Blame SOURCES/kvm-file-posix-Mitigate-file-fragmentation-with-extent-s.patch

6e7d01
From 7ee01b5ccb7fc660dafaf3fdb1578649d17fbddf Mon Sep 17 00:00:00 2001
6e7d01
From: Kevin Wolf <kwolf@redhat.com>
6e7d01
Date: Wed, 26 May 2021 09:05:52 -0400
6e7d01
Subject: [PATCH 1/4] file-posix: Mitigate file fragmentation with extent size
6e7d01
 hints
6e7d01
6e7d01
RH-Author: Kevin Wolf <kwolf@redhat.com>
6e7d01
Message-id: <20210526090552.155820-2-kwolf@redhat.com>
6e7d01
Patchwork-id: 101638
6e7d01
O-Subject: [RHEL-8.5.0 qemu-kvm PATCH 1/1] file-posix: Mitigate file fragmentation with extent size hints
6e7d01
Bugzilla: 1877163
6e7d01
RH-Acked-by: Sergio Lopez Pascual <slp@redhat.com>
6e7d01
RH-Acked-by: Stefan Hajnoczi <stefanha@redhat.com>
6e7d01
RH-Acked-by: Max Reitz <mreitz@redhat.com>
6e7d01
6e7d01
Especially when O_DIRECT is used with image files so that the page cache
6e7d01
indirection can't cause a merge of allocating requests, the file will
6e7d01
fragment on the file system layer, with a potentially very small
6e7d01
fragment size (this depends on the requests the guest sent).
6e7d01
6e7d01
On Linux, fragmentation can be reduced by setting an extent size hint
6e7d01
when creating the file (at least on XFS, it can't be set any more after
6e7d01
the first extent has been allocated), basically giving raw files a
6e7d01
"cluster size" for allocation.
6e7d01
6e7d01
This adds a create option to set the extent size hint, and changes the
6e7d01
default from not setting a hint to setting it to 1 MB. The main reason
6e7d01
why qcow2 defaults to smaller cluster sizes is that COW becomes more
6e7d01
expensive, which is not an issue with raw files, so we can choose a
6e7d01
larger size. The tradeoff here is only potentially wasted disk space.
6e7d01
6e7d01
For qcow2 (or other image formats) over file-posix, the advantage should
6e7d01
even be greater because they grow sequentially without leaving holes, so
6e7d01
there won't be wasted space. Setting even larger extent size hints for
6e7d01
such images may make sense. This can be done with the new option, but
6e7d01
let's keep the default conservative for now.
6e7d01
6e7d01
The effect is very visible with a test that intentionally creates a
6e7d01
badly fragmented file with qemu-img bench (the time difference while
6e7d01
creating the file is already remarkable) and then looks at the number of
6e7d01
extents and the time a simple "qemu-img map" takes.
6e7d01
6e7d01
Without an extent size hint:
6e7d01
6e7d01
    $ ./qemu-img create -f raw -o extent_size_hint=0 ~/tmp/test.raw 10G
6e7d01
    Formatting '/home/kwolf/tmp/test.raw', fmt=raw size=10737418240 extent_size_hint=0
6e7d01
    $ ./qemu-img bench -f raw -t none -n -w ~/tmp/test.raw -c 1000000 -S 8192 -o 0
6e7d01
    Sending 1000000 write requests, 4096 bytes each, 64 in parallel (starting at offset 0, step size 8192)
6e7d01
    Run completed in 25.848 seconds.
6e7d01
    $ ./qemu-img bench -f raw -t none -n -w ~/tmp/test.raw -c 1000000 -S 8192 -o 4096
6e7d01
    Sending 1000000 write requests, 4096 bytes each, 64 in parallel (starting at offset 4096, step size 8192)
6e7d01
    Run completed in 19.616 seconds.
6e7d01
    $ filefrag ~/tmp/test.raw
6e7d01
    /home/kwolf/tmp/test.raw: 2000000 extents found
6e7d01
    $ time ./qemu-img map ~/tmp/test.raw
6e7d01
    Offset          Length          Mapped to       File
6e7d01
    0               0x1e8480000     0               /home/kwolf/tmp/test.raw
6e7d01
6e7d01
    real    0m1,279s
6e7d01
    user    0m0,043s
6e7d01
    sys     0m1,226s
6e7d01
6e7d01
With the new default extent size hint of 1 MB:
6e7d01
6e7d01
    $ ./qemu-img create -f raw -o extent_size_hint=1M ~/tmp/test.raw 10G
6e7d01
    Formatting '/home/kwolf/tmp/test.raw', fmt=raw size=10737418240 extent_size_hint=1048576
6e7d01
    $ ./qemu-img bench -f raw -t none -n -w ~/tmp/test.raw -c 1000000 -S 8192 -o 0
6e7d01
    Sending 1000000 write requests, 4096 bytes each, 64 in parallel (starting at offset 0, step size 8192)
6e7d01
    Run completed in 11.833 seconds.
6e7d01
    $ ./qemu-img bench -f raw -t none -n -w ~/tmp/test.raw -c 1000000 -S 8192 -o 4096
6e7d01
    Sending 1000000 write requests, 4096 bytes each, 64 in parallel (starting at offset 4096, step size 8192)
6e7d01
    Run completed in 10.155 seconds.
6e7d01
    $ filefrag ~/tmp/test.raw
6e7d01
    /home/kwolf/tmp/test.raw: 178 extents found
6e7d01
    $ time ./qemu-img map ~/tmp/test.raw
6e7d01
    Offset          Length          Mapped to       File
6e7d01
    0               0x1e8480000     0               /home/kwolf/tmp/test.raw
6e7d01
6e7d01
    real    0m0,061s
6e7d01
    user    0m0,040s
6e7d01
    sys     0m0,014s
6e7d01
6e7d01
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
6e7d01
Message-Id: <20200707142329.48303-1-kwolf@redhat.com>
6e7d01
Reviewed-by: Eric Blake <eblake@redhat.com>
6e7d01
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
6e7d01
(cherry picked from commit ffa244c84a1a30dff69ecc80b0137a2b6d428ecb)
6e7d01
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
6e7d01
Signed-off-by: Danilo C. L. de Paula <ddepaula@redhat.com>
6e7d01
---
6e7d01
 block/file-posix.c               | 44 ++++++++++++++++++++++++++++++++
6e7d01
 include/block/block_int.h        |  1 +
6e7d01
 qapi/block-core.json             | 11 +++++---
6e7d01
 tests/qemu-iotests/082.out       | 16 ++++++++++++
6e7d01
 tests/qemu-iotests/106           |  7 +++--
6e7d01
 tests/qemu-iotests/175           |  6 ++---
6e7d01
 tests/qemu-iotests/243           |  6 ++---
6e7d01
 tests/qemu-iotests/common.filter |  1 +
6e7d01
 8 files changed, 80 insertions(+), 12 deletions(-)
6e7d01
6e7d01
diff --git a/block/file-posix.c b/block/file-posix.c
6e7d01
index 2d834fbdf6..62a463229f 100644
6e7d01
--- a/block/file-posix.c
6e7d01
+++ b/block/file-posix.c
6e7d01
@@ -30,6 +30,7 @@
6e7d01
 #include "block/block_int.h"
6e7d01
 #include "qemu/module.h"
6e7d01
 #include "qemu/option.h"
6e7d01
+#include "qemu/units.h"
6e7d01
 #include "trace.h"
6e7d01
 #include "block/thread-pool.h"
6e7d01
 #include "qemu/iov.h"
6e7d01
@@ -2289,6 +2290,14 @@ raw_co_create(BlockdevCreateOptions *options, Error **errp)
6e7d01
     if (!file_opts->has_preallocation) {
6e7d01
         file_opts->preallocation = PREALLOC_MODE_OFF;
6e7d01
     }
6e7d01
+    if (!file_opts->has_extent_size_hint) {
6e7d01
+        file_opts->extent_size_hint = 1 * MiB;
6e7d01
+    }
6e7d01
+    if (file_opts->extent_size_hint > UINT32_MAX) {
6e7d01
+        result = -EINVAL;
6e7d01
+        error_setg(errp, "Extent size hint is too large");
6e7d01
+        goto out;
6e7d01
+    }
6e7d01
 
