Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 2620:137:e000::1:20 as permitted sender) client-ip=2620:137:e000::1:20;
Date:   Thu, 28 Apr 2022 14:34:28 -0400
From:   Johannes Weiner <hannes@cmpxchg.org>
To:     Minchan Kim <minchan@kernel.org>
Cc:     Andrew Morton <akpm@linux-foundation.org>,
        Michal Hocko <mhocko@suse.com>, Roman Gushchin <guro@fb.com>,
        Shakeel Butt <shakeelb@google.com>,
        Seth Jennings <sjenning@redhat.com>,
        Dan Streetman <ddstreet@ieee.org>, linux-mm@kvack.org,
        cgroups@vger.kernel.org, linux-kernel@vger.kernel.org,
        kernel-team@fb.com
Subject: Re: [PATCH 4/5] mm: zswap: add basic meminfo and vmstat coverage
Message-ID: <YmreNLge7b+FBC30@cmpxchg.org>
References: <20220427160016.144237-1-hannes@cmpxchg.org>
 <20220427160016.144237-5-hannes@cmpxchg.org>
 <Ymmnrkn0mSWcuvmH@google.com>
 <YmmznQ8AO5RLxicA@cmpxchg.org>
 <YmnA0Da90IURbxrM@google.com>
 <Ymqj93gEEzu2Gb3U@cmpxchg.org>
 <YmrICbP6bDJqDv5R@google.com>
 <YmrNiY3VhdMbfTq1@cmpxchg.org>
 <YmrPgWs5WPabIBQk@google.com>
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
In-Reply-To: <YmrPgWs5WPabIBQk@google.com>
Precedence: bulk

On Thu, Apr 28, 2022 at 10:31:45AM -0700, Minchan Kim wrote:
> On Thu, Apr 28, 2022 at 01:23:21PM -0400, Johannes Weiner wrote:
> > On Thu, Apr 28, 2022 at 09:59:53AM -0700, Minchan Kim wrote:
> > > On Thu, Apr 28, 2022 at 10:25:59AM -0400, Johannes Weiner wrote:
> > > > On Wed, Apr 27, 2022 at 03:16:48PM -0700, Minchan Kim wrote:
> > > > > On Wed, Apr 27, 2022 at 05:20:29PM -0400, Johannes Weiner wrote:
> > > > > > On Wed, Apr 27, 2022 at 01:29:34PM -0700, Minchan Kim wrote:
> > > > > > > Hi Johannes,
> > > > > > > 
> > > > > > > On Wed, Apr 27, 2022 at 12:00:15PM -0400, Johannes Weiner wrote:
> > > > > > > > Currently it requires poking at debugfs to figure out the size and
> > > > > > > > population of the zswap cache on a host. There are no counters for
> > > > > > > > reads and writes against the cache. As a result, it's difficult to
> > > > > > > > understand zswap behavior on production systems.
> > > > > > > > 
> > > > > > > > Print zswap memory consumption and how many pages are zswapped out in
> > > > > > > > /proc/meminfo. Count zswapouts and zswapins in /proc/vmstat.
> > > > > > > > 
> > > > > > > > Signed-off-by: Johannes Weiner <hannes@cmpxchg.org>
> > > > > > > > ---
> > > > > > > >  fs/proc/meminfo.c             |  7 +++++++
> > > > > > > >  include/linux/swap.h          |  5 +++++
> > > > > > > >  include/linux/vm_event_item.h |  4 ++++
> > > > > > > >  mm/vmstat.c                   |  4 ++++
> > > > > > > >  mm/zswap.c                    | 13 ++++++-------
> > > > > > > >  5 files changed, 26 insertions(+), 7 deletions(-)
> > > > > > > > 
> > > > > > > > diff --git a/fs/proc/meminfo.c b/fs/proc/meminfo.c
> > > > > > > > index 6fa761c9cc78..6e89f0e2fd20 100644
> > > > > > > > --- a/fs/proc/meminfo.c
> > > > > > > > +++ b/fs/proc/meminfo.c
> > > > > > > > @@ -86,6 +86,13 @@ static int meminfo_proc_show(struct seq_file *m, void *v)
> > > > > > > >  
> > > > > > > >  	show_val_kb(m, "SwapTotal:      ", i.totalswap);
> > > > > > > >  	show_val_kb(m, "SwapFree:       ", i.freeswap);
> > > > > > > > +#ifdef CONFIG_ZSWAP
> > > > > > > > +	seq_printf(m,  "Zswap:          %8lu kB\n",
> > > > > > > > +		   (unsigned long)(zswap_pool_total_size >> 10));
> > > > > > > > +	seq_printf(m,  "Zswapped:       %8lu kB\n",
> > > > > > > > +		   (unsigned long)atomic_read(&zswap_stored_pages) <<
> > > > > > > > +		   (PAGE_SHIFT - 10));
> > > > > > > > +#endif
> > > > > > > 
> > > > > > > I agree it would be very handy to have the memory consumption in meminfo
> > > > > > > 
> > > > > > > https://lore.kernel.org/all/YYwZXrL3Fu8%2FvLZw@google.com/
> > > > > > > 
> > > > > > > If we really go this Zswap only metric instead of general term
> > > > > > > "Compressed", I'd like to post maybe "Zram:" with same reason
> > > > > > > in this patchset. Do you think that's better idea instead of
> > > > > > > introducing general term like "Compressed:" or something else?
> > > > > > 
> > > > > > I'm fine with changing it to Compressed. If somebody cares about a
> > > > > > more detailed breakdown, we can add Zswap, Zram subsets as needed.
> > > > > 
> > > > > Thanks! Please consider ZSWPIN to rename more general term, too.
> > > > 
> > > > That doesn't make sense to me.
> > > > 
> > > > Zram is a swap backend, its traffic is accounted in PSWPIN/OUT. Zswap
> > > > is a writeback cache on top of the swap backend. It has pages
> > > > entering, refaulting, and being written back to the swap backend
> > > > (PSWPOUT). A zswpout and a zramout are different things.
> > > 
> > > Think about that system has two swap devices (storage + zram).
> > > I think it's useful to know how many swap IO comes from zram
> > > and rest of them are storage.
> > 
> > Hm, isn't this comparable to having one swap on flash and one swap on
> > a rotating disk? /sys/block/*/stat should be able to tell you how
> > traffic is distributed, no?
> 
> That raises me a same question. Could you also look at the zswap stat
> instead of adding it into vmstat? (If zswap doesn't have the counter,
> couldn't we simply add new stat in sysfs?)

My point is that for regular swap backends there is already
PSWP*. Distinguishing traffic between two swap backends is legitimate
of course, but zram is not really special compared to other backends
from that POV. It's only special in its memory consumption.

zswap *is* special, though. Even though some people use it *like* a
swap backend, it's also a cache on top of swap. zswap loads and stores
do not show up in PSWP*. And they shouldn't, because in a cache
configuration, you still need the separate PSWP* stats to understand
cache eviction behavior and cache miss ratio. memory -> zswap is
ZSWPOUT; zswap -> disk is PSWPOUT; PSWPIN is a cache miss etc.

> I thought the patch aims for exposting statistics to grab easier
> using popular meminfo and vmstat and wanted to leverage it for
> zram, too.

Right. zram and zswap overlap in their functionality and have similar
deficits in their stats. Both should be fixed, I'm not opposing
that. But IMO we should be careful about conflating
them. Fundamentally, one is a block device, the other is an MM-native
cache layer that sits on top of block devices. Drawing false
equivalencies between them will come back to haunt us.