[PATCH] mm/userfaultfd: Support operation on multiple VMAs (Muhammad Usama Anjum) https://lkml.kernel.org/r/20230213104323.1792839-1-usama.anjum@collabora.com
mwriteprotect_range() handle only single VMA, but there are usecases which need to handle multiple VMAs in a memory range of interest. This commit fixes the case.
[v2 PATCH 0/5] Introduce mempool pages bulk allocator and use it in dm-crypt (Yang Shi) https://lkml.kernel.org/r/20230214190221.1156876-1-shy828301@gmail.com
Noticeable overhead on disc encryption due to page allocations have found. The overhead was because the allocation logic in dm-crypt allocates one page in each pass of a loop.
[LSF/MM/BPF TOPIC] sframe: An orc like stack unwinder for the kernel to get a user space stacktrace (Steven Rostedt) https://lkml.kernel.org/r/20230206103828.6efcb28f@rorschach.local.home
Steven Rostedt proposes yet another LSF/MM/BPF topic for using sframe section, which is introduced by binutils, for more efficient user space stack tracing.
[PATCH v4 00/14] Introduce Copy-On-Write to Page Table (Chih-En Lin) https://lkml.kernel.org/r/20230207035139.272707-1-shiyn.lin@gmail.com
Fourth version of the patchset making copy-on-write be used for page table, while it is currently only used for the mapped memory.
[LSF/MM/BPF TOPIC] CXL Fabric Manager (FM) architecture (Viacheslav A.Dubeyko) https://lkml.kernel.org/r/7F001EAF-C512-436A-A9DD-E08730C91214@bytedance.com
A topic suggestion for LSF/MM/BPF. It suggests to discuss about Fabric Manager (FM) architecture, which requires configuration tool, daemon, and QEMU emulation of CXL hardware features.
[PATCH V2 0/3] sched/numa: Enhance vma scanning (Ragghavendra K T) https://lkml.kernel.org/r/cover.1675159422.git.raghavendra.kt@amd.com
A patchset for improving access scanning for NUMA balancing. This uses per-thread VMA scanning idea from Mel.
LSFMMBPF proposal [MM]: Lazy RCU memory reclaim (Joel Fernandes) https://lkml.
[RFC PATCH 00/19] mm: Introduce a cgroup to limit the amount of locked and pinned memory (Alistair Popple) https://lkml.kernel.org/r/cover.f52b9eb2792bccb8a9ecd6bc95055705cfe2ae03.1674538665.git-series.apopple@nvidia.com
This RFC patchset introduces a cgroup limit like RLIMIT_MEMLOCKED but that for pinned pages.
[RFC PATCH 0/4] Fix excessive CPU usage during compaction (Mel Gorman) https://lkml.kernel.org/r/20230125134434.18017-1-mgorman@techsingularity.net
A patch for not ignoring fast_find_migrateblock() found pageblocks made compaction shows high CPU and stalls. It therefore once merged in mainline, than reverted, and again merged in mm-unstable.
[RFC] memory pressure detection in VMs using PSI mechanism for dynamically inflating/deflating VM memory (Sudarshan Rajagopalan) https://lkml.kernel.org/r/DS0PR02MB90787835F5B9CB9771A20329C4C09@DS0PR02MB9078.namprd02.prod.outlook.com
Qualcomm’s Linux memory team asks upstream’s opinion about their userspace deamon approach for controlling VM memory size based on memory demands via monitoring PSI.
[PATCH v1 0/3] Introduce per NUMA node memory error statistics (Jiaqi Yan) https://lkml.kernel.org/r/20230116193902.1315236-1-jiaqiyan@google.com
On huge memory systems, memory error is inevitable, so detailed statistics of it helps administrators to operate systems more efficiently.
!! NOTE !!
This post has migrated to https://damonitor.github.io/posts/damon_evaluation. This out-dated post will be removed soon.
DAMON is lightweight. It increases system memory usage by 0.39% and slows target workloads down by 1.16%.
DAMON is accurate and useful for memory management optimizations. An experimental DAMON-based operation scheme for THP, namely ‘ethp’, removes 76.15% of THP memory overheads while preserving 51.25% of THP speedup. Another experimental DAMON-based ‘proactive reclamation’ implementation, namely ‘prcl’, reduces 93.
[PATCH 00/41] Per-VMA locks (Suren Baghdasaryan) https://lkml.kernel.org/r/20230109205336.3665937-1-surenb@google.com
For the mmap_lock scalability issue, yet another approach called per-VMA locks has posted.
[PATCH v2 0/2] docs/mm: start filling out new structure (Mike Rapoport) https://lkml.kernel.org/r/20230110152358.2641910-1-rppt@kernel.org
MM kernel documentation has restructured to be somewhat more like textbook that can better explaining overall concept and how it manage memory, but the content is empty. Mike starts filling the content.
LSFMMBPF proposal [General/MM]: SRCU: a case study in memory ordering (Joel Fernandes) https://lkml.
[PATCH] zram: correctly handle all next_arg() cases (Sergey Senozhatsky) https://lkml.kernel.org/r/20230103030119.1496358-1-senozhatsky@chromium.org
Handle NULL val pointer case from zram’s arguments handling.
[RFC PATCH 00/25] Upstream kvx Linux port (Yann Sionneau) https://lkml.kernel.org/r/20230103164359.24347-1-ysionneau@kalray.eu
This patchset adds support of Kalray’s kvx family kv3-1 CPU architecture.
[PATCH 00/46] Based on latest mm-unstable (85b44c25cd1e). (James Houghton) https://lkml.kernel.org/r/20230105101844.1893104-1-jthoughton@google.com
This patchset introduces HugeTLB high-Granularity Mapping (HGM), which teaches HugeTLB how to map HugeTLB pages at high granularity like how THPs PTE-mapped.
As today is the last day of 2022, I checked some open source commits statistics using my simple and buggy script[1], and github.
Linux kernel statistics I ran my buggy script[1] to show some 2022 statistics for memory management subsystem and whole tree of Linux kernel as below.
Memory Management In 2022, 305 people participated in Linux kernel memory management subsystem development by making 81,631 lines of changes with 2,135 commits.
A summary of DAMON development in 2022 has posted: https://lore.kernel.org/damon/20221229171209.162356-1-sj@kernel.org/
2022 was a year of active and healthy DAMON development.
Seven new DAMON major features were delivered to users. Some of those were featured in articles and academic papers.
It was possible thanks to the DAMON community. The community has expanded with its own mailing list and an open bi-weekly chat series. 40 people contributed their great code to DAMON via making their 275 commits merged into the mainline.