lkml

LKML News v6.13

[LSF/MM/BPF TOPIC] Predictive readahead of dentries (Shyam Prasad N) https://lore.kernel.org/CANT5p=rxLH-D9qSoOWgjYeD87uahmZJMwXp8uNKW66mbv8hmDg@mail.gmail.com Shyam proposes an LSFMM topic for dentries readahead, like that of filesystems. [PATCH v9 00/17] reimplement per-vma lock as a refcount (Suren Baghdasaryan) https://lore.kernel.org/20250111042604.3230628-1-surenb@google.com Suren continues revisioning of his per-vma lock optimization, and posted the ninth version of it, with changed title. [Feedback Needed] Hypervisor live update biweekly meeting time (David Rientjes) https://lore.kernel.org/2908e4ab-abc4-ddd0-b191-fe820856cfb4@google.com David announces start of a bi-weekly syncup for hypervisor live update, which was a topic of last year’s MM alignment session, and asking feedback about the time proposal.

LKML News v6.13-rc7

[PATCH] tools/mm: Introduce a tool to handle entries in allocinfo (Hao Ge) https://lore.kernel.org/20250106112103.25401-1-hao.ge@linux.dev Hao posted a patch adding a tool for handling of the kernel’s memory allocation profilier output. [Invitation] Linux MM Alignment Session on Cgroup-based THP Control on Wednesday (David Rientjes) https://lore.kernel.org/5535fc79-c566-93a0-6134-bfdf1cf5171a@google.com David invites people for next Linux MM Alignment session. The topic will be cgroup-based THP control, led by Gutierez Asier, who proposed the idea. Swap Min Odrer (Daniel Gomez) https://lore.

LKML News v6.13-rc6

[PATCH v3 00/12] AMD broadcast TLB invalidation (Rik van Riel) https://lore.kernel.org/20241230175550.4046587-1-riel@surriel.com Rik posted the third version of the patch series for AMD TLB invalidastion broadcasting support. [PATCH mm-unstable v4 0/7] mm/mglru: performance optimizations (Yu Zhao) https://lore.kernel.org/20241231043538.4075764-1-yuzhao@google.com Yu posted the fourth version of MGLRU performance optimization, which fixes two more regressions. [RFC PATCH 0/9 v2] cgroup: separate per-subsystem rstat trees (JP Kobryn) https://lore.kernel.org/20250103015020.78547-1-inwardvessel@gmail.com JP posted second version of the RFC patch sereis that reduce per-subsystem stat reading overhead by separating the data structure per subsystem.

LKML News v6.13-rc5

Linus 6.13-rc5 (Linus Torvalds) https://lore.kernel.org/<CAHk-=wh=sk+uzFj5edCwdSmUuhfe40mH+G04XQXZtzbx_XBnvQ@mail.gmail.com After yet another calm week of holidays, Linus released the fifth candidate for Linux 6.13. It it again a tiny one, having less than a hundred non-merge commits.

LKML News v6.13-rc4

[RFC PATCH 00/12] khugepaged: Asynchronous mTHP collapse (Dev Jain) https://lore.kernel.org/20241216165105.56185-1-dev.jain@arm.com Dev Jain posted an RFC patch series that makes khugepaged collapses multi-sized THP. [PATCH] mm/page_idle: Constify ‘struct bin_attribute’ (Thomas Weißschuh) https://lore.kernel.org/20241216-sysfs-const-bin_attr-page_idle-v1-1-cc01ecc55196@weissschuh.net Thomas marks struct bin_attribute objects that used in page_idle to be marked as const, since it is now supported. Slow-tier Page Promotion discussion recap and open questions (David Rientjes) https://lore.kernel.org/6d582bb6-3ba5-1768-92f2-6025340a3cd4@google.com David shares his summary of the discussions around asynchronous page promotion approach from AMD that was held on last MM alignment session.

