[PATCH v2 0/4] VMD fixups

[PATCH v2 0/4] VMD fixups

[Jon Derrick]

Mostly just cleanup in this revision, eg, trying to limit scope of vmd code to
x86

Previous:
https://patchwork.kernel.org/patch/9886095/
https://patchwork.kernel.org/patch/9886097/
https://patchwork.kernel.org/patch/9886101/


Jon Derrick (4):
  MAINTAINERS: Add Jonathan Derrick as VMD maintainer
  pci/x86: Move VMD quirks to x86 fixups
  x86/PCI: Use is_vmd rather than relying on the domain number
  iommu: Prevent VMD child devices from being remapping targets

 MAINTAINERS                 |  1 +
 arch/x86/pci/fixup.c        | 18 ++++++++++++++++++
 drivers/iommu/intel-iommu.c |  5 +++++
 drivers/pci/quirks.c        | 17 -----------------
 4 files changed, 24 insertions(+), 17 deletions(-)

-- 
2.9.4

[PATCH v2 1/4] MAINTAINERS: Add Jonathan Derrick as VMD maintainer

[Jon Derrick]

Add myself as VMD maintainer

Signed-off-by: Jon Derrick <jonathan.derrick-ral2JQCrhuEAvxtiuMwx3w@public.gmane.org>
Acked-by: Keith Busch <keith.busch-ral2JQCrhuEAvxtiuMwx3w@public.gmane.org>
---
 MAINTAINERS | 1 +
 1 file changed, 1 insertion(+)

diff --git a/MAINTAINERS b/MAINTAINERS
index f66488d..3ec39df 100644
--- a/MAINTAINERS
+++ b/MAINTAINERS
@@ -10090,6 +10090,7 @@ F:	drivers/pci/dwc/*imx6*
 
 PCI DRIVER FOR INTEL VOLUME MANAGEMENT DEVICE (VMD)
 M:	Keith Busch <keith.busch-ral2JQCrhuEAvxtiuMwx3w@public.gmane.org>
+M:	Jonathan Derrick <jonathan.derrick-ral2JQCrhuEAvxtiuMwx3w@public.gmane.org>
 L:	linux-pci-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
 S:	Supported
 F:	drivers/pci/host/vmd.c
-- 
2.9.4

[PATCH v2 3/4] x86/PCI: Use is_vmd rather than relying on the domain number

[Jon Derrick]

Signed-off-by: Jon Derrick <jonathan.derrick-ral2JQCrhuEAvxtiuMwx3w@public.gmane.org>
---
 arch/x86/pci/fixup.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/arch/x86/pci/fixup.c b/arch/x86/pci/fixup.c
index ca4b02e5..1ed2fbf 100644
--- a/arch/x86/pci/fixup.c
+++ b/arch/x86/pci/fixup.c
@@ -628,7 +628,7 @@ DECLARE_PCI_FIXUP_HEADER(PCI_VENDOR_ID_INTEL, 0x8c10, quirk_apple_mbp_poweroff);
 static void quirk_no_aersid(struct pci_dev *pdev)
 {
 	/* VMD Domain */
-	if (pdev->bus->sysdata && pci_domain_nr(pdev->bus) >= 0x10000)
+	if (is_vmd(pdev->bus))
 		pdev->bus->bus_flags |= PCI_BUS_FLAGS_NO_AERSID;
 }
 DECLARE_PCI_FIXUP_EARLY(PCI_VENDOR_ID_INTEL, 0x2030, quirk_no_aersid);
-- 
2.9.4

Re: [PATCH v2 0/4] VMD fixups

[Bjorn Helgaas]

[+cc Joerg]

On Thu, Aug 17, 2017 at 12:10:10PM -0600, Jon Derrick wrote:
> Mostly just cleanup in this revision, eg, trying to limit scope of vmd code to
> x86
> 
> Previous:
> https://patchwork.kernel.org/patch/9886095/
> https://patchwork.kernel.org/patch/9886097/
> https://patchwork.kernel.org/patch/9886101/
> 
> 
> Jon Derrick (4):
>   MAINTAINERS: Add Jonathan Derrick as VMD maintainer
>   pci/x86: Move VMD quirks to x86 fixups
>   x86/PCI: Use is_vmd rather than relying on the domain number
>   iommu: Prevent VMD child devices from being remapping targets
> 
>  MAINTAINERS                 |  1 +
>  arch/x86/pci/fixup.c        | 18 ++++++++++++++++++
>  drivers/iommu/intel-iommu.c |  5 +++++
>  drivers/pci/quirks.c        | 17 -----------------
>  4 files changed, 24 insertions(+), 17 deletions(-)

I haven't heard anything from IOMMU folks, so I applied these to
pci/host-vmd for v4.14.  Let me know if there's any objection.