6e7d01
     /* Create file */
6e7d01
     fd = qemu_open(file_opts->filename, O_RDWR | O_CREAT | O_BINARY, 0644);
6e7d01
@@ -2346,6 +2355,27 @@ raw_co_create(BlockdevCreateOptions *options, Error **errp)
6e7d01
         }
6e7d01
 #endif
6e7d01
     }
6e7d01
+#ifdef FS_IOC_FSSETXATTR
6e7d01
+    /*
6e7d01
+     * Try to set the extent size hint. Failure is not fatal, and a warning is
6e7d01
+     * only printed if the option was explicitly specified.
6e7d01
+     */
6e7d01
+    {
6e7d01
+        struct fsxattr attr;
6e7d01
+        result = ioctl(fd, FS_IOC_FSGETXATTR, &attr);
6e7d01
+        if (result == 0) {
6e7d01
+            attr.fsx_xflags |= FS_XFLAG_EXTSIZE;
6e7d01
+            attr.fsx_extsize = file_opts->extent_size_hint;
6e7d01
+            result = ioctl(fd, FS_IOC_FSSETXATTR, &attr);
6e7d01
+        }
6e7d01
+        if (result < 0 && file_opts->has_extent_size_hint &&
6e7d01
+            file_opts->extent_size_hint)
6e7d01
+        {
6e7d01
+            warn_report("Failed to set extent size hint: %s",
6e7d01
+                        strerror(errno));
6e7d01
+        }
6e7d01
+    }
6e7d01
+#endif
6e7d01
 
