linux

LKML News v6.5-rc2

[Invitation] Linux MM Alignment Session on mshare on Wednesday (David Rientjes) https://lkml.kernel.org/r/CAE-26VAZZ+buEX0wbqHTz0k1hshYzYP3KzMmXT6jc6=NucPa0Q@mail.gmail.com David from Google is hosting yet another Linux MM Alignment session on Wednesday. The topic of this week is the process page table sharing, which could be a clear efficiency win for situations where huge number of processes sharing same pages. [PATCH v3 0/7] Add support for memmap on memory feature on ppc64 (Aneesh Kumar K.V) https://lkml.kernel.org/r/20230711044834.72809-1-aneesh.kumar@linux.ibm.com This makes the memmap on memory feature more useful for ppc64-like architectures having alignment rules that different with 64K page size.

I will have a talk at the Open Source Summit Europe 2023

DAMON talk for OSSummit EU 2023 has accepted and scheduled. The talk will focus on its user-space tool, damo. It’s mainly for helping more audiences to digest the content easier, and also for accelerating the development of damo. By the time, hopefully the version of damo would reach to, or exceed 2.0.0, and newer interface and features will be introduced together. The title of the talk is “Data Access Monitoring Operator (DAMO): User-Space Tool/Python Library for Access-Aware Profiling and Optimization of Your Linux Systems”.

LKML News v6.5-rc1

[GIT PULL] MM updates for 6.5-rc1 (Andrew Morton) https://lkml.kernel.org/r/20230626085035.e66992e96b4c6d37dad54bd9@linux-foundation.org Andrew Morton sent the memory management subsystem pull request for v6.5-rc1. Among the change, cachestata() syscall and Mel Gorman’s compaction optimization looks interesting to me. [GIT PULL] slab updates for 6.5 (Vlastimil Babka) https://lkml.kernel.org/r/1c39c9b0-ec37-f910-2b09-cedf7acf6e91@suse.cz Vlastimil sent the slab pull request for v6.5 to Linus Torvalds. This contains the SLAB config deprecation renaming (s/CONFIG_SLAB/CONFIG_SLAB_DEPRECATED/). [PATCH v1 0/5] mm/memory_hotplug: make offline_and_remove_memory() timeout instead of failing on fatal signals (David Hildenbrand) https://lkml.

LKML News v6.4

[PATCH] mm/memory_hotplug.c: don’t fail hot unplug quite so eagerly (John Hubbard) https://lkml.kernel.org/r/20230620011719.155379-1-jhubbard@nvidia.com Some device drivers add and remove memory to the system via memory hot [un]plog. Nevertheless, the ungplugging could fail easily, especially for the pending signal on the process. This patch simply remove the signal check error. [PATCH v2] mm/min_free_kbytes: modify min_free_kbytes calculation rules (liuq) https://lkml.kernel.org/r/20230621073912.21747-1-liuq131@chinatelecom.cn This patch adjusts min_free_kbytes calculation so that ZONE_NORMAL is unnecessarily be too small when ZONE_MOVABLE exists.

LKML News v6.4-rc7

[GIT PULL] hotfixes for 6.4-rc7 (Andrew Morton) https://lkml.kernel.org/r/20230612122712.f495e24b80f88f58557ea470@linux-foundation.org Andrew sent hot fixes pull request to Linus Torvalds. The one for DAMON is also contained. [PATCH stable 5.10 0/1] Fix memleak during hotremove memory (Ma Wupeng) https://lkml.kernel.org/r/20230614061900.3296725-1-mawupeng1@huawei.com Ma found a memory leak in memory hotplugging, and providing a patch. [ANNOUNCE/CFP] CXL Microconference at LPC 2023 (Davidlohr Bueso) https://lkml.kernel.org/r/a4c2gx2tnm4ckax7qkx2trnvmqjssfytc45sb2zikuayd2marc@rpsjp4icgsvn Davidlohr announces CXL microconference has accepted for LPC this year, and inviting people in the call for participation.

