linux-kernel - Re: [PATCH -next] cpuset: treate root invalid trialcs as exclusive

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite for Android: free password hash cracker in your pocket

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Message-Id: <20251117065347.1052678-1-sunshaojie@kylinos.cn>
Date: Mon, 17 Nov 2025 14:53:47 +0800
From: Sun Shaojie <sunshaojie@...inos.cn>
To: chenridong@...weicloud.com
Cc: cgroups@...r.kernel.org,
	hannes@...xchg.org,
	linux-kernel@...r.kernel.org,
	longman@...hat.com,
	mkoutny@...e.com,
	sunshaojie@...inos.cn,
	tj@...nel.org
Subject: Re: [PATCH -next] cpuset: treate root invalid trialcs as exclusive

On 2025/11/15 14:23, Chen Ridong wrote:
>On 2025/11/17 12:35, Sun Shaojie wrote:
>> Hi, Ridong,
>> 
>> Maybe, this patch does not apply to the following cases:
>>  Step
>>  #1> echo "root" > A1/cpuset.cpus.partition
>>  #1> echo "0-1" > B1/cpuset.cpus
>>  #2> echo "1-2" > A1/cpuset.cpus.exclusive  -> return error
>>  It should return success here.
>> 
>> Please consider the following modification.
>> 
>
>If A1 will automatically change to a valid partition, I think it should return error.

Hi, Ridong,

A1 will not automatically change to a valid partition.

Perhaps this example is more intuitive.

For example:

 Before apply this patch:
 #1> echo "0-1" > B1/cpuset.cpus
 #2> echo "root" > A1/cpuset.cpus.partition -> A1's prstate is "root invalid"
 #3> echo "1-2" > A1/cpuset.cpus.exclusive
 Return success, and A1's prstate is "root invalid"

 After apply this patch:
 #1> echo "0-1" > B1/cpuset.cpus
 #2> echo "root" > A1/cpuset.cpus.partition -> A1's prstate is "root invalid"
 #3> echo "1-2" > A1/cpuset.cpus.exclusive
 Return error, and A1's prstate is "root invalid"

 It should return success here. Because A1's prstate is "root invalid.
 
For this example, the behavior should remain consistent before and after 
applying the patch. This is because when A1 is in the "root invalid" state,
its behavior is equivalent to that of a "member," meaning A1's 
cpuset.cpus.exclusive and B1's cpuset.cpus are allowed to overlap.

Thanks,
Sun Shaojie