6e7d01
     /* Resize and potentially preallocate the file to the desired
6e7d01
      * final size */
6e7d01
@@ -2381,6 +2411,8 @@ static int coroutine_fn raw_co_create_opts(BlockDriver *drv,
6e7d01
 {
6e7d01
     BlockdevCreateOptions options;
6e7d01
     int64_t total_size = 0;
6e7d01
+    int64_t extent_size_hint = 0;
6e7d01
+    bool has_extent_size_hint = false;
6e7d01
     bool nocow = false;
6e7d01
     PreallocMode prealloc;
6e7d01
     char *buf = NULL;
6e7d01
@@ -2392,6 +2424,11 @@ static int coroutine_fn raw_co_create_opts(BlockDriver *drv,
6e7d01
     /* Read out options */
6e7d01
     total_size = ROUND_UP(qemu_opt_get_size_del(opts, BLOCK_OPT_SIZE, 0),
6e7d01
                           BDRV_SECTOR_SIZE);
6e7d01
+    if (qemu_opt_get(opts, BLOCK_OPT_EXTENT_SIZE_HINT)) {
6e7d01
+        has_extent_size_hint = true;
6e7d01
+        extent_size_hint =
6e7d01
+            qemu_opt_get_size_del(opts, BLOCK_OPT_EXTENT_SIZE_HINT, -1);
6e7d01
+    }
6e7d01
     nocow = qemu_opt_get_bool(opts, BLOCK_OPT_NOCOW, false);
6e7d01
     buf = qemu_opt_get_del(opts, BLOCK_OPT_PREALLOC);
6e7d01
     prealloc = qapi_enum_parse(&PreallocMode_lookup, buf,
6e7d01
@@ -2411,6 +2448,8 @@ static int coroutine_fn raw_co_create_opts(BlockDriver *drv,
6e7d01
             .preallocation      = prealloc,
6e7d01
             .has_nocow          = true,
6e7d01
             .nocow              = nocow,
6e7d01
+            .has_extent_size_hint = has_extent_size_hint,
6e7d01
+            .extent_size_hint   = extent_size_hint,
6e7d01
         },
6e7d01
     };
6e7d01
     return raw_co_create(&options, errp);
6e7d01
@@ -2902,6 +2941,11 @@ static QemuOptsList raw_create_opts = {
6e7d01
 #endif
6e7d01
                     ", full)"
6e7d01
         },
6e7d01
+        {
6e7d01
+            .name = BLOCK_OPT_EXTENT_SIZE_HINT,
6e7d01
+            .type = QEMU_OPT_SIZE,
6e7d01
+            .help = "Extent size hint for the image file, 0 to disable"
6e7d01
+        },
6e7d01
         { /* end of list */ }
6e7d01
     }
6e7d01
 };
6e7d01
diff --git a/include/block/block_int.h b/include/block/block_int.h
6e7d01
index 41f13ecbed..4b23da2eb0 100644
6e7d01
--- a/include/block/block_int.h
6e7d01
+++ b/include/block/block_int.h
6e7d01
@@ -53,6 +53,7 @@
6e7d01
 #define BLOCK_OPT_ADAPTER_TYPE      "adapter_type"
6e7d01
 #define BLOCK_OPT_REDUNDANCY        "redundancy"
6e7d01
 #define BLOCK_OPT_NOCOW             "nocow"
6e7d01
+#define BLOCK_OPT_EXTENT_SIZE_HINT  "extent_size_hint"
6e7d01
 #define BLOCK_OPT_OBJECT_SIZE       "object_size"
6e7d01
 #define BLOCK_OPT_REFCOUNT_BITS     "refcount_bits"
6e7d01
 #define BLOCK_OPT_DATA_FILE         "data_file"
6e7d01
diff --git a/qapi/block-core.json b/qapi/block-core.json
6e7d01
index 289320902d..c7aa919fa3 100644
6e7d01
--- a/qapi/block-core.json
6e7d01
+++ b/qapi/block-core.json
6e7d01
@@ -4272,14 +4272,17 @@
6e7d01
 #                   falloc (if defined CONFIG_POSIX_FALLOCATE),
6e7d01
 #                   full (if defined CONFIG_POSIX))
6e7d01
 # @nocow            Turn off copy-on-write (valid only on btrfs; default: off)
6e7d01
+# @extent-size-hint: Extent size hint to add to the image file; 0 for not
6e7d01
+#                    adding an extent size hint (default: 1 MB, since 5.1)
6e7d01
 #
6e7d01
 # Since: 2.12
6e7d01
 ##
6e7d01
 { 'struct': 'BlockdevCreateOptionsFile',
6e7d01
-  'data': { 'filename':         'str',
6e7d01
-            'size':             'size',
6e7d01
-            '*preallocation':   'PreallocMode',
6e7d01
-            '*nocow':           'bool' } }
6e7d01
+  'data': { 'filename':             'str',
6e7d01
+            'size':                 'size',
6e7d01
+            '*preallocation':       'PreallocMode',
6e7d01
+            '*nocow':               'bool',
6e7d01
+            '*extent-size-hint':    'size'} }
6e7d01
 
