[PULL][SRU Zesty] PCI: Use relaxed ordering on platforms that support it

Previous Topic Next Topic
 
classic Classic list List threaded Threaded
3 messages Options
Reply | Threaded
Open this post in threaded view
|

[PULL][SRU Zesty] PCI: Use relaxed ordering on platforms that support it

dann frazier-4
BugLink: https://bugs.launchpad.net/bugs/1721365

Other than some minor context adjustments for Kconfig, these are clean
cherry-picks from upstream. See the above bug for details of the issue.

Tested w/ igbxe driver on a pair of HiSilicon D05 boards.

The following changes since commit 2124f29359ff51985f93defce3f73989fbe3b004:

  UBUNTU: Ubuntu-4.10.0-39.43 (2017-11-01 12:43:05 +0100)

are available in the Git repository at:

  git://git.launchpad.net/~dannf/ubuntu/+source/linux/+git/linux lp1721365-z

for you to fetch changes up to 2741a88b07f5f56790bed48dad39d52116c918ae:

  net: ixgbe: Use new PCI_DEV_FLAGS_NO_RELAXED_ORDERING flag (2017-11-03 12:32:56 -0600)

----------------------------------------------------------------
Casey Leedom (2):
      net/cxgb4: Use new PCI_DEV_FLAGS_NO_RELAXED_ORDERING flag
      net/cxgb4vf: Use new PCI_DEV_FLAGS_NO_RELAXED_ORDERING flag

Ding Tianhong (1):
      net: ixgbe: Use new PCI_DEV_FLAGS_NO_RELAXED_ORDERING flag

Thierry Reding (1):
      PCI: Allow PCI express root ports to find themselves

dingtianhong (4):
      PCI: Disable PCIe Relaxed Ordering if unsupported
      PCI: Disable Relaxed Ordering for some Intel processors
      PCI: Disable Relaxed Ordering Attributes for AMD A1100
      PCI: fix oops when try to find Root Port for a PCI device

 drivers/net/ethernet/chelsio/cxgb4/cxgb4.h         |  1 +
 drivers/net/ethernet/chelsio/cxgb4/cxgb4_main.c    | 23 ++++--
 drivers/net/ethernet/chelsio/cxgb4/sge.c           |  5 +-
 drivers/net/ethernet/chelsio/cxgb4vf/adapter.h     |  1 +
 .../net/ethernet/chelsio/cxgb4vf/cxgb4vf_main.c    | 18 +++++
 drivers/net/ethernet/chelsio/cxgb4vf/sge.c         |  3 +
 drivers/net/ethernet/intel/ixgbe/ixgbe_82598.c     | 22 ------
 drivers/net/ethernet/intel/ixgbe/ixgbe_common.c    | 19 -----
 drivers/pci/pci.c                                  |  2 +-
 drivers/pci/probe.c                                | 43 +++++++++++
 drivers/pci/quirks.c                               | 89 ++++++++++++++++++++++
 include/linux/pci.h                                |  3 +
 12 files changed, 179 insertions(+), 50 deletions(-)

--
kernel-team mailing list
[hidden email]
https://lists.ubuntu.com/mailman/listinfo/kernel-team
Reply | Threaded
Open this post in threaded view
|

ACK: [PULL][SRU Zesty] PCI: Use relaxed ordering on platforms that support it

Kleber Souza
On 11/08/17 00:03, dann frazier wrote:

> BugLink: https://bugs.launchpad.net/bugs/1721365
>
> Other than some minor context adjustments for Kconfig, these are clean
> cherry-picks from upstream. See the above bug for details of the issue.
>
> Tested w/ igbxe driver on a pair of HiSilicon D05 boards.
>
> The following changes since commit 2124f29359ff51985f93defce3f73989fbe3b004:
>
>   UBUNTU: Ubuntu-4.10.0-39.43 (2017-11-01 12:43:05 +0100)
>
> are available in the Git repository at:
>
>   git://git.launchpad.net/~dannf/ubuntu/+source/linux/+git/linux lp1721365-z
>
> for you to fetch changes up to 2741a88b07f5f56790bed48dad39d52116c918ae:
>
>   net: ixgbe: Use new PCI_DEV_FLAGS_NO_RELAXED_ORDERING flag (2017-11-03 12:32:56 -0600)
>
> ----------------------------------------------------------------
> Casey Leedom (2):
>       net/cxgb4: Use new PCI_DEV_FLAGS_NO_RELAXED_ORDERING flag
>       net/cxgb4vf: Use new PCI_DEV_FLAGS_NO_RELAXED_ORDERING flag
>
> Ding Tianhong (1):
>       net: ixgbe: Use new PCI_DEV_FLAGS_NO_RELAXED_ORDERING flag
>
> Thierry Reding (1):
>       PCI: Allow PCI express root ports to find themselves
>
> dingtianhong (4):
>       PCI: Disable PCIe Relaxed Ordering if unsupported
>       PCI: Disable Relaxed Ordering for some Intel processors
>       PCI: Disable Relaxed Ordering Attributes for AMD A1100
>       PCI: fix oops when try to find Root Port for a PCI device
>
>  drivers/net/ethernet/chelsio/cxgb4/cxgb4.h         |  1 +
>  drivers/net/ethernet/chelsio/cxgb4/cxgb4_main.c    | 23 ++++--
>  drivers/net/ethernet/chelsio/cxgb4/sge.c           |  5 +-
>  drivers/net/ethernet/chelsio/cxgb4vf/adapter.h     |  1 +
>  .../net/ethernet/chelsio/cxgb4vf/cxgb4vf_main.c    | 18 +++++
>  drivers/net/ethernet/chelsio/cxgb4vf/sge.c         |  3 +
>  drivers/net/ethernet/intel/ixgbe/ixgbe_82598.c     | 22 ------
>  drivers/net/ethernet/intel/ixgbe/ixgbe_common.c    | 19 -----
>  drivers/pci/pci.c                                  |  2 +-
>  drivers/pci/probe.c                                | 43 +++++++++++
>  drivers/pci/quirks.c                               | 89 ++++++++++++++++++++++
>  include/linux/pci.h                                |  3 +
>  12 files changed, 179 insertions(+), 50 deletions(-)
>

