kernel

LKML News v5.11

[PATCH v2] mm: cma: support sysfs (Minchan Kim) https://lkml.kernel.org/r/20210208180142.2765456-1-minchan@kernel.org This patch introduces sysfs interface for CMA that shows stats including number of CMA allocation attempts and failures. [PATCH v10 00/14] prohibit pinning pages in ZONE_MOVABLE (Pavel Tatashin) https://lkml.kernel.org/r/20210211162427.618913-1-pasha.tatashin@soleen.com The 10th version of the patchset making ZONE_MOVABLE really movable. Linux 5.11 (Linus Torvalds) https://lkml.kernel.org/r/CAHk-=wg8LpRi9+P2_V+ehJ2_EoJifNh2QmVf47aLXfkCd-1UAQ@mail.gmail.com Nothing strange happened last week, so Linus released v5.11 as expected. The number of changes from the rc7 to this release is actually even smaller than average.

LKML News v5.11-rc7

[RFC 00/20] TLB batching consolidation and enhancements (Nadav Amit) https://lkml.kernel.org/r/20210131001132.3368247-1-namit@vmware.com This patchset prepares more aggressive TLB flushing batching by consolidating a few TLB batching methods. [PATCH v9 00/14] prohibit pinning pages in ZONE_MOVABLE (Pavel Tatashin) https://lkml.kernel.org/r/20210201153827.444374-1-pasha.tatashin@soleen.com This patchset makes ZONE_MOVABLE really movable by prohibiting pinning of pages in the zone. Augmented Page Reclaim (Yu Zhao) https://lkml.kernel.org/r/YBkT6175GmMWBvw3@google.com This mail shares a part of decade researh/experimentation in memory overcommit at Google. The author wants to start submitting patch based on responses.

LKML News v5.11-rc6

[RFC][PATCH 00/13] [v5] Migrate Pages in lieu of discard (Dave Hansen) https://lkml.kernel.org/r/20210126003411.2AC51464@viggo.jf.intel.com If a system has a small DRAM and a large persistent memory (PMEM), reclamation starts when DRAM fills up, and newer allocations will done on PMEM. This patchset makes the reclaim logic to migrate pages to PMEM instead of just discarding it. [PATCH v3 00/25] Page folios (Matthew Wilcox) https://lkml.kernel.org/r/20210128070404.1922318-1-willy@infradead.org This patchset introduces a new data structure called page folio.

LKML News v5.11-rc5

[PATCH v2 00/27] Page folios (Matthew Wilcox) https://lkml.kernel.org/r/20210118170148.3126186-1-willy@infradead.org This is the second version of the page folios patchset. This introduces a new struct named folio. Functions that receive folio as an argument always receives head page, so the function don’t need to do compound_head(). [PATCH -V9 0/3] numa balancing: Migrate on fault among multiple bound nodes (Huang Ying) https://lkml.kernek.org/r/20210120061235.148637-1-ying.huang@intel.com This patch make optimization of cross-socket memory accesses for AutoNUMA in case of application is bound to multiple NUMA nodes possible.

LKML News v5.11-rc4

[PATCH] mm: thp: introduce NR_PARTIAL_THPS (Bin Wang) https://lkml.kernel.org/r/20210112021208.1875-1-wangbin224@huawei.com Because partially unmapped THP is not split, the memory stat becomes confusing. This patch add another counter called NR_PARTIAL_THPS in node_stat_item to reduce such confusions. [PATCH v2 00/70] RFC mm: Introducing the Maple Tree (Liam R. Howlett) https://lkml.kernel.org/r/20210112161240.2024684-1-Liam.Howlett@Oracle.com Second RFC of the Maple tree. It is an RCU-safe B-tree that aims to be used as the vma tree with better scalability. [PATCH] mm/debug: Improve memcg debugging (Matthew Wilcox) https://lkml.

