nvdimm.lists.linux.dev archive mirror
 help / color / mirror / Atom feed
* [PATCH v3 00/11] Copy Offload in NVMe Fabrics with P2P PCI Memory
@ 2018-03-12 19:35 Logan Gunthorpe
  2018-03-12 19:35 ` [PATCH v3 01/11] PCI/P2PDMA: Support peer-to-peer memory Logan Gunthorpe
                   ` (10 more replies)
  0 siblings, 11 replies; 68+ messages in thread
From: Logan Gunthorpe @ 2018-03-12 19:35 UTC (permalink / raw)
  To: linux-kernel, linux-pci, linux-nvme, linux-rdma, linux-nvdimm,
	linux-block
  Cc: Jens Axboe, Benjamin Herrenschmidt, Alex Williamson, Keith Busch,
	Jérôme Glisse, Jason Gunthorpe, Bjorn Helgaas,
	Max Gurtovoy, Christoph Hellwig

Hi Everyone,

Here's v3 of our series to introduce P2P based copy offload to NVMe
fabrics. This version has been rebased onto v4.16-rc5.

Thanks,

Logan


Changes in v3:

* Many more fixes and minor cleanups that were spotted by Bjorn

* Additional explanation of the ACS change in both the commit message
  and Kconfig doc. Also, the code that disables the ACS bits is surrounded
  explicitly by an #ifdef

* Removed the flag we added to rdma_rw_ctx() in favour of using
  is_pci_p2pdma_page(), as suggested by Sagi.

* Adjust pci_p2pmem_find() so that it prefers P2P providers that
  are closest to (or the same as) the clients using them. In cases
  of ties, the provider is randomly chosen.

* Modify the NVMe Target code so that the PCI device name of the provider
  may be explicitly specified, bypassing the logic in pci_p2pmem_find().
  (Note: it's still enforced that the provider must be behind the
   same switch as the clients).

* As requested by Bjorn, added documentation for driver writers.


Changes in v2:

* Renamed everything to 'p2pdma' per the suggestion from Bjorn as well
  as a bunch of cleanup and spelling fixes he pointed out in the last
  series.

* To address Alex's ACS concerns, we change to a simpler method of
  just disabling ACS behind switches for any kernel that has
  CONFIG_PCI_P2PDMA.

* We also reject using devices that employ 'dma_virt_ops' which should
  fairly simply handle Jason's concerns that this work might break with
  the HFI, QIB and rxe drivers that use the virtual ops to implement
  their own special DMA operations.

--

This is a continuation of our work to enable using Peer-to-Peer PCI
memory in NVMe fabrics targets. Many thanks go to Christoph Hellwig who
provided valuable feedback to get these patches to where they are today.

The concept here is to use memory that's exposed on a PCI BAR as
data buffers in the NVME target code such that data can be transferred
from an RDMA NIC to the special memory and then directly to an NVMe
device avoiding system memory entirely. The upside of this is better
QoS for applications running on the CPU utilizing memory and lower
PCI bandwidth required to the CPU (such that systems could be designed
with fewer lanes connected to the CPU). However, presently, the
trade-off is currently a reduction in overall throughput. (Largely due
to hardware issues that would certainly improve in the future).

Due to these trade-offs we've designed the system to only enable using
the PCI memory in cases where the NIC, NVMe devices and memory are all
behind the same PCI switch. This will mean many setups that could likely
work well will not be supported so that we can be more confident it
will work and not place any responsibility on the user to understand
their topology. (We chose to go this route based on feedback we
received at the last LSF). Future work may enable these transfers behind
a fabric of PCI switches or perhaps using a white list of known good
root complexes.

In order to enable this functionality, we introduce a few new PCI
functions such that a driver can register P2P memory with the system.
Struct pages are created for this memory using devm_memremap_pages()
and the PCI bus offset is stored in the corresponding pagemap structure.

Another set of functions allow a client driver to create a list of
client devices that will be used in a given P2P transactions and then
use that list to find any P2P memory that is supported by all the
client devices. This list is then also used to selectively disable the
ACS bits for the downstream ports behind these devices.

In the block layer, we also introduce a P2P request flag to indicate a
given request targets P2P memory as well as a flag for a request queue
to indicate a given queue supports targeting P2P memory. P2P requests
will only be accepted by queues that support it. Also, P2P requests
are marked to not be merged seeing a non-homogenous request would
complicate the DMA mapping requirements.

