreallocMode prealloc = PREALLOC_MODE_OFF;
     int version = 3;
     Error *local_err = NULL;
     int ret;
@@ -1645,9 +1645,9 @@ static int qcow2_create(const char *filename, QEMUOptionParameter *options,
             }
         } else if (!strcmp(options->name, BLOCK_OPT_PREALLOC)) {
             if (!options->value.s || !strcmp(options->value.s, "off")) {
-                prealloc = 0;
+                prealloc = PREALLOC_MODE_OFF;
             } else if (!strcmp(options->value.s, "metadata")) {
-                prealloc = 1;
+                prealloc = PREALLOC_MODE_METADATA;
             } else {
                 error_setg(errp, "Invalid preallocation mode: '%s'",
                            options->value.s);
diff --git a/qapi-schema.json b/qapi-schema.json
index 7cfb5e5..540d396 100644
--- a/qapi-schema.json
+++ b/qapi-schema.json
@@ -4420,3 +4420,17 @@
 # Since: 1.7
 ##
 { 'command': 'blockdev-add', 'data': { 'options': 'BlockdevOptions' } }
+
+##
+# @PreallocMode
+#
+# Preallocation mode of QEMU image file
+#
+# @off: no preallocation
+# @metadata: preallocate only for metadata
+# @full: preallocate all data, including metadata
+#
+# Since 2.0
+##
+{ 'enum': 'PreallocMode',
+  'data': [ 'off', 'metadata', 'full' ] }
-- 
1.8.0

[Qemu-devel] [PATCH v6 2/4] raw, qcow2: don't convert file size to sector size

[Hu Tao]

and avoid converting it back later.

Signed-off-by: Hu Tao <hutao@cn.fujitsu.com>
---
 block/qcow2.c     | 8 ++++----
 block/raw-posix.c | 4 ++--
 block/raw-win32.c | 4 ++--
 3 files changed, 8 insertions(+), 8 deletions(-)

diff --git a/block/qcow2.c b/block/qcow2.c
index d54d289..59c01f7 100644
--- a/block/qcow2.c
+++ b/block/qcow2.c
@@ -1571,7 +1571,7 @@ static int qcow2_create2(const char *filename, int64_t total_size,
     }
 
     /* Okay, now that we have a valid image, let's give it the right size */
-    ret = bdrv_truncate(bs, total_size * BDRV_SECTOR_SIZE);
+    ret = bdrv_truncate(bs, total_size);
     if (ret < 0) {
         error_setg_errno(errp, -ret, "Could not resize image");
         goto out;
@@ -1621,7 +1621,7 @@ static int qcow2_create(const char *filename, QEMUOptionParameter *options,
 {
     const char *backing_file = NULL;
     const char *backing_fmt = NULL;
-    uint64_t sectors = 0;
+    uint64_t size = 0;
     int flags = 0;
     size_t cluster_size = DEFAULT_CLUSTER_SIZE;
     PreallocMode prealloc = PREALLOC_MODE_OFF;
@@ -1632,7 +1632,7 @@ static int qcow2_create(const char *filename, QEMUOptionParameter *options,
     /* Read out options */
     while (options && options->name) {
         if (!strcmp(options->name, BLOCK_OPT_SIZE)) {
-            sectors = options->value.n / 512;
+            size = options->value.n & BDRV_SECTOR_MASK;
         } else if (!strcmp(options->name, BLOCK_OPT_BACKING_FILE)) {
             backing_file = options->value.s;
         } else if (!strcmp(options->name, BLOCK_OPT_BACKING_FMT)) {
@@ -1683,7 +1683,7 @@ static int qcow2_create(const char *filename, QEMUOptionParameter *options,
         return -EINVAL;
     }
 
