On Tue, Apr 06, 2021 at 08:36:13PM +0200, Igor Mammedov wrote:
> On Tue, 6 Apr 2021 09:42:50 +0200
> Andrew Jones <drjo...@redhat.com> wrote:
>
> > On Thu, Apr 01, 2021 at 11:32:25PM +0200, Igor Mammedov wrote:
> > > On Thu, 01 Apr 2021 23:07:06 +0200
> > > Vincent Bernat <vinc...@bernat.ch> wrote:
> > >
> > > > ❦ 1 avril 2021 22:58 +02, Igor Mammedov:
> > > >
> > > > >> This can be invoked with:
> > > > >>
> > > > >> $QEMU -netdev user,id=internet
> > > > >> -device
> > > > >> virtio-net-pci,mac=50:54:00:00:00:42,netdev=internet,id=internet-dev
> > > > >> \
> > > > >> -smbios type=41,designation='Onboard
> > > > >> LAN',instance=1,kind=ethernet,pcidev=internet-dev
> > > > >
> > > > > an ACPI alternative was merged recently (current master).
> > > > > assigning 'designation=' wasn't implemented there, but important part
> > > > > of giving users control over PCI devices 'eno' index is implemented.
> > > > >
> > > > > When I looked into the issue, smbios way was a bit over-kill for the
> > > > > task
> > > > > and didn't really work if hotplug were used.
> > > > >
> > > > > See, for example how to use new feature:
> > > > > https://www.mail-archive.com/qemu-devel@nongnu.org/msg794164.html
> > > >
> > > > It seems simpler this way. I don't think my patch is needed then.
> > >
> > > SMBIOS ways is fine for static configs where no hot-plug is involved.
> > > Also potentially SMBIOS way may be used by arm/virt board,
> > > since acpi-index shares a lot with ACPI PCI hotplug infrastructure
> > > and we haven't ported that to arm/virt impl. yet.
> > >
> > > It also won't work for Q35 at the moment, but Julia is working
> > > on adding support for ACPI PCI hotplug to it, and once it arrives
> > > acpi-index will become available there.
> > >
> > > Perhaps we should also add support for ACPI PCI hotplug to virt/arm,
> > > along with Q35.
> > >
> >
> > What's required of the guest kernel for ACPI PCI hotplug? If there are
> > arch-specific aspects to that, then do we know if Linux for AArch64
> > has the support?
>
> I could only guess, it could be just a matter turning on HOTPLUG_PCI_ACPI
> in Kconfig.
Thanks for the pointer. It looks like it might be worth experimenting with
that
commit 23237ef3725de6e4f6f68cf11ae7cb52f8a5d60e
Author: Timur Tabi <ti...@codeaurora.org>
Date: Wed Jun 21 12:30:43 2017 -0500
arm64: defconfig: enable support for PCIe hotplug
Some ARM64 server systems support PCIe hotplug, so enable the options
for that.
Signed-off-by: Timur Tabi <ti...@codeaurora.org>
Signed-off-by: Arnd Bergmann <a...@arndb.de>
diff --git a/arch/arm64/configs/defconfig b/arch/arm64/configs/defconfig
index 71d77d2d69e0..41f827393651 100644
--- a/arch/arm64/configs/defconfig
+++ b/arch/arm64/configs/defconfig
@@ -61,7 +61,10 @@ CONFIG_ARCH_XGENE=y
CONFIG_ARCH_ZX=y
CONFIG_ARCH_ZYNQMP=y
CONFIG_PCI=y
+CONFIG_HOTPLUG_PCI_PCIE=y
CONFIG_PCI_IOV=y
+CONFIG_HOTPLUG_PCI=y
+CONFIG_HOTPLUG_PCI_ACPI=y
CONFIG_PCI_LAYERSCAPE=y
CONFIG_PCI_HISI=y
CONFIG_PCIE_QCOM=y
