On 09.10.2013 17:15, Michal Privoznik wrote:
On 10.09.2013 20:59, Oskari Saarenmaa wrote:
> This commit adds support for btrfs subvolumes as directory volumes. The
> directory storage pool can be used to manage btrfs subvolumes on an existing
> btrfs filesystem. The driver can create new blank subvolumes and snapshots
> of existing subvolumes as well as delete existing subvolumes.
>
> The subvolumes created are automatically made visible on the host side
> and can be attached to domains using the <filesystem> tags as defined in
> 'format domain' documentation.
>
> Subvolumes do not implement quotas at the moment because the current
> (btrfs-progs-0.20.rc1.20130501git7854c8b-4.fc20.x86_64) support for quota
> management in btrfs-progs is lacking the necessary features, for example
> it's not possible to see the quota assigned to a certain subvolume and
> usage information is only updated on syncfs(2). Quota support will be
> implemented once the tools gain the necessary features.
>
> Signed-off-by: Oskari Saarenmaa <os(a)ohmu.fi>
> ---
> configure.ac | 25 ++-
> docs/schemas/storagevol.rng | 1 +
> docs/storage.html.in | 30 ++-
> libvirt.spec.in | 4 +
> src/conf/storage_conf.c | 3 +
> src/conf/storage_conf.h | 1 +
> src/storage/storage_backend.c | 15 +-
> src/storage/storage_backend_fs.c | 222 ++++++++++++++++++++--
> src/util/virstoragefile.c | 4 +-
> src/util/virstoragefile.h | 1 +
> tests/storagevolxml2xmlin/vol-btrfs-snapshot.xml | 13 ++
> tests/storagevolxml2xmlin/vol-btrfs.xml | 9 +
> tests/storagevolxml2xmlout/vol-btrfs-snapshot.xml | 26 +++
> tests/storagevolxml2xmlout/vol-btrfs.xml | 17 ++
> tests/storagevolxml2xmltest.c | 2 +
> 15 files changed, 343 insertions(+), 30 deletions(-)
> create mode 100644 tests/storagevolxml2xmlin/vol-btrfs-snapshot.xml
> create mode 100644 tests/storagevolxml2xmlin/vol-btrfs.xml
> create mode 100644 tests/storagevolxml2xmlout/vol-btrfs-snapshot.xml
> create mode 100644 tests/storagevolxml2xmlout/vol-btrfs.xml
>
Just to let you and everybody else's know - I'll wait for you to post v2
to this patch (probably worth resending the set) and push the whole new
patch set even though the 1/2 is ACKed. The rationale is, the
virDirIsExecutable is worth a separate patch - that is the new set will
have 3 patches (at least).
Michal