Arch Linux User Repository

• Leaving eclairvoyant's comment pinned because it contains some useful information.
• Patch policy is different, but similar.
  • Primary purpose of this package are the ACS override and i915 VGA arbiter patches.
  • Upstream Arch patches will be applied to maintain parity with the standard kernel.
  • No other patches will be added.
• Package does have options that can be changed, like building with clang or disabling Arch patches.
  • Defaults will match Arch packages, except when incompatible with this package's primary patches.
  • Options are subject to change. Read PKGBUILD for details.
• Avoid commenting and flagging at the same time for the same issue.
  • Flag for common issues with standard solutions, like new releases, key changes, etc.
  • Comment for issues requiring explanation or debugging.
    • Use a pastebin for blocks of text more than a few lines.

This package exists for the specific purpose of adding ported patches based on those originally created by Alex Williamson for:

• Bypassing the default IOMMU grouping by overriding PCIe ACS support, and

• Enabling VGA arbitration support on Intel iGPUs.

Arbitrary patches will not be added.

Refer to the wiki on PCI passthrough and this blog post on IOMMU groups for risks/caveats before using this package.

Regular AUR etiquette applies as well (knowledge of makepkg and searching the wiki/Arch forums is expected, and AUR helpers or Arch-based distros that are not Arch Linux are unsupported).

5.9.1 is up now, but note that nvidia is only partially supported on 5.9.x; nvidia expects to have an update ready in november. this seems to only affect cuda and opencl.

https://forums.developer.nvidia.com/t/nvidia-driver-not-yet-supported-for-linux-kernel-5-9/157263

there are supposedly some patches available that workaround this issue; you'll need to find/apply those yourself if you need cuda/opencl on 5.9.x.

Do you have any custom gpg configs or are you perhaps using an alternative keyring? I also am not able to reproduce the problem. Make sure your GPG keyring has all the listed signatures from your first post...

:: PGP keys need importing:
 -> ABAF11C65A2970B130ABE3C479BE3E4300411886, required by: linux-vfio
 -> 647F28654894E3BD457199BE38DBBDC86092693E, required by: linux-vfio
 -> A2FF3A36AAA56654109064AB19802F8B0D70FC30, required by: linux-vfio

gpg --recv-keys 3B94A80E50A477C7
git clone https://aur.archlinux.org/linux-vfio.git
cd linux-vfio
makepkg -si

These steps do not work for me, without helpers. Even after having a clean repo/folder.

==> Verifying source file signatures with gpg...
    archlinux-linux git repo ... SIGNATURE NOT FOUND
==> ERROR: One or more PGP signatures could not be verified!

gpg --recv-keys 3B94A80E50A477C7

git clone https://aur.archlinux.org/linux-vfio.git

cd linux-vfio

makepkg -si

Works here. I suggest reading the pinned comment as AUR helpers are not supported.

:: PGP keys need importing:
 -> ABAF11C65A2970B130ABE3C479BE3E4300411886, required by: linux-vfio
 -> 647F28654894E3BD457199BE38DBBDC86092693E, required by: linux-vfio
 -> A2FF3A36AAA56654109064AB19802F8B0D70FC30, required by: linux-vfio
==> Import? [Y/n] y
:: Importing keys with gpg...
gpg: key 19802F8B0D70FC30: 6 duplicate signatures removed
gpg: key 19802F8B0D70FC30: 6 signatures reordered
gpg: key 19802F8B0D70FC30: public key "Jan Alexander Steffens (heftig) <jan.steffens@gmail.com>" imported
gpg: key 38DBBDC86092693E: public key "Greg Kroah-Hartman <gregkh@linuxfoundation.org>" imported
gpg: key 79BE3E4300411886: public key "Linus Torvalds <torvalds@kernel.org>" imported
gpg: Total number processed: 3
gpg:               imported: 3
==> Making package: linux-vfio 5.8.12.arch1-1
==> Retrieving sources...
  -> Cloning archlinux-linux git repo...
Cloning into bare repository '/home/user/.cache/yay/linux-vfio/archlinux-linux'...
remote: Enumerating objects: 19718, done.
remote: Counting objects: 100% (19718/19718), done.
remote: Compressing objects: 100% (9357/9357), done.
remote: Total 7741632 (delta 13062), reused 13650 (delta 10325), pack-reused 7721914
Receiving objects: 100% (7741632/7741632), 2.29 GiB | 6.63 MiB/s, done.
Resolving deltas: 100% (6468837/6468837), done.
  -> Found config
  -> Found add-acs-overrides.patch
  -> Found i915-vga-arbiter.patch
  -> Found sphinx-workaround.patch
==> Validating source files with sha256sums...
    archlinux-linux ... Skipped
    config ... Passed
    add-acs-overrides.patch ... Passed
    i915-vga-arbiter.patch ... Passed
    sphinx-workaround.patch ... Passed
==> Verifying source file signatures with gpg...
    archlinux-linux git repo ... SIGNATURE NOT FOUND
==> ERROR: One or more PGP signatures could not be verified!
error downloading sources: linux-vfio

Getting all of heftig's keys from KeyBase does not work either. Suggestion gpg --recv-keys 3B94A80E50A477C7 also does not work. What is going on here?

Just threw 5.6.15 up, sorry for the delay. Note that heftig's GPG key was rotated recently, so you'll probably need to gpg --recv-keys 3B94A80E50A477C7 before building.

5.6.10 Is no longer supported by kernel modules, and 5.7 is testing, at-least update to 5.6.13 for module support so the nvidia drivers work, the current patch-set works with 5.6.13 no change required.

with gcc 10 i have a compile error on virt/kvm/kvm main.c:2236

it seems upstream bug: https://bugzilla.kernel.org/show_bug.cgi?id=207173 should be fixed with 5.7 upstream mentioned there

commit: https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/commit/?id=78a5255ffb6a1

adding that patch makes it compile again

It wouldn't matter much anyways because I'm almost certain that X does not support hot plugging display devices. Heck, if you notice in the 0/6 patch, the submitter mentions that he restarts X when hot plugging, which is what I would expect.

The patch seems to be for people who are hotplugging PCI-e devices (which I would expect to be over Thunderbolt), which is something I would bet a vast majority of users do not attempt. I do not expect this patch would improve AMDGPU users' experiences while running a "normal" setup with just a AMD video device plugged directly into one of the available PCI-e slots inside their machines and not touching it while the machine is running (as is a very good idea to not touch an active device on a PCI-e slot/device that isn't specifically designed for hot plugging).

Also, it appears that the issues discussed in the earlier comments are related to certain AMD processors and not video devices.