Amazon's DAMON Might Finally Be Ready For Upstreaming Into The Linux Kernel

At the start of 2020 Amazon engineers sent out a "request for comments" around DAMON as a new kernel feature to monitor data accesses and can be used for purposes like analyzing over-committed memory use, debugging, and other optimizations. DAMON has now seen thirty-four revisions to the patches but it looks like it could finally be in a state for mainlining in the Linux kernel.

Sent out this morning were the DAMON v34 patches that are re-based on the Linux 5.14-rc1 state as found in the memory management development tree. At this stage the DAMON code itself isn't seeing much churn.

Most importantly about the v34 patches is that all the patches have at least one reviewed-by or acknowledged-by tag on each patch. As such, SeongJae Park who has been sending out the patches is hoping now DAMON can be considered for merging into the kernel's -mm (memory management) tree. In turn this would make DAMON land for Linux 5.15 if it's indeed deemed acceptable at this stage.

Building off of DAMON, Amazon engineers are already working on additional functionality like DAMON-based proactive memory reclamation.

As for the current overall impact of using DAMON, per the patch series:

DAMON is lightweight. It increases system memory usage by 0.39% and slows target workloads down by 1.16%.

DAMON is accurate and useful for memory management optimizations. An experimental DAMON-based operation scheme for THP, namely 'ethp', removes 76.15% of THP memory overheads while preserving 51.25% of THP speedup. Another experimental DAMON-based 'proactive reclamation' implementation, 'prcl', reduces 93.38% of residential sets and 23.63% of system memory footprint while incurring only 1.22% runtime overhead in the best case (parsec3/freqmine).

More details on the current DAMON implementation for Linux via the kernel mailing list.