6e7d01
 ##
6e7d01
 # @BlockdevCreateOptionsGluster:
6e7d01
diff --git a/tests/qemu-iotests/082.out b/tests/qemu-iotests/082.out
6e7d01
index 9d4ed4dc9d..7a87946fa2 100644
6e7d01
--- a/tests/qemu-iotests/082.out
6e7d01
+++ b/tests/qemu-iotests/082.out
6e7d01
@@ -59,6 +59,7 @@ Supported options:
6e7d01
   encrypt.ivgen-hash-alg=<str> - Name of IV generator hash algorithm
6e7d01
   encrypt.key-secret=<str> - ID of secret providing qcow AES key or LUKS passphrase
6e7d01
   encryption=<bool (on/off)> - Encrypt the image with format 'aes'. (Deprecated in favor of encrypt.format=aes)
6e7d01
+  extent_size_hint=<size> - Extent size hint for the image file, 0 to disable
6e7d01
   lazy_refcounts=<bool (on/off)> - Postpone refcount updates
6e7d01
   nocow=<bool (on/off)>  - Turn off copy-on-write (valid only on btrfs)
6e7d01
   preallocation=<str>    - Preallocation mode (allowed values: off, metadata, falloc, full)
6e7d01
@@ -82,6 +83,7 @@ Supported options:
6e7d01
   encrypt.ivgen-hash-alg=<str> - Name of IV generator hash algorithm
6e7d01
   encrypt.key-secret=<str> - ID of secret providing qcow AES key or LUKS passphrase
6e7d01
   encryption=<bool (on/off)> - Encrypt the image with format 'aes'. (Deprecated in favor of encrypt.format=aes)
6e7d01
+  extent_size_hint=<size> - Extent size hint for the image file, 0 to disable
6e7d01
   lazy_refcounts=<bool (on/off)> - Postpone refcount updates
6e7d01
   nocow=<bool (on/off)>  - Turn off copy-on-write (valid only on btrfs)
6e7d01
   preallocation=<str>    - Preallocation mode (allowed values: off, metadata, falloc, full)
6e7d01
@@ -105,6 +107,7 @@ Supported options:
6e7d01
   encrypt.ivgen-hash-alg=<str> - Name of IV generator hash algorithm
6e7d01
   encrypt.key-secret=<str> - ID of secret providing qcow AES key or LUKS passphrase
6e7d01
   encryption=<bool (on/off)> - Encrypt the image with format 'aes'. (Deprecated in favor of encrypt.format=aes)
6e7d01
+  extent_size_hint=<size> - Extent size hint for the image file, 0 to disable
6e7d01
   lazy_refcounts=<bool (on/off)> - Postpone refcount updates
6e7d01
   nocow=<bool (on/off)>  - Turn off copy-on-write (valid only on btrfs)
6e7d01
   preallocation=<str>    - Preallocation mode (allowed values: off, metadata, falloc, full)
6e7d01
@@ -128,6 +131,7 @@ Supported options:
6e7d01
   encrypt.ivgen-hash-alg=<str> - Name of IV generator hash algorithm
6e7d01
   encrypt.key-secret=<str> - ID of secret providing qcow AES key or LUKS passphrase
6e7d01
   encryption=<bool (on/off)> - Encrypt the image with format 'aes'. (Deprecated in favor of encrypt.format=aes)
6e7d01
+  extent_size_hint=<size> - Extent size hint for the image file, 0 to disable
6e7d01
   lazy_refcounts=<bool (on/off)> - Postpone refcount updates
6e7d01
   nocow=<bool (on/off)>  - Turn off copy-on-write (valid only on btrfs)
6e7d01
   preallocation=<str>    - Preallocation mode (allowed values: off, metadata, falloc, full)
6e7d01
@@ -151,6 +155,7 @@ Supported options:
6e7d01
   encrypt.ivgen-hash-alg=<str> - Name of IV generator hash algorithm
6e7d01
   encrypt.key-secret=<str> - ID of secret providing qcow AES key or LUKS passphrase
6e7d01
   encryption=<bool (on/off)> - Encrypt the image with format 'aes'. (Deprecated in favor of encrypt.format=aes)
6e7d01
+  extent_size_hint=<size> - Extent size hint for the image file, 0 to disable
6e7d01
   lazy_refcounts=<bool (on/off)> - Postpone refcount updates
6e7d01
   nocow=<bool (on/off)>  - Turn off copy-on-write (valid only on btrfs)
6e7d01
   preallocation=<str>    - Preallocation mode (allowed values: off, metadata, falloc, full)
6e7d01
@@ -174,6 +179,7 @@ Supported options:
6e7d01
   encrypt.ivgen-hash-alg=<str> - Name of IV generator hash algorithm
6e7d01
   encrypt.key-secret=<str> - ID of secret providing qcow AES key or LUKS passphrase
