[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <3fb41114-4ccf-bfc2-dc02-9379aec0415e@bytedance.com>
Date: Fri, 4 Nov 2022 00:46:55 +0800
From: wuqiang <wuqiang.matt@...edance.com>
To: Andrew Morton <akpm@...ux-foundation.org>
Cc: mhiramat@...nel.org, davem@...emloft.net,
anil.s.keshavamurthy@...el.com, naveen.n.rao@...ux.ibm.com,
rostedt@...dmis.org, peterz@...radead.org, sander@...nheule.net,
ebiggers@...gle.com, dan.j.williams@...el.com, jpoimboe@...nel.org,
linux-kernel@...r.kernel.org, mattwu@....com,
kernel test robot <lkp@...el.com>
Subject: Re: [PATCH v4] kprobes,lib: kretprobe scalability improvement
On 2022/11/3 05:33, Andrew Morton wrote:
> On Wed, 2 Nov 2022 10:30:12 +0800 wuqiang <wuqiang.matt@...edance.com> wrote:
>
>> Tests of
>> kretprobe throughput show the biggest ratio as 333.9x of the original
>> freelist.
>
> Seriously.
>
> I'll add this for some runtime testing.
Thanks.
>
> Are you able to identify other parts of the kernel which could use
> (and benefit from) the new objpool?
The scalability issue is caused by freelist. Currently kretprobe and rethook
are the only use cases.
I'm working on the evaluation of bpf percpu-freelist, which scales well but
uses raw_spinlock and needs local irq disabled.
Powered by blists - more mailing lists