[<prev] [next>] [<thread-prev] [day] [month] [year] [list]
Message-ID: <fc1465ab-3df8-4272-a5a7-101887b9a82c@intel.com>
Date: Thu, 20 Nov 2025 14:17:44 -0800
From: Reinette Chatre <reinette.chatre@...el.com>
To: "Luck, Tony" <tony.luck@...el.com>
CC: <james.morse@....com>, <Dave.Martin@....com>, <babu.moger@....com>,
<bp@...en8.de>, <tglx@...utronix.de>, <dave.hansen@...ux.intel.com>,
<x86@...nel.org>, <hpa@...or.com>, <fustini@...nel.org>,
<fenghuay@...dia.com>, <peternewman@...gle.com>,
<linux-kernel@...r.kernel.org>, <patches@...ts.linux.dev>,
<christophe.jaillet@...adoo.fr>
Subject: Re: [PATCH] fs/resctrl: Consider sparse masks when initializing new
group's allocation
Hi Tony,
On 11/20/25 11:35 AM, Luck, Tony wrote:
> On Mon, Nov 17, 2025 at 04:42:45PM -0800, Reinette Chatre wrote:
>> A new resource group is intended to be created with sane defaults. For
>> a cache resource this means all cache portions the new group could possibly
>> allocate into. This includes unused cache portions and shareable cache
>> portions used by other groups and hardware.
>>
>> New resource group creation does not take sparse masks into account. After
>> determining the bitmask reflecting the new group's possible allocations the
>> bitmask is forced to be contiguous even if the system supports sparse masks.
>> For example, a new group could by default allocate into a large portion of
>> cache represented by 0xff0f, but it is instead created with a mask of 0xf.
>>
>> Do not force a contiguous allocation range if the system supports sparse
>> masks.
>
> Note that the allocated mask on systems that require contiguous bit allocation
> is not optimal. In your example where the available bits for the new group
> are 0xff0f, resctrl will focus on the least significant contguous range and
> set 0x000f, rather than the larger group of bits 0xff00.
Right. The priority, as supported by aptly named cbm_ensure_valid(), is to ensure
the mask can be supported by the hardware.
>
> Fixing this would be more complex, and I don't see a lot of value as the
> user is very likley to rewrite the schemata immediatley after creating
> the new group.
I agree this can be optimized and that it is not a priority (would appreciate
to hear if there are different opinions on this). Users that may encounter
the "less than optimal" initial allocation are the ones using exclusive and/or
pseudo-locking groups where I expect user space controls allocations with care.
>
> So for this patch as-is:
>
> Reviewed-by: Tony Luck <tony.luck@...el.com>
Thank you very much.
Reinette
Powered by blists - more mailing lists