hw/i386/intel_iommu.c | 9 + hw/vfio/Makefile.objs | 2 +- hw/vfio/common.c | 303 +++++++++++++++- hw/vfio/migration.c | 788 ++++++++++++++++++++++++++++++++++++++++++ hw/vfio/pci.c | 203 +++++++++-- hw/vfio/pci.h | 1 - hw/vfio/trace-events | 19 + include/exec/memory.h | 19 + include/hw/vfio/vfio-common.h | 20 ++ linux-headers/linux/vfio.h | 297 +++++++++++++++- memory.c | 13 +- 11 files changed, 1639 insertions(+), 35 deletions(-) create mode 100644 hw/vfio/migration.c
Hi, This Patch set adds migration support for VFIO devices in QEMU. This Patch set include patches as below: Patch 1: - Define KABI for VFIO device for migration support for device state and newly added ioctl definations to get dirty pages bitmap. This is a placeholder patch. Patch 2-4: - Few code refactor - Added save and restore functions for PCI configuration space Patch 5-10: - Generic migration functionality for VFIO device. * This patch set adds functionality only for PCI devices, but can be extended to other VFIO devices. * Added all the basic functions required for pre-copy, stop-and-copy and resume phases of migration. * Added state change notifier and from that notifier function, VFIO device's state changed is conveyed to VFIO device driver. * During save setup phase and resume/load setup phase, migration region is queried and is used to read/write VFIO device data. * .save_live_pending and .save_live_iterate are implemented to use QEMU's functionality of iteration during pre-copy phase. * In .save_live_complete_precopy, that is in stop-and-copy phase, iteration to read data from VFIO device driver is implemented till pending bytes returned by driver are not zero. Patch 11-12 - Add helper function for migration with vIOMMU enabled to get address limit IOMMU supports. - Set DIRTY_MEMORY_MIGRATION flag in dirty log mask for migration with vIOMMU enabled. Patch 13-14: - Add function to start and stop dirty pages tracking. - Add vfio_listerner_log_sync to mark dirty pages. Dirty pages bitmap is queried per container. All pages pinned by vendor driver through vfio_pin_pages external API has to be marked as dirty during migration. When there are CPU writes, CPU dirty page tracking can identify dirtied pages, but any page pinned by vendor driver can also be written by device. As of now there is no device which has hardware support for dirty page tracking. So all pages which are pinned by vendor driver should be considered as dirty. In Qemu, marking pages dirty is only done when device is in stop-and-copy phase because if pages are marked dirty during pre-copy phase and content is transfered from source to distination, there is no way to know newly dirtied pages from the point they were copied earlier until device stops. To avoid repeated copy of same content, pinned pages are marked dirty only during stop-and-copy phase. Patch 15: - With vIOMMU, IO virtual address range can get unmapped while in pre-copy phase of migration. In that case, unmap ioctl should return pages pinned in that range and QEMU should report corresponding guest physical pages dirty. Patch 16: - Make VFIO PCI device migration capable. If migration region is not provided by driver, migration is blocked. Yet TODO: Since there is no device which has hardware support for system memmory dirty bitmap tracking, right now there is no other API from vendor driver to VFIO IOMMU module to report dirty pages. In future, when such hardware support will be implemented, an API will be required in kernel such that vendor driver could report dirty pages to VFIO module during migration phases. Below is the flow of state change for live migration where states in brackets represent VM state, migration state and VFIO device state as: (VM state, MIGRATION_STATUS, VFIO_DEVICE_STATE) Live migration save path: QEMU normal running state (RUNNING, _NONE, _RUNNING) | migrate_init spawns migration_thread. (RUNNING, _SETUP, _RUNNING|_SAVING) Migration thread then calls each device's .save_setup() | (RUNNING, _ACTIVE, _RUNNING|_SAVING) If device is active, get pending bytes by .save_live_pending() if pending bytes >= threshold_size, call save_live_iterate() Data of VFIO device for pre-copy phase is copied. Iterate till pending bytes converge and are less than threshold | On migration completion, vCPUs stops and calls .save_live_complete_precopy for each active device. VFIO device is then transitioned in _SAVING