On 2020/1/10 下午6:08, Jonathan Cameron wrote:
On Fri, 10 Jan 2020 15:03:25 +0800
zhangfei <zhangfei.gao@xxxxxxxxxx> wrote:
On 2020/1/10 上午1:49, Jonathan Cameron wrote:
On Mon, 16 Dec 2019 11:08:13 +0800
Zhangfei Gao <zhangfei.gao@xxxxxxxxxx> wrote:
Uacce (Unified/User-space-access-intended Accelerator Framework) targets to
provide Shared Virtual Addressing (SVA) between accelerators and processes.
So accelerator can access any data structure of the main cpu.
This differs from the data sharing between cpu and io device, which share
data content rather than address.
Because of unified address, hardware and user space of process can share
the same virtual address in the communication.
Uacce is intended to be used with Jean Philippe Brucker's SVA
patchset[1], which enables IO side page fault and PASID support.
We have keep verifying with Jean's sva patchset [2]
We also keep verifying with Eric's SMMUv3 Nested Stage patches [3]
Hi Zhangfei Gao,
Just to check my understanding...
This patch set is not dependent on either 2 or 3?
To use it on our hardware, we need 2, but the interfaces used are already
upstream, so this could move forwards in parallel.
Yes,
patch 1, 2 is for uacce.
patch 3, 4 is an example using uacce, which happen to be crypto.
Sorry, I wasn't clear enough.
Question is whether we need Jean's sva patch set [2] to merge this?
Since the sva api are already merged, the patchset can be build without
problem, so no dependency.
Though in order to make sva work, we need Jean's sva patch together,
which is in reviewing
https://lore.kernel.org/linux-iommu/20191219163033.2608177-1-jean-philippe@xxxxxxxxxx/
And with this patchset, we can also easier verify sva.
So I think they can go independently.
Thanks