LKML News v6.4-rc6

As of 2023-06-06, for some reason https://lore.kernel.org/linux-mm seems stopped being updated since 2023-05-31. On the day of 2023-06-06, the update has resumed. [PATCH v3 0/5] riscv: Introduce KASLR (Alexandre Ghiti) https://lkml.kernel.org/r/20230606123242.20804-1-alexghiti@rivosinc.com This patchset introduces KASLR, which allows kernel mapping randomization, for riscv. Question about oom-killer (Gou Hao) https://lore.kernel.org/linux-mm/A31B3EA06AF77F13+c7278bd0-2695-4cc0-5a56-531a06ac2389@uniontech.com/ A question about oom-killer behavior that suspected to be a kernel bug. [PATCH v3] Documentation/mm: Initial page table documentation (Linus Walleij) https://lkml.kernel.org/r/20230608125501.3960093-1-linus.walleij@linaro.org

LKML News v6.4-rc5

[PATCH 00/12] mm: free retracted page table by RCU (Hugh Dickins) https://lkml.kernel.org/r/35e983f5-7ed3-b310-d949-9ae8b130cdab@google.com This patchset aims to reduce mmap_lock contention by freeing retracted page table by RCU, with preceding and succeeding patchsets. This patchset was in Google data center kernel for three years. [PATCH] mm: madvise: fix uneven accounting of psi (Charan Teja Kalla) https://lkml.kernel.org/r/1685531374-6091-1-git-send-email-quic_charante@quicinc.com For madvise(MADV_PAGEOUT) is called for folios, PSI for the folio is not always accounted. This patchset makes it be consistent by always not considering PSI for folios that are reclaimed through madvise(MADV_PAGEOUT).

LKML News v6.4-rc4

[PATCH 0/9] Mitigate a vmap lock contention (Uladzislau Rezki) https://lkml.kernel.org/r/20230522110849.2921-1-urezki@gmail.com Uladzislau sent a patchset for vmap lock contention mitigation, as suggested at LSFMM+BPF. [PATCH] mm: deduct the number of pages reclaimed by madvise from workingset (Zhaoyang Huang) https://lkml.kernel.org/r/1684919574-28368-1-git-send-email-zhaoyang.huang@unisoc.com Pages reclaimed by madvise() with MADV_PAGEOUT hint doesn’t handle the workingset, hence leads to large refault distance than it should be. This patch addresses the case by making reclaim_pages() to receive mm_struct and deduct the pages from workingset.

LKML News v6.4-rc3

[PATCH v3 00/11] Introduce cmpxchg128() – aka. the demise of cmpxchg_double() (Peter Zijlstra) https://lkml.kernel.org/r/20230515075659.118447996@infradead.org This patchset introduces an alternative for cmpxchg_double(), namely cmpxchg128(), and replace the old usages of cmpxchg_double(). This is because Linus hates cmpxchg_double()[1]. [1] https://lore.kernel.org/all/Y2U3WdU61FvYlpUh@hirez.programming.kicks-ass.net/T/#u [PATCH 0/6] Memory Mapping (VMA) protection using PKU - set 1 (Jeff Xu) https://lkml.kernel.org/r/20230515130553.2311248-1-jeffxu@chromium.org [Invitation] Linux MM Alignment Session on Code Tagging on Wednesday (David Rientjes) https://lkml.kernel.org/r/CAE-26VDomF_pZybGjgCBJZuWusX0z3MTEJmkwtuQXx3PyrN6fw@mail.gmail.com Google is hosting Linux MM alignment session.

LKML News v6.4-rc2

[PATCH 0/2] Optimization of CMA allocation and reclaiming (Zhaoyang Huang) https://lkml.kernel.org/r/1683538383-19685-1-git-send-email-zhaoyang.huang@unisoc.com This patchset avoids being OOMed when all free pages are composed of CMA. [RFC PATCH 0/2] mm: Working Set Reporting (Yuanchu Xie) https://lkml.kernel.org/r/20230509185419.1088297-1-yuanchu@google.com This RFC patch adds working set size reporting structure for each node and total, using MGLRU as its core, and extends virtio-balloon for letting the host get the information. [PATCH] MAINTAINERS: repair pattern in DIALOG SEMICONDUCTOR DRIVERS (Lukas Bulwahn) https://lkml.