state. (FINISH_MIGRATE, _DEVICE, _SAVING) For VFIO device, iterate in .save_live_complete_precopy until pending data is 0. (FINISH_MIGRATE, _DEVICE, _STOPPED) | (FINISH_MIGRATE, _COMPLETED, STOPPED) Migraton thread schedule cleanup bottom half and exit Live migration resume path: Incomming migration calls .load_setup for each device (RESTORE_VM, _ACTIVE, STOPPED) | For each device, .load_state is called for that device section data | At the end, called .load_cleanup for each device and vCPUs are started. | (RUNNING, _NONE, _RUNNING) Note that: - Migration post copy is not supported. v9 -> v16 - KABI almost finalised on kernel patches. - Added support for migration with vIOMMU enabled. v8 -> v9: - Split patch set in 2 sets, Kernel and QEMU sets. - Dirty pages bitmap is queried from IOMMU container rather than from vendor driver for per device. Added 2 ioctls to achieve this. v7 -> v8: - Updated comments for KABI - Added BAR address validation check during PCI device's config space load as suggested by Dr. David Alan Gilbert. - Changed vfio_migration_set_state() to set or clear device state flags. - Some nit fixes. v6 -> v7: - Fix build failures. v5 -> v6: - Fix build failure. v4 -> v5: - Added decriptive comment about the sequence of access of members of structure vfio_device_migration_info to be followed based on Alex's suggestion - Updated get dirty pages sequence. - As per Cornelia Huck's suggestion, added callbacks to VFIODeviceOps to get_object, save_config and load_config. - Fixed multiple nit picks. - Tested live migration with multiple vfio device assigned to a VM. v3 -> v4: - Added one more bit for _RESUMING flag to be set explicitly. - data_offset field is read-only for user space application. - data_size is read for every iteration before reading data from migration, that is removed assumption that data will be till end of migration region. - If vendor driver supports mappable sparsed region, map those region during setup state of save/load, similarly unmap those from cleanup routines. - Handles race condition that causes data corruption in migration region during save device state by adding mutex and serialiaing save_buffer and get_dirty_pages routines. - Skip called get_dirty_pages routine for mapped MMIO region of device. - Added trace events. - Splitted into multiple functional patches. v2 -> v3: - Removed enum of VFIO device states. Defined VFIO device state with 2 bits. - Re-structured vfio_device_migration_info to keep it minimal and defined action on read and write access on its members. v1 -> v2: - Defined MIGRATION region type and sub-type which should be used with region type capability. - Re-structured vfio_device_migration_info. This structure will be placed at 0th offset of migration region. - Replaced ioctl with read/write for trapped part of migration region. - Added both type of access support, trapped or mmapped, for data section of the region. - Moved PCI device functions to pci file. - Added iteration to get dirty page bitmap until bitmap for all requested pages are copied. Thanks, Kirti Kirti Wankhede (16): vfio: KABI for migration interface - Kernel header placeholder vfio: Add function to unmap VFIO region vfio: Add vfio_get_object callback to VFIODeviceOps vfio: Add save and load functions for VFIO PCI devices vfio: Add migration region initialization and finalize function vfio: Add VM state change handler to know state of VM vfio: Add migration state change notifier vfio: Register SaveVMHandlers for VFIO device vfio: Add save state functions to SaveVMHandlers vfio: Add load state functions to SaveVMHandlers iommu: add callback to get address limit IOMMU supports memory: Set DIRTY_MEMORY_MIGRATION when IOMMU is enabled vfio: Add function to start and stop dirty pages tracking vfio: Add vfio_listener_log_sync to mark dirty pages vfio: Add ioctl to get dirty pages bitmap during dma unmap. vfio: Make vfio-pci device migration capable hw/i386/intel_iommu.c | 9 + hw/vfio/Makefile.objs | 2 +- hw/vfio/common.c | 303 +++++++++++++++- hw/vfio/migration.c | 788 ++++++++++++++++++++++++++++++++++++++++++ hw/vfio/pci.c | 203 +++++++++-- hw/vfio/pci.h | 1 - hw/vfio/trace-events | 19 + include/exec/memory.h | 19 + include/hw/vfio/vfio-common.h | 20 ++ linux-headers/linux/vfio.h | 297 +++++++++++++++- memory.c | 13 +- 11 files changed, 1639 insertions(+), 35 deletions(-) create mode 100644 hw/vfio/migration.c -- 2.7.0
