On Thu, Feb 01, 2024 at 01:12:40PM -0500, Johannes Weiner wrote: > On Thu, Feb 01, 2024 at 03:49:05PM +0000, Chengming Zhou wrote: > > The !zswap_exclusive_loads_enabled mode will leave compressed copy in > > the zswap tree and lru list after the folio swapin. > > > > There are some disadvantages in this mode: > > 1. It's a waste of memory since there are two copies of data, one is > > folio, the other one is compressed data in zswap. And it's unlikely > > the compressed data is useful in the near future. > > > > 2. If that folio is dirtied, the compressed data must be not useful, > > but we don't know and don't invalidate the trashy memory in zswap. > > > > 3. It's not reclaimable from zswap shrinker since zswap_writeback_entry() > > will always return -EEXIST and terminate the shrinking process. > > > > On the other hand, the only downside of zswap_exclusive_loads_enabled > > is a little more cpu usage/latency when compression, and the same if > > the folio is removed from swapcache or dirtied. > > > > Not sure if we should accept the above disadvantages in the case of > > !zswap_exclusive_loads_enabled, so send this out for disscusion. > > > > Signed-off-by: Chengming Zhou <zhouchengming@xxxxxxxxxxxxx> > > This is interesting. > > First, I will say that I never liked this config option, because it's > nearly impossible for a user to answer this question. Much better to > just pick a reasonable default. > > What should the default be? > > Caching "swapout work" is helpful when the system is thrashing. Then > recently swapped in pages might get swapped out again very soon. It > certainly makes sense with conventional swap, because keeping a clean > copy on the disk saves IO work and doesn't cost any additional memory. > > But with zswap, it's different. It saves some compression work on a > thrashing page. But the act of keeping compressed memory contributes > to a higher rate of thrashing. And that can cause IO in other places > like zswap writeback and file memory. Agreed. At Google, we have been using exclusive loads for a very long time in production, so I have no objections to this. The user interface is also relatively new, so I don't think it will have accumulated users. > > It would be useful to have an A/B test to confirm that not caching is > better. Can you run your test with and without keeping the cache, and > in addition to the timings also compare the deltas for pgscan_anon, > pgscan_file, workingset_refault_anon, workingset_refault_file? That would be interesting :)
