LKML News v5.6-rc2

[RFC PATCH 0/3] DAMON: Implement The Data Access Pattern Awared Memory Management Rules

https://lkml.kernel.org/r/20200210150921.32482-1-sjpark@amazon.com

This is an RFC patch from me that implementing simple way to run data access pattern awared memory system using DAMON. Proactive reclamation or access awared THP promotion/demotion will be easily available using this change.

[PATCH 0/9] workingset protection/detection on the anonymous LRU list

https://lkml.kernel.org/r/1581401993-20041-1-git-send-email-iamjoonsoo.kim@lge.com

This patchset is for improving kernel’s reclamation scheme in a corner case where pages which used only once evicting hot pages. For example, if there are hot working set and periodically a big region of memory is scanned, the hot working set will be swapped out for each of the scan. This patchset avoid the problem by letting the scanning-like pages to be inserted into the inactive list first.

Of course, for this change, accurate working set detection is necessary. This patchset detects the working sets mostly based on the previous scheme, but extends it to deal with LRU well.

[PATCH v4 0/8] introduce memory hinting API for external process

https://lkml.kernel.org/r/20200212233946.246210-1-minchan@kernel.org

This is the fourth version of the patchset for madvise_common() system call. Seems like the justification for the API itself has finished.

[LSF/MM/BPF TOPIC] AutoNUMA enhancements to optimize Tiered Memory

https://lkml.kernel.org/r/87mu9mhn10.fsf@yhuang-dev.intel.com

This is another LSF/MM session proposal. As current kernel reclamation scheme is designed for universal DRAMs, it doesn’t work well with tiered memory. Author has worked for AutoNUMA, which can find recently-accessed pages regardless of memory pressure and even frequently-access pages. The authors want to discuss about this in the summit.

[LFS/MM/BPF Topic] User space OOM management on Chrome OS

https://lkml.kernel.org/r/CAPz6YkUci9=d-nkoN9e8pb1eO3Wgchh1GTNSNdWc3kYuvoQdBg@mail.gmail.com

Another proposal for LSF/MM. The author wants to share his experience of memory management in Chrom OS development of 8 year, including their use of Zram and per-process reclaim.

[LSF/MM/BPF TOPIC] Replacing mmap_sem with finer grained locks

https://lkml.kernel.org/r/CANN689EippvdwsHuhU-Aq=kNdk8-cAaCzPC7v=zbe6X302NteA@mail.gmail.com

A lot of LSF/MM proposals this week, maybe because the due is this week. This is old topic, but not solved yet. Authors want to keep the discussion about mmap splitting for multi CPU this year LSF/MM.

[LSF/MM TOPIC] Guest memory without struct page

https://lkml.kernel.org/r/1be38ae3-d51e-2661-d0ab-6ad8baefe804@oracle.com

struct page consumes about 1.5% of total memory. Because the structure could be unnecessary for hypervisors, the author want to save the 1.5%.

Linux 5.6-rc2

https://lkml.kernel.org/r/CAHk-=wgqwiBLGvwTqU2kJEPNmafPpPe_K0XgBU-A58M+mkwpgQ@mail.gmail.com

rc2 was calm as usual. More than half of the changes are kvm documentation, due to its conversion to ReST format. ‘perf’ change was also remarkable.