[PATCH v2 2/4] pci/x86: Move VMD quirks to x86 fixups

[Jon Derrick]

VMD currently only exists for Intel x86 products

Signed-off-by: Jon Derrick <jonathan.derrick@intel.com>
---
 arch/x86/pci/fixup.c | 18 ++++++++++++++++++
 drivers/pci/quirks.c | 17 -----------------
 2 files changed, 18 insertions(+), 17 deletions(-)

diff --git a/arch/x86/pci/fixup.c b/arch/x86/pci/fixup.c
index 11e4074..ca4b02e5 100644
--- a/arch/x86/pci/fixup.c
+++ b/arch/x86/pci/fixup.c
@@ -618,3 +618,21 @@ static void quirk_apple_mbp_poweroff(struct pci_dev *pdev)
 		dev_info(dev, "can't work around MacBook Pro poweroff issue\n");
 }
 DECLARE_PCI_FIXUP_HEADER(PCI_VENDOR_ID_INTEL, 0x8c10, quirk_apple_mbp_poweroff);
+
+/*
+ * VMD-enabled root ports will change the source ID for all messages
+ * to the VMD device. Rather than doing device matching with the source
+ * ID, the AER driver should traverse the child device tree, reading
+ * AER registers to find the faulting device.
+ */
+static void quirk_no_aersid(struct pci_dev *pdev)
+{
+	/* VMD Domain */
+	if (pdev->bus->sysdata && pci_domain_nr(pdev->bus) >= 0x10000)
+		pdev->bus->bus_flags |= PCI_BUS_FLAGS_NO_AERSID;
+}
+DECLARE_PCI_FIXUP_EARLY(PCI_VENDOR_ID_INTEL, 0x2030, quirk_no_aersid);
+DECLARE_PCI_FIXUP_EARLY(PCI_VENDOR_ID_INTEL, 0x2031, quirk_no_aersid);
+DECLARE_PCI_FIXUP_EARLY(PCI_VENDOR_ID_INTEL, 0x2032, quirk_no_aersid);
+DECLARE_PCI_FIXUP_EARLY(PCI_VENDOR_ID_INTEL, 0x2033, quirk_no_aersid);
+
diff --git a/drivers/pci/quirks.c b/drivers/pci/quirks.c
index f1c9852..073baba 100644
--- a/drivers/pci/quirks.c
+++ b/drivers/pci/quirks.c
@@ -4671,23 +4671,6 @@ static void quirk_intel_qat_vf_cap(struct pci_dev *pdev)
 }
 DECLARE_PCI_FIXUP_EARLY(PCI_VENDOR_ID_INTEL, 0x443, quirk_intel_qat_vf_cap);
 
-/*
- * VMD-enabled root ports will change the source ID for all messages
- * to the VMD device. Rather than doing device matching with the source
- * ID, the AER driver should traverse the child device tree, reading
- * AER registers to find the faulting device.
- */
-static void quirk_no_aersid(struct pci_dev *pdev)
-{
-	/* VMD Domain */
-	if (pdev->bus->sysdata && pci_domain_nr(pdev->bus) >= 0x10000)
-		pdev->bus->bus_flags |= PCI_BUS_FLAGS_NO_AERSID;
-}
-DECLARE_PCI_FIXUP_EARLY(PCI_VENDOR_ID_INTEL, 0x2030, quirk_no_aersid);
-DECLARE_PCI_FIXUP_EARLY(PCI_VENDOR_ID_INTEL, 0x2031, quirk_no_aersid);
-DECLARE_PCI_FIXUP_EARLY(PCI_VENDOR_ID_INTEL, 0x2032, quirk_no_aersid);
-DECLARE_PCI_FIXUP_EARLY(PCI_VENDOR_ID_INTEL, 0x2033, quirk_no_aersid);
-
 /* FLR may cause some 82579 devices to hang. */
 static void quirk_intel_no_flr(struct pci_dev *dev)
 {
-- 
2.9.4

[PATCH v2 4/4] iommu: Prevent VMD child devices from being remapping targets

[Jon Derrick]

VMD child devices must use the VMD endpoint's ID as the requester.
Because of this, there needs to be a way to link the parent VMD
endpoint's iommu group and associated mappings to the VMD child devices
such that attaching and detaching child devices modify the endpoint's
mappings, while preventing early detaching on a singular device removal
or unbinding.

The reassignment of individual VMD child devices devices to VMs is
outside the scope of VMD, but may be implemented in the future. For now
it is best to prevent any such attempts.

