Re: [PATCH v2] mm: count zeromap read and set for swapout and swapin

[Nhat Pham <nphamcs@gmail.com>]

On Sat, Nov 2, 2024 at 3:12 AM Barry Song <21cnbao@gmail.com> wrote:
>
> From: Barry Song <v-songbaohua@oppo.com>
>
> When the proportion of folios from the zero map is small, missing their
> accounting may not significantly impact profiling. However, it’s easy
> to construct a scenario where this becomes an issue—for example,
> allocating 1 GB of memory, writing zeros from userspace, followed by
> MADV_PAGEOUT, and then swapping it back in. In this case, the swap-out
> and swap-in counts seem to vanish into a black hole, potentially
> causing semantic ambiguity.
>
> We have two ways to address this:
>
> 1. Add a separate counter specifically for the zero map.
> 2. Continue using the current accounting, treating the zero map like
> a normal backend. (This aligns with the current behavior of zRAM
> when supporting same-page fills at the device level.)
>
> This patch adopts option 1 as pswpin/pswpout counters are that they
> only apply to IO done directly to the backend device (as noted by
> Nhat Pham).
>
> We can find these counters from /proc/vmstat (counters for the whole
> system) and memcg's memory.stat (counters for the interested memcg).
>
> For example:
>
> $ grep -E 'swpin_zero|swpout_zero' /proc/vmstat
> swpin_zero 1648
> swpout_zero 33536
>
> $ grep -E 'swpin_zero|swpout_zero' /sys/fs/cgroup/system.slice/memory.stat
> swpin_zero 3905
> swpout_zero 3985
>

LGTM FWIW, so I'll leave my review tag here:

Reviewed-by: Nhat Pham <nphamcs@gmail.com>

Too many emails in this thread, but my opinions is:

1. A fix tag is appropriate. It's not a kernel bug per se, but it's
incredibly confusing, and can potentially throw off user space agents
who rely on the rate of change of these counters as signals.

2. I do think we should use a separate set of counters for this
optimization. No strong opinions regarding combining this with the
zswap counters, but it can get confusing for users when they
enable/disable zswap.

If we are to combine, I'd be much more comfortable if we have a
generic name, like the one David suggested in v1 ("swpin_skip" /
"swpout_skip"). This would still require some API change tho, so not
sure if this is the best approach? :)

It would also be appropriate if we bring back the same-filled
optimization (which should be doable in the swap ID world, but I
digress).