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.

[PATCH 1/1] process_madvise.2: Add process_madvise man page (Suren Baghdasaryan)

https://lkml.kernel.org/r/20210120202337.1481402-1-surenb@google.com

This patch is for the man page of the process_madvise() system call.

[PATCH v2 0/5] create hugetlb flags to consolidate state (Mike Kravetz)

https://lkml.kernel.org/r/20210120013049.311822-1-mike.kravetz@oracle.com

Hugetlb specific page state information is stored in hard to read/understand/maintain way. This patch series store hugetlb specific page flags in the ‘private’ field of ‘struct page’ to mitigate the problem.

[PATCH v6 00/14] prohibit pinning pages in ZONE_MOVABLE (Pavel Tatashin)

https://lkml.kernel.org/r/20210120014333.222547-1-pasha.tatashin@soleen.com

This patch makes movable pages really movable by prohibiting pinning them in ZONE_MOVABLE.

[PATCH RFC 0/6] fix the negative dentres bloating system memory usage (Gautham Ananthakrishna)

https://lkml.kernel.org/r/1611235185-1685-1-git-send-email-gautham.ananthakrishna@oracle.com

This commit mitigates the memory bloat caused by ‘negative dentries’ by moving those to end of siblings list and keep dcache at most three unreferenced negative dentries in each hash bucket.

Linux 5.11-rc5

https://lkml.kernel.org/r/CAHk-=wgmJ0q1URHrOb-2iCOdZ8gYybiH6LY2Gq7cosXu6kxAnA@mail.gmail.com

The fifthe release candidate for v5.11 is slightly larger than usual. Mainly because of about 40~% commits for -rc5 which came in the last day. Noticeable change is regression due to get rid of set_fs() change that made for v5.10. The regression was under rare-to-happen condition, but seems now v5.10 is widespread enough to let people notice such problems.

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 2019-01-25 | ~/lazybox/gnuplot/plot.py \
    --data_fmt table --type labeled-lines --xtics_rotate -90 \
    --font "Times New Roman, 5pt" --ylog --pointsize 0.3

And, below is the diffstat of the -rc2 releases in the last two years.