12.08.2019 17:48, Max Reitz wrote: > On 10.08.19 21:31, Vladimir Sementsov-Ogievskiy wrote: >> copy_range ignores these limitations, let's improve it. >> >> Signed-off-by: Vladimir Sementsov-Ogievskiy <vsement...@virtuozzo.com> >> --- >> block/io.c | 44 ++++++++++++++++++++++++++++++++++++-------- >> 1 file changed, 36 insertions(+), 8 deletions(-) > > Reviewed-by: Max Reitz <mre...@redhat.com> >
Hmm. Now I think that next patch is arguable, and I still don't see true way of organizing limitation of request length and memory allocation in conjunction with async requests in backup. So, I'll send next version of "improvements" without this (there are already enough simpler patches). And this patch becomes something separate. Do you think we need it anyway? If yes, please queue it in separate. It may be better to return ENOTSUP on too big requests too, to keep simpler code and make callers optimize their copying loops by themselves. -- Best regards, Vladimir
