Re: [PATCH v6 2/7] xen/arm: add pci-domain for disabled devices

[Julien Grall <julien@xxxxxxx>]

On 18/11/2021 07:13, Oleksandr Andrushchenko wrote:
> Hi, Julien!
Hi Oleksandr,

> On 17.11.21 23:33, Julien Grall wrote:
> > Hi Oleksandr,
> >
> > On 17/11/2021 06:56, Oleksandr Andrushchenko wrote:
> > > Hi, Julien!
> > >
> > > On 16.11.21 20:48, Julien Grall wrote:
> > > > Hi Oleksander,
> > > >
> > > > On 05/11/2021 06:33, Oleksandr Andrushchenko wrote:
> > > > > From: Oleksandr Andrushchenko <oleksandr_andrushchenko@xxxxxxxx>
> > > > >
> > > > > If a PCI host bridge device is present in the device tree, but is
> > > > > disabled, then its PCI host bridge driver was not instantiated.
> > > > > This results in the failure of the pci_get_host_bridge_segment()
> > > > > and the following panic during Xen start:
> > > > >
> > > > > (XEN) Device tree generation failed (-22).
> > > > > (XEN)
> > > > > (XEN) ****************************************
> > > > > (XEN) Panic on CPU 0:
> > > > > (XEN) Could not set up DOM0 guest OS
> > > > > (XEN) ****************************************
> > > > >
> > > > > Fix this by adding "linux,pci-domain" property for all device tree nodes
> > > > > which have "pci" device type, so we know which segments will be used by
> > > > > the guest for which bridges.
> > > > >
> > > > > Fixes: 4cfab4425d39 ("xen/arm: Add linux,pci-domain property for hwdom if not 
> > > > > available.")
> > > > >
> > > > > Signed-off-by: Oleksandr Andrushchenko <oleksandr_andrushchenko@xxxxxxxx>
> > > > >
> > > > > ---
> > > > > New in v6
> > > > > ---
> > > > >     xen/arch/arm/domain_build.c        | 15 ++++++++++++++-
> > > > >     xen/arch/arm/pci/pci-host-common.c |  2 +-
> > > > >     xen/include/asm-arm/pci.h          |  8 ++++++++
> > > > >     3 files changed, 23 insertions(+), 2 deletions(-)
> > > > >
> > > > > diff --git a/xen/arch/arm/domain_build.c b/xen/arch/arm/domain_build.c
> > > > > index 491f5e2c316e..f7fcb1400c19 100644
> > > > > --- a/xen/arch/arm/domain_build.c
> > > > > +++ b/xen/arch/arm/domain_build.c
> > > > > @@ -753,9 +753,22 @@ static int __init write_properties(struct domain *d, 
> > > > > struct kernel_info *kinfo,
> > > > >             {
> > > > >                 uint16_t segment;
> > > > >     +            /*
> > > > > +             * The node doesn't have "linux,pci-domain" property and it is
> > > > > +             * possible that:
> > > > > +             *  - Xen only has drivers for a part of the host bridges
> > > > > +             *  - some host bridges are disabled
> > > > > +             * Make sure we insert the correct "linux,pci-domain" property
> > > > > +             * in any case, so we know which segments will be used
> > > > > +             * by Linux for which bridges.
> > > > The check above will check the node type is "pci". AFAICT, this would also cover PCI 
> > > > devices. I am not aware of any issue to add "linux,pci-domain" for them. However, this 
> > > > feels a bit odd.
> > > >
> > > >   From my understanding, a PCI device would always be described as a child of the 
> > > > hostbridges. So I would rework the 'if' to also check if the parent type is not 
> > > > "pci".
> > > We may have "bridge -> bridge -> device" topology as well.
> > Do you have an example of Device-Tree?
> No, I don't have at hand, but I can imagine this can relatively easy be done with 
> QEMU > Even if not, do you assume this topology can never happen?
I think it is not possible to describe them in the Device-Tree because 
you wouldn't be able to translate the "regs" of the nested hostbridge to 
an actual MMIO address.
In fact, looking at dt_device_get_address(), I think Xen would reject 
such DT.
> > > So, I prefer to have the check as it is.
> > I don't really like the idea to spuriously add "linux,pci-domain" to PCI DT 
> > node. But if there are no other solution, then this should at least be mentionned in the 
> > commit message and code.
> I am fine with any solution here, I just want that to be defined and 
> implemented.
> Please let me know the final decision on this and how we proceed
I would prefer to go with my way. This can be refined in the future if 
we find Device-Tree that matches what you wrote.
Cheers,

--
Julien Grall