On 2017-10-24 10:56, Anton Nefedov wrote: > during discard, signed integer overflow leads to end_offset possibly > be less than offset, > that in turn makes nb_clusters (to discard) much bigger than expected, > leading to a possible data loss, operation taking very long and > image growing large as discard will allocate L2 tables for all that > huge amount of clusters > > One possibility where nb_sectors is large enough (>=2^21) is vm_state > discard at snapshot creation. > > Fixed on master in 2.10 with: > > d2cb36af2b0040d421b347e6e4e803e07220f78d > Author: Eric Blake <ebl...@redhat.com> > Date: Sat May 6 19:05:52 2017 -0500 > > qcow2: Discard/zero clusters by byte count > > Signed-off-by: Anton Nefedov <anton.nefe...@virtuozzo.com> > Signed-off-by: Denis V. Lunev <d...@openvz.org> > --- > block/qcow2.h | 6 +++--- > block/qcow2-cluster.c | 6 +++--- > 2 files changed, 6 insertions(+), 6 deletions(-)
FWIW: Reviewed-by: Max Reitz <mre...@redhat.com>
signature.asc
Description: OpenPGP digital signature