[Qemu-stable] [PATCH] problem may apparently compromise disk images

[Peter Mattern]

Running QEMU 2.4.0 on Arch Linux which is providing quite a vanilla 
build there's strong evidence that the problem addressed by this fix may 
compromise disk images:

Say a snapshot was created by 'qemu-img create -f qcow2 -b 
baseimage.qcow2 snapshot.qcow2'.
Then committing this snapshot may yield
     qemu-img commit snapshot.qcow2
     Co-routine re-entered recursively
     Aborted (core dumped)
If the base image gets launched thereafter (qemu-system-i386 [...] 
-snapshot [...]) there are lots of error messages like
     EXT4-fs error (device vda3): ext4_mb_generate_buddy:757: group 17, 
block \
     bitmap and bg descriptor inconsistent: 18082 vs 18059 free clusters
One time a VM even crashed during boot.

Neither those messages nor crashes were ever seen on any of the VMs in 
question, i686 and x86_64, running QEMU versions prior to 2.4.0 when 
exactly same actions were performed on the very same disk images.
After applying the fix proposed by this thread all problems cease.