It causes to a memory corruption in kernel. It has never been triggered since the user-space driver has never requested it on DCT (another bug). Corresponding to this fix, additional fix in user-space driver is introduced. > -----Original Message----- > From: linux-rdma-owner@xxxxxxxxxxxxxxx <linux-rdma-owner@xxxxxxxxxxxxxxx> > On Behalf Of Leon Romanovsky > Sent: Wednesday, April 10, 2019 10:20 PM > To: Jason Gunthorpe <jgg@xxxxxxxxxxxx> > Cc: Doug Ledford <dledford@xxxxxxxxxx>; Guy Levi(SW) > <guyle@xxxxxxxxxxxx>; RDMA mailing list <linux-rdma@xxxxxxxxxxxxxxx>; > Yonatan Cohen (SW) <yonatanc@xxxxxxxxxxxx> > Subject: Re: [PATCH rdma-rc] IB/mlx5: Fix scatter to CQE in DCT QP creation > > On Wed, Apr 10, 2019 at 06:13:14PM +0000, Jason Gunthorpe wrote: > > On Wed, Apr 10, 2019 at 10:59:45AM +0300, Leon Romanovsky wrote: > > > From: Guy Levi <guyle@xxxxxxxxxxxx> > > > > > > The scatter to CQE over DCT QP actually has never been enabled since > > > the driver refers to device QP context layout instead of device DCT > > > context layout when it is configure the scatter to CQE relevant > > > bits. It causes a memory corruption. > > > Additionally, A new capability is exposed to user-space which lets > > > it to ensure that it will not enable the feature without this fix. > > > > I'm not sure what this is trying to say, it needs to be reworded to be > > a -rc patch.. > > I'm fine with -next too. > > > > > Is it that mixing scatter to cqe and DCT causes memory corruption in > > the kernel? > > Not, it causes to memory corruption in application, because it assumes something > that it is not correct. Kernel doesn't use this knowledge. > > Thanks > > > > > Jason
