Re: Kernel 5.9 IOMMU groups regression/change

[Bjorn Helgaas <helgaas@kernel.org>]

[+cc Rajat, LKML]

On Tue, Oct 27, 2020 at 08:31:09PM +0100, Boris V. wrote:
> On 25/10/2020 20:45, Boris V. wrote:
> > With upgrade to kernel 5.9 my VMs stopped working, because some devices
> > can't be passed through.
> > This is caused by different IOMMU groups and devices being in the same
> > group.
> > 
> > For ex. with kernel 5.8 this are IOMMU groups:
> > IOMMU Group 40:
> >         08:01.0 PCI bridge [0604]: ASMedia Technology Inc. Device
> > [1b21:118f]
> >         09:00.0 Ethernet controller [0200]: Intel Corporation I211
> > Gigabit Network Connection [8086:1539] (rev 03)
> > IOMMU Group 43:
> >         0c:00.0 SATA controller [0106]: ASMedia Technology Inc. ASM1062
> > Serial ATA Controller [1b21:0612] (rev 02)
> > IOMMU Group 44:
> >         0d:00.0 USB controller [0c03]: ASMedia Technology Inc. ASM1042A
> > USB 3.0 Host Controller [1b21:1142]
> > 
> > Ethernet, SATA and USB controller in its own group.
> > 
> > And with 5.9, everything is in one group:
> > IOMMU Group 29:
> >         00:1c.0 PCI bridge [0604]: Intel Corporation C610/X99 series
> > chipset PCI Express Root Port #1 [8086:8d10] (rev d5)
> >         00:1c.3 PCI bridge [0604]: Intel Corporation C610/X99 series
> > chipset PCI Express Root Port #4 [8086:8d16] (rev d5)
> >         00:1c.4 PCI bridge [0604]: Intel Corporation C610/X99 series
> > chipset PCI Express Root Port #5 [8086:8d18] (rev d5)
> >         00:1c.6 PCI bridge [0604]: Intel Corporation C610/X99 series
> > chipset PCI Express Root Port #7 [8086:8d1c] (rev d5)
> >         07:00.0 PCI bridge [0604]: ASMedia Technology Inc. Device
> > [1b21:118f]
> >         08:01.0 PCI bridge [0604]: ASMedia Technology Inc. Device
> > [1b21:118f]
> >         08:03.0 PCI bridge [0604]: ASMedia Technology Inc. Device
> > [1b21:118f]
> >         08:04.0 PCI bridge [0604]: ASMedia Technology Inc. Device
> > [1b21:118f]
> >         09:00.0 Ethernet controller [0200]: Intel Corporation I211
> > Gigabit Network Connection [8086:1539] (rev 03)
> >         0c:00.0 SATA controller [0106]: ASMedia Technology Inc. ASM1062
> > Serial ATA Controller [1b21:0612] (rev 02)
> >         0d:00.0 USB controller [0c03]: ASMedia Technology Inc. ASM1042A
> > USB 3.0 Host Controller [1b21:1142]
> > 
> > 
> > This seems to be caused by commit
> > 52fbf5bdeeef415b28b8e6cdade1e48927927f60.
> > commit 52fbf5bdeeef415b28b8e6cdade1e48927927f60
> > Author: Rajat Jain <rajatja@google.com>
> > Date:   Tue Jul 7 15:46:02 2020 -0700
> > 
> >     PCI: Cache ACS capability offset in device
> > 
> >     Currently the ACS capability is being looked up at a number of
> > places. Read
> >     and store it once at enumeration so that it can be used by all
> > later.  No
> >     functional change intended.
> > 
> >     Link:
> > https://lore.kernel.org/r/20200707224604.3737893-2-rajatja@google.com
> >     Signed-off-by: Rajat Jain <rajatja@google.com>
> >     Signed-off-by: Bjorn Helgaas <bhelgaas@google.com>
> > 
> >  drivers/pci/p2pdma.c |  2 +-
> >  drivers/pci/pci.c    | 20 ++++++++++++++++----
> >  drivers/pci/pci.h    |  2 +-
> >  drivers/pci/probe.c  |  2 +-
> >  drivers/pci/quirks.c |  8 ++++----
> >  include/linux/pci.h  |  1 +
> >  6 files changed, 24 insertions(+), 11 deletions(-)
> > 
> > 
> > If I revert this commit, I get back old groups.
> > 
> > In commit log there is message 'No functional change intended'. But
> > there is functional change.
> > 
> > This is Intel Core i7-5930K CPU and X99 chipset. But I see the same
> > thing on other Intel systems (didn't test on AMD).
> 
> Some more info.
> Problem seems to be that pci_dev_specific_enable_acs() is not called
> anymore.
> Before, pci_enable_acs() was called from pci_init_capabilities() and in
> pci_enable_acs(), pci_dev_specific_enable_acs() was called.
> I don't know anything about PCI and this stuff, but I'm guessing that this
> function enable ACS for some Intel devices.
> But after this commit, pci_acs_init() is called from pci_init_capabilities()
> and if pci_find_ext_capability(dev, PCI_EXT_CAP_ID_ACS) returns 0,
> pci_enable_acs() and pci_dev_specific_enable_acs() is not called anymore.
> If I apply for ex. this patch bellow, groups are right again and everything
> works as before.

Thanks very much for the report and the debugging.  Maybe we can get
this sorted and fixed for v5.10-rc2 or -rc3.

> diff -ur linux-5.9.1.orig/drivers/pci/pci.c linux-5.9.1/drivers/pci/pci.c
> --- linux-5.9.1.orig/drivers/pci/pci.c  2020-10-17 08:31:22.000000000 +0200
> +++ linux-5.9.1/drivers/pci/pci.c       2020-10-27 19:01:32.650010803 +0100
> @@ -3502,9 +3502,7 @@
>  void pci_acs_init(struct pci_dev *dev)
>  {
>         dev->acs_cap = pci_find_ext_capability(dev, PCI_EXT_CAP_ID_ACS);
> -
> -       if (dev->acs_cap)
> -               pci_enable_acs(dev);
> +       pci_enable_acs(dev);
>  }
> 
>  /**
> 
>