[<prev] [next>] [<thread-prev] [day] [month] [year] [list]
Message-ID: <ed67836b-3235-4a56-a923-1a5c7a14db52@kernel.org>
Date: Mon, 6 May 2024 21:54:13 +0200
From: Jesper Dangaard Brouer <hawk@...nel.org>
To: Shakeel Butt <shakeel.butt@...ux.dev>
Cc: Waiman Long <longman@...hat.com>, tj@...nel.org, hannes@...xchg.org,
lizefan.x@...edance.com, cgroups@...r.kernel.org, yosryahmed@...gle.com,
netdev@...r.kernel.org, linux-mm@...ck.org, kernel-team@...udflare.com,
Arnaldo Carvalho de Melo <acme@...nel.org>,
Sebastian Andrzej Siewior <bigeasy@...utronix.de>,
Daniel Dao <dqminh@...udflare.com>, Ivan Babrou <ivan@...udflare.com>,
jr@...udflare.com
Subject: Re: [PATCH v1] cgroup/rstat: add cgroup_rstat_cpu_lock helpers and
tracepoints
On 06/05/2024 18.22, Shakeel Butt wrote:
> Anyways did the mutex change helped your production workload regarding
> latencies?
Yes, see attached NET_RX softirq latency improvement with mutex patch[1].
Showing two grafana heatmaps over softirq wait (no-able-to run) time.
It is very clear that most of the latency spikes are gone with mutex
patch applied.
--Jesper
[1]
https://lore.kernel.org/all/171328989335.3930751.3091577850420501533.stgit@firesoul/
Download attachment "softirq_wait_time_compare01.png" of type "image/png" (119601 bytes)
Powered by blists - more mailing lists