[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <a17428e9-0221-44ff-8a10-6cf93a00c172@intel.com>
Date: Mon, 1 Sep 2025 13:10:19 +0800
From: "Chen, Yu C" <yu.c.chen@...el.com>
To: Pan Deng <pan.deng@...el.com>
CC: <linux-kernel@...r.kernel.org>, <tianyou.li@...el.com>,
<tim.c.chen@...ux.intel.com>, <peterz@...radead.org>, <mingo@...nel.org>,
Chen Yu <yu.chen.surf@...il.com>
Subject: Re: [PATCH 1/4] sched/rt: Optimize cpupri_vec layout to mitigate
cache line contention
On 7/7/2025 10:35 AM, Pan Deng wrote:
> When running a multi-instance FFmpeg workload on an HCC system, significant
> cache line contention is observed around `cpupri_vec->count` and `mask` in
> struct root_domain.
>
[it seems that my last reply did not make it to the lkml][snip]
> diff --git a/kernel/sched/cpupri.h b/kernel/sched/cpupri.h
> index d6cba0020064..245b0fa626be 100644
> --- a/kernel/sched/cpupri.h
> +++ b/kernel/sched/cpupri.h
> @@ -9,7 +9,7 @@
>
> struct cpupri_vec {
> atomic_t count;
> - cpumask_var_t mask;
> + cpumask_var_t mask ____cacheline_aligned;
Just curious, since this is to avoid cache contention among CPUs,
is it better to use ____cacheline_aligned_in_smp, so the single
CPU system is not impacted.
thanks,
Chenyu> };
>
> struct cpupri {
Powered by blists - more mailing lists