On Wed, May 31, 2023 at 3:21 PM Yu Zhao <yuzhao@xxxxxxxxxx> wrote: > > On Wed, May 31, 2023 at 4:13 PM Yosry Ahmed <yosryahmed@xxxxxxxxxx> wrote: > > > > Support using multiple zpools of the same type in zswap, for concurrency > > purposes. Add CONFIG_ZSWAP_NR_ZPOOLS_ORDER to control the number of > > zpools. The order is specific by the config rather than the absolute > > number to guarantee a power of 2. This is useful so that we can use > > deterministically link each entry to a zpool by hashing the zswap_entry > > pointer. > > > > On a setup with zswap and zsmalloc, comparing a single zpool (current > > default) to 32 zpools (by setting CONFIG_ZSWAP_NR_ZPOOLS_ORDER=32) shows > > improvements in the zsmalloc lock contention, especially on the swap out > > path. > > > > The following shows the perf analysis of the swapout path when 10 > > workloads are simulatenously reclaiming and refaulting tmpfs pages. > > simultaneously > > > There are some improvements on the swap in path as well, but much less > > significant. > > ... > > > --- > > mm/Kconfig | 12 +++++++ > > mm/zswap.c | 91 ++++++++++++++++++++++++++++++++++++------------------ > > 2 files changed, 73 insertions(+), 30 deletions(-) > > > > diff --git a/mm/Kconfig b/mm/Kconfig > > index 92c30879bf67..de1da56d2c07 100644 > > --- a/mm/Kconfig > > +++ b/mm/Kconfig > > @@ -59,6 +59,18 @@ config ZSWAP_EXCLUSIVE_LOADS > > The cost is that if the page was never dirtied and needs to be > > swapped out again, it will be re-compressed. > > > > +config ZSWAP_NR_ZPOOLS_ORDER > > + int "Number of zpools in zswap, as power of 2" > > + default 0 > > + depends on ZSWAP > > + help > > + This options determines the number of zpools to use for zswap, it > > option Thanks for taking a look. Will correct both in v3 (if any). Otherwise, I hope Andrew will fix them in-place :)