In the PCI NVMe driver, we modify the existing CMB support to utilize
the new PCI P2P memory infrastructure and also add support for P2P
memory in its request queue. When a P2P request is received it uses the
pci_p2pmem_map_sg() function which applies the necessary transformation
to get the corrent pci_bus_addr_t for the DMA transactions.

In the RDMA core, we also adjust rdma_rw_ctx_init() and
rdma_rw_ctx_destroy() to take a flags argument which indicates whether
to use the PCI P2P mapping functions or not.

Finally, in the NVMe fabrics target port we introduce a new
configuration boolean: 'allow_p2pmem'. When set, the port will attempt
to find P2P memory supported by the RDMA NIC and all namespaces. If
supported memory is found, it will be used in all IO transfers. And if
a port is using P2P memory, adding new namespaces that are not supported
by that memory will fail.

Logan Gunthorpe (11):
  PCI/P2PDMA: Support peer-to-peer memory
  PCI/P2PDMA: Add sysfs group to display p2pmem stats
  PCI/P2PDMA: Add PCI p2pmem dma mappings to adjust the bus offset
  PCI/P2PDMA: Clear ACS P2P flags for all devices behind switches
  PCI/P2PDMA: Add P2P DMA driver writer's documentation
  block: Introduce PCI P2P flags for request and request queue
  IB/core: Ensure we map P2P memory correctly in
    rdma_rw_ctx_[init|destroy]()
  nvme-pci: Use PCI p2pmem subsystem to manage the CMB
  nvme-pci: Add support for P2P memory in requests
  nvme-pci: Add a quirk for a pseudo CMB
  nvmet: Optionally use PCI P2P memory

 Documentation/ABI/testing/sysfs-bus-pci |  25 +
 Documentation/PCI/index.rst             |  14 +
 Documentation/PCI/p2pdma.rst            | 164 +++++++
 Documentation/index.rst                 |   3 +-
 block/blk-core.c                        |   3 +
 drivers/infiniband/core/rw.c            |  13 +-
 drivers/nvme/host/core.c                |   4 +
 drivers/nvme/host/nvme.h                |   8 +
 drivers/nvme/host/pci.c                 | 118 +++--
 drivers/nvme/target/configfs.c          |  67 +++
 drivers/nvme/target/core.c              | 106 +++-
 drivers/nvme/target/io-cmd.c            |   3 +
 drivers/nvme/target/nvmet.h             |  12 +
 drivers/nvme/target/rdma.c              |  32 +-
 drivers/pci/Kconfig                     |  25 +
 drivers/pci/Makefile                    |   1 +
 drivers/pci/p2pdma.c                    | 828 ++++++++++++++++++++++++++++++++
 drivers/pci/pci.c                       |   6 +
 include/linux/blk_types.h               |  18 +-
 include/linux/blkdev.h                  |   3 +
 include/linux/memremap.h                |  19 +
 include/linux/pci-p2pdma.h              | 119 +++++
 include/linux/pci.h                     |   4 +
 23 files changed, 1543 insertions(+), 52 deletions(-)
 create mode 100644 Documentation/PCI/index.rst
 create mode 100644 Documentation/PCI/p2pdma.rst
 create mode 100644 drivers/pci/p2pdma.c
 create mode 100644 include/linux/pci-p2pdma.h

--
2.11.0
_______________________________________________
Linux-nvdimm mailing list
Linux-nvdimm@lists.01.org
https://lists.01.org/mailman/listinfo/linux-nvdimm

^ permalink raw reply	[flat|nested] 68+ messages in thread

end of thread, other threads:[~2018-04-13 21:56 UTC | newest]

