On Wed 26 Aug 2015 04:53:06 PM CEST, Max Reitz wrote:
Yet another thing is the problem described in the patch's commit
message. Why and how is the driver option inherited by the snapshot?
I think you're right and my description was wrong, this happens before
the snapshot is created, when
On 25.08.2015 09:03, Alberto Garcia wrote:
On Mon 24 Aug 2015 08:54:56 PM CEST, Max Reitz wrote:
[bdrv_fill_options()]
User-specified options should always have precedence over any other
option. The thing is, we consider the filename to be specified by the
user.
For user-specified
On Mon 24 Aug 2015 08:54:56 PM CEST, Max Reitz wrote:
[bdrv_fill_options()]
User-specified options should always have precedence over any other
option. The thing is, we consider the filename to be specified by the
user.
For user-specified options like the lazy-refcounts case that I
If an image is opened with driver-specific options then attempting to
use snapshot_blkdev will fail with Driver specified twice.
The reason is that bs-filename is replaced with a full JSON object by
bdrv_refresh_filename() when such options are present:
-drive
-BEGIN PGP SIGNED MESSAGE-
Hash: SHA256
On 24.08.2015 15:05, Alberto Garcia wrote:
If an image is opened with driver-specific options then attempting
to use snapshot_blkdev will fail with Driver specified twice.
The reason is that bs-filename is replaced with a full JSON object
by