The changes look sane to me. My only concern is regarding the changes on
the PCI core, but since the bulk of the code is in Artful already and
fixes for the known issues are included on the patch set, I believe the
regression potential is low.

Acked-by: Kleber Sacilotto de Souza <[hidden email]>

--
kernel-team mailing list
[hidden email]
https://lists.ubuntu.com/mailman/listinfo/kernel-team
Reply | Threaded
Open this post in threaded view
|

NACK: [PULL][SRU Zesty] PCI: Use relaxed ordering on platforms that support it

Stefan Bader-2
In reply to this post by dann frazier-4
On 08.11.2017 00:03, dann frazier wrote:

> BugLink: https://bugs.launchpad.net/bugs/1721365
>
> Other than some minor context adjustments for Kconfig, these are clean
> cherry-picks from upstream. See the above bug for details of the issue.
>
> Tested w/ igbxe driver on a pair of HiSilicon D05 boards.
>
> The following changes since commit 2124f29359ff51985f93defce3f73989fbe3b004:
>
>   UBUNTU: Ubuntu-4.10.0-39.43 (2017-11-01 12:43:05 +0100)
>
> are available in the Git repository at:
>
>   git://git.launchpad.net/~dannf/ubuntu/+source/linux/+git/linux lp1721365-z
>
> for you to fetch changes up to 2741a88b07f5f56790bed48dad39d52116c918ae:
>
>   net: ixgbe: Use new PCI_DEV_FLAGS_NO_RELAXED_ORDERING flag (2017-11-03 12:32:56 -0600)
>
> ----------------------------------------------------------------
> Casey Leedom (2):
>       net/cxgb4: Use new PCI_DEV_FLAGS_NO_RELAXED_ORDERING flag
>       net/cxgb4vf: Use new PCI_DEV_FLAGS_NO_RELAXED_ORDERING flag
>
> Ding Tianhong (1):
>       net: ixgbe: Use new PCI_DEV_FLAGS_NO_RELAXED_ORDERING flag
>
> Thierry Reding (1):
>       PCI: Allow PCI express root ports to find themselves
>
> dingtianhong (4):
>       PCI: Disable PCIe Relaxed Ordering if unsupported
>       PCI: Disable Relaxed Ordering for some Intel processors
>       PCI: Disable Relaxed Ordering Attributes for AMD A1100
>       PCI: fix oops when try to find Root Port for a PCI device
>
>  drivers/net/ethernet/chelsio/cxgb4/cxgb4.h         |  1 +
>  drivers/net/ethernet/chelsio/cxgb4/cxgb4_main.c    | 23 ++++--
>  drivers/net/ethernet/chelsio/cxgb4/sge.c           |  5 +-
>  drivers/net/ethernet/chelsio/cxgb4vf/adapter.h     |  1 +
>  .../net/ethernet/chelsio/cxgb4vf/cxgb4vf_main.c    | 18 +++++
>  drivers/net/ethernet/chelsio/cxgb4vf/sge.c         |  3 +
>  drivers/net/ethernet/intel/ixgbe/ixgbe_82598.c     | 22 ------
>  drivers/net/ethernet/intel/ixgbe/ixgbe_common.c    | 19 -----
>  drivers/pci/pci.c                                  |  2 +-
>  drivers/pci/probe.c                                | 43 +++++++++++
>  drivers/pci/quirks.c                               | 89 ++++++++++++++++++++++
>  include/linux/pci.h                                |  3 +
>  12 files changed, 179 insertions(+), 50 deletions(-)
>
Since this is addressing performance issues rather than something not working, I
would see this as something that should not be picked up for Zesty. For one
thing it requires changes to common code which always bears some risk of
impacting other things. The other reason is that Zesty heads towards being EOLed
soon (Jan 2018) and we should be replacing the Zesty kernel by the Artful kernel
rather soonish. So anything performance or enablement like should rather get
applied there. For this case this even has the advantage of being much less
change there.

-Stefan


--
kernel-team mailing list
[hidden email]
https://lists.ubuntu.com/mailman/listinfo/kernel-team

signature.asc (836 bytes) Download Attachment