[ Cc: Fam, qemu-stable ] Am 25.11.2016 um 11:06 hat QingFeng Hao geschrieben: > The problem was triggered by qemu-iotests case 055. It failed when it > was comparing the compressed vmdk image with original test.img. > > The cause is that buf_len in vmdk_write_extent wasn't converted to > little-endian before it was stored to disk. But later vmdk_read_extent > read it and converted it from little-endian to cpu endian. > If the cpu is big-endian like s390, the problem will happen and > the data length read by vmdk_read_extent will become invalid! > The fix is to add the conversion in vmdk_write_extent. > > Signed-off-by: QingFeng Hao <ha...@linux.vnet.ibm.com> > Signed-off-by: Jing Liu <liuj...@linux.vnet.ibm.com>
Sounds like something that should still be fixed for 2.8 and in the stable branches. > diff --git a/block/vmdk.c b/block/vmdk.c > index a11c27a..bf6667f 100644 > --- a/block/vmdk.c > +++ b/block/vmdk.c > @@ -1355,7 +1355,7 @@ static int vmdk_write_extent(VmdkExtent *extent, > int64_t cluster_offset, > } > > data->lba = offset >> BDRV_SECTOR_BITS; > - data->size = buf_len; > + data->size = cpu_to_le32(buf_len); At least data->lba needs to be fixed, too, both here and in vmdk_read_extent(). Host endianness in an image file is always wrong. Maybe we should audit the whole driver for endianness problems. Kevin