LKML News v6.13-rc3

[Invitation] Linux MM Alignment Session on Slow-tier Page Promotion on Wednesday (David Rientjes) https://lore.kernel.org/52a4867a-f330-4d3f-2da8-0919bc49c6ea@google.com David Rientjes invites MM people to the next alignment session. The topic is slow-tier page promotion, led by RaghavendraKT from AMD. They have RFC, too: https://lore.kernel.org/linux-mm/20241201153818.2633616-1-raghavendra.kt@amd.com/T/#t [PATCH mm-unstable 00/17] mm: Introduce arch_mmap_hint() (Kalesh Singh) https://lore.kernel.org/20241210024119.2488608-1-kaleshsingh@google.com Kalesh posted a patch that introducing per-architecture mmap_hint() implementation and fixing the mmap hint being ignored due to THP in some cases.

LKML News v6.13-rc2

[PATCH] MAINTAINERS: group all VMA-related files into the VMA section (Lorenzo Stoakes) https://lore.kernel.org/20241206191600.45119-1-lorenzo.stoakes@oracle.com Lorenzo posted a patch adding more source files for VMA section of MAINTAINERS file, to avoid people sending patches without Cc-ing VMA maintainers. Linux 6.13-rc2 (Linus Torvalds) https://lore.kernel.org/CAHk-=wgDe9JfGynY9KQ1exkL0DM-cyLYdeYdOxbCXNHm=qAgSQ@mail.gmail.com Linus released the second candidate of Linux v6.13. Some autoamted cleanup changes made the diff bit noisy, and there are some real fixes for drivers, networking and bpf.

LKML News v6.13-rc1

[GIT PULL] MM updates for 6.13-rc1 (Andrew Morton) https://lore.kernel.org/20241118193001.6aefcadd7426feafedf824e1@linux-foundation.org Andrew posted memory management subsystem pull request for v6.13-rc1 to Linus Torvalds. [PATCHSET v5 0/17] Uncached buffered IO (Jens Axboe) https://lore.kernel.org/20241114152743.2381672-2-axboe@kernel.dk Jens is continuing his work on support of buffered IO that isn’t page cache persistent. The motivation is the fact that device speed is getting faster while reclaim is not. [RFCv1 0/6] Page Detective (Pasha Tatashin) https://lore.kernel.org/20241116175922.3265872-1-pasha.tatashin@soleen.com Pasha posted an RFC patch series for better debugging of page status.

LKML News v6.12

[PATCH 0/4] move per-vma lock into vm_area_struct (Suren Baghdasaryan) https://lore.kernel.org/20241111205506.3404479-1-surenb@google.com Suren posted a patch moving vm_lock back into vm_area_struct after confirming the performance regression found before happens only on limited cases (old Broadwell microarchitecture). [PATCH] lib/alloc_tag: Remove the sysctl configuration to prevent users from disabling it at runtime (Hao Ge) https://lore.kernel.org/20241108075004.131911-1-hao.ge@linux.dev Hao tries to remove runtime memory allocation profiling on/off knob, since he believes it is only making users confused.

LKML News v6.12-rc7

[PATCH 0/3] Introduce acctmem (Matthew Wilcox) https://lore.kernel.org/20241104210602.374975-1-willy@infradead.org Willy tries to split out page->memcg_data reference counting using a new abstract, acctmem. [linus:master] [mm, mmap] d4148aeab4: will-it-scale.per_process_ops 3888.9% improvement (kernel test robot) https://lore.kernel.org/202411072132.a8d2cf0f-oliver.sang@intel.com Kernel test robot reports huge improvement on will-it-scale benchmark, coming from commit d4148aeab412 (“mm, mmap: limit THP alignment of anonymous mappings to PMD-aligned sizes”). Linux 6.12-rc7 (Linus Torvalds) https://lore.kernel.org/CAHk-=wiWK-GCmFGNqknDZzWMZM-u58tmP=jZ9ooTo9b8NURvgQ@mail.gmail.com Linus released the seventh candidate for Linux v6.12. Linus says no big surprises found, so he anticipates final 6.