[PATCH v3 3/5] conf: introduce acpi-hotplug-bridge and acpi-root-hotplug pm options

[Laine Stump]

On 9/28/21 12:54 PM, Ani Sinha wrote:
> 
> 
> On Tue, Sep 28, 2021 at 10:17 PM Laine Stump <laine at redhat.com 
> <mailto:laine at redhat.com>> wrote:
> 
>     On 9/28/21 4:44 AM, Daniel P. Berrangé wrote:
>      > On Sun, Sep 12, 2021 at 08:56:29AM +0530, Ani Sinha wrote:
>      >> This change introduces libvirt xml support for the following two
>     pm options:
>      >>
>      >> <pm>
>      >>    <acpi-hotplug-bridge enabled='no'/>
>      >>    <acpi-root-hotplug enabled='yes'/>
>      >> </pm>
>      >
>      >
>      >> +``acpi-hotplug-bridge``
>      >> +   :since:`Since 7.8.0` This option enables or disables BIOS
>     ACPI based hotplug support
>      >> +   for cold plugged bridges. It is available only for x86
>     guests, both for q35 and pc
>      >> +   machine types. For pc machines, the support is available
>     from `QEMU 2.12`. For q35
>      >> +   machines, the support is available from `QEMU 6.1`. Examples
>     of cold plugged bridges
>      >> +   include PCI-PCI bridges for pc machine types (pci-bridge
>     controller). For q35 machines,
>      >> +   it includes PCIE root ports (pcie-root-port controller).
>     This is a global option that
>      >> +   affects all bridges. No other bridge specific option is
>     required to be specified.
>      >
>      > Can you confirm my understanding of the situation..
>      >
>      >   - i440fx / PCI topology - hotplug always uses ACPI
>      >
>      >   - q35 / PCIe topology - hotplug historically used native PCIe
>     hotplug,
>      >                           but in 6.1 switched to ACPI
>      >
>      > Given, the name "acpi-hotplug-bridge",  am I right that this option
>      > has *no* effect, if the q35 machine is using native PCIe hotplug
>      > approach ?  IOW, is it a no-op until 6.1 based machine types for
>     q35 ?
> 
>     I *think* that in machinetypes where the default is native-pcie
>     hotplug,
>     setting acpi-hotplug-bridge=on 
> 
> 
> 6.1 not only introduced this option/command line in Qemu but also 
> flipped the switch to make ACPI hotplug default for pcie root ports. So 
> there are  no officially released  version of Qemu where this command 
> line exists and the default is native pcie hotplug.

You're assuming that everyone will use the canonical "q35" machinetype 
rather than a specific versioned machinetype (e.g. "pc-q35-6.0"). For 
pre-6.1 machinetypes, the default will still be native-pcie hotplug, 
even when running qemu-6.1+.

> 
> 
>     will simultaneously enable ACPI hotplug
>     and disable native-pcie hotplug for all pcie-root-ports and
>     pcie-downstream-ports.
> 
> 
> This is for 6.1 based machines as well.
> 
>     Similarly on 6.1-based machinetypes, setting
>     acpi-hotplug-bridge=off will disable ACPI hotplug and enable
>     native-pcie
>     hotplug.
> 
>     On 440fx, where the default has always been ACPI (and where SHPC
>     hotplug
>     has been disabled), acpi-hotplug-bridge=on will be a NOP, and
>     acpi-hotplug-bridge=off will completely disable hotplug on any
>     pci-bridge (but *not* on pci-root).
> 
>     As for the acpi-hotplug-root option, that is only valid for 440fx, is
>     "on" by default, and when acpi-hotplug-root=off it will completely
>     disable hotplug to any slot on pci-root.
> 
>     (for completeness - when a pcie-root-port or pcie-downstream-port has
>     <target hotplug='off'/>, that will disable whatever hotplug mode would
>     have been enabled for the controller - no hotplug at all will be
>     possible on that controller. QEMU also has a "native_hotplug" option
>     (not supported in libvirt) for pcie-root-ports and
>     pcie-downstream-ports
>     which can be used to enable native-pcie hotplug on a specific
>     controller
>     when it had been disabled (in favor of ACPI) with the global
>     acpi-hotplug-bridge ).
>