mbox series

[v8,0/5] Fix PF/VF dependency issue

Message ID 20190905193146.90250-1-helgaas@kernel.org
Headers show
Series Fix PF/VF dependency issue | expand

Message

Bjorn Helgaas Sept. 5, 2019, 7:31 p.m. UTC
From: Bjorn Helgaas <bhelgaas@google.com>

The current implementation of ATS, PASID, and PRI does not handle VF
dependencies correctly.  These are essentially Kuppuswamy's patches; I
reordered and tweaked them slightly.

Please treat this as a proposal, not a done deal, and post a v9 with any
changes needed.

Changes since v7:
 * Moved PRI & PASID capability offset caching to last since they're just
   optimizations
 * Cached offsets at first use instead of adding _init() functions
 * Dropped complicated pci_prg_resp_pasid_required() #ifdefs by just
   searching for the capability again
 * Dropped EA/VF validation since it only enforces spec language without
   fixing any known issues

Changes since v6:
 * Removed locking interfaces used in v6.
 * Made necessary changes to PRI/PASID enable/disable calls to allow
   register changes only for PF.

Changes since v5:
 * Created new patches for PRI/PASID capability caching.
 * Removed individual locks (pri_lock, pasid_lock) and added common
   resource lock to synchronize PRI/PASID updates between PF/VF.
 * Addressed comments from Bjorn Helgaas.

Changes since v4:
 * Defined empty functions for pci_pri_init() and pci_pasid_init() for cases
   where CONFIG_PCI_PRI and CONFIG_PCI_PASID are not enabled.

Changes since v3:
 * Fixed critical path (lock context) in pci_restore_*_state functions.

Changes since v2:
 * Added locking mechanism to synchronize accessing PF registers in VF.
 * Removed spec compliance checks in patches.
 * Addressed comments from Bjorn Helgaas.

Changes since v1:
 * Added more details about the patches in commit log.
 * Removed bulk spec check patch.
 * Addressed comments from Bjorn Helgaas.


Link: https://lore.kernel.org/r/cover.1567029860.git.sathyanarayanan.kuppuswamy@linux.intel.com v7

Kuppuswamy Sathyanarayanan (5):
  PCI/ATS: Handle sharing of PF PRI Capability with all VFs
  PCI/ATS: Handle sharing of PF PASID Capability with all VFs
  PCI/ATS: Disable PF/VF ATS service independently
  PCI/ATS: Cache PRI Capability offset
  PCI/ATS: Cache PASID Capability offset

 drivers/pci/ats.c   | 153 +++++++++++++++++++++++++++-----------------
 include/linux/pci.h |   3 +-
 2 files changed, 96 insertions(+), 60 deletions(-)

Comments

Bjorn Helgaas Sept. 27, 2019, 10:30 a.m. UTC | #1
On Thu, Sep 05, 2019 at 02:31:41PM -0500, Bjorn Helgaas wrote:
> From: Bjorn Helgaas <bhelgaas@google.com>
> 
> The current implementation of ATS, PASID, and PRI does not handle VF
> dependencies correctly.  These are essentially Kuppuswamy's patches; I
> reordered and tweaked them slightly.
> 
> Please treat this as a proposal, not a done deal, and post a v9 with any
> changes needed.
> 
> Changes since v7:
>  * Moved PRI & PASID capability offset caching to last since they're just
>    optimizations
>  * Cached offsets at first use instead of adding _init() functions
>  * Dropped complicated pci_prg_resp_pasid_required() #ifdefs by just
>    searching for the capability again
>  * Dropped EA/VF validation since it only enforces spec language without
>    fixing any known issues
> 
> Changes since v6:
>  * Removed locking interfaces used in v6.
>  * Made necessary changes to PRI/PASID enable/disable calls to allow
>    register changes only for PF.
> 
> Changes since v5:
>  * Created new patches for PRI/PASID capability caching.
>  * Removed individual locks (pri_lock, pasid_lock) and added common
>    resource lock to synchronize PRI/PASID updates between PF/VF.
>  * Addressed comments from Bjorn Helgaas.
> 
> Changes since v4:
>  * Defined empty functions for pci_pri_init() and pci_pasid_init() for cases
>    where CONFIG_PCI_PRI and CONFIG_PCI_PASID are not enabled.
> 
> Changes since v3:
>  * Fixed critical path (lock context) in pci_restore_*_state functions.
> 
> Changes since v2:
>  * Added locking mechanism to synchronize accessing PF registers in VF.
>  * Removed spec compliance checks in patches.
>  * Addressed comments from Bjorn Helgaas.
> 
> Changes since v1:
>  * Added more details about the patches in commit log.
>  * Removed bulk spec check patch.
>  * Addressed comments from Bjorn Helgaas.
> 
> 
> Link: https://lore.kernel.org/r/cover.1567029860.git.sathyanarayanan.kuppuswamy@linux.intel.com v7
> 
> Kuppuswamy Sathyanarayanan (5):
>   PCI/ATS: Handle sharing of PF PRI Capability with all VFs
>   PCI/ATS: Handle sharing of PF PASID Capability with all VFs
>   PCI/ATS: Disable PF/VF ATS service independently
>   PCI/ATS: Cache PRI Capability offset
>   PCI/ATS: Cache PASID Capability offset
> 
>  drivers/pci/ats.c   | 153 +++++++++++++++++++++++++++-----------------
>  include/linux/pci.h |   3 +-
>  2 files changed, 96 insertions(+), 60 deletions(-)

Applied to pci/virtualization for v5.5.