Thread overview: 68+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2018-03-12 19:35 [PATCH v3 00/11] Copy Offload in NVMe Fabrics with P2P PCI Memory Logan Gunthorpe
2018-03-12 19:35 ` [PATCH v3 01/11] PCI/P2PDMA: Support peer-to-peer memory Logan Gunthorpe
2018-03-13  3:28   ` Sinan Kaya
2018-03-13 16:43     ` Logan Gunthorpe
2018-03-13 17:49       ` Sinan Kaya
2018-03-13 18:44         ` Logan Gunthorpe
2018-03-13 19:10           ` Sinan Kaya
2018-03-13 19:19             ` Logan Gunthorpe
2018-03-13 19:53               ` Sinan Kaya
2018-03-13 20:46                 ` Logan Gunthorpe
2018-03-13 21:22                   ` Sinan Kaya
2018-03-13 22:00                     ` Logan Gunthorpe
2018-03-13 22:29                       ` Sinan Kaya
2018-03-13 22:45                         ` Stephen  Bates
2018-03-13 22:48                         ` Logan Gunthorpe
2018-03-13 23:19                           ` Sinan Kaya
2018-03-13 23:45                             ` Logan Gunthorpe
2018-03-14 12:16                               ` David Laight
2018-03-14 16:23                                 ` Logan Gunthorpe
2018-03-13 22:31                       ` Stephen  Bates
2018-03-13 23:08                         ` Bjorn Helgaas
2018-03-13 23:21                           ` Logan Gunthorpe
2018-03-14  2:56                             ` Bjorn Helgaas
2018-03-14 14:05                               ` Stephen  Bates
2018-03-14 16:17                               ` Logan Gunthorpe
2018-03-14 18:51                                 ` Bjorn Helgaas
2018-03-14 19:03                                   ` Logan Gunthorpe
2018-03-14 19:28                                     ` Dan Williams
2018-03-14 19:30                                       ` Logan Gunthorpe
2018-03-14 19:34                                       ` Stephen  Bates
2018-03-15  4:00                                         ` Martin K. Petersen
2018-03-15  4:30                                         ` Dan Williams
2018-03-22 22:57                           ` Stephen  Bates
2018-03-23 21:50                             ` Bjorn Helgaas
2018-03-23 21:59                               ` Logan Gunthorpe
2018-03-24  3:49                                 ` Bjorn Helgaas
2018-03-24 15:28                                   ` Stephen  Bates
2018-03-26 15:43                                     ` Logan Gunthorpe
2018-03-26 11:11       ` Jonathan Cameron
2018-03-26 14:01         ` Bjorn Helgaas
2018-03-26 15:46           ` Logan Gunthorpe
2018-03-27  8:47             ` Jonathan Cameron
2018-03-27 15:37               ` Logan Gunthorpe
2018-04-13 21:56               ` Stephen  Bates
2018-03-26 16:41         ` Jason Gunthorpe
2018-03-26 17:30           ` Logan Gunthorpe
2018-03-26 19:35             ` Jason Gunthorpe
2018-03-26 20:42               ` Logan Gunthorpe
2018-03-13 18:40     ` Logan Gunthorpe
2018-03-12 19:35 ` [PATCH v3 02/11] PCI/P2PDMA: Add sysfs group to display p2pmem stats Logan Gunthorpe
2018-03-12 19:35 ` [PATCH v3 03/11] PCI/P2PDMA: Add PCI p2pmem dma mappings to adjust the bus offset Logan Gunthorpe
2018-03-12 19:35 ` [PATCH v3 04/11] PCI/P2PDMA: Clear ACS P2P flags for all devices behind switches Logan Gunthorpe
2018-03-12 19:35 ` [PATCH v3 05/11] PCI/P2PDMA: Add P2P DMA driver writer's documentation Logan Gunthorpe
2018-03-12 19:41   ` Jonathan Corbet
2018-03-12 21:18     ` Logan Gunthorpe
2018-03-12 19:35 ` [PATCH v3 06/11] block: Introduce PCI P2P flags for request and request queue Logan Gunthorpe
2018-03-21  9:27   ` Christoph Hellwig
2018-03-12 19:35 ` [PATCH v3 07/11] IB/core: Ensure we map P2P memory correctly in rdma_rw_ctx_[init|destroy]() Logan Gunthorpe
2018-03-21  9:27   ` Christoph Hellwig
2018-03-12 19:35 ` [PATCH v3 08/11] nvme-pci: Use PCI p2pmem subsystem to manage the CMB Logan Gunthorpe
2018-03-13  1:55   ` Sinan Kaya
2018-03-13  1:58     ` Sinan Kaya
2018-03-12 19:35 ` [PATCH v3 09/11] nvme-pci: Add support for P2P memory in requests Logan Gunthorpe
2018-03-21  9:23   ` Christoph Hellwig
2018-03-12 19:35 ` [PATCH v3 10/11] nvme-pci: Add a quirk for a pseudo CMB Logan Gunthorpe
2018-03-12 19:35 ` [PATCH v3 11/11] nvmet: Optionally use PCI P2P memory Logan Gunthorpe
2018-03-21  9:27   ` Christoph Hellwig
2018-03-21 16:52     ` Logan Gunthorpe

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).