LKML News v6.11-rc2
[RFC 0/2] mm: introduce THP deferred setting (Nico Pache)
https://lore.kernel.org/20240729222727.64319-1-npache@redhat.com
Nico found increased memory footprint from THP after their customers switching
from RHEL7 to RHEL8. To mitigate the issue, Nico is proposing adding a new THP
configuration called deferred
. It is same to madvise
but allows
khugepaged
to do the work.
[PATCH v2 0/4] mm: clarify nofail memory allocation (Barry Song)
https://lore.kernel.org/20240731000155.109583-1-21cnbao@gmail.com
Barry posted the second version of their patchset for __GFP_NOFAIL
clarification. The patchset is now merged in mm tree.
[PATCH 0/6] mm: split underutilized THPs (Usama Arif)
https://lore.kernel.org/20240730125346.1580150-1-usamaarif642@gmail.com
Usama reworks underutilized THPs split work for reducing memory overhead.
Race condition observed between page migration and page fault handling on arm64 machines (Dev Jain)
https://lore.kernel.org/20240801081657.1386743-1-dev.jain@arm.com
Dev Jain and Ryan found a race between page migration and fault handling with a hacky fix. Dev Jain further share it with the community to get feedback.
[PATCH v2 0/1] mm: introduce MADV_DEMOTE/MADV_PROMOTE (BiscuitOS Broiler)
https://lore.kernel.org/20240801075610.12351-1-zhang.renze@h3c.com
BiscuitOS Brolers proposes Scalable Tiered Memory Control (STMC) mechanism,
which let the application to do the promotion and demotion of memory. For
this, they implement two new madvise()
hints for promotion and demotion.
[PATCH v1 1/2] lib/htree: Implementation of new Hash Tree (JaeJoon Jung)
https://lore.kernel.org/20240802051635.8179-1-rgbi3307@gmail.com
new Hash Tree Features
-------------------------------------------------------------------------------
* Very small hash tree structure. [16 Bytes]
* Dynamic memory allocation and free.
* Both 32-bit and 64-bit indexes are possible
* Generate hash keys uniformly based on the index.
* Hash trees are balanced by hash keys, and have no rotation costs.
* Standard deviation of hash key is 4 or less.
* Algorithm O(n) is depth(d) x nodes(c)
* Finding walks is (d x c), min:4, avg:12, max:20
* First hash table has smallest, largest index, algorithm O(1).
* The codes implementing of the algorithm is simple.
* Adjust hash tree depth according to system memory and index nr.
* Hash list nodes use include/linux/list.h, hlist as their base.
-------------------------------------------------------------------------------
[PATCH] Hitshield : Something new eviction process for MGLRU (Minwoo Jo)
https://lore.kernel.org/20240802000546.322036-1-chminoo@g.cbnu.ac.kr
This patch introduces a way to improve page space occupancy issue based on the finding that the current scheme is not accounting the state of the folio during the eviction.
Linux 6.11-rc2 (Linus Torvalds)
https://lore.kernel.org/CAHk-=wh01xPAWUT_=J1TehFOu3SST12UTNuB=QQTeRw+1N4pDQ@mail.gmail.com
Linus Torvalds releases the second candidate for Linux v6.11. He mentions small build or test failures on rc1, which were annoying but never been the show stopper.