On Wed, Mar 12, 2014 at 12:16:04PM +0100, Kevin Wolf wrote: > Am 24.01.2014 um 14:48 hat Jeff Cody geschrieben: > > On Fri, Jan 24, 2014 at 02:33:19PM +0100, Kevin Wolf wrote: > > > Am 14.01.2014 um 20:12 hat Jeff Cody geschrieben: > > > > Having both read-only=on and snapshot=on together does not make sense; > > > > currently, the read-only argument is effectively ignored for the > > > > temporary snapshot. To prevent confusion, disallow the usage of both > > > > 'snapshot=on' and 'read-only=on'. > > > > > > > > Signed-off-by: Jeff Cody <jc...@redhat.com> > > > > > > This breaks in a surprising way: > > > > > > $ softmmu/qemu-system-x86_64 -drive file=~/images/hd.img,snapshot=on > > > ... works fine ... > > > > > > $ qemu-system-x86_64 -drive file=~/images/hd.img -snapshot > > > qemu-system-x86_64: invalid option combination: read-only and snapshot > > > > > > qemu-iotests case 051 catches this. I'll have to remove this patch and > > > the follow-up from the queue for now. > > > > > > > Odd - OK, I'll follow up on this and submit a series with both patches > > (well, likely squashed together), and whatever is needed to fix this > > as well. > > Jeff, what happened with this? I found the two patches in an old git > branch and wondered why they didn't disappear in the rebase. But > apparently we still allow read-only and snapshot at the same time. >
The test case 051 failed because when -snapshot was specified, it was enabled for all the default drives as well, which included default_cdrom. The solutions for that seemed a little hacky, and I wasn't sure it was actually worth it, given that this is probably not a real problem to begin with. If you think it is worth it, I can resurrect the patch and make sure it works with both snapshot=on and -snapshot.