This patch prevents VMD child devices from returning an IOMMU, which
prevents it from exposing an iommu_group sysfs directories and allowing
subsequent binding by userspace-access drivers such as VFIO.

Signed-off-by: Jon Derrick <jonathan.derrick@intel.com>
---
 drivers/iommu/intel-iommu.c | 5 +++++
 1 file changed, 5 insertions(+)

diff --git a/drivers/iommu/intel-iommu.c b/drivers/iommu/intel-iommu.c
index 687f18f..94353a6e 100644
--- a/drivers/iommu/intel-iommu.c
+++ b/drivers/iommu/intel-iommu.c
@@ -901,6 +901,11 @@ static struct intel_iommu *device_to_iommu(struct device *dev, u8 *bus, u8 *devf
 		struct pci_dev *pf_pdev;
 
 		pdev = to_pci_dev(dev);
+
+		/* VMD child devices currently cannot be handled individually */
+		if (is_vmd(pdev->bus))
+			return NULL;
+
 		/* VFs aren't listed in scope tables; we need to look up
 		 * the PF instead to find the IOMMU. */
 		pf_pdev = pci_physfn(pdev);
-- 
2.9.4

Re: [PATCH v2 4/4] iommu: Prevent VMD child devices from being remapping targets

[Bjorn Helgaas]

[+cc Robin]

This series looks fine to me as far as PCI is concerned, and I'd be
happy to take it via my tree given an ack from David for this IOMMU
piece.  Alternatively, you can add my

Acked-by: Bjorn Helgaas <bhelgaas-hpIqsD4AKlfQT0dZR+AlfA@public.gmane.org>

to the other patches if you want to take it via another tree.

Robin raised a question about basically this same patch the first time
around.  Not sure whether there's still an objection there.

On Thu, Aug 17, 2017 at 12:10:14PM -0600, Jon Derrick wrote:
> VMD child devices must use the VMD endpoint's ID as the requester.
> Because of this, there needs to be a way to link the parent VMD
> endpoint's iommu group and associated mappings to the VMD child devices
> such that attaching and detaching child devices modify the endpoint's
> mappings, while preventing early detaching on a singular device removal
> or unbinding.
> 
> The reassignment of individual VMD child devices devices to VMs is
> outside the scope of VMD, but may be implemented in the future. For now
> it is best to prevent any such attempts.
> 
> This patch prevents VMD child devices from returning an IOMMU, which
> prevents it from exposing an iommu_group sysfs directories and allowing
> subsequent binding by userspace-access drivers such as VFIO.
> 
> Signed-off-by: Jon Derrick <jonathan.derrick-ral2JQCrhuEAvxtiuMwx3w@public.gmane.org>
> ---
>  drivers/iommu/intel-iommu.c | 5 +++++
>  1 file changed, 5 insertions(+)
> 
> diff --git a/drivers/iommu/intel-iommu.c b/drivers/iommu/intel-iommu.c
> index 687f18f..94353a6e 100644
> --- a/drivers/iommu/intel-iommu.c
> +++ b/drivers/iommu/intel-iommu.c
> @@ -901,6 +901,11 @@ static struct intel_iommu *device_to_iommu(struct device *dev, u8 *bus, u8 *devf
>  		struct pci_dev *pf_pdev;
>  
>  		pdev = to_pci_dev(dev);
> +
> +		/* VMD child devices currently cannot be handled individually */
> +		if (is_vmd(pdev->bus))
> +			return NULL;
> +
>  		/* VFs aren't listed in scope tables; we need to look up
>  		 * the PF instead to find the IOMMU. */
>  		pf_pdev = pci_physfn(pdev);
> -- 
> 2.9.4
> 

Re: [PATCH v2 4/4] iommu: Prevent VMD child devices from being remapping targets

[Robin Murphy]

On 18/08/17 17:04, Bjorn Helgaas wrote:
> [+cc Robin]
> 
> This series looks fine to me as far as PCI is concerned, and I'd be
> happy to take it via my tree given an ack from David for this IOMMU
> piece.  Alternatively, you can add my
> 
> Acked-by: Bjorn Helgaas <bhelgaas@google.com>
> 
> to the other patches if you want to take it via another tree.
> 
> Robin raised a question about basically this same patch the first time
> around.  Not sure whether there's still an objection there.

Oh, I don't actually mind the patch as-is - implementing proper IOMMU
group support for VMD almost certainly should be the longer-term goal,
but for all I know that may be non-trivial so in the meantime a quick
and simple way to make things safe is good, however heavy-handed.