-    ret = qcow2_create2(filename, sectors, backing_file, backing_fmt, flags,
+    ret = qcow2_create2(filename, size, backing_file, backing_fmt, flags,
                         cluster_size, prealloc, options, version, &local_err);
     if (error_is_set(&local_err)) {
         error_propagate(errp, local_err);
diff --git a/block/raw-posix.c b/block/raw-posix.c
index 126a634..01fb41a 100644
--- a/block/raw-posix.c
+++ b/block/raw-posix.c
@@ -1233,7 +1233,7 @@ static int raw_create(const char *filename, QEMUOptionParameter *options,
     /* Read out options */
     while (options && options->name) {
         if (!strcmp(options->name, BLOCK_OPT_SIZE)) {
-            total_size = options->value.n / BDRV_SECTOR_SIZE;
+            total_size = options->value.n & BDRV_SECTOR_MASK;
         }
         options++;
     }
@@ -1244,7 +1244,7 @@ static int raw_create(const char *filename, QEMUOptionParameter *options,
         result = -errno;
         error_setg_errno(errp, -result, "Could not create file");
     } else {
-        if (ftruncate(fd, total_size * BDRV_SECTOR_SIZE) != 0) {
+        if (ftruncate(fd, total_size) != 0) {
             result = -errno;
             error_setg_errno(errp, -result, "Could not resize file");
         }
diff --git a/block/raw-win32.c b/block/raw-win32.c
index beb7f23..35d1207 100644
--- a/block/raw-win32.c
+++ b/block/raw-win32.c
@@ -473,7 +473,7 @@ static int raw_create(const char *filename, QEMUOptionParameter *options,
     /* Read out options */
     while (options && options->name) {
         if (!strcmp(options->name, BLOCK_OPT_SIZE)) {
-            total_size = options->value.n / 512;
+            total_size = options->value.n & BDRV_SECTOR_MASK;
         }
         options++;
     }
@@ -485,7 +485,7 @@ static int raw_create(const char *filename, QEMUOptionParameter *options,
         return -EIO;
     }
     set_sparse(fd);
-    ftruncate(fd, total_size * 512);
+    ftruncate(fd, total_size);
     qemu_close(fd);
     return 0;
 }
-- 
1.8.0

[Qemu-devel] qemu-img cannot convert 280G VHD file to qcow2 rightly

[redtone]

Vhd file uslinux01-1.vhd is uploaded to KVM node, the original VHD's size is
290G
But when I convert to QCOW2, I find the disk size is 127G, and then I also
found the uslinux01-1.vhd disk size is 127G by qemu-img info.

Qemu version is 1.7.0 stable

[root@raknode01 disk01]# qemu-img convert -f vpc uslinux01-2.vhd -O qcow2
kvm570_data
[root@raknode01 disk01]# qemu-img info kvm570_os
image: kvm570_os
file format: qcow2
virtual size: 10.0G (10737377280 bytes)
disk size: 3.9G
cluster_size: 65536
Format specific information:
    compat: 1.1
    lazy refcounts: false
[root@raknode01 disk01]# qemu-img info kvm570_data
image: kvm570_data
file format: qcow2
virtual size: 127G (136899993600 bytes)
disk size: 113M
cluster_size: 65536
Format specific information:
    compat: 1.1
    lazy refcounts: false
[root@raknode01 disk01]# qemu-img info uslinux01-2.vhd
image: uslinux01-2.vhd
file format: vpc
virtual size: 127G (136899993600 bytes)
disk size: 5.0G

Re: [Qemu-devel] qemu-img cannot convert 280G VHD file to qcow2 rightly

[Stefan Hajnoczi]

On Fri, Feb 28, 2014 at 07:10:18PM +0800, redtone wrote:
> Vhd file uslinux01-1.vhd is uploaded to KVM node, the original VHD's size is
> 290G
> But when I convert to QCOW2, I find the disk size is 127G, and then I also
> found the uslinux01-1.vhd disk size is 127G by qemu-img info.
> 
> Qemu version is 1.7.0 stable

This bug report is unrelated to this patch series.  In the future,
please start new email threads for unrelated discussions.

> [root@raknode01 disk01]# qemu-img convert -f vpc uslinux01-2.vhd -O qcow2
> kvm570_data
[...]
> [root@raknode01 disk01]# qemu-img info kvm570_data
> image: kvm570_data
> file format: qcow2
> virtual size: 127G (136899993600 bytes)
> disk size: 113M
> cluster_size: 65536
> Format specific information:
>     compat: 1.1
>     lazy refcounts: false
> [root@raknode01 disk01]# qemu-img info uslinux01-2.vhd
> image: uslinux01-2.vhd
> file format: vpc
> virtual size: 127G (136899993600 bytes)
> disk size: 5.0G

This output doesn't show anything unexpected.  You converted a VHD file
to qcow2.  The guest sees a 127 GB virtual disk for both the VHD and the
qcow2 file.

You didn't show the part where the VHD is 290 GB.

Can you post step-by-step instructions for reproducing the problem?

Thanks,
Stefan

[Qemu-devel] [PATCH v6 3/4] raw-posix: Add full image preallocation option

[Hu Tao]

This patch adds a new option preallocation for raw format, and implements
full preallocation.

