Pmem December 2021

pmem@lists.openanolis.cn

3 参与者
3 讨论

[PATCH 00/15] More optimization and improvements for tiered memory system

by Baolin Wang

Hi, This patch set backported more optimization and improvements for tiered memory system from git branch [1]. Please find details in each patch. Thanks. Note: Almost patches have no logic conflicts except patch 4, since we should promote pages one by one in K4.19. [1] https://git.kernel.org/pub/scm/linux/kernel/git/vishal/tiering.git/log/?h=t… Feng Tang (3): mm/migrate: make migrate_misplaced_page works for unmapped file pages memory tiering: support unmapped file cache pages promotion mm/migrate: add vm counters to track reasons of page migration failure Huang Ying (12): memory tiering: add trace point for threshold adjustment memory tiering: add node-vmstat for promote threshold NUMA balancing: migrate dirty file cache pages memory tiering: add file pages promotion counter memory tiering: add file pages demotion counter memory tiering: fix THP failed to be isolated live lock memory tiering: check order in migrate_balanced_pgdat() memory tiering: avoid kswap to be keep in failure state for long memory tiering: accelerate promotion threshold adjustment memory tiering: extend promotion threshold range memory tiering: loosen per second rate limit memory tiering: double hot threshold for write hint page fault include/linux/mempolicy.h | 4 +- include/linux/migrate.h | 11 +++- include/linux/mmzone.h | 5 ++ include/linux/sched/numa_balancing.h | 5 +- include/linux/vm_event_item.h | 5 ++ include/trace/events/sched.h | 24 +++++++++ kernel/sched/fair.c | 63 +++++++++++++++-------- mm/filemap.c | 26 +++++++++- mm/huge_memory.c | 2 +- mm/memory.c | 5 +- mm/mempolicy.c | 6 ++- mm/migrate.c | 98 +++++++++++++++++++++++++----------- mm/mprotect.c | 8 --- mm/page_alloc.c | 3 ++ mm/vmscan.c | 23 ++++++++- mm/vmstat.c | 8 +++ 16 files changed, 227 insertions(+), 69 deletions(-) -- 1.8.3.1

2 years, 11 months

[PATCH Anolis tiered-memory 0/6] NUMA balancing: optimize memory placement for memory tiering system

by zhongjiang-ali

The patch come from upstream contributed by Huang Ying to optimize memory placement for memory tiering system. I review and backport the patches to anolis-4.19. More details can see in following link. https://lkml.org/lkml/2021/12/6/2203 Huang Ying (6): NUMA Balancing: add page promotion counter NUMA balancing: optimize page placement for memory tiering system memory tiering: skip to scan fast memory memory tiering: hot page selection with hint page fault latency memory tiering: rate limit NUMA migration throughput memory tiering: adjust hot threshold automatically Documentation/sysctl/kernel.txt | 32 ++++++--- include/linux/mm.h | 29 ++++++++ include/linux/mmzone.h | 11 +++ include/linux/node.h | 5 ++ include/linux/sched/sysctl.h | 16 +++++ kernel/sched/core.c | 36 ++++++++-- kernel/sched/fair.c | 146 ++++++++++++++++++++++++++++++++++++++++ kernel/sysctl.c | 19 +++++- mm/huge_memory.c | 42 +++++++++--- mm/memory.c | 11 ++- mm/migrate.c | 44 ++++++++++-- mm/mmzone.c | 17 +++++ mm/mprotect.c | 19 +++++- mm/vmscan.c | 16 +++++ mm/vmstat.c | 4 ++ 15 files changed, 414 insertions(+), 33 deletions(-) -- 1.8.3.1

3 years

[PATCH 00/17] Support page demotion on tiered memory system

by Baolin Wang

Hi, This patch set backported several patches from mainline to support page demotion for tiered memory system. Please find more details in each patch. Thanks. Baolin Wang (3): anolis: mm: migrate: introduce a standard migration target allocation function mm: migrate: support multiple target nodes demotion mm: migrate: add more comments for selecting target node randomly Dave Hansen (7): mm/numa: automatically generate node migration order mm/migrate: update node demotion order on hotplug events mm/migrate: demote pages during reclaim mm/vmscan: add helper for querying ability to age anonymous pages mm/vmscan: never demote for memcg reclaim mm/migrate: optimize hotplug-time demotion order updates mm/migrate: add CPU hotplug to demotion #ifdef Huang Ying (3): mm/migrate: add sysfs interface to enable reclaim migration mm/migrate: fix CPUHP state to update node demotion order mm/migrate: move node demotion code to near its user Keith Busch (1): mm/vmscan: Consider anonymous pages without swap Yang Shi (3): mm/migrate: enable returning precise migrate_pages() success count mm/vmscan: add page demotion counter mm: migrate: make demotion knob depend on migration Documentation/ABI/testing/sysfs-kernel-mm-numa | 24 ++ include/linux/cpuhotplug.h | 4 + include/linux/memory.h | 5 +- include/linux/mempolicy.h | 1 + include/linux/migrate.h | 18 +- include/linux/vm_event_item.h | 2 + include/trace/events/migrate.h | 3 +- mm/compaction.c | 2 +- mm/internal.h | 6 + mm/memory-failure.c | 4 +- mm/memory_hotplug.c | 2 +- mm/mempolicy.c | 4 +- mm/migrate.c | 530 ++++++++++++++++++++++++- mm/page_alloc.c | 4 +- mm/page_ext.c | 4 +- mm/slab.c | 4 +- mm/vmscan.c | 136 ++++++- mm/vmstat.c | 2 + 18 files changed, 730 insertions(+), 25 deletions(-) create mode 100644 Documentation/ABI/testing/sysfs-kernel-mm-numa -- 1.8.3.1

3 years

2024

2023

2022

2021

Pmem December 2021