On Fri, 11/14 09:29, Stefan Hajnoczi wrote: > On Fri, Nov 14, 2014 at 12:09:21PM +0800, Fam Zheng wrote: > > When extent types don't match, we return -ENOTSUP. In this case, be > > polite to the caller and don't modify bdi. > > > > Signed-off-by: Fam Zheng <f...@redhat.com> > > --- > > block/vmdk.c | 20 +++++++++++++------- > > 1 file changed, 13 insertions(+), 7 deletions(-) > > At this stage of the release process it would be very helpful to include > some context for this patch: is it a pure cleanup or actually a bug fix > that needs to go into QEMU 2.2?
This one is more of a cleanup than bug fix: although there is a small logical change, it's only visible to user on abnormally structured VMDK (mixed extent type, etc.) that require manual editing image file, which is AFAIK not a valid use case and will not cause anything worse than an error. I sent the patch mostly for the function semantic purpose: don't change caller's data unless it's necessary to. > > Since you are a regular contributor and I therefore trust you, I have > merged this patch. But please be clear in commit descriptions about > whether or not a patch is a bug fix (how to reproduce the bug, if a > previous commit's regression is being fixed, etc). OK. I'll remember next time. Thanks for pointing out. Fam
