From: Jason Gunthorpe <jgg@ziepe.ca>
To: Logan Gunthorpe <logang@deltatee.com>
Cc: Christoph Hellwig <hch@lst.de>,
linux-kernel@vger.kernel.org, linux-block@vger.kernel.org,
linux-nvme@lists.infradead.org, linux-pci@vger.kernel.org,
linux-rdma@vger.kernel.org, Jens Axboe <axboe@kernel.dk>,
Bjorn Helgaas <bhelgaas@google.com>,
Dan Williams <dan.j.williams@intel.com>,
Sagi Grimberg <sagi@grimberg.me>, Keith Busch <kbusch@kernel.org>,
Stephen Bates <sbates@raithlin.com>
Subject: Re: [RFC PATCH 00/28] Removing struct page from P2PDMA
Date: Fri, 28 Jun 2019 01:57:05 -0300 [thread overview]
Message-ID: <20190628045705.GD3705@ziepe.ca> (raw)
In-Reply-To: <4894142c-3233-a3bb-f9a3-4a4985136e9b@deltatee.com>
On Thu, Jun 27, 2019 at 10:49:43AM -0600, Logan Gunthorpe wrote:
> > I don't think a GPU/FPGA driver will be involved, this would enter the
> > block layer through the O_DIRECT path or something generic.. This the
> > general flow I was suggesting to Dan earlier
>
> I would say the O_DIRECT path has to somehow call into the driver
> backing the VMA to get an address to appropriate memory (in some way
> vaguely similar to how we were discussing at LSF/MM)
Maybe, maybe no. For something like VFIO the PTE already has the
correct phys_addr_t and we don't need to do anything..
For DEVICE_PRIVATE we need to get the phys_addr_t out - presumably
through a new pagemap op?
> If P2P can't be done at that point, then the provider driver would
> do the copy to system memory, in the most appropriate way, and
> return regular pages for O_DIRECT to submit to the block device.
That only makes sense for the migratable DEVICE_PRIVATE case, it
doesn't help the VFIO-like case, there you'd need to bounce buffer.
> >> I think it would be a larger layering violation to have the NVMe driver
> >> (for example) memcpy data off a GPU's bar during a dma_map step to
> >> support this bouncing. And it's even crazier to expect a DMA transfer to
> >> be setup in the map step.
> >
> > Why? Don't we already expect the DMA mapper to handle bouncing for
> > lots of cases, how is this case different? This is the best place to
> > place it to make it shared.
>
> This is different because it's special memory where the DMA mapper
> can't possibly know the best way to transfer the data.
Why not? If we have a 'bar info' structure that could have data
transfer op callbacks, infact, I think we might already have similar
callbacks for migrating to/from DEVICE_PRIVATE memory with DMA..
> One could argue that the hook to the GPU/FPGA driver could be in the
> mapping step but then we'd have to do lookups based on an address --
> where as the VMA could more easily have a hook back to whatever driver
> exported it.
The trouble with a VMA hook is that it is only really avaiable when
working with the VA, and it is not actually available during GUP, you
have to have a GUP-like thing such as hmm_range_snapshot that is
specifically VMA based. And it is certainly not available during dma_map.
When working with VMA's/etc it seems there are some good reasons to
drive things off of the PTE content (either via struct page & pgmap or
via phys_addr_t & barmap)
I think the best reason to prefer a uniform phys_addr_t is that it
does give us the option to copy the data to/from CPU memory. That
option goes away as soon as the bio sometimes provides a dma_addr_t.
At least for RDMA, we do have some cases (like siw/rxe, hfi) where
they sometimes need to do that copy. I suspect the block stack is
similar, in the general case.
