[v3] x86/dom0: be less restrictive with the Interrupt Address Range

[PATCH v3 0/3] x86/dom0: be less restrictive with the Interrupt Address Range

Posted by Roger Pau Monne 8 months, 2 weeks ago

Hello,

First two patches are preparatory changes to reduce the changes required
in patch 3.  I would have wanted those to go in 4.20 to fix the issues
on Lenovo Thinkpads, but it's too late now.

Thanks, Roger.

Roger Pau Monne (3):
  x86/dom0: correctly set the maximum ->iomem_caps bound for PVH
  x86/iommu: account for IOMEM caps when populating dom0 IOMMU
    page-tables
  x86/dom0: be less restrictive with the Interrupt Address Range

 xen/arch/x86/dom0_build.c           | 24 ++++++++---
 xen/arch/x86/hvm/dom0_build.c       | 14 +++---
 xen/arch/x86/hvm/io.c               |  6 +--
 xen/arch/x86/include/asm/hvm/io.h   |  4 +-
 xen/drivers/passthrough/x86/iommu.c | 67 ++++++++++++-----------------
 5 files changed, 57 insertions(+), 58 deletions(-)

-- 
2.46.0

Re: [PATCH v3 0/3] x86/dom0: be less restrictive with the Interrupt Address Range

Posted by Jan Beulich 8 months ago

On 19.02.2025 17:48, Roger Pau Monne wrote:
> Hello,
> 
> First two patches are preparatory changes to reduce the changes required
> in patch 3.  I would have wanted those to go in 4.20 to fix the issues
> on Lenovo Thinkpads, but it's too late now.
> 
> Thanks, Roger.
> 
> Roger Pau Monne (3):
>   x86/dom0: correctly set the maximum ->iomem_caps bound for PVH
>   x86/iommu: account for IOMEM caps when populating dom0 IOMMU
>     page-tables
>   x86/dom0: be less restrictive with the Interrupt Address Range

I'm uncertain whether to take this and "x86/pvh: workaround missing MMIO
regions in dom0 p2m" for backport. The sole Fixes: tag is in patch 1 here.
Thoughts?

Thanks, Jan

Re: [PATCH v3 0/3] x86/dom0: be less restrictive with the Interrupt Address Range

Posted by Roger Pau Monné 8 months ago

On Wed, Mar 05, 2025 at 03:27:18PM +0100, Jan Beulich wrote:
> On 19.02.2025 17:48, Roger Pau Monne wrote:
> > Hello,
> > 
> > First two patches are preparatory changes to reduce the changes required
> > in patch 3.  I would have wanted those to go in 4.20 to fix the issues
> > on Lenovo Thinkpads, but it's too late now.
> > 
> > Thanks, Roger.
> > 
> > Roger Pau Monne (3):
> >   x86/dom0: correctly set the maximum ->iomem_caps bound for PVH
> >   x86/iommu: account for IOMEM caps when populating dom0 IOMMU
> >     page-tables
> >   x86/dom0: be less restrictive with the Interrupt Address Range
> 
> I'm uncertain whether to take this and "x86/pvh: workaround missing MMIO
> regions in dom0 p2m" for backport. The sole Fixes: tag is in patch 1 here.
> Thoughts?

At least the ones here would be helpful for the reported Lenovo
Thinkpad issue.  The PVH p2m addition would be nice IMO.

Thanks, Roger.

Re: [PATCH v3 0/3] x86/dom0: be less restrictive with the Interrupt Address Range

Posted by Jan Beulich 8 months ago

On 05.03.2025 15:35, Roger Pau Monné wrote:
> On Wed, Mar 05, 2025 at 03:27:18PM +0100, Jan Beulich wrote:
>> On 19.02.2025 17:48, Roger Pau Monne wrote:
>>> Hello,
>>>
>>> First two patches are preparatory changes to reduce the changes required
>>> in patch 3.  I would have wanted those to go in 4.20 to fix the issues
>>> on Lenovo Thinkpads, but it's too late now.
>>>
>>> Thanks, Roger.
>>>
>>> Roger Pau Monne (3):
>>>   x86/dom0: correctly set the maximum ->iomem_caps bound for PVH
>>>   x86/iommu: account for IOMEM caps when populating dom0 IOMMU
>>>     page-tables
>>>   x86/dom0: be less restrictive with the Interrupt Address Range
>>
>> I'm uncertain whether to take this and "x86/pvh: workaround missing MMIO
>> regions in dom0 p2m" for backport. The sole Fixes: tag is in patch 1 here.
>> Thoughts?
> 
> At least the ones here would be helpful for the reported Lenovo
> Thinkpad issue.  The PVH p2m addition would be nice IMO.

Are the ones here sufficient to deal with that issue? IOW iasn't the other
2-patch series also necessary?

Jan

Re: [PATCH v3 0/3] x86/dom0: be less restrictive with the Interrupt Address Range

Posted by Roger Pau Monné 8 months ago

On Wed, Mar 05, 2025 at 03:54:56PM +0100, Jan Beulich wrote:
> On 05.03.2025 15:35, Roger Pau Monné wrote:
> > On Wed, Mar 05, 2025 at 03:27:18PM +0100, Jan Beulich wrote:
> >> On 19.02.2025 17:48, Roger Pau Monne wrote:
> >>> Hello,
> >>>
> >>> First two patches are preparatory changes to reduce the changes required
> >>> in patch 3.  I would have wanted those to go in 4.20 to fix the issues
> >>> on Lenovo Thinkpads, but it's too late now.
> >>>
> >>> Thanks, Roger.
> >>>
> >>> Roger Pau Monne (3):
> >>>   x86/dom0: correctly set the maximum ->iomem_caps bound for PVH
> >>>   x86/iommu: account for IOMEM caps when populating dom0 IOMMU
> >>>     page-tables
> >>>   x86/dom0: be less restrictive with the Interrupt Address Range
> >>
> >> I'm uncertain whether to take this and "x86/pvh: workaround missing MMIO
> >> regions in dom0 p2m" for backport. The sole Fixes: tag is in patch 1 here.
> >> Thoughts?
> > 
> > At least the ones here would be helpful for the reported Lenovo
> > Thinkpad issue.  The PVH p2m addition would be nice IMO.
> 
> Are the ones here sufficient to deal with that issue? IOW iasn't the other
> 2-patch series also necessary?

For a PV dom0, yes, the patches here are enough.  For a PVH dom0 you
also need "x86/pvh: workaround missing MMIO regions in dom0 p2m".
Given that we now officially support PVH I think we would need to
backport the latter, to have parity between PV and PVH dom0.

Thanks, Roger.