Am 23.10.2014 um 00:18 hat Eric Blake geschrieben: > On 10/21/2014 03:24 AM, Max Reitz wrote: > > On 2014-10-21 at 10:04, Zhang Haoyu wrote: > >> Use local variable to bdrv_pwrite_sync L1 table, > >> needless to make conversion of cached L1 table between > >> big-endian and host style. > >> > >> Signed-off-by: Zhang Haoyu <zhan...@sangfor.com> > >> --- > >> block/qcow2-refcount.c | 22 +++++++--------------- > >> 1 file changed, 7 insertions(+), 15 deletions(-) > >> > > I know we're up to v5 and that Max already took it into his branch, but... > > > >> l1_size2 = l1_size * sizeof(uint64_t); > >> + l1_table = g_try_malloc0(align_offset(l1_size2, 512)); > > > > I wanted to propose using qemu_try_blockalign(), but since it'd require > > a memset() afterwards, it gets rather ugly. > > Not after this recent patch: > > https://lists.gnu.org/archive/html/qemu-devel/2014-10/msg02499.html
We can switch to qemu_try_blockalign0() in a follow-up patch. But this is far from being a fast path, so there is very little to gain anyway. Kevin
pgpUkQBKWDYRQ.pgp
Description: PGP signature