Re: [PATCH 06/12] IB/core: Add optional PCI P2P flag to rdma_rw_ctx_[init|destroy]()

From: Jason Gunthorpe
Date: Mon Jan 08 2018 - 14:01:32 EST


On Mon, Jan 08, 2018 at 07:34:34PM +0100, Christoph Hellwig wrote:
> > > > And on that topic, does this scheme work with HFI?
> > >
> > > No, and I guess we need an opt-out. HFI generally seems to be
> > > extremely weird.
> >
> > This series needs some kind of fix so HFI, QIB, rxe, etc don't get
> > broken, and it shouldn't be 'fixed' at the RDMA level.
>
> I don't think rxe is a problem as it won't show up a pci device.

Right today's restrictions save us..

> HFI and QIB do show as PCI devices, and could be used for P2P transfers
> from the PCI point of view. It's just that they have a layer of
> software indirection between their hardware and what is exposed at
> the RDMA layer.
>
> So I very much disagree about where to place that workaround - the
> RDMA code is exactly the right place.

But why? RDMA is using core code to do this. It uses dma_ops in struct
device and it uses normal dma_map SG. How is it RDMA's problem that
some PCI drivers provide strange DMA ops?

Admittedly they are RDMA drivers, but it is a core mechanism they
(ab)use these days..

> > It could, if we had a DMA op for p2p then the drivers that provide
> > their own ops can implement it appropriately or not at all.
> >
> > Eg the correct implementation for rxe to support p2p memory is
> > probably somewhat straightfoward.
>
> But P2P is _not_ a factor of the dma_ops implementation at all,
> it is something that happens behind the dma_map implementation.

Only as long as the !ACS and switch limitations are present.

Those limitations are fine to get things started, but there is going
to a be a push improve the system to remove them.

> > Very long term the IOMMUs under the ops will need to care about this,
> > so the wrapper is not an optimal place to put it - but I wouldn't
> > object if it gets it out of RDMA :)
>
> Unless you have an IOMMU on your PCIe switch and not before/inside
> the root complex that is not correct.

I understand the proposed patches restrict things to require a switch
and not transit the IOMMU.

But *very long term* P2P will need to work with paths that transit the
system IOMMU and root complex.

This already exists as out-of-tree funtionality that has been deployed
in production for years and years that does P2P through the root
complex with the IOMMU turned off.

Jason