From: Alex Williamson <alex.williamson@redhat.com>
To: Yi Liu <yi.l.liu@intel.com>
Cc: Jason Gunthorpe <jgg@nvidia.com>,
"Tian, Kevin" <kevin.tian@intel.com>,
"joro@8bytes.org" <joro@8bytes.org>,
"robin.murphy@arm.com" <robin.murphy@arm.com>,
"eric.auger@redhat.com" <eric.auger@redhat.com>,
"nicolinc@nvidia.com" <nicolinc@nvidia.com>,
"kvm@vger.kernel.org" <kvm@vger.kernel.org>,
"chao.p.peng@linux.intel.com" <chao.p.peng@linux.intel.com>,
"iommu@lists.linux.dev" <iommu@lists.linux.dev>,
"baolu.lu@linux.intel.com" <baolu.lu@linux.intel.com>,
"Duan, Zhenzhong" <zhenzhong.duan@intel.com>,
"Pan, Jacob jun" <jacob.jun.pan@intel.com>
Subject: Re: [PATCH v2 0/4] vfio-pci support pasid attach/detach
Date: Thu, 25 Apr 2024 06:58:27 -0600 [thread overview]
Message-ID: <20240425065827.66b3b9b8.alex.williamson@redhat.com> (raw)
In-Reply-To: <07fbea50-b88d-46d8-b438-b4abda0447bb@intel.com>
On Thu, 25 Apr 2024 17:26:54 +0800
Yi Liu <yi.l.liu@intel.com> wrote:
> On 2024/4/25 02:24, Alex Williamson wrote:
> > On Tue, 23 Apr 2024 21:12:21 -0300
> > Jason Gunthorpe <jgg@nvidia.com> wrote:
> >
> >> On Tue, Apr 23, 2024 at 11:47:50PM +0000, Tian, Kevin wrote:
> >>>> From: Jason Gunthorpe <jgg@nvidia.com>
> >>>> Sent: Tuesday, April 23, 2024 8:02 PM
> >>>>
> >>>> On Tue, Apr 23, 2024 at 07:43:27AM +0000, Tian, Kevin wrote:
> >>>>> I'm not sure how userspace can fully handle this w/o certain assistance
> >>>>> from the kernel.
> >>>>>
> >>>>> So I kind of agree that emulated PASID capability is probably the only
> >>>>> contract which the kernel should provide:
> >>>>> - mapped 1:1 at the physical location, or
> >>>>> - constructed at an offset according to DVSEC, or
> >>>>> - constructed at an offset according to a look-up table
> >>>>>
> >>>>> The VMM always scans the vfio pci config space to expose vPASID.
> >>>>>
> >>>>> Then the remaining open is what VMM could do when a VF supports
> >>>>> PASID but unfortunately it's not reported by vfio. W/o the capability
> >>>>> of inspecting the PASID state of PF, probably the only feasible option
> >>>>> is to maintain a look-up table in VMM itself and assumes the kernel
> >>>>> always enables the PASID cap on PF.
> >>>>
> >>>> I'm still not sure I like doing this in the kernel - we need to do the
> >>>> same sort of thing for ATS too, right?
> >>>
> >>> VF is allowed to implement ATS.
> >>>
> >>> PRI has the same problem as PASID.
> >>
> >> I'm surprised by this, I would have guessed ATS would be the device
> >> global one, PRI not being per-VF seems problematic??? How do you
> >> disable PRI generation to get a clean shutdown?
> >>
> >>>> It feels simpler if the indicates if PASID and ATS can be supported
> >>>> and userspace builds the capability blocks.
> >>>
> >>> this routes back to Alex's original question about using different
> >>> interfaces (a device feature vs. PCI PASID cap) for VF and PF.
> >>
> >> I'm not sure it is different interfaces..
> >>
> >> The only reason to pass the PF's PASID cap is to give free space to
> >> the VMM. If we are saying that gaps are free space (excluding a list
> >> of bad devices) then we don't acutally need to do that anymore.
> >
> > Are we saying that now?? That's new.
> >
> >> VMM will always create a synthetic PASID cap and kernel will always
> >> suppress a real one.
> >>
> >> An iommufd query will indicate if the vIOMMU can support vPASID on
> >> that device.
> >>
> >> Same for all the troublesome non-physical caps.
> >>
> >>>> There are migration considerations too - the blocks need to be
> >>>> migrated over and end up in the same place as well..