6e7d01
   encryption=<bool (on/off)> - Encrypt the image with format 'aes'. (Deprecated in favor of encrypt.format=aes)
6e7d01
+  extent_size_hint=<size> - Extent size hint for the image file, 0 to disable
6e7d01
   lazy_refcounts=<bool (on/off)> - Postpone refcount updates
6e7d01
   nocow=<bool (on/off)>  - Turn off copy-on-write (valid only on btrfs)
6e7d01
   preallocation=<str>    - Preallocation mode (allowed values: off, metadata, falloc, full)
6e7d01
@@ -197,6 +203,7 @@ Supported options:
6e7d01
   encrypt.ivgen-hash-alg=<str> - Name of IV generator hash algorithm
6e7d01
   encrypt.key-secret=<str> - ID of secret providing qcow AES key or LUKS passphrase
6e7d01
   encryption=<bool (on/off)> - Encrypt the image with format 'aes'. (Deprecated in favor of encrypt.format=aes)
6e7d01
+  extent_size_hint=<size> - Extent size hint for the image file, 0 to disable
6e7d01
   lazy_refcounts=<bool (on/off)> - Postpone refcount updates
6e7d01
   nocow=<bool (on/off)>  - Turn off copy-on-write (valid only on btrfs)
6e7d01
   preallocation=<str>    - Preallocation mode (allowed values: off, metadata, falloc, full)
6e7d01
@@ -220,6 +227,7 @@ Supported options:
6e7d01
   encrypt.ivgen-hash-alg=<str> - Name of IV generator hash algorithm
6e7d01
   encrypt.key-secret=<str> - ID of secret providing qcow AES key or LUKS passphrase
6e7d01
   encryption=<bool (on/off)> - Encrypt the image with format 'aes'. (Deprecated in favor of encrypt.format=aes)
6e7d01
+  extent_size_hint=<size> - Extent size hint for the image file, 0 to disable
6e7d01
   lazy_refcounts=<bool (on/off)> - Postpone refcount updates
6e7d01
   nocow=<bool (on/off)>  - Turn off copy-on-write (valid only on btrfs)
6e7d01
   preallocation=<str>    - Preallocation mode (allowed values: off, metadata, falloc, full)
6e7d01
@@ -339,6 +347,7 @@ Supported options:
6e7d01
   encrypt.ivgen-hash-alg=<str> - Name of IV generator hash algorithm
6e7d01
   encrypt.key-secret=<str> - ID of secret providing qcow AES key or LUKS passphrase
6e7d01
   encryption=<bool (on/off)> - Encrypt the image with format 'aes'. (Deprecated in favor of encrypt.format=aes)
6e7d01
+  extent_size_hint=<size> - Extent size hint for the image file, 0 to disable
6e7d01
   lazy_refcounts=<bool (on/off)> - Postpone refcount updates
6e7d01
   nocow=<bool (on/off)>  - Turn off copy-on-write (valid only on btrfs)
6e7d01
   preallocation=<str>    - Preallocation mode (allowed values: off, metadata, falloc, full)
6e7d01
@@ -362,6 +371,7 @@ Supported options:
6e7d01
   encrypt.ivgen-hash-alg=<str> - Name of IV generator hash algorithm
6e7d01
   encrypt.key-secret=<str> - ID of secret providing qcow AES key or LUKS passphrase
6e7d01
   encryption=<bool (on/off)> - Encrypt the image with format 'aes'. (Deprecated in favor of encrypt.format=aes)
6e7d01
+  extent_size_hint=<size> - Extent size hint for the image file, 0 to disable
6e7d01
   lazy_refcounts=<bool (on/off)> - Postpone refcount updates
6e7d01
   nocow=<bool (on/off)>  - Turn off copy-on-write (valid only on btrfs)
6e7d01
   preallocation=<str>    - Preallocation mode (allowed values: off, metadata, falloc, full)
6e7d01
@@ -385,6 +395,7 @@ Supported options:
6e7d01
   encrypt.ivgen-hash-alg=<str> - Name of IV generator hash algorithm
6e7d01
   encrypt.key-secret=<str> - ID of secret providing qcow AES key or LUKS passphrase
6e7d01
   encryption=<bool (on/off)> - Encrypt the image with format 'aes'. (Deprecated in favor of encrypt.format=aes)
6e7d01
+  extent_size_hint=<size> - Extent size hint for the image file, 0 to disable
6e7d01
   lazy_refcounts=<bool (on/off)> - Postpone refcount updates
6e7d01
   nocow=<bool (on/off)>  - Turn off copy-on-write (valid only on btrfs)
6e7d01