Patchew URL: https://patchew.org/QEMU/1585084154-29461-1-git-send-email-kwankhede@nvidia.com/ Hi, This series failed the docker-quick@centos7 build test. Please find the testing commands and their output below. If you have Docker installed, you can probably reproduce it locally. === TEST SCRIPT BEGIN === #!/bin/bash make docker-image-centos7 V=1 NETWORK=1 time make docker-test-quick@centos7 SHOW_ENV=1 J=14 NETWORK=1 === TEST SCRIPT END === CC x86_64-softmmu/hw/vfio/pci-quirks.o CC aarch64-softmmu/hw/intc/exynos4210_combiner.o /tmp/qemu-test/src/hw/vfio/common.c: In function 'vfio_listerner_log_sync': /tmp/qemu-test/src/hw/vfio/common.c:945:66: error: 'giommu' may be used uninitialized in this function [-Werror=maybe-uninitialized] memory_region_iommu_get_address_limit(giommu->iommu, ^ /tmp/qemu-test/src/hw/vfio/common.c:923:21: note: 'giommu' was declared here VFIOGuestIOMMU *giommu; ^ cc1: all warnings being treated as errors make[1]: *** [hw/vfio/common.o] Error 1 make[1]: *** Waiting for unfinished jobs.... CC aarch64-softmmu/hw/intc/omap_intc.o CC aarch64-softmmu/hw/intc/bcm2835_ic.o --- CC aarch64-softmmu/hw/vfio/amd-xgbe.o CC aarch64-softmmu/hw/virtio/virtio.o CC aarch64-softmmu/hw/virtio/vhost.o make: *** [x86_64-softmmu/all] Error 2 make: *** Waiting for unfinished jobs.... CC aarch64-softmmu/hw/virtio/vhost-backend.o CC aarch64-softmmu/hw/virtio/vhost-user.o --- CC aarch64-softmmu/hw/virtio/virtio-iommu.o CC aarch64-softmmu/hw/virtio/vhost-vsock.o /tmp/qemu-test/src/hw/vfio/common.c: In function 'vfio_listerner_log_sync': /tmp/qemu-test/src/hw/vfio/common.c:945:66: error: 'giommu' may be used uninitialized in this function [-Werror=maybe-uninitialized] memory_region_iommu_get_address_limit(giommu->iommu, ^ /tmp/qemu-test/src/hw/vfio/common.c:923:21: note: 'giommu' was declared here VFIOGuestIOMMU *giommu; ^ cc1: all warnings being treated as errors make[1]: *** [hw/vfio/common.o] Error 1 make[1]: *** Waiting for unfinished jobs.... make: *** [aarch64-softmmu/all] Error 2 Traceback (most recent call last): File "./tests/docker/docker.py", line 664, in <module> sys.exit(main()) --- raise CalledProcessError(retcode, cmd) subprocess.CalledProcessError: Command '['sudo', '-n', 'docker', 'run', '--label', 'com.qemu.instance.uuid=c9b01bcc7fc04e2d8f5e74bf460f0d7a', '-u', '1001', '--security-opt', 'seccomp=unconfined', '--rm', '-e', 'TARGET_LIST=', '-e', 'EXTRA_CONFIGURE_OPTS=', '-e', 'V=', '-e', 'J=14', '-e', 'DEBUG=', '-e', 'SHOW_ENV=1', '-e', 'CCACHE_DIR=/var/tmp/ccache', '-v', '/home/patchew/.cache/qemu-docker-ccache:/var/tmp/ccache:z', '-v', '/var/tmp/patchew-tester-tmp-lne31pn7/src/docker-src.2020-03-24-19.33.46.14149:/var/tmp/qemu:z,ro', 'qemu:centos7', '/var/tmp/qemu/run', 'test-quick']' returned non-zero exit status 2. filter=--filter=label=com.qemu.instance.uuid=c9b01bcc7fc04e2d8f5e74bf460f0d7a make[1]: *** [docker-run] Error 1 make[1]: Leaving directory `/var/tmp/patchew-tester-tmp-lne31pn7/src' make: *** [docker-run-test-quick@centos7] Error 2 real 3m5.634s user 0m8.335s The full log is available at http://patchew.org/logs/1585084154-29461-1-git-send-email-kwankhede@nvidia.com/testing.docker-quick@centos7/?type=message. --- Email generated automatically by Patchew [https://patchew.org/]. Please send your feedback to patchew-devel@redhat.com
On Wed, 25 Mar 2020 02:38:58 +0530 Kirti Wankhede <kwankhede@nvidia.com> wrote: > Hi, > > This Patch set adds migration support for VFIO devices in QEMU. Hi Kirti, Do you have any migration data you can share to show that this solution is viable and useful? I was chatting with Dave Gilbert and there still seems to be a concern that we actually have a real-world practical solution. We know this is inefficient with QEMU today, vendor pinned memory will get copied multiple times if we're lucky. If we're not lucky we may be copying all of guest RAM repeatedly. There are known inefficiencies with vIOMMU, etc. QEMU could learn new heuristics to account for some of this and we could potentially report different bitmaps in different phases through vfio, but let's make sure that there are useful cases enabled by this first implementation. With a reasonably sized VM, running a reasonable graphics demo or workload, can we achieve reasonably live migration? What kind of downtime do we achieve and what is the working set size of the pinned memory? Intel folks, if you've been able to port to this or similar code base, please report your results as well, open source consumers are arguably even more important. Thanks, Alex