Signed-off-by: Hu Tao <hutao@cn.fujitsu.com>
---
 block/raw-posix.c | 43 +++++++++++++++++++++++++++++++++++--------
 1 file changed, 35 insertions(+), 8 deletions(-)

diff --git a/block/raw-posix.c b/block/raw-posix.c
index 01fb41a..1961b74 100644
--- a/block/raw-posix.c
+++ b/block/raw-posix.c
@@ -1229,11 +1229,22 @@ static int raw_create(const char *filename, QEMUOptionParameter *options,
     int fd;
     int result = 0;
     int64_t total_size = 0;
+    PreallocMode prealloc = PREALLOC_MODE_OFF;
 
     /* Read out options */
     while (options && options->name) {
         if (!strcmp(options->name, BLOCK_OPT_SIZE)) {
             total_size = options->value.n & BDRV_SECTOR_MASK;
+        } else if (!strcmp(options->name, BLOCK_OPT_PREALLOC)) {
+            if (!options->value.s || !strcmp(options->value.s, "off")) {
+                prealloc = PREALLOC_MODE_OFF;
+            } else if (!strcmp(options->value.s, "full")) {
+                prealloc = PREALLOC_MODE_FULL;
+            } else {
+                error_setg(errp, "Invalid preallocation mode: '%s'",
+                           options->value.s);
+                return -EINVAL;
+            }
         }
         options++;
     }
@@ -1243,16 +1254,27 @@ static int raw_create(const char *filename, QEMUOptionParameter *options,
     if (fd < 0) {
         result = -errno;
         error_setg_errno(errp, -result, "Could not create file");
-    } else {
-        if (ftruncate(fd, total_size) != 0) {
-            result = -errno;
-            error_setg_errno(errp, -result, "Could not resize file");
-        }
-        if (qemu_close(fd) != 0) {
-            result = -errno;
-            error_setg_errno(errp, -result, "Could not close the new file");
+        goto out;
+    }
+    if (ftruncate(fd, total_size) != 0) {
+        result = -errno;
+        error_setg_errno(errp, -result, "Could not resize file");
+        goto out_close;
+    }
+    if (prealloc == PREALLOC_MODE_FULL) {
+        /* posix_fallocate() doesn't set errno. */
+        result = -posix_fallocate(fd, 0, total_size);
+        if (result != 0) {
+            error_setg_errno(errp, -result,
+                             "Could not preallocate data for the new file");
         }
     }
+out_close:
+    if (qemu_close(fd) != 0) {
+        result = -errno;
+        error_setg_errno(errp, -result, "Could not close the new file");
+    }
+out:
     return result;
 }
 
@@ -1403,6 +1425,11 @@ static QEMUOptionParameter raw_create_options[] = {
         .type = OPT_SIZE,
         .help = "Virtual disk size"
     },
+    {
+        .name = BLOCK_OPT_PREALLOC,
+        .type = OPT_STRING,
+        .help = "Preallocation mode (allowed values: off, full)"
+    },
     { NULL }
 };
 
-- 
1.8.0

[Qemu-devel] [PATCH v6 4/4] qcow2: Add full image preallocation option

[Hu Tao]

This adds a preallocation=full mode to qcow2 image creation, which
creates a non-sparse image file.

Signed-off-by: Hu Tao <hutao@cn.fujitsu.com>
---
 block/qcow2.c | 75 ++++++++++++++++++++++++++++++++++++++++++++++++++++++++---
 1 file changed, 72 insertions(+), 3 deletions(-)

diff --git a/block/qcow2.c b/block/qcow2.c
index 59c01f7..2aa6efa 100644
--- a/block/qcow2.c
+++ b/block/qcow2.c
@@ -1458,6 +1458,7 @@ static int qcow2_create2(const char *filename, int64_t total_size,
                          QEMUOptionParameter *options, int version,
                          Error **errp)
 {
+    QEMUOptionParameter *alloc_options = NULL;
     /* Calculate cluster_bits */
     int cluster_bits;
     cluster_bits = ffs(cluster_size) - 1;
@@ -1487,16 +1488,80 @@ static int qcow2_create2(const char *filename, int64_t total_size,
     Error *local_err = NULL;
     int ret;
 
