Re: [PATCH RFC] cpuset: change error code to -EPERM from -ENOSPC

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

 



Hi Chengguang,

On 2018/3/30 8:50, cgxu519@xxxxxxx wrote:
> 在 2018年3月29日,下午7:26,Marian Marinov <kernel@xxxxxxxx> 写道:
>>
>> On 03/27/2018 06:20 PM, Chengguang Xu wrote:
>>>> Sent: Tuesday, March 27, 2018 at 9:58 PM
>>>> From: "Tejun Heo" <tj@xxxxxxxxxx>
>>>> To: "Chengguang Xu" <cgxu519@xxxxxxx>
>>>> Cc: lizefan@xxxxxxxxxx, cgroups@xxxxxxxxxxxxxxx
>>>> Subject: Re: [PATCH RFC] cpuset: change error code to -EPERM from -ENOSPC
>>>>
>>>> On Tue, Mar 27, 2018 at 06:02:24PM +0800, Chengguang Xu wrote:
>>>>> When adding task to cpuset with empty cpus_allowed or empty mems_allowed,
>>>>> there is an error with code -ENOSPC, it is a little bit confusing,
>>>>> so change error code to -EPERM for better understanding.
>>>>>
>>>>> Signed-off-by: Chengguang Xu <cgxu519@xxxxxxx>
>>>>
>>>> I don't see why we'd change this or how EPERM is necessarily more
>>>> intuitive here.
>>>
>>> After changing the error code, error message showed to user will change
>>> from
>>>
>>> $ echo $$ > /sys/fs/cgroup/cpuset/2/tasks
>>> -bash: echo: write error: No space left on device
>>>
>>> to
>>>
>>> $ echo $$ > /sys/fs/cgroup/cpuset/2/tasks
>>> -bash: echo: write error: Operation not permitted
>>>
>>> Isn't it more understandable? IIUC, the logic here is for prohibiting attaching or 
>>> changing operations.
>>
>> I somewhat agree that "Operation not permitted" is a bit more meaningful.
>> However this is something that has been in the kernel for a lot of years and changing it to another error message, that is also not the actual cause is not solving anyone's problems with this ;(
> 
> Actually, the motivation of this patch is from my experience. 
> I encountered -ENOSPC when I try to manipulate cpuset, then I had no idea about the reason
> at all, so I had to dive into the code to analyze the root cause, but I think there are also
> many sysadmins do not understand the code, so meaningful error message is much helpful to
> them.
> 
> IMHO, it’s a trivial change and it will not bring side effect so the change would not be so
> difficult though the -ENOSPC has been in kernel for a long time. 
> 

You don't have to dive into cpuset internal implmentation. There's document
you can refer to.

http://man7.org/linux/man-pages/man7/cpuset.7.html

       ENOSPC Attempted to write(2) the process ID (PID) of a process to a
              cpuset tasks file when the cpuset had an empty cpuset.cpus or
              empty cpuset.mems setting.

> 
>>
>> The other thing is, that "Operation not permitted" may push the sysadmins in completely different search, like is it a SELinux related problem.
> 
> I also propose to add additional warning message here to give some hints to sysadmins,
> but I guess mismatched warning/error messages will also cause users’ confusing.
> 
> 
> Thanks,
> Chengguang.
> 
> .
> 

--
To unsubscribe from this list: send the line "unsubscribe cgroups" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html



[Index of Archives]     [Linux ARM Kernel]     [Linux ARM]     [