Hi Kirti, what do you think of the pre-copy interface in this series?
Thanks,
Yulei
> -Original Message-
> From: Zhang, Yulei
> Sent: Tuesday, April 10, 2018 2:02 PM
> To: qemu-devel@nongnu.org
> Cc: Tian, Kevin ; joonas.lahti...@linux.intel.com;
> zhen...@linux.intel.com; kwankh...@nvidia.com; Wang, Zhi A
> ; alex.william...@redhat.com;
> dgilb...@redhat.com; quint...@redhat.com; Zhang, Yulei
>
> Subject: [RFC V4 PATCH 0/4] vfio: Introduce live migation capability to
>
> Summary
>
> This series RFC would like to resume the discussion about how to
> introduce the live migration capability to vfio mdev device.
>
> A new subtype region VFIO_REGION_SUBTYPE_DEVICE_STATE is introduced
> for vfio device status migrate, during the initialization it will
> check if the region is supported by the vfio device, otherwise it
> will remain non-migratable.
>
> The intention to add the new region is using it for mdev device status
> save and restore during the migration. The access to this region
> will be trapped and forward to the mdev device driver, it also uses
> the first byte in the new region to control the running state of mdev
> device, so during the migration after stop the mdev driver, qemu could
> retrieve the specific device status from this region and transfer to
> the target VM side for the mdev device restore.
>
> In addition, during the pre-copy period, it will be able to fetch the
> dirty bitmap of vfio device through ioctl VFIO_DEVICE_GET_DIRTY_BITMAP
> iteratively, which will be able to shorten the system downtime during
> the static copy.
>
> Below is the vfio mdev device migration sequence
> Source VM side:
> start migration
> |
> V
> in pre-copy stage, fetch the device dirty bitmap
> and add into qemu dirty list for migrate iteratively.
> |
> V
>get the cpu state change callback, write to the
>subregion's first byte to stop the mdev device
> |
> V
>quary the dirty page bitmap from iommu container
>and add into qemu dirty list for last synchronization
> |
> V
>save the deivce status into Qemufile which is
> read from the vfio device subregion
>
> Target VM side:
>restore the mdev device after get the
>saved status context from Qemufile
> |
> V
> get the cpu state change callback write to
> subregion's first byte to start the mdev device
> to put it in running status
> |
> V
> finish migration
>
> V3->V4:
> 1. add migration_blocker if device state region isnot supported.
> 2. instead of using vmsd, register SaveVMHandlers for VFIO device
>to leverage the pro-copy facility, and add new ioctl for VFIO
>device to fetch dirty bitmap during pro-copy.
> 3. remove the intel vendor ID dependence for the device state
>subregion.
>
> V2->V3:
> 1. rebase the patch to Qemu stable 2.10 branch.
> 2. use a common name for the subregion instead of specific for
>intel IGD.
>
> V1->V2:
> Per Alex's suggestion:
> 1. use device subtype region instead of VFIO PCI fixed region.
> 2. remove unnecessary ioctl, use the first byte of subregion to
>control the running state of mdev device.
> 3. for dirty page synchronization, implement the interface with
>VFIOContainer instead of vfio pci device.
>
> Yulei Zhang (4):
> vfio: introduce a new VFIO subregion for mdev device migration support
> vfio: Add vm status change callback to stop/restart the mdev device
> vfio: Add SaveVMHanlders for VFIO device to support live migration
> vifo: introduce new VFIO ioctl VFIO_IOMMU_GET_DIRTY_BITMAP
>
> hw/vfio/common.c | 34 ++
> hw/vfio/pci.c | 240
> --
> hw/vfio/pci.h | 2 +
> include/hw/vfio/vfio-common.h | 1 +
> linux-headers/linux/vfio.h| 43 +++-
> roms/seabios | 2 +-
> 6 files changed, 312 insertions(+), 10 deletions(-)
>
> --
> 2.7.4