On Wed, Apr 05, 2017 at 02:26:53PM -0500, Eric Blake wrote: > On 04/05/2017 02:20 PM, John Snow wrote: > > > Conceptually straightforward. > > > > looks like this might change behavior for... RBD and vvfat, right? > > RBD is the subject of this series so we'll just assume that was broken > > and stupid. > >
Yes on RBD, and that change is intentional. > > What's vvfat's story? It always set the read-only property to false > > regardless of what you asked for? > > vvfat is even stupider than that - it has its own independent property > 'rw' that determines whether to allow write operations, separate from > the inherited BDS readonly property. > Yes, it is very odd. But if we have copy_on_read enabled, or explicitly set the block device to read-only via QAPI or -drive, I think that those should take precedence.
