On Thu, Feb 18, 2016 at 06:17:09PM +0300, Olga Krishtal wrote:
> On 18/02/16 16:57, Ján Tomko wrote:
> > On Wed, Feb 17, 2016 at 02:40:02PM +0300, Olga Krishtal wrote:
> >> In case of ploop volume, target path of the volume is the path to the
> >> directory that contains image file named root.hds and DiskDescriptor.xml.
> >> While using uploadVol and downloadVol callbacks we need to open root.hds
> >> itself. To accomplish this goal we must change path from
> >> path/to/ploop directory to path/to/ploop/root.hds
> >>
> >> In case of .uploadVol, we have to additionaly update DiskDescriptor.xml
> >>
> >> Signed-off-by: Olga Krishtal <okrishtal@xxxxxxxxxxxxx>
> >> ---
> >> src/storage/storage_backend.c | 66 +++++++++++++++++++++++++++++++++++++++++--
> >> 1 file changed, 64 insertions(+), 2 deletions(-)
> >>
> >> diff --git a/src/storage/storage_backend.c b/src/storage/storage_backend.c
> >> index 9f0e020..ac44fdf 100644
> >> --- a/src/storage/storage_backend.c
> >> +++ b/src/storage/storage_backend.c
> >> @@ -2023,12 +2023,63 @@ virStorageBackendVolUploadLocal(virConnectPtr conn ATTRIBUTE_UNUSED,
> >> unsigned long long len,
> >> unsigned int flags)
> >> {
> >> + char *path = NULL;
> >> + char *target_path = vol->target.path;
> >> + int ret;
> >> + virCommandPtr cmd = NULL;
> >> + char *create_tool = NULL;
> >> +
> >> virCheckFlags(0, -1);
> >>
> >> /* Not using O_CREAT because the file is required to already exist at
> >> * this point */
> >> - return virFDStreamOpenBlockDevice(stream, vol->target.path,
> >> + if (vol->target.format != VIR_STORAGE_FILE_PLOOP) {
> >> + return virFDStreamOpenBlockDevice(stream, target_path,
> >> offset, len, O_WRONLY);
> >> + } else {
> > Looking at the last patch, it seems a volume could be detected as
> > VIR_STORAGE_FILE_PLOOP if it's a disk image matching the magic,
> > but this code assumes it's a directory with "root.hds" image and the
> > XML.
> >
> >> + if (virAsprintf(&path, "%s/root.hds", vol->target.path) < 0)
> >> + return -1;
> > I thought the target.path was already pointing to the image, not
> > directory.
> >
> > Jan
> The thing is that that volume target.path is path to folder with
> root.hds and DiskDescriptor.xml.
I don't like VOL_TYPE_FILE pointing to a directory.
> And after resize we have to update DiskDescriptor.xml. This can be done
> only after volume will be fully
> upload, I mean after stream operations will be finished. Now I am
> looking for the correct place to do it.
If DiskDescriptor.xml contains the headers, then they should be updated
by the same tool that changes the data.
> I think about refreshPool. But not quite sure yet.
The purpose of refreshPool is to update libvirtd's knowledge of the
files by reading the on-disk metadata, not chaning the on-disk metadata.
Jan
--
libvir-list mailing list
libvir-list@xxxxxxxxxx
https://www.redhat.com/mailman/listinfo/libvir-list