Jason
next prev parent reply other threads:[~2019-06-28 4:57 UTC|newest]
Thread overview: 89+ messages / expand[flat|nested] mbox.gz Atom feed top
2019-06-20 16:12 [RFC PATCH 00/28] Removing struct page from P2PDMA Logan Gunthorpe
2019-06-20 16:12 ` [RFC PATCH 01/28] block: Introduce DMA direct request type Logan Gunthorpe
2019-06-20 16:12 ` [RFC PATCH 02/28] block: Add dma_vec structure Logan Gunthorpe
2019-06-20 16:12 ` [RFC PATCH 03/28] block: Warn on mis-use of dma-direct bios Logan Gunthorpe
2019-06-20 16:12 ` [RFC PATCH 04/28] block: Never bounce " Logan Gunthorpe
2019-06-20 17:23 ` Jason Gunthorpe
2019-06-20 18:38 ` Logan Gunthorpe
2019-06-20 16:12 ` [RFC PATCH 05/28] block: Skip dma-direct bios in bio_integrity_prep() Logan Gunthorpe
2019-06-20 16:12 ` [RFC PATCH 06/28] block: Support dma-direct bios in bio_advance_iter() Logan Gunthorpe
2019-06-20 16:12 ` [RFC PATCH 07/28] block: Use dma_vec length in bio_cur_bytes() for dma-direct bios Logan Gunthorpe
2019-06-20 16:12 ` [RFC PATCH 08/28] block: Introduce dmavec_phys_mergeable() Logan Gunthorpe
2019-06-20 16:12 ` [RFC PATCH 09/28] block: Introduce vec_gap_to_prev() Logan Gunthorpe
2019-06-20 16:12 ` [RFC PATCH 10/28] block: Create generic vec_split_segs() from bvec_split_segs() Logan Gunthorpe
2019-06-20 16:12 ` [RFC PATCH 11/28] block: Create blk_segment_split_ctx Logan Gunthorpe
2019-06-20 16:12 ` [RFC PATCH 12/28] block: Create helper for bvec_should_split() Logan Gunthorpe
2019-06-20 16:12 ` [RFC PATCH 13/28] block: Generalize bvec_should_split() Logan Gunthorpe
2019-06-20 16:12 ` [RFC PATCH 14/28] block: Support splitting dma-direct bios Logan Gunthorpe
2019-06-20 16:12 ` [RFC PATCH 15/28] block: Support counting dma-direct bio segments Logan Gunthorpe
2019-06-20 16:12 ` [RFC PATCH 16/28] block: Implement mapping dma-direct requests to SGs in blk_rq_map_sg() Logan Gunthorpe
2019-06-20 16:12 ` [RFC PATCH 17/28] block: Introduce queue flag to indicate support for dma-direct bios Logan Gunthorpe
2019-06-20 16:12 ` [RFC PATCH 18/28] block: Introduce bio_add_dma_addr() Logan Gunthorpe
2019-06-20 16:12 ` [RFC PATCH 19/28] nvme-pci: Support dma-direct bios Logan Gunthorpe
2019-06-20 16:12 ` [RFC PATCH 20/28] IB/core: Introduce API for initializing a RW ctx from a DMA address Logan Gunthorpe
2019-06-20 16:49 ` Jason Gunthorpe
2019-06-20 16:59 ` Logan Gunthorpe
2019-06-20 17:11 ` Jason Gunthorpe
2019-06-20 18:24 ` Logan Gunthorpe
2019-06-20 16:12 ` [RFC PATCH 21/28] nvmet: Split nvmet_bdev_execute_rw() into a helper function Logan Gunthorpe
2019-06-20 16:12 ` [RFC PATCH 22/28] nvmet: Use DMA addresses instead of struct pages for P2P Logan Gunthorpe
2019-06-20 16:12 ` [RFC PATCH 23/28] nvme-pci: Remove support for PCI_P2PDMA requests Logan Gunthorpe
2019-06-20 16:12 ` [RFC PATCH 24/28] block: Remove PCI_P2PDMA queue flag Logan Gunthorpe