Robin.

> On Thu, Aug 17, 2017 at 12:10:14PM -0600, Jon Derrick wrote:
>> VMD child devices must use the VMD endpoint's ID as the requester.
>> Because of this, there needs to be a way to link the parent VMD
>> endpoint's iommu group and associated mappings to the VMD child devices
>> such that attaching and detaching child devices modify the endpoint's
>> mappings, while preventing early detaching on a singular device removal
>> or unbinding.
>>
>> The reassignment of individual VMD child devices devices to VMs is
>> outside the scope of VMD, but may be implemented in the future. For now
>> it is best to prevent any such attempts.
>>
>> This patch prevents VMD child devices from returning an IOMMU, which
>> prevents it from exposing an iommu_group sysfs directories and allowing
>> subsequent binding by userspace-access drivers such as VFIO.
>>
>> Signed-off-by: Jon Derrick <jonathan.derrick@intel.com>
>> ---
>>  drivers/iommu/intel-iommu.c | 5 +++++
>>  1 file changed, 5 insertions(+)
>>
>> diff --git a/drivers/iommu/intel-iommu.c b/drivers/iommu/intel-iommu.c
>> index 687f18f..94353a6e 100644
>> --- a/drivers/iommu/intel-iommu.c
>> +++ b/drivers/iommu/intel-iommu.c
>> @@ -901,6 +901,11 @@ static struct intel_iommu *device_to_iommu(struct device *dev, u8 *bus, u8 *devf
>>  		struct pci_dev *pf_pdev;
>>  
>>  		pdev = to_pci_dev(dev);
>> +
>> +		/* VMD child devices currently cannot be handled individually */
>> +		if (is_vmd(pdev->bus))
>> +			return NULL;
>> +
>>  		/* VFs aren't listed in scope tables; we need to look up
>>  		 * the PF instead to find the IOMMU. */
>>  		pf_pdev = pci_physfn(pdev);
>> -- 
>> 2.9.4
>>

Re: [PATCH v2 4/4] iommu: Prevent VMD child devices from being remapping targets

[Bjorn Helgaas]

On Fri, Aug 18, 2017 at 11:04:33AM -0500, Bjorn Helgaas wrote:
> [+cc Robin]
> 
> This series looks fine to me as far as PCI is concerned, and I'd be
> happy to take it via my tree given an ack from David for this IOMMU
> piece.  Alternatively, you can add my
> 
> Acked-by: Bjorn Helgaas <bhelgaas@google.com>
> 
> to the other patches if you want to take it via another tree.
> 
> Robin raised a question about basically this same patch the first time
> around.  Not sure whether there's still an objection there.

Ping, David, any thoughts on this patch?

> On Thu, Aug 17, 2017 at 12:10:14PM -0600, Jon Derrick wrote:
> > VMD child devices must use the VMD endpoint's ID as the requester.
> > Because of this, there needs to be a way to link the parent VMD
> > endpoint's iommu group and associated mappings to the VMD child devices
> > such that attaching and detaching child devices modify the endpoint's
> > mappings, while preventing early detaching on a singular device removal
> > or unbinding.
> > 
> > The reassignment of individual VMD child devices devices to VMs is
> > outside the scope of VMD, but may be implemented in the future. For now
> > it is best to prevent any such attempts.
> > 
> > This patch prevents VMD child devices from returning an IOMMU, which
> > prevents it from exposing an iommu_group sysfs directories and allowing
> > subsequent binding by userspace-access drivers such as VFIO.
> > 
> > Signed-off-by: Jon Derrick <jonathan.derrick@intel.com>
> > ---
> >  drivers/iommu/intel-iommu.c | 5 +++++
> >  1 file changed, 5 insertions(+)
> > 
> > diff --git a/drivers/iommu/intel-iommu.c b/drivers/iommu/intel-iommu.c
> > index 687f18f..94353a6e 100644
> > --- a/drivers/iommu/intel-iommu.c
> > +++ b/drivers/iommu/intel-iommu.c
> > @@ -901,6 +901,11 @@ static struct intel_iommu *device_to_iommu(struct device *dev, u8 *bus, u8 *devf
> >  		struct pci_dev *pf_pdev;
> >  
> >  		pdev = to_pci_dev(dev);
> > +
> > +		/* VMD child devices currently cannot be handled individually */
> > +		if (is_vmd(pdev->bus))
> > +			return NULL;
> > +
> >  		/* VFs aren't listed in scope tables; we need to look up
> >  		 * the PF instead to find the IOMMU. */
> >  		pf_pdev = pci_physfn(pdev);
> > -- 
> > 2.9.4
> >