> >>>
> >>> Can you elaborate what is the problem with the kernel emulating
> >>> the PASID cap in this consideration?
> >>
> >> If the kernel changes the algorithm, say it wants to do PASID, PRI,
> >> something_new then it might change the layout
> >>
> >> We can't just have the kernel decide without also providing a way for
> >> userspace to say what the right layout actually is. :\
> >
> > The capability layout is only relevant to migration, right? A variant
> > driver that supports migration is a prerequisite and would also be
> > responsible for exposing the PASID capability. This isn't as disjoint
> > as it's being portrayed.
> >
> >>> Does it talk about a case where the devices between src/dest are
> >>> different versions (but backward compatible) with different unused
> >>> space layout and the kernel approach may pick up different offsets
> >>> while the VMM can guarantee the same offset?
> >>
> >> That is also a concern where the PCI cap layout may change a bit but
> >> they are still migration compatible, but my bigger worry is that the
> >> kernel just lays out the fake caps in a different way because the
> >> kernel changes.
> >
> > Outside of migration, what does it matter if the cap layout is
^^^^^^^^^^^^^^^^^^^^
> > different? A driver should never hard code the address for a
> > capability.
> >
>
> But it may store the offset of capability to make next cap access more
> convenient. I noticted struct pci_dev stores the offset of PRI and PASID
> cap. So if the layout of config space changed between src and dst, it may
> result in problem in guest when guest driver uses the offsets to access
> PRI/PASID cap. I can see pci_dev stores offsets of other caps (acs, msi,
> msix). So there is already a problem even put aside the PRI and PASID cap.
Yes, I had noted "outside of migration" above. Config space must be
consistent to a running VM. But the possibility of config space
changing like this only exists in the case where the driver supports
migration, so I think we're inventing an unrealistic concern that a
driver that supports migration would arbitrarily modify the config space
layout in order to make an argument for VMM managed layout. Thanks,
Alex
> #ifdef CONFIG_PCI_PRI
> u16 pri_cap; /* PRI Capability offset */
> u32 pri_reqs_alloc; /* Number of PRI requests allocated */
> unsigned int pasid_required:1; /* PRG Response PASID Required */
> #endif
> #ifdef CONFIG_PCI_PASID
> u16 pasid_cap; /* PASID Capability offset */
> u16 pasid_features;
> #endif
> #ifdef CONFIG_PCI_P2PDMA
> struct pci_p2pdma __rcu *p2pdma;
> #endif
> #ifdef CONFIG_PCI_DOE
> struct xarray doe_mbs; /* Data Object Exchange mailboxes */
> #endif
> u16 acs_cap; /* ACS Capability offset */
>
> https://github.com/torvalds/linux/blob/master/include/linux/pci.h#L350
>
> >> At least if the VMM is doing this then the VMM can include the
> >> information in its migration scheme and use it to recreate the PCI
> >> layout withotu having to create a bunch of uAPI to do so.
> >
> > We're again back to migration compatibility, where again the capability
> > layout would be governed by the migration support in the in-kernel
> > variant driver. Once migration is involved the location of a PASID
> > shouldn't be arbitrary, whether it's provided by the kernel or the VMM.
