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).

[PATCH 00/12] Introduce cmpxchg128() – aka. the demise of cmpxchg_double() (Peter Zijlstra)

https://lkml.kernel.org/r/20230531130833.635651916@infradead.org

Peter sent the third version of the patchset for introducing cmpxchg128(). It is for replacing cmpxchg_double().

[PATCH 0/4] vmstat bug fixes for nohz_full CPUs (Marcelo Tosatti)

https://lkml.kernel.org/r/20230530145234.968927611@redhat.com

This patchset fixes vmstat bugs on nohz_full CPUs that caused process stalls under direct reclaim, and latency increases.

Linux 6.4-rc5 (Linus Torvalds)

https://lkml.kernel.org/r/CAHk-=wifuPqAFXQQTTLkp_+FMzxGFHpSG-hEtZazG-46s=noAw@mail.gmail.com

There was a timing dependent module loading hiccup in rc4. The comit that introduced the bug has reverted. Other than that, no special thing in rc5. Torvalds says it looks like in good shape.

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 2021-06-05 | ~/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 -rc5 releases in the last two years.