+    if (prealloc == PREALLOC_MODE_FULL) {
+        int64_t meta_size = 0;
+        unsigned nreftablee, nrefblocke, nl1e, nl2e;
+        BlockDriver *drv;
+
+        total_size = align_offset(total_size, cluster_size);
+
+        drv = bdrv_find_protocol(filename, true);
+        if (drv == NULL) {
+            error_setg(errp, "Could not find protocol for file '%s'", filename);
+            return -ENOENT;
+        }
+
+        alloc_options = append_option_parameters(alloc_options,
+                                                 drv->create_options);
+        alloc_options = append_option_parameters(alloc_options, options);
+
+        /* header: 1 cluster */
+        meta_size += cluster_size;
+
+        /* total size of L2 tables */
+        nl2e = total_size / cluster_size;
+        nl2e = align_offset(nl2e, cluster_size / sizeof(uint64_t));
+        meta_size += nl2e * sizeof(uint64_t);
+
+        /* total size of L1 tables */
+        nl1e = nl2e * sizeof(uint64_t) / cluster_size;
+        nl1e = align_offset(nl1e, cluster_size / sizeof(uint64_t));
+        meta_size += nl1e * sizeof(uint64_t);
+
+        /* total size of refcount blocks
+         *
+         * note: every host cluster is reference-counted, including metadata
+         * (even refcount blocks are recursively included).
+         * Let:
+         *   a = total_size (this is the guest disk size)
+         *   m = meta size not including refcount blocks and refcount tables
+         *   c = cluster size
+         *   y1 = number of refcount blocks entries
+         *   y2 = meta size including everything
+         * then,
+         *   y1 = (y2 + a)/c
+         *   y2 = y1 * sizeof(u16) + y1 * sizeof(u16) * sizeof(u64) / c + m
+         * we can get y1:
+         *   y1 = (a + m) / (c - sizeof(u16) - sizeof(u16) * sizeof(u64) / c)
+         */
+        nrefblocke = (total_size + meta_size + cluster_size) /
+            (cluster_size - sizeof(uint16_t) -
+             1.0 * sizeof(uint16_t) * sizeof(uint64_t) / cluster_size);
+        nrefblocke = align_offset(nrefblocke, cluster_size / sizeof(uint16_t));
+        meta_size += nrefblocke * sizeof(uint16_t);
+
+        /* total size of refcount tables */
+        nreftablee = nrefblocke * sizeof(uint16_t) / cluster_size;
+        nreftablee = align_offset(nreftablee, cluster_size / sizeof(uint64_t));
+        meta_size += nreftablee * sizeof(uint64_t);
+
+        set_option_parameter_int(alloc_options, BLOCK_OPT_SIZE,
+                                 total_size + meta_size);
+        set_option_parameter(alloc_options, BLOCK_OPT_PREALLOC, "full");
+
+        options = alloc_options;
+    }
+
     ret = bdrv_create_file(filename, options, &local_err);
     if (ret < 0) {
         error_propagate(errp, local_err);
-        return ret;
+        goto out_options;
     }
 
     ret = bdrv_file_open(&bs, filename, NULL, NULL, BDRV_O_RDWR, &local_err);
     if (ret < 0) {
         error_propagate(errp, local_err);
-        return ret;
+        goto out_options;
     }
 
     /* Write the header */
@@ -1613,6 +1678,8 @@ static int qcow2_create2(const char *filename, int64_t total_size,
     ret = 0;
 out:
     bdrv_unref(bs);
+out_options:
+    free_option_parameters(alloc_options);
     return ret;
 }
 
@@ -1648,6 +1715,8 @@ static int qcow2_create(const char *filename, QEMUOptionParameter *options,
                 prealloc = PREALLOC_MODE_OFF;
             } else if (!strcmp(options->value.s, "metadata")) {
                 prealloc = PREALLOC_MODE_METADATA;
+            } else if (!strcmp(options->value.s, "full")) {
+                prealloc = PREALLOC_MODE_FULL;
             } else {
                 error_setg(errp, "Invalid preallocation mode: '%s'",
                            options->value.s);
@@ -2213,7 +2282,7 @@ static QEMUOptionParameter qcow2_create_options[] = {
     {
         .name = BLOCK_OPT_PREALLOC,
         .type = OPT_STRING,
-        .help = "Preallocation mode (allowed values: off, metadata)"
+        .help = "Preallocation mode (allowed values: off, metadata, full)"
     },
     {
         .name = BLOCK_OPT_LAZY_REFCOUNTS,
-- 
1.8.0