LKML News v5.11-rc3

[PATCH] mm: vmscan: support complete shrinker reclaim (Sudarshan Rajagopalan) https://lkml.kernel.org/r/2d1f1dbb7e018ad02a9e7af36a8c86397a1598a7.1609892546.git.sudaraja@codeaurora.org This patch makes caches can dropped even when those are smaller than the batch size. [PATCH 0/6] hugetlbfs: support free page reporting (Liang Li) https://lkml.kernel.org/r/20210106034623.GA1128@open-light-1.localdomain Because hugetlbfs is using reserved memory, it’s not reported by the free page reporting feature. This patch makes it able to be reported. Linux 5.11-rc3 (Linus Torvalds) https://lkml.kernel.org/r/CAHk-=wizR7--Ub-rasYx2S5XURkooCQ63Sat23BpvTAuN0scQA@mail.gmail.com The prediction of Linus for this release was wrong.

LKML News v5.11-rc2

Linux 5.11-rc2 (Linus Torvalds) https://lkml.kernel.org/r/CAHk-=whS+rzNMSsU6vRoLMzrm0JPN-OVg+BxhybMet3NpqpR-Q@mail.gmail.com Due to the holiday season, rc2 ends up fairly small. Maybe we will see rc8 for v5.11, but Torvalds just saying let’s watch more. Below is the diffstat of the releases in the last two years. Note that the y-axis is in logarithm. I draw it using https://github.com/sjp38/relstat and https://github.com/sjp38/lazybox using below command: $ relstat.py --since 2019-01-04 | ~/lazybox/gnuplot/plot.py \ --data_fmt table --type labeled-lines --xtics_rotate -90 \ --font "Times New Roman, 5pt" --ylog --pointsize 0.

LKML News v5.10-rc1

incoming (Andrew Morton) https://lkml.kernel.org/r/20201214190237.a17b70ae14f129e2dca3d204@linux-foundation.org The pull request for v5.11 from Andrew Morton. The mmap_lock tracepoint patchset series is included in here. incoming (Andrew Morton) https://lkml.kernel.org/r/20201215123253.954eca9a5ef4c0d52fd381fa@linux-foundation.org Another pull request from Andrew Morton for memcg scalability improvement. This makes per-memcg lru lock. [PATCH 00/25] Page folios (Matthew Wilcox) https://lkml.kernel.org/r/20201216182335.27227-1-willy@infradead.org This is the patchset for the page folios, which Matthew proposed a few days ago and discussed via zoom. Do we still need skip swapcache logic in do_swap_page() for SWP_SYNCHRONOUS_IO?

LKML News v5.10

[RFC V2 00/37] Enhance memory utilization with DMEMFS (Yuley Zhang) https://lkml.kernel.org/r/cover.1607332046.git.yuleixzhang@tencent.com This patchset aims to remove the ‘struct page’ from user-sepcified region of memory and use a file system called ‘dmemfs’ to minimize the memory region dedicated for ‘struct page’s. [PATCH] mm: don’t SetPageWorkingset unconditionally during swapin (Yu Zhao) https://lkml.kernel.org/r/20201209012400.1771150-1-yuzhao@google.com Unconditional ‘SetPageWorkingset’ miscategorizes read-ahead pages or pages that never belong to the working set as working set because it’s called unconditionally.

LKML News v5.10-rc7

[PATCH] mm: mmap_lock: fix use-after-free race and css ref leak in tracepoints (Axel Rasmussen) https://lkml.kernel.org/r/20201130233504.3725241-1-axelrasmussen@google.com This patch fixes a use-after-free bug that reported by syzbot. [PATCH 0/6] prohibit pinning pages in ZONE_MOVABLE (Pavel Tatashin) https://lkml.kernel.org/r/20201202052330.474592-1-pasha.tatashin@soleen.com Pinned pages in ZONE_MOVABLE is a wellknown problem. This patchset fixes the issue by forcing pinning new page to allocated from a zone other than ZONE_MOVABLE and migrating out existing pages from the zone during pinning.