> >
> > Regardless, the VMM ultimately has the authority what the guest
> > sees in config space. The VMM is not bound to expose the PASID at the
> > offset provided by the kernel, or bound to expose it at all. The
> > kernel exposed PASID can simply provide an available location and set
> > of enabled capabilities. Thanks,
> >
> > Alex
> >
>
next prev parent reply other threads:[~2024-04-25 12:58 UTC|newest]
Thread overview: 72+ messages / expand[flat|nested] mbox.gz Atom feed top
2024-04-12 8:21 [PATCH v2 0/4] vfio-pci support pasid attach/detach Yi Liu
2024-04-12 8:21 ` [PATCH v2 1/4] ida: Add ida_get_lowest() Yi Liu
2024-04-16 16:03 ` Alex Williamson
2024-04-18 7:02 ` Yi Liu
2024-04-18 16:23 ` Alex Williamson
2024-04-18 17:12 ` Jason Gunthorpe
2024-04-19 13:43 ` Yi Liu
2024-04-19 13:55 ` Alex Williamson
2024-04-19 14:00 ` Jason Gunthorpe
2024-04-23 7:19 ` Yi Liu
2024-04-19 13:40 ` Yi Liu
2024-04-12 8:21 ` [PATCH v2 2/4] vfio-iommufd: Support pasid [at|de]tach for physical VFIO devices Yi Liu
2024-04-16 9:01 ` Tian, Kevin
2024-04-16 9:24 ` Yi Liu
2024-04-16 9:47 ` Tian, Kevin
2024-04-18 7:04 ` Yi Liu
2024-04-23 12:43 ` Jason Gunthorpe
2024-04-24 0:33 ` Tian, Kevin
2024-04-24 4:48 ` Yi Liu
2024-04-12 8:21 ` [PATCH v2 3/4] vfio: Add VFIO_DEVICE_PASID_[AT|DE]TACH_IOMMUFD_PT Yi Liu
2024-04-16 9:13 ` Tian, Kevin
2024-04-16 9:36 ` Yi Liu
2024-04-23 12:45 ` Jason Gunthorpe
2024-04-12 8:21 ` [PATCH v2 4/4] vfio: Report PASID capability via VFIO_DEVICE_FEATURE ioctl Yi Liu
2024-04-16 9:40 ` Tian, Kevin
2024-04-16 17:57 ` Alex Williamson
2024-04-17 7:09 ` Tian, Kevin
2024-04-17 20:25 ` Alex Williamson
2024-04-18 0:21 ` Tian, Kevin
2024-04-18 8:23 ` Yi Liu
2024-04-18 16:34 ` Alex Williamson
2024-04-23 12:39 ` Jason Gunthorpe
2024-04-24 0:24 ` Tian, Kevin
2024-04-24 13:59 ` Jason Gunthorpe
2024-04-16 8:38 ` [PATCH v2 0/4] vfio-pci support pasid attach/detach Tian, Kevin
2024-04-16 17:50 ` Jason Gunthorpe
2024-04-17 7:16 ` Tian, Kevin
2024-04-17 12:20 ` Jason Gunthorpe
2024-04-17 23:02 ` Alex Williamson
2024-04-18 0:06 ` Tian, Kevin
2024-04-18 9:03 ` Yi Liu
2024-04-18 20:37 ` Alex Williamson
2024-04-19 5:52 ` Tian, Kevin
2024-04-19 16:35 ` Alex Williamson
2024-04-23 7:43 ` Tian, Kevin
2024-04-23 12:01 ` Jason Gunthorpe
2024-04-23 23:47 ` Tian, Kevin
2024-04-24 0:12 ` Jason Gunthorpe
2024-04-24 2:57 ` Tian, Kevin
2024-04-24 12:29 ` Baolu Lu
2024-04-24 14:04 ` Jason Gunthorpe
2024-04-24 5:19 ` Tian, Kevin
2024-04-24 14:15 ` Jason Gunthorpe
2024-04-24 18:38 ` Alex Williamson
2024-04-24 18:45 ` Jason Gunthorpe
2024-04-24 18:24 ` Alex Williamson
2024-04-24 18:36 ` Jason Gunthorpe
2024-04-24 20:13 ` Alex Williamson
2024-04-26 14:11 ` Jason Gunthorpe
2024-04-26 20:13 ` Alex Williamson
2024-04-28 6:19 ` Tian, Kevin
2024-04-29 7:43 ` Yi Liu
2024-04-29 17:15 ` Jason Gunthorpe
2024-04-29 17:44 ` Jason Gunthorpe
2024-04-27 5:05 ` Christoph Hellwig
2024-04-25 9:26 ` Yi Liu
2024-04-25 12:58 ` Alex Williamson [this message]
2024-04-26 9:01 ` Yi Liu
2024-04-19 13:59 ` Jason Gunthorpe
2024-04-23 7:58 ` Yi Liu
2024-04-23 12:05 ` Jason Gunthorpe
2024-04-19 13:34 ` Jason Gunthorpe
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=20240425065827.66b3b9b8.alex.williamson@redhat.com \
--to=alex.williamson@redhat.com \
--cc=baolu.lu@linux.intel.com \
--cc=chao.p.peng@linux.intel.com \
--cc=eric.auger@redhat.com \
--cc=iommu@lists.linux.dev \
--cc=jacob.jun.pan@intel.com \
--cc=jgg@nvidia.com \
--cc=joro@8bytes.org \
--cc=kevin.tian@intel.com \
--cc=kvm@vger.kernel.org \
--cc=nicolinc@nvidia.com \
--cc=robin.murphy@arm.com \
--cc=yi.l.liu@intel.com \
--cc=zhenzhong.duan@intel.com \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.