On Wed, Apr 01, 2020 at 02:34:24AM +0800, Alex Williamson wrote: > On Wed, 25 Mar 2020 02:38:58 +0530 > Kirti Wankhede <kwankhede@nvidia.com> wrote: > > > Hi, > > > > This Patch set adds migration support for VFIO devices in QEMU. > > Hi Kirti, > > Do you have any migration data you can share to show that this solution > is viable and useful? I was chatting with Dave Gilbert and there still > seems to be a concern that we actually have a real-world practical > solution. We know this is inefficient with QEMU today, vendor pinned > memory will get copied multiple times if we're lucky. If we're not > lucky we may be copying all of guest RAM repeatedly. There are known > inefficiencies with vIOMMU, etc. QEMU could learn new heuristics to > account for some of this and we could potentially report different > bitmaps in different phases through vfio, but let's make sure that > there are useful cases enabled by this first implementation. > > With a reasonably sized VM, running a reasonable graphics demo or > workload, can we achieve reasonably live migration? What kind of > downtime do we achieve and what is the working set size of the pinned > memory? Intel folks, if you've been able to port to this or similar > code base, please report your results as well, open source consumers > are arguably even more important. Thanks, > hi Alex we're in the process of porting to this code, and now it's able to migrate successfully without dirty pages. when there're dirty pages, we met several issues. one of them is reported here (https://lists.gnu.org/archive/html/qemu-devel/2020-04/msg00004.html). dirty pages for some regions are not able to be collected correctly, especially for memory range from 3G to 4G. even without this bug, qemu still got stuck in middle before reaching stop-and-copy phase and cannot be killed by admin. still in debugging of this problem. Thanks Yan
On Wed, 1 Apr 2020 02:41:54 -0400 Yan Zhao <yan.y.zhao@intel.com> wrote: > On Wed, Apr 01, 2020 at 02:34:24AM +0800, Alex Williamson wrote: > > On Wed, 25 Mar 2020 02:38:58 +0530 > > Kirti Wankhede <kwankhede@nvidia.com> wrote: > > > > > Hi, > > > > > > This Patch set adds migration support for VFIO devices in QEMU. > > > > Hi Kirti, > > > > Do you have any migration data you can share to show that this solution > > is viable and useful? I was chatting with Dave Gilbert and there still > > seems to be a concern that we actually have a real-world practical > > solution. We know this is inefficient with QEMU today, vendor pinned > > memory will get copied multiple times if we're lucky. If we're not > > lucky we may be copying all of guest RAM repeatedly. There are known > > inefficiencies with vIOMMU, etc. QEMU could learn new heuristics to > > account for some of this and we could potentially report different > > bitmaps in different phases through vfio, but let's make sure that > > there are useful cases enabled by this first implementation. > > > > With a reasonably sized VM, running a reasonable graphics demo or > > workload, can we achieve reasonably live migration? What kind of > > downtime do we achieve and what is the working set size of the pinned > > memory? Intel folks, if you've been able to port to this or similar > > code base, please report your results as well, open source consumers > > are arguably even more important. Thanks, > > > hi Alex > we're in the process of porting to this code, and now it's able to > migrate successfully without dirty pages. > > when there're dirty pages, we met several issues. > one of them is reported here > (https://lists.gnu.org/archive/html/qemu-devel/2020-04/msg00004.html). > dirty pages for some regions are not able to be collected correctly, > especially for memory range from 3G to 4G. > > even without this bug, qemu still got stuck in middle before > reaching stop-and-copy phase and cannot be killed by admin. > still in debugging of this problem. Thanks, Yan. So it seems we have various bugs, known limitations, and we haven't actually proven that this implementation provides a useful feature, at least for the open source consumer. This doesn't give me much confidence to consider the kernel portion ready for v5.7 given how late we are already :-\ Thanks, Alex
© 2016 - 2024 Red Hat, Inc.