   preallocation=<str>    - Preallocation mode (allowed values: off, metadata, falloc, full)
6e7d01
@@ -408,6 +419,7 @@ Supported options:
6e7d01
   encrypt.ivgen-hash-alg=<str> - Name of IV generator hash algorithm
6e7d01
   encrypt.key-secret=<str> - ID of secret providing qcow AES key or LUKS passphrase
6e7d01
   encryption=<bool (on/off)> - Encrypt the image with format 'aes'. (Deprecated in favor of encrypt.format=aes)
6e7d01
+  extent_size_hint=<size> - Extent size hint for the image file, 0 to disable
6e7d01
   lazy_refcounts=<bool (on/off)> - Postpone refcount updates
6e7d01
   nocow=<bool (on/off)>  - Turn off copy-on-write (valid only on btrfs)
6e7d01
   preallocation=<str>    - Preallocation mode (allowed values: off, metadata, falloc, full)
6e7d01
@@ -431,6 +443,7 @@ Supported options:
6e7d01
   encrypt.ivgen-hash-alg=<str> - Name of IV generator hash algorithm
6e7d01
   encrypt.key-secret=<str> - ID of secret providing qcow AES key or LUKS passphrase
6e7d01
   encryption=<bool (on/off)> - Encrypt the image with format 'aes'. (Deprecated in favor of encrypt.format=aes)
6e7d01
+  extent_size_hint=<size> - Extent size hint for the image file, 0 to disable
6e7d01
   lazy_refcounts=<bool (on/off)> - Postpone refcount updates
6e7d01
   nocow=<bool (on/off)>  - Turn off copy-on-write (valid only on btrfs)
6e7d01
   preallocation=<str>    - Preallocation mode (allowed values: off, metadata, falloc, full)
6e7d01
@@ -454,6 +467,7 @@ Supported options:
6e7d01
   encrypt.ivgen-hash-alg=<str> - Name of IV generator hash algorithm
6e7d01
   encrypt.key-secret=<str> - ID of secret providing qcow AES key or LUKS passphrase
6e7d01
   encryption=<bool (on/off)> - Encrypt the image with format 'aes'. (Deprecated in favor of encrypt.format=aes)
6e7d01
+  extent_size_hint=<size> - Extent size hint for the image file, 0 to disable
6e7d01
   lazy_refcounts=<bool (on/off)> - Postpone refcount updates
6e7d01
   nocow=<bool (on/off)>  - Turn off copy-on-write (valid only on btrfs)
6e7d01
   preallocation=<str>    - Preallocation mode (allowed values: off, metadata, falloc, full)
6e7d01
@@ -477,6 +491,7 @@ Supported options:
6e7d01
   encrypt.ivgen-hash-alg=<str> - Name of IV generator hash algorithm
6e7d01
   encrypt.key-secret=<str> - ID of secret providing qcow AES key or LUKS passphrase
6e7d01
   encryption=<bool (on/off)> - Encrypt the image with format 'aes'. (Deprecated in favor of encrypt.format=aes)
6e7d01
+  extent_size_hint=<size> - Extent size hint for the image file, 0 to disable
6e7d01
   lazy_refcounts=<bool (on/off)> - Postpone refcount updates
6e7d01
   nocow=<bool (on/off)>  - Turn off copy-on-write (valid only on btrfs)
6e7d01
   preallocation=<str>    - Preallocation mode (allowed values: off, metadata, falloc, full)
6e7d01
@@ -500,6 +515,7 @@ Supported options:
6e7d01
   encrypt.ivgen-hash-alg=<str> - Name of IV generator hash algorithm
6e7d01
   encrypt.key-secret=<str> - ID of secret providing qcow AES key or LUKS passphrase
6e7d01
   encryption=<bool (on/off)> - Encrypt the image with format 'aes'. (Deprecated in favor of encrypt.format=aes)
6e7d01
+  extent_size_hint=<size> - Extent size hint for the image file, 0 to disable
6e7d01
   lazy_refcounts=<bool (on/off)> - Postpone refcount updates
6e7d01
   nocow=<bool (on/off)>  - Turn off copy-on-write (valid only on btrfs)
6e7d01
   preallocation=<str>    - Preallocation mode (allowed values: off, metadata, falloc, full)
6e7d01
diff --git a/tests/qemu-iotests/106 b/tests/qemu-iotests/106
6e7d01
index ac47eaa0f5..ee6f51d08b 100755
6e7d01
--- a/tests/qemu-iotests/106
6e7d01
+++ b/tests/qemu-iotests/106
6e7d01
@@ -51,7 +51,10 @@ for create_mode in off falloc full; do
6e7d01
         echo
6e7d01
         echo "--- create_mode=$create_mode growth_mode=$growth_mode ---"
6e7d01
 
