Re: [External] Re: [PATCH v1] PCI: Set no io resource for bridges that behind VMD controller

[Bjorn Helgaas]

On Wed, Nov 16, 2022 at 03:34:34AM +0000, Xiaochun XC17 Li wrote:
> On Sat, Nov 11, 2022 at 07:54:25, "Bjorn Helgaas" <helgaas@xxxxxxxxxx> wrote:
> > On Tue, Sep 13, 2022 at 09:24:45PM +0800, Xiaochun Lee wrote:
> > > From: Xiaochun Lee <lixc17@xxxxxxxxxx>
> > >
> > > When enable VMDs on Intel CPUs, VMD controllers(8086:28c0) be
> > > recognized by VMD driver and there are many failed messages of BAR 13
> > > when scan the bridges and assign IO resource behind it as listed
> > > below, the bridge wants to get 0x6000 as its IO resource, but there is
> > > no IO resources on the host bridge.
> > >
> > > VMD host bridge resources:
> > > vmd 0000:64:00.5: PCI host bridge to bus 10000:80 pci_bus 10000:80:
> > > root bus resource [bus 80-9f] pci_bus 10000:80: root bus resource [mem
> > > 0xe0000000-0xe1ffffff] pci_bus 10000:80: root bus resource [mem
> > > 0x24ffff02010-0x24fffffffff 64bit]
> > >
> > > Failed messages of BAR#13:
> > > pci 10000:80:02.0: BAR 13: no space for [io  size 0x1000] pci
> > > 10000:80:02.0: BAR 13: failed to assign [io  size 0x1000] pci
> > > 10000:80:03.0: BAR 13: no space for [io  size 0x1000] pci
> > > 10000:80:03.0: BAR 13: failed to assign [io  size 0x1000]
> > >
> > > VMD-enabled root ports use
> > > Enhanced Configuration Access Mechanism (ECAM) access PCI Express
> > > configuration space, and offer VMD_CFGBAR as base of PCI Express
> > > configuration space for the bridges behind it. The configuration space
> > > includes IO resources, but these IO resources are not actually used on
> > > X86, especially the NVMes as device connected on this hot plug
> > > bridges, and it can result in BAR#13 assign IO resource failed. So we
> > > clear IO resources by setting an IO base value greater than limit to
> > > these bridges. Hence, we can leverage kernel parameter
> > > "pci=hpiosize=0KB" to avoid this failed messages show out.
> > >
> > > Signed-off-by: Xiaochun Lee <lixc17@xxxxxxxxxx>
> > 
> > Some of the discussion here got lost because of email issues.  Lore has
> > some:
> > https://apc01.safelinks.protection.outlook.com/?url=https%3A%2F%2Flore.
> > kernel.org%2Fall%2F1663075485-20591-1-git-send-email-
> > lixiaochun.2888%40163.com%2FT%2F%23u&amp;data=05%7C01%7Clixc17
> > %40lenovo.com%7C9cd095ffdb584e492dec08dac440139b%7C5c7d0b28bdf
> > 8410caa934df372b16203%7C0%7C0%7C638038076734438158%7CUnknown
> > %7CTWFpbGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1ha
> > WwiLCJXVCI6Mn0%3D%7C3000%7C%7C%7C&amp;sdata=JzWYkIsaEfINofiqo
> > XyjEh43VjXO3HZw2JLSsmhpUiQ%3D&amp;reserved=0,
> > and patchwork has a v2 with a little more discussion:
> > https://apc01.safelinks.protection.outlook.com/?url=https%3A%2F%2Fpatc
> > hwork.kernel.org%2Fproject%2Flinux-pci%2Fpatch%2F1664288166-7432-1-
> > git-send-email-
> > lixiaochun.2888%40163.com%2F&amp;data=05%7C01%7Clixc17%40lenovo.
> > com%7C9cd095ffdb584e492dec08dac440139b%7C5c7d0b28bdf8410caa934
> > df372b16203%7C0%7C0%7C638038076734438158%7CUnknown%7CTWFpb
> > GZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI
> > 6Mn0%3D%7C3000%7C%7C%7C&amp;sdata=y%2BtIsepTpyLeoHW6CrgkZz2
> > tiMjY0TONfK7zNCKXQ90%3D&amp;reserved=0
> > 
> > But the v2 patch doesn't seem to have made it to the mailing lists or to
> > lore
> > (https://apc01.safelinks.protection.outlook.com/?url=https%3A%2F%2Flor
> > e.kernel.org%2Fall%2F%3Fq%3Df%253Alixc17&amp;data=05%7C01%7Clixc1
> > 7%40lenovo.com%7C9cd095ffdb584e492dec08dac440139b%7C5c7d0b28bd
> > f8410caa934df372b16203%7C0%7C0%7C638038076734438158%7CUnknow
> > n%7CTWFpbGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1h
> > aWwiLCJXVCI6Mn0%3D%7C3000%7C%7C%7C&amp;sdata=zxGz1hZOD2tvQP
> > EsbxQTzjHwQvXvqeO%2FUd6I9S%2Fj314%3D&amp;reserved=0) and I don't
> > apply things until they appear on the mailing list.
> > 
> > I *would* like to get rid of those "no space" and "failed to assign"
> > messages.  This is an issue for platforms other than VMD, too.  Just an FYI
> > that you need to follow up on this if we want make progress.
> 
> Thanks for your comments, so do you mean we'd better come up with a
> solution to avoid messages like "no space" and "failed to assign" using
> a common way, for both VMD and other platforms?

I tried to say two separate things:

1) It doesn't seem like a VMD-specific thing, so it would be ideal if
the solution were generic instead of being VMD-specific.

2) Some of your previous patches didn't make it to the mailing list,
so we couldn't really do anything with them.

Bjorn