On Mon, May 10, 2021 at 2:18 PM Leon Romanovsky <leon@xxxxxxxxxx> wrote: > > On Mon, May 10, 2021 at 02:06:55PM +0200, Haris Iqbal wrote: > > On Mon, May 10, 2021 at 1:53 PM Leon Romanovsky <leon@xxxxxxxxxx> wrote: > > > > > > On Mon, May 10, 2021 at 01:00:33PM +0200, Haris Iqbal wrote: > > > > On Sun, May 9, 2021 at 1:24 PM Leon Romanovsky <leon@xxxxxxxxxx> wrote: > > > > > > > > > > On Mon, May 03, 2021 at 01:48:01PM +0200, Gioh Kim wrote: > > > > > > From: Gioh Kim <gi-oh.kim@xxxxxxxxxxxxxxx> > > > > > > > > > > > > The queue_depth size is sent from server and > > > > > > server already checks validity of the value. > > > > > > > > > > Do you trust server? What will be if server is not reliable and sends > > > > > garbage? > > > > > > > > Hi Leon, > > > > > > > > The server code checks for the queue_depth before sending. If the > > > > server is really running malicious code, then the queue_depth is the > > > > last thing that the client needs to worry about. > > > > > > Like what? for an example? > > > > Like accessing compromised block devices. If the queue_depth is > > garbage, the client would fail at allocation with ENOMEM; thats it. > > The client will get wrong data, check it and discard. The case of ENOMEM triggered > by remote side is different. It can cause to DDOS on the client. True. This makes sense. Will drop this patch. Thanks. > > Thanks > > > > > > > > > Thanks
