Re: [PATCH v9 09/24] x86/resctrl: Use __set_bit()/__clear_bit() instead of open coding

From: James Morse <james.morse@arm.com>
To: David Hildenbrand <david@redhat.com>,
	x86@kernel.org, linux-kernel@vger.kernel.org
Cc: Fenghua Yu <fenghua.yu@intel.com>,
	Reinette Chatre <reinette.chatre@intel.com>,
	Thomas Gleixner <tglx@linutronix.de>,
	Ingo Molnar <mingo@redhat.com>, Borislav Petkov <bp@alien8.de>,
	H Peter Anvin <hpa@zytor.com>, Babu Moger <Babu.Moger@amd.com>,
	shameerali.kolothum.thodi@huawei.com,
	D Scott Phillips OS <scott@os.amperecomputing.com>,
	carl@os.amperecomputing.com, lcherian@marvell.com,
	bobo.shaobowang@huawei.com, tan.shaopeng@fujitsu.com,
	baolin.wang@linux.alibaba.com,
	Jamie Iles <quic_jiles@quicinc.com>,
	Xin Hao <xhao@linux.alibaba.com>,
	peternewman@google.com, dfustini@baylibre.com,
	amitsinght@marvell.com
Subject: Re: [PATCH v9 09/24] x86/resctrl: Use __set_bit()/__clear_bit() instead of open coding
Date: Tue, 20 Feb 2024 16:27:19 +0000	[thread overview]
Message-ID: <91b58253-f682-49f7-914d-8faf1ce181ef@arm.com> (raw)
In-Reply-To: <b5bc883a-8a8a-4f07-a6ff-b1bd02ffc99d@redhat.com>

Hi David,

On 20/02/2024 16:00, David Hildenbrand wrote:
> On 13.02.24 19:44, James Morse wrote:
>> The resctrl CLOSID allocator uses a single 32bit word to track which
>> CLOSID are free. The setting and clearing of bits is open coded.
>>
>> Convert the existing open coded bit manipulations of closid_free_map
>> to use __set_bit() and friends. These don't need to be atomic as this
>> list is protected by the mutex.

>> diff --git a/arch/x86/kernel/cpu/resctrl/rdtgroup.c
>> b/arch/x86/kernel/cpu/resctrl/rdtgroup.c
>> index dcffd1c4a476..bc6e0f83c847 100644
>> --- a/arch/x86/kernel/cpu/resctrl/rdtgroup.c
>> +++ b/arch/x86/kernel/cpu/resctrl/rdtgroup.c
>> @@ -111,7 +111,7 @@ void rdt_staged_configs_clear(void)
>>    * - Our choices on how to configure each resource become progressively more
>>    *   limited as the number of resources grows.
>>    */
> 
> That comment talks about "free CLOSIDs in a single integer". Once could think about
> rephrasing that to "free CLOSIDs in a simple bitmap."
> 
>> -static int closid_free_map;
>> +static unsigned long closid_free_map;
>>   static int closid_free_map_len;
>>     int closids_supported(void)
>> @@ -130,8 +130,8 @@ static void closid_init(void)
>>         closid_free_map = BIT_MASK(rdt_min_closid) - 1;
> 
> Now that we use "unsigned long", I was wondering for a second if we should use "proper"
> bitmap functions here like
> 
>     bitmap_fill(closid_free_map, rdt_min_closid);
> 
> and converting the alloc path (replace ffs() in closid_alloc()):
> 
>     closid = find_first_bit(closid_free_map, closid_free_map_len);
>     if (closid == closid_free_map_len)
>         return -ENOSPC;
>     __clear_bit(closid, &closid_free_map);
> 
> (would get rid of the closid-- in closid_alloc())

Yup. I have this as something to post after all the MPAM changes as it's not necessary to
get MPAM going. The patch[0] uses the bitmap APIs you suggest to remove the fixed limit on
the number of CLOSID/PARTID.
MPAM systems are being built with more than 32, but will work without that patch.

> Just a thought, so
> 
> Reviewed-by: David Hildenbrand <david@redhat.com>

Thanks!

James

[0]
https://git.kernel.org/pub/scm/linux/kernel/git/morse/linux.git/commit/?h=mpam/snapshot/v6.7-rc2&id=b530deed244d9b45f3bce3cccde91f6ed0ebf7ea