Re: [PATCH v2] zswap: replace RB tree with xarray

From: Yosry Ahmed <yosryahmed@google.com>
To: David Laight <David.Laight@aculab.com>
Cc: "'Chris Li'" <chrisl@kernel.org>,
	Andrew Morton <akpm@linux-foundation.org>,
	 "linux-kernel@vger.kernel.org" <linux-kernel@vger.kernel.org>,
	"linux-mm@kvack.org" <linux-mm@kvack.org>,
	 Nhat Pham <nphamcs@gmail.com>,
	Johannes Weiner <hannes@cmpxchg.org>,
	 "Matthew Wilcox (Oracle)" <willy@infradead.org>,
	Chengming Zhou <zhouchengming@bytedance.com>,
	 Barry Song <v-songbaohua@oppo.com>
Subject: Re: [PATCH v2] zswap: replace RB tree with xarray
Date: Tue, 5 Mar 2024 03:08:15 +0000	[thread overview]
Message-ID: <ZeaMn488S_xVNtAh@google.com> (raw)
In-Reply-To: <26e77602326d4e169a9484314cac2465@AcuMS.aculab.com>

On Fri, Mar 01, 2024 at 09:58:43PM +0000, David Laight wrote:
> From: Chris Li
> > Sent: 29 February 2024 08:46
> > 
> > Very deep RB tree requires rebalance at times. That
> > contributes to the zswap fault latencies. Xarray does not
> > need to perform tree rebalance. Replacing RB tree to xarray
> > can have some small performance gain.
> > 
> > One small difference is that xarray insert might fail with
> > ENOMEM, while RB tree insert does not allocate additional
> > memory.
> 
> What is the difference in kernel memory use?
> IIRC someone pointed out (in the rosebush thread) that xarray
> uses a lot of kernel memory if the items are randomly distributed.

I often thought it would be useful if there is an API for the xarray to
report how much memory it's using. For zswap, we could plumb this to a
debugfs interface or so to measure the tree usage and compare to the
rbtree. The cost is fixed with the rbtree IIUC, 3 words per zswap entry.