Re: [Qemu-devel] [PATCH v6 0/4] qemu-img: add preallocation=full

[Stefan Hajnoczi]

On Fri, Feb 28, 2014 at 10:29:29AM +0800, Hu Tao wrote:
> This series implements full image preallocation to create a non-sparse image
> file at creation time, both for raw and qcow2 format. The purpose is to avoid
> performance deterioration of the guest cause by sparse image.
> 
> v6:
>   - add `Since 2.0' to PreallocMode 
>   - apply total_size change to raw-win32.c as well
> 
> Hu Tao (4):
>   qapi: introduce PreallocMode and a new PreallocMode full.
>   raw, qcow2: don't convert file size to sector size
>   raw-posix: Add full image preallocation option
>   qcow2: Add full image preallocation option
> 
>  block/qcow2.c     | 91 ++++++++++++++++++++++++++++++++++++++++++++++++-------
>  block/raw-posix.c | 45 +++++++++++++++++++++------
>  block/raw-win32.c |  4 +--
>  qapi-schema.json  | 14 +++++++++
>  4 files changed, 132 insertions(+), 22 deletions(-)

Acked-by: Stefan Hajnoczi <stefanha@redhat.com>

Re: [Qemu-devel] [PATCH v6 0/4] qemu-img: add preallocation=full

[Hu Tao]

On Mon, Mar 03, 2014 at 04:00:37PM +0100, Stefan Hajnoczi wrote:
> On Fri, Feb 28, 2014 at 10:29:29AM +0800, Hu Tao wrote:
> > This series implements full image preallocation to create a non-sparse image
> > file at creation time, both for raw and qcow2 format. The purpose is to avoid
> > performance deterioration of the guest cause by sparse image.
> > 
> > v6:
> >   - add `Since 2.0' to PreallocMode 
> >   - apply total_size change to raw-win32.c as well
> > 
> > Hu Tao (4):
> >   qapi: introduce PreallocMode and a new PreallocMode full.
> >   raw, qcow2: don't convert file size to sector size
> >   raw-posix: Add full image preallocation option
> >   qcow2: Add full image preallocation option
> > 
> >  block/qcow2.c     | 91 ++++++++++++++++++++++++++++++++++++++++++++++++-------
> >  block/raw-posix.c | 45 +++++++++++++++++++++------
> >  block/raw-win32.c |  4 +--
> >  qapi-schema.json  | 14 +++++++++
> >  4 files changed, 132 insertions(+), 22 deletions(-)
> 
> Acked-by: Stefan Hajnoczi <stefanha@redhat.com>

Stefan, Kevin,

Is this series in time for 2.0?

Hu

Re: [Qemu-devel] [PATCH v6 0/4] qemu-img: add preallocation=full

[Stefan Hajnoczi]

On Tue, Mar 04, 2014 at 03:31:31PM +0800, Hu Tao wrote:
> On Mon, Mar 03, 2014 at 04:00:37PM +0100, Stefan Hajnoczi wrote:
> > On Fri, Feb 28, 2014 at 10:29:29AM +0800, Hu Tao wrote:
> > > This series implements full image preallocation to create a non-sparse image
> > > file at creation time, both for raw and qcow2 format. The purpose is to avoid
> > > performance deterioration of the guest cause by sparse image.
> > > 
> > > v6:
> > >   - add `Since 2.0' to PreallocMode 
> > >   - apply total_size change to raw-win32.c as well
> > > 
> > > Hu Tao (4):
> > >   qapi: introduce PreallocMode and a new PreallocMode full.
> > >   raw, qcow2: don't convert file size to sector size
> > >   raw-posix: Add full image preallocation option
> > >   qcow2: Add full image preallocation option
> > > 
> > >  block/qcow2.c     | 91 ++++++++++++++++++++++++++++++++++++++++++++++++-------
> > >  block/raw-posix.c | 45 +++++++++++++++++++++------
> > >  block/raw-win32.c |  4 +--
> > >  qapi-schema.json  | 14 +++++++++
> > >  4 files changed, 132 insertions(+), 22 deletions(-)
> > 
> > Acked-by: Stefan Hajnoczi <stefanha@redhat.com>
> 
> Stefan, Kevin,
> 
> Is this series in time for 2.0?

Yes, it was on the mailing list before soft freeze so it is a candidate
for 2.0.  In some cases maintainers may decide to hold back if they feel
a patch series imposes a high risk of breaking things - but I think this
series should be okay.

Stefan