LKML News v5.14-rc3

[PATCH 0/7] Free user PTE page table pages (Qi Zheng)

https://lkml.kernel.org/r/20210718043034.76431-1-zhengqi.arch@bytedance.com

This patchset frees user PTE page table pages when all PTE entries are empty, to reduce the memory overhead due to page tables.

[PATCH v15 00/17] Folio support in block + iomap layers (Matthew Wilcox)

https://lkml.kernel.org/r/20210719184001.1750630-1-willy@infradead.org

Seems memory folios patchset is now merged in -mm tree and -next tree. This patchset further updates iomap to use folios.

Runtime Memory Validation in Intel-TDX and AMD-SNP (Joerg Roedel)

https://lkml.kernel.org/r/YPV27hDPZUoVsIZt@suse.de

This is a proposal mail describing how to implement runtime memory validation for confidential guests.

[RFC -V7 0/6] NUMA balancing: optimize memory placement for memory tiering system (Huang Ying)

https://lkml.kernel.org/r/20210722031819.3446711-1-ying.huang@intel.com

Tiered-memory system usually uses RAM and PMEM. Commit c221c0b0308f (“device-dax: “Hotplug” persistent memory for use like normal RAM”) made PMEM could be accessed as simply a very slow NUMA node. After that, ‘[PATCH -V10 0/9] Migrate Pages in lieu of discard’ made a mechanism to demote cold DRAM pages to PMEM node under memory pressure. This patchset further optimizes the mechanism and achieves up to 119.1% improvement with pmbench.

[PATCH v3 1/2] mm: introduce process_mrelease system call (Suren Baghdasaryan)

https://lkml.kernel.org/r/20210723011436.60960-1-surenb@google.com

Facebook’s oomd or Android’s lmkd uses a strategy of killing processes under memory pressure. However, the time to free memory depends on each process. This patchset introduces a new system call that can be used for faster memory release.

[PATCH resend] mm: compaction: optimize proactive compaction deferrals (Charan Teja Reddy)

When proactive compaction is making no progress, it waits some time before trying again. For this, it repeatedly sleeps, awakes, and decrement a counter. If the counter becomes 0, it retries the compaction. The wakeup for only counter decrement makes no sense, so this patch makes it unnecessary.

Linux 5.14-rc3 (Linus Torvalds)

https://lkml.kernel.org/r/CAHk-=wgvwZiLbZpFUApWFDkwyjC3+EmRoEBAuP+VzNQaMd630g@mail.gmail.com

Another week, another rc. Last release was something worrisome, becuase it was quite big. Seems things calmed down, as rc3 looks pretty normal.

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 2019-07-26 | ~/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 -rc3 releases in the last two years.