6e7d01
-        IMGOPTS="preallocation=$create_mode" _make_test_img ${CREATION_SIZE}K
6e7d01
+        # Our calculation below assumes kilobytes as unit for the actual size.
6e7d01
+        # Disable the extent size hint because it would give us a result in
6e7d01
+        # megabytes.
6e7d01
+        IMGOPTS="preallocation=$create_mode,extent_size_hint=0" _make_test_img ${CREATION_SIZE}K
6e7d01
         $QEMU_IMG resize -f "$IMGFMT" --preallocation=$growth_mode "$TEST_IMG" +${GROWTH_SIZE}K
6e7d01
 
6e7d01
         expected_size=0
6e7d01
@@ -98,7 +101,7 @@ for growth_mode in falloc full; do
6e7d01
     # plain int.  We should use the correct type for the result, and
6e7d01
     # this tests we do.
6e7d01
 
6e7d01
-    _make_test_img 2G
6e7d01
+    _make_test_img -o "extent_size_hint=0" 2G
6e7d01
     $QEMU_IMG resize -f "$IMGFMT" --preallocation=$growth_mode "$TEST_IMG" +${GROWTH_SIZE}K
6e7d01
 
6e7d01
     actual_size=$($QEMU_IMG info -f "$IMGFMT" "$TEST_IMG" | grep 'disk size')
6e7d01
diff --git a/tests/qemu-iotests/175 b/tests/qemu-iotests/175
6e7d01
index 55db2803ed..8a8494aeb6 100755
6e7d01
--- a/tests/qemu-iotests/175
6e7d01
+++ b/tests/qemu-iotests/175
6e7d01
@@ -89,20 +89,20 @@ min_blocks=$(stat -c '%b' "$TEST_DIR/empty")
6e7d01
 
