* Vladimir Sementsov-Ogievskiy (vsement...@yandex-team.ru) wrote:
> [add migration maintainers]
Thanks,
> On 8/11/22 19:49, Andrey Ryabinin wrote:
> > pxb/pxb-pcie/pxb-cxl devices currently doesn't have vmstate description
> > So the state of device is not preserved during migration and
> > guest can notice that as change of PCI_COMMAND_* registers state.
> >
> > The diff of lspci output before and after migration:
> >
> > 00:03.0 Host bridge [0600]: Red Hat, Inc. QEMU PCIe Expander bridge
> > [1b36:000b]
> > Subsystem: Red Hat, Inc QEMU PCIe Expander bridge [1af4:1100]
> > - Control: I/O+ Mem+ BusMaster- SpecCycle- MemWINV- VGASnoop-
> > ParErr- Stepping- SERR+ FastB2B- DisINTx-
> > + Control: I/O- Mem- BusMaster- SpecCycle- MemWINV- VGASnoop-
> > ParErr- Stepping- SERR- FastB2B- DisINTx-
> > Status: Cap- 66MHz+ UDF- FastB2B+ ParErr- DEVSEL=fast >TAbort-
> > <TAbort- <MAbort- >SERR- <PERR- INTx-
Do you notice any other symptoms other than that minor diff?
> > Add VMStateDescription to pxb devices so their state is transferred
> > during migrations. For saving migration compatibility add
> > 'x-config-reg-migration-enabled' property to pxb devices and disable
> > it for all released machine types.
> >
> > Signed-off-by: Andrey Ryabinin <a...@yandex-team.com>
> > ---
> > hw/core/machine.c | 3 +++
> > hw/pci-bridge/pci_expander_bridge.c | 24 ++++++++++++++++++++++++
> > include/hw/pci/pci_bridge.h | 1 +
> > 3 files changed, 28 insertions(+)
> >
> > diff --git a/hw/core/machine.c b/hw/core/machine.c
> > index a673302ccec..071853469e2 100644
> > --- a/hw/core/machine.c
> > +++ b/hw/core/machine.c
> > @@ -43,6 +43,9 @@
> > GlobalProperty hw_compat_7_0[] = {
> > { "arm-gicv3-common", "force-8-bit-prio", "on" },
> > { "nvme-ns", "eui64-default", "on"},
> > + { "pxb", "x-config-reg-migration-enabled", "off" },
> > + { "pxb-cxl", "x-config-reg-migration-enabled", "off" },
> > + { "pxb-pcie", "x-config-reg-migration-enabled", "off" },
>
>
> Seems that it's too late for 7.1, rc2 is out and rc3 is coming soon. And
> that's not a degradation.
Yeh
> Up to maintainers, but I think we'd better start new hw_compat_7_1 for this
> thing.
Yep.
Dave
>
> > };
> > const size_t hw_compat_7_0_len = G_N_ELEMENTS(hw_compat_7_0);
> > diff --git a/hw/pci-bridge/pci_expander_bridge.c
> > b/hw/pci-bridge/pci_expander_bridge.c
> > index c9e817aa586..a3680d4d045 100644
> > --- a/hw/pci-bridge/pci_expander_bridge.c
> > +++ b/hw/pci-bridge/pci_expander_bridge.c
> > @@ -23,6 +23,7 @@
> > #include "qemu/error-report.h"
> > #include "qemu/module.h"
> > #include "sysemu/numa.h"
> > +#include "migration/vmstate.h"
> > #include "hw/boards.h"
> > #include "qom/object.h"
> > @@ -404,9 +405,29 @@ static Property pxb_dev_properties[] = {
> > DEFINE_PROP_UINT8("bus_nr", PXBDev, bus_nr, 0),
> > DEFINE_PROP_UINT16("numa_node", PXBDev, numa_node,
> > NUMA_NODE_UNASSIGNED),
> > DEFINE_PROP_BOOL("bypass_iommu", PXBDev, bypass_iommu, false),
> > + DEFINE_PROP_BOOL("x-config-reg-migration-enabled", PXBDev,
> > + migratable, true),
>
> With it, do we create a user-visible property? If so, is it possible to avoid
> it?
>
> > DEFINE_PROP_END_OF_LIST(),
> > };
> > +static bool pxb_vmstate_needed(void *opaque)
> > +{
> > + PXBDev *pxb = opaque;
> > +
> > + return pxb->migratable;
> > +}
> > +
> > +static const VMStateDescription vmstate_pxb_device = {
> > + .name = "pxb-pci",
> > + .needed = pxb_vmstate_needed,
> > + .version_id = 1,
> > + .minimum_version_id = 1,
> > + .fields = (VMStateField[]) {
> > + VMSTATE_PCI_DEVICE(parent_obj, PXBDev),
> > + VMSTATE_END_OF_LIST()
> > + }
> > +};
> > +
> > static void pxb_dev_class_init(ObjectClass *klass, void *data)
> > {
> > DeviceClass *dc = DEVICE_CLASS(klass);
> > @@ -422,6 +443,7 @@ static void pxb_dev_class_init(ObjectClass *klass, void
> > *data)
> > device_class_set_props(dc, pxb_dev_properties);
> > dc->hotpluggable = false;
> > set_bit(DEVICE_CATEGORY_BRIDGE, dc->categories);
> > + dc->vmsd = &vmstate_pxb_device;
> > }
> > static const TypeInfo pxb_dev_info = {
> > @@ -460,6 +482,7 @@ static void pxb_pcie_dev_class_init(ObjectClass *klass,
> > void *data)
> > device_class_set_props(dc, pxb_dev_properties);
> > dc->hotpluggable = false;
> > set_bit(DEVICE_CATEGORY_BRIDGE, dc->categories);
> > + dc->vmsd = &vmstate_pxb_device;
> > }
> > static const TypeInfo pxb_pcie_dev_info = {
> > @@ -504,6 +527,7 @@ static void pxb_cxl_dev_class_init(ObjectClass *klass,
> > void *data)
> > /* Host bridges aren't hotpluggable. FIXME: spec reference */
> > dc->hotpluggable = false;
> > dc->reset = pxb_dev_reset;
> > + dc->vmsd = &vmstate_pxb_device;
> > }
> > static const TypeInfo pxb_cxl_dev_info = {
> > diff --git a/include/hw/pci/pci_bridge.h b/include/hw/pci/pci_bridge.h
> > index ba4bafac7c2..404dc02e36e 100644
> > --- a/include/hw/pci/pci_bridge.h
> > +++ b/include/hw/pci/pci_bridge.h
> > @@ -91,6 +91,7 @@ struct PXBDev {
> > uint8_t bus_nr;
> > uint16_t numa_node;
> > bool bypass_iommu;
> > + bool migratable;
> > struct cxl_dev {
> > CXLHost *cxl_host_bridge; /* Pointer to a CXLHost */
> > } cxl;
>
>
> --
> Best regards,
> Vladimir
>
--
Dr. David Alan Gilbert / dgilb...@redhat.com / Manchester, UK
