On Tue, Jan 11, 2022 at 3:55 AM Marc Zyngier <maz@xxxxxxxxxx> wrote: > > On Mon, 10 Jan 2022 21:04:38 +0000, > Jing Zhang <jingzhangos@xxxxxxxxxx> wrote: > > > > This patch is to reduce the performance degradation of guest workload during > > dirty logging on ARM64. A fast path is added to handle permission relaxation > > during dirty logging. The MMU lock is replaced with rwlock, by which all > > permision relaxations on leaf pte can be performed under the read lock. This > > greatly reduces the MMU lock contention during dirty logging. With this > > solution, the source guest workload performance degradation can be improved > > by more than 60%. > > > > Problem: > > * A Google internal live migration test shows that the source guest workload > > performance has >99% degradation for about 105 seconds, >50% degradation > > for about 112 seconds, >10% degradation for about 112 seconds on ARM64. > > This shows that most of the time, the guest workload degradtion is above > > 99%, which obviously needs some improvement compared to the test result > > on x86 (>99% for 6s, >50% for 9s, >10% for 27s). > > * Tested H/W: Ampere Altra 3GHz, #CPU: 64, #Mem: 256GB > > * VM spec: #vCPU: 48, #Mem/vCPU: 4GB > > What are the host and guest page sizes? Both are 4K and guest mem is 2M hugepage backed. Will add the info for future posts. > > > > > Analysis: > > * We enabled CONFIG_LOCK_STAT in kernel and used dirty_log_perf_test to get > > the number of contentions of MMU lock and the "dirty memory time" on > > various VM spec. > > By using test command > > ./dirty_log_perf_test -b 2G -m 2 -i 2 -s anonymous_hugetlb_2mb -v [#vCPU] > > How is this test representative of the internal live migration test > you mention above? '-m 2' indicates a mode that varies depending on > the HW and revision of the test (I just added a bunch of supported > modes). Which one is it? The "dirty memory time" is the time vCPU threads spent in KVM after fault. Higher "dirty memory time" means higher degradation to guest workload. '-m 2' indicates mode "PA-bits:48, VA-bits:48, 4K pages". Will add this for future posts. > > > Below are the results: > > +-------+------------------------+-----------------------+ > > | #vCPU | dirty memory time (ms) | number of contentions | > > +-------+------------------------+-----------------------+ > > | 1 | 926 | 0 | > > +-------+------------------------+-----------------------+ > > | 2 | 1189 | 4732558 | > > +-------+------------------------+-----------------------+ > > | 4 | 2503 | 11527185 | > > +-------+------------------------+-----------------------+ > > | 8 | 5069 | 24881677 | > > +-------+------------------------+-----------------------+ > > | 16 | 10340 | 50347956 | > > +-------+------------------------+-----------------------+ > > | 32 | 20351 | 100605720 | > > +-------+------------------------+-----------------------+ > > | 64 | 40994 | 201442478 | > > +-------+------------------------+-----------------------+ > > > > * From the test results above, the "dirty memory time" and the number of > > MMU lock contention scale with the number of vCPUs. That means all the > > dirty memory operations from all vCPU threads have been serialized by > > the MMU lock. Further analysis also shows that the permission relaxation > > during dirty logging is where vCPU threads get serialized. > > > > Solution: > > * On ARM64, there is no mechanism as PML (Page Modification Logging) and > > the dirty-bit solution for dirty logging is much complicated compared to > > the write-protection solution. The straight way to reduce the guest > > performance degradation is to enhance the concurrency for the permission > > fault path during dirty logging. > > * In this patch, we only put leaf PTE permission relaxation for dirty > > logging under read lock, all others would go under write lock. > > Below are the results based on the solution: > > +-------+------------------------+ > > | #vCPU | dirty memory time (ms) | > > +-------+------------------------+ > > | 1 | 803 | > > +-------+------------------------+ > > | 2 | 843 | > > +-------+------------------------+ > > | 4 | 942 | > > +-------+------------------------+ > > | 8 | 1458 | > > +-------+------------------------+ > > | 16 | 2853 | > > +-------+------------------------+ > > | 32 | 5886 | > > +-------+------------------------+ > > | 64 | 12190 | > > +-------+------------------------+ > > All "dirty memory time" have been reduced by more than 60% when the > > number of vCPU grows. > > How does that translate to the original problem statement with your > live migration test? Based on the solution, the test results from the Google internal live migration test also shows more than 60% improvement with >99% for 30s, >50% for 58s and >10% for 76s. Will add this info in to future posts. > > Thanks, > > M. > > -- > Without deviation from the norm, progress is not possible. Thanks, Jing
