[PATCH v3 01/11] PCI/P2PDMA: Support peer-to-peer memory

Jonathan Cameron Jonathan.Cameron at huawei.com
Mon Mar 26 04:11:38 PDT 2018


On Tue, 13 Mar 2018 10:43:55 -0600
Logan Gunthorpe <logang at deltatee.com> wrote:

> On 12/03/18 09:28 PM, Sinan Kaya wrote:
> > On 3/12/2018 3:35 PM, Logan Gunthorpe wrote:
> > Regarding the switch business, It is amazing how much trouble you went into
> > limit this functionality into very specific hardware.
> > 
> > I thought that we reached to an agreement that code would not impose
> > any limits on what user wants.
> > 
> > What happened to all the emails we exchanged?  
> 
> It turns out that root ports that support P2P are far less common than 
> anyone thought. So it will likely have to be a white list.

This came as a bit of a surprise to our PCIe architect.

His follow up was whether it was worth raising an ECR for the PCIe spec
to add a capability bit to allow this to be discovered.  This might
long term avoid the need to maintain the white list for new devices.

So is it worth having a long term solution for making this discoverable?

Jonathan

> Nobody else 
> seems keen on allowing the user to enable this on hardware that doesn't 
> work. The easiest solution is still limiting it to using a switch. From 
> there, if someone wants to start creating a white-list then that's 
> probably the way forward to support root ports.
> 
> And there's also the ACS problem which means if you want to use P2P on 
> the root ports you'll have to disable ACS on the entire system. (Or 
> preferably, the IOMMU groups need to get more sophisticated to allow for 
> dynamic changes).
> 
> Additionally, once you allow for root ports you may find the IOMMU 
> getting in the way.
> 
> So there are great deal more issues to sort out if you don't restrict to 
> devices behind switches.
> 
> Logan




More information about the Linux-nvme mailing list