6e7d01
 echo
6e7d01
 echo "== creating image with default preallocation =="
6e7d01
-_make_test_img $size | _filter_imgfmt
6e7d01
+_make_test_img -o extent_size_hint=0 $size | _filter_imgfmt
6e7d01
 stat -c "size=%s, blocks=%b" $TEST_IMG | _filter_blocks $extra_blocks $min_blocks $size
6e7d01
 
6e7d01
 for mode in off full falloc; do
6e7d01
     echo
6e7d01
     echo "== creating image with preallocation $mode =="
6e7d01
-    IMGOPTS=preallocation=$mode _make_test_img $size | _filter_imgfmt
6e7d01
+    IMGOPTS="preallocation=$mode,extent_size_hint=0" _make_test_img $size | _filter_imgfmt
6e7d01
     stat -c "size=%s, blocks=%b" $TEST_IMG | _filter_blocks $extra_blocks $min_blocks $size
6e7d01
 done
6e7d01
 
6e7d01
 for new_size in 4096 1048576; do
6e7d01
     echo
6e7d01
     echo "== resize empty image with block_resize =="
6e7d01
-    _make_test_img 0 | _filter_imgfmt
6e7d01
+    _make_test_img -o extent_size_hint=0 0 | _filter_imgfmt
6e7d01
     _block_resize $TEST_IMG $new_size >/dev/null
6e7d01
     stat -c "size=%s, blocks=%b" $TEST_IMG | _filter_blocks $extra_blocks $min_blocks $new_size
6e7d01
 done
6e7d01
diff --git a/tests/qemu-iotests/243 b/tests/qemu-iotests/243
6e7d01
index e563761307..104c7256c4 100755
6e7d01
--- a/tests/qemu-iotests/243
6e7d01
+++ b/tests/qemu-iotests/243
6e7d01
@@ -47,7 +47,7 @@ for mode in off metadata falloc full; do
6e7d01
     echo "=== preallocation=$mode ==="
6e7d01
     echo
6e7d01
 
6e7d01
-    IMGOPTS="preallocation=$mode" _make_test_img 64M
6e7d01
+    IMGOPTS="preallocation=$mode,extent_size_hint=0" _make_test_img 64M
6e7d01
 
6e7d01
     printf "File size: "
6e7d01
     du -b $TEST_IMG | cut -f1
6e7d01
@@ -64,7 +64,7 @@ for mode in off metadata falloc full; do
6e7d01
     echo "=== External data file: preallocation=$mode ==="
6e7d01
     echo
6e7d01
 
6e7d01
-    IMGOPTS="data_file=$TEST_IMG.data,preallocation=$mode" _make_test_img 64M
6e7d01
+    IMGOPTS="data_file=$TEST_IMG.data,preallocation=$mode,extent_size_hint=0" _make_test_img 64M
6e7d01
 
6e7d01
     echo -n "qcow2 file size: "
6e7d01
     du -b $TEST_IMG | cut -f1
6e7d01
@@ -75,7 +75,7 @@ for mode in off metadata falloc full; do
6e7d01
     echo -n "qcow2 disk usage: "
6e7d01
     [ $(du -B1 $TEST_IMG | cut -f1) -lt 1048576 ] && echo "low" || echo "high"
6e7d01
     echo -n "data disk usage:  "
6e7d01
-    [ $(du -B1 $TEST_IMG.data | cut -f1) -lt 1048576 ] && echo "low" || echo "high"
6e7d01
+    [ $(du -B1 $TEST_IMG.data | cut -f1) -lt 2097152 ] && echo "low" || echo "high"
6e7d01
 
6e7d01
 done
6e7d01
 
6e7d01
diff --git a/tests/qemu-iotests/common.filter b/tests/qemu-iotests/common.filter
6e7d01
index c8e8663665..f29c1d3238 100644
6e7d01
--- a/tests/qemu-iotests/common.filter
6e7d01
+++ b/tests/qemu-iotests/common.filter
6e7d01
@@ -146,6 +146,7 @@ _filter_img_create()
6e7d01
         -e "s# refcount_bits=[0-9]\\+##g" \
6e7d01
         -e "s# key-secret=[a-zA-Z0-9]\\+##g" \
6e7d01
         -e "s# iter-time=[0-9]\\+##g" \
6e7d01
+        -e "s# extent_size_hint=[0-9]\\+##g" \
6e7d01
         -e "s# force_size=\\(on\\|off\\)##g"
6e7d01
 }
6e7d01
 
6e7d01
-- 
6e7d01
2.27.0
6e7d01