[v2,0/3] Fix broken PCIe device after migration

Message

Igor Mammedov March 1, 2022, 3:11 p.m. UTC

Changelog:
   v2:
     * instead of disabling power control on slot and letting ACPI
       PCI hotplug  module in guest to deal with it, set
       PCI_EXP_SLTCTL_PWR_ON on PCIe slot from acpi_pcihp_device_plug_cb()
       when a device plugged into it.

Currently ACPI PCI hotplug is enabled by default for Q35 machine
type and overrides native PCIe hotplug. It works as expected when
a PCIe device is hotplugged into slot, however the device becomes
in-operational after migration. Which is caused by BARs being
disabled on target due to powered off status of the slot.

Proposed fix turns on power on PCIe slot when a device is
hotplugged into it, then that state is migrated and device
stays powred on after migration. 

Tested with (seabios: FC34, Win2012; ovmf: RHEL8)

RHBZ: https://bugzilla.redhat.com/show_bug.cgi?id=2053584
CC: [e-mail]
CC: [e-mail]

Ref to v1:
 https://lore.kernel.org/all/[e-mail]/T/
Gitlab link:
 https://gitlab.com/imammedo/qemu/-/tree/pcie_poweroff_acpi_regression_rhbz2053584_V2

Igor Mammedov (3):
  pci: expose TYPE_XIO3130_DOWNSTREAM name
  acpi: pcihp: pcie: set power on cap on parent slot
  q35: compat: keep hotplugged PCIe device broken after migration for
    6.2 and older machine types

 include/hw/acpi/pcihp.h                    |  1 +
 include/hw/pci-bridge/xio3130_downstream.h | 15 +++++++++++++++
 include/hw/pci/pcie.h                      |  1 +
 hw/acpi/ich9.c                             | 20 ++++++++++++++++++++
 hw/acpi/pcihp.c                            | 15 ++++++++++++++-
 hw/core/machine.c                          |  4 +++-
 hw/pci-bridge/xio3130_downstream.c         |  3 ++-
 hw/pci/pcie.c                              | 11 +++++++++++
 8 files changed, 67 insertions(+), 3 deletions(-)
 create mode 100644 include/hw/pci-bridge/xio3130_downstream.h

Comments

Michael S. Tsirkin March 1, 2022, 3:20 p.m. UTC | #1

On Tue, Mar 01, 2022 at 10:11:57AM -0500, Igor Mammedov wrote:
> Changelog:
>    v2:
>      * instead of disabling power control on slot and letting ACPI
>        PCI hotplug  module in guest to deal with it, set
>        PCI_EXP_SLTCTL_PWR_ON on PCIe slot from acpi_pcihp_device_plug_cb()
>        when a device plugged into it.
> 
> Currently ACPI PCI hotplug is enabled by default for Q35 machine
> type and overrides native PCIe hotplug. It works as expected when
> a PCIe device is hotplugged into slot, however the device becomes
> in-operational after migration. Which is caused by BARs being
> disabled on target due to powered off status of the slot.
> 
> Proposed fix turns on power on PCIe slot when a device is
> hotplugged into it, then that state is migrated and device
> stays powred on after migration. 


Looks good for 1-2. Some questions on 3.
Thanks!

> Tested with (seabios: FC34, Win2012; ovmf: RHEL8)
> 
> RHBZ: https://bugzilla.redhat.com/show_bug.cgi?id=2053584
> CC: [e-mail]
> CC: [e-mail]
> 
> Ref to v1:
>  https://lore.kernel.org/all/[e-mail]/T/
> Gitlab link:
>  https://gitlab.com/imammedo/qemu/-/tree/pcie_poweroff_acpi_regression_rhbz2053584_V2
> 
> Igor Mammedov (3):
>   pci: expose TYPE_XIO3130_DOWNSTREAM name
>   acpi: pcihp: pcie: set power on cap on parent slot
>   q35: compat: keep hotplugged PCIe device broken after migration for
>     6.2 and older machine types
> 
>  include/hw/acpi/pcihp.h                    |  1 +
>  include/hw/pci-bridge/xio3130_downstream.h | 15 +++++++++++++++
>  include/hw/pci/pcie.h                      |  1 +
>  hw/acpi/ich9.c                             | 20 ++++++++++++++++++++
>  hw/acpi/pcihp.c                            | 15 ++++++++++++++-
>  hw/core/machine.c                          |  4 +++-
>  hw/pci-bridge/xio3130_downstream.c         |  3 ++-
>  hw/pci/pcie.c                              | 11 +++++++++++
>  8 files changed, 67 insertions(+), 3 deletions(-)
>  create mode 100644 include/hw/pci-bridge/xio3130_downstream.h
> 
> -- 
> 2.31.1