| Message ID | 1544516693-5395-1-git-send-email-wei.w.wang@intel.com (mailing list archive) |
|---|---|
| Headers | show |
| Series | virtio-balloon: free page hint support | expand |
On Tue, Dec 11, 2018 at 04:24:46PM +0800, Wei Wang wrote: > This is the deivce part implementation to add a new feature, > VIRTIO_BALLOON_F_FREE_PAGE_HINT to the virtio-balloon device. The device > receives the guest free page hints from the driver and clears the > corresponding bits in the dirty bitmap, so that those free pages are > not sent by the migration thread to the destination. For virtio bits: Reviewed-by: Michael S. Tsirkin <mst@redhat.com> I think this is primarily a migration feature so please merge through the migration tree. > *Tests > 1 Test Environment > Host: Intel(R) Xeon(R) CPU E5-2699 v4 @ 2.20GHz > Migration setup: migrate_set_speed 100G, migrate_set_downtime 400ms > > 2 Test Results (results are averaged over several repeated runs) > 2.1 Guest setup: 8G RAM, 4 vCPU > 2.1.1 Idle guest live migration time > Optimization v.s. Legacy = 620ms vs 2970ms > --> ~79% reduction > 2.1.2 Guest live migration with Linux compilation workload > (i.e. make bzImage -j4) running > 1) Live Migration Time: > Optimization v.s. Legacy = 2273ms v.s. 4502ms > --> ~50% reduction > 2) Linux Compilation Time: > Optimization v.s. Legacy = 8min42s v.s. 8min43s > --> no obvious difference > > 2.2 Guest setup: 128G RAM, 4 vCPU > 2.2.1 Idle guest live migration time > Optimization v.s. Legacy = 5294ms vs 41651ms > --> ~87% reduction > 2.2.2 Guest live migration with Linux compilation workload > 1) Live Migration Time: > Optimization v.s. Legacy = 8816ms v.s. 54201ms > --> 84% reduction > 2) Linux Compilation Time: > Optimization v.s. Legacy = 8min30s v.s. 8min36s > --> no obvious difference > > ChangeLog: > v10->v11: > migration: > - qemu_guest_free_page_hint: > - "offset >= block->used_length", instead of > "offset > block->used_length"; > - RAMState: enable the "fpo_enabled" flag, when the free page > optimization feature is used, instead of disabling ram_bulk_stage. > Please see patch 6 commit log for details. > > Previous changelog: > http://lists.nongnu.org/archive/html/qemu-devel/2018-12/msg00055.html > > Wei Wang (7): > bitmap: fix bitmap_count_one > bitmap: bitmap_count_one_with_offset > migration: use bitmap_mutex in migration_bitmap_clear_dirty > migration: API to clear bits of guest free pages from the dirty bitmap > migration/ram.c: add a notifier chain for precopy > migration/ram.c: add the free page optimization enable flag > virtio-balloon: VIRTIO_BALLOON_F_FREE_PAGE_HINT > > hw/virtio/virtio-balloon.c | 263 ++++++++++++++++++++++++ > include/hw/virtio/virtio-balloon.h | 28 ++- > include/migration/misc.h | 22 ++ > include/qemu/bitmap.h | 17 ++ > include/standard-headers/linux/virtio_balloon.h | 5 + > migration/ram.c | 121 ++++++++++- > migration/savevm.c | 15 ++ > vl.c | 1 + > 8 files changed, 466 insertions(+), 6 deletions(-) > > -- > 1.8.3.1
This is the deivce part implementation to add a new feature, VIRTIO_BALLOON_F_FREE_PAGE_HINT to the virtio-balloon device. The device receives the guest free page hints from the driver and clears the corresponding bits in the dirty bitmap, so that those free pages are not sent by the migration thread to the destination. *Tests 1 Test Environment Host: Intel(R) Xeon(R) CPU E5-2699 v4 @ 2.20GHz Migration setup: migrate_set_speed 100G, migrate_set_downtime 400ms 2 Test Results (results are averaged over several repeated runs) 2.1 Guest setup: 8G RAM, 4 vCPU 2.1.1 Idle guest live migration time Optimization v.s. Legacy = 620ms vs 2970ms --> ~79% reduction 2.1.2 Guest live migration with Linux compilation workload (i.e. make bzImage -j4) running 1) Live Migration Time: Optimization v.s. Legacy = 2273ms v.s. 4502ms --> ~50% reduction 2) Linux Compilation Time: Optimization v.s. Legacy = 8min42s v.s. 8min43s --> no obvious difference 2.2 Guest setup: 128G RAM, 4 vCPU 2.2.1 Idle guest live migration time Optimization v.s. Legacy = 5294ms vs 41651ms --> ~87% reduction 2.2.2 Guest live migration with Linux compilation workload 1) Live Migration Time: Optimization v.s. Legacy = 8816ms v.s. 54201ms --> 84% reduction 2) Linux Compilation Time: Optimization v.s. Legacy = 8min30s v.s. 8min36s --> no obvious difference ChangeLog: v10->v11: migration: - qemu_guest_free_page_hint: - "offset >= block->used_length", instead of "offset > block->used_length"; - RAMState: enable the "fpo_enabled" flag, when the free page optimization feature is used, instead of disabling ram_bulk_stage. Please see patch 6 commit log for details. Previous changelog: http://lists.nongnu.org/archive/html/qemu-devel/2018-12/msg00055.html Wei Wang (7): bitmap: fix bitmap_count_one bitmap: bitmap_count_one_with_offset migration: use bitmap_mutex in migration_bitmap_clear_dirty migration: API to clear bits of guest free pages from the dirty bitmap migration/ram.c: add a notifier chain for precopy migration/ram.c: add the free page optimization enable flag virtio-balloon: VIRTIO_BALLOON_F_FREE_PAGE_HINT hw/virtio/virtio-balloon.c | 263 ++++++++++++++++++++++++ include/hw/virtio/virtio-balloon.h | 28 ++- include/migration/misc.h | 22 ++ include/qemu/bitmap.h | 17 ++ include/standard-headers/linux/virtio_balloon.h | 5 + migration/ram.c | 121 ++++++++++- migration/savevm.c | 15 ++ vl.c | 1 + 8 files changed, 466 insertions(+), 6 deletions(-)