[Public]
-----Original Message----- From: Thomas Monjalon <tho...@monjalon.net> Sent: Friday, May 28, 2021 12:51 PM To: Somalapuram, Amaranath <amaranath.somalapu...@amd.com> Cc: David Marchand <david.march...@redhat.com>; dev <dev@dpdk.org>; Akhil Goyal <gak...@marvell.com>; Kumar, Ravi1 <ravi1.ku...@amd.com>; Song, Keesang <keesang.s...@amd.com> Subject: Re: [dpdk-dev] [PATCH v3] crypto/ccp: enable IOMMU for CCP [CAUTION: External Email] 28/05/2021 07:02, Somalapuram, Amaranath: > From: Thomas Monjalon <tho...@monjalon.net> > > 27/05/2021 15:24, David Marchand: > > > On Fri, Dec 25, 2020 at 9:06 AM <asoma...@amd.com> wrote: > > > > From: Amaranath Somalapuram <amaranath.somalapu...@amd.com> > > > > > > > > CCP use vdev framework, and vdev framework don’t support IOMMU. > > > > Adding custom IOMMU support for AMD CCP driver. > > > > > > I am currently looking at pci bus patches/cleanups. > > > I ended up looking at crypto/ccp. > > > This driver code contains a lot of features duplicated with the pci bus. > > > > > > Why is the ccp driver not a PCI driver? > > > > Indeed it looks abusing vdev. > > We should drop all the code duplicating the PCI bus driver. > > If nothing else is done, it would mean breaking the probing of this > > driver. > > > > Adding more people in Cc list to have a fix before it is broken, thanks. > > Enable IOMMU for vdev was not supported in DPDK. > I can remove all the duplicating code after I test the CCP with IOMMU. I think you didn't get it. It should not be a vdev. We want to switch the driver to a true PCI device, and remove all the code copied from the PCI bus driver. We will implement CCP as true PCI device.