2019-06-20 16:12 ` [RFC PATCH 25/28] IB/core: Remove P2PDMA mapping support in rdma_rw_ctx Logan Gunthorpe
2019-06-20 16:12 ` [RFC PATCH 26/28] PCI/P2PDMA: Remove SGL helpers Logan Gunthorpe
2019-06-20 16:12 ` [RFC PATCH 27/28] PCI/P2PDMA: Remove struct pages that back P2PDMA memory Logan Gunthorpe
2019-06-20 16:12 ` [RFC PATCH 28/28] memremap: Remove PCI P2PDMA page memory type Logan Gunthorpe
2019-06-20 18:45 ` [RFC PATCH 00/28] Removing struct page from P2PDMA Dan Williams
2019-06-20 19:33 ` Jason Gunthorpe
2019-06-20 20:18 ` Dan Williams
2019-06-20 20:51 ` Logan Gunthorpe
2019-06-21 17:47 ` Jason Gunthorpe
2019-06-21 17:54 ` Dan Williams
2019-06-24 7:31 ` Christoph Hellwig
2019-06-24 13:46 ` Jason Gunthorpe
2019-06-24 13:50 ` Christoph Hellwig
2019-06-24 13:55 ` Jason Gunthorpe
2019-06-24 16:53 ` Logan Gunthorpe
2019-06-24 18:16 ` Jason Gunthorpe
2019-06-24 18:28 ` Logan Gunthorpe
2019-06-24 18:54 ` Jason Gunthorpe
2019-06-24 19:37 ` Logan Gunthorpe
2019-06-24 16:10 ` Logan Gunthorpe
2019-06-25 7:18 ` Christoph Hellwig
2019-06-20 19:34 ` Logan Gunthorpe
2019-06-20 23:40 ` Dan Williams
2019-06-20 23:42 ` Logan Gunthorpe
2019-06-24 7:27 ` Christoph Hellwig
2019-06-24 16:07 ` Logan Gunthorpe
2019-06-25 7:20 ` Christoph Hellwig
2019-06-25 15:57 ` Logan Gunthorpe
2019-06-25 17:01 ` Christoph Hellwig
2019-06-25 19:54 ` Logan Gunthorpe
2019-06-26 6:57 ` Christoph Hellwig
2019-06-26 18:31 ` Logan Gunthorpe
2019-06-26 20:21 ` Jason Gunthorpe
2019-06-26 20:39 ` Dan Williams
2019-06-26 20:54 ` Jason Gunthorpe
2019-06-26 20:55 ` Logan Gunthorpe
2019-06-26 20:45 ` Logan Gunthorpe
2019-06-26 21:00 ` Jason Gunthorpe
2019-06-26 21:18 ` Logan Gunthorpe
2019-06-27 6:32 ` Jason Gunthorpe
2019-06-27 16:09 ` Logan Gunthorpe
2019-06-27 16:35 ` Jason Gunthorpe
2019-06-27 16:49 ` Logan Gunthorpe
2019-06-28 4:57 ` Jason Gunthorpe [this message]
2019-06-28 16:22 ` Logan Gunthorpe
2019-06-28 17:29 ` Jason Gunthorpe
2019-06-28 18:29 ` Logan Gunthorpe
2019-06-28 19:09 ` Jason Gunthorpe
2019-06-28 19:35 ` Logan Gunthorpe
2019-07-02 22:45 ` Jason Gunthorpe
2019-07-02 22:52 ` Logan Gunthorpe
2019-06-27 9:08 ` Christoph Hellwig
2019-06-27 16:30 ` Logan Gunthorpe
2019-06-27 17:00 ` Christoph Hellwig
2019-06-27 18:00 ` Logan Gunthorpe
2019-06-28 13:38 ` Christoph Hellwig
2019-06-28 15:54 ` Logan Gunthorpe
2019-06-27 9:01 ` Christoph Hellwig
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=20190628045705.GD3705@ziepe.ca \
--to=jgg@ziepe.ca \
--cc=axboe@kernel.dk \
--cc=bhelgaas@google.com \
--cc=dan.j.williams@intel.com \
--cc=hch@lst.de \
--cc=kbusch@kernel.org \
--cc=linux-block@vger.kernel.org \
--cc=linux-kernel@vger.kernel.org \
--cc=linux-nvme@lists.infradead.org \
--cc=linux-pci@vger.kernel.org \
--cc=linux-rdma@vger.kernel.org \
--cc=logang@deltatee.com \
--cc=sagi@grimberg.me \
--cc=sbates@raithlin.com \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).