[PATCH v1 03/14] mm: add iomem vma selection for memory migration

Christoph Hellwig hch at lst.de
Thu Sep 2 08:18:26 UTC 2021


On Wed, Sep 01, 2021 at 11:40:43AM -0400, Felix Kuehling wrote:
> >>> It looks like I'm totally misunderstanding what you are adding here
> >>> then.  Why do we need any special treatment at all for memory that
> >>> has normal struct pages and is part of the direct kernel map?
> >> The pages are like normal memory for purposes of mapping them in CPU
> >> page tables and for coherent access from the CPU.
> > That's the user page tables.  What about the kernel direct map?
> > If there is a normal kernel struct page backing there really should
> > be no need for the pgmap.
> 
> I'm not sure. The physical address ranges are in the UEFI system address
> map as special-purpose memory. Does Linux create the struct pages and
> kernel direct map for that without a pgmap call? I didn't see that last
> time I went digging through that code.

So doing some googling finds a patch from Dan that claims to hand EFI
special purpose memory to the device dax driver.  But when I try to
follow the version that got merged it looks it is treated simply as an
MMIO region to be claimed by drivers, which would not get a struct page.

Dan, did I misunderstand how E820_TYPE_SOFT_RESERVED works?

> >> From an application
> >> perspective, we want file-backed and anonymous mappings to be able to
> >> use DEVICE_PUBLIC pages with coherent CPU access. The goal is to
> >> optimize performance for GPU heavy workloads while minimizing the need
> >> to migrate data back-and-forth between system memory and device memory.
> > I don't really understand that part.  file backed pages are always
> > allocated by the file system using the pagecache helpers, that is
> > using the page allocator.  Anonymouns memory also always comes from
> > the page allocator.
> 
> I'm coming at this from my experience with DEVICE_PRIVATE. Both
> anonymous and file-backed pages should be migrateable to DEVICE_PRIVATE
> memory by the migrate_vma_* helpers for more efficient access by our
> GPU. (*) It's part of the basic premise of HMM as I understand it. I
> would expect the same thing to work for DEVICE_PUBLIC memory.

Ok, so you want to migrate to and from them.  Not use DEVICE_PUBLIC
for the actual page cache pages.  That maks a lot more sense.

> I see DEVICE_PUBLIC as an improved version of DEVICE_PRIVATE that allows
> the CPU to map the device memory coherently to minimize the need for
> migrations when CPU and GPU access the same memory concurrently or
> alternatingly. But we're not going as far as putting that memory
> entirely under the management of the Linux memory manager and VM
> subsystem. Our (and HPE's) system architects decided that this memory is
> not suitable to be used like regular NUMA system memory by the Linux
> memory manager.

So yes.  It is a Memory Mapped I/O region, which unlike the PCIe BARs
that people typically deal with is fully cache coherent.  I think this
does make more sense as a description.

But to go back to what start this discussion:  If these are memory
mapped I/O pfn_valid should generally not return true for them.

And as you already pointed out in reply to Alex we need to tighten the
selection criteria one way or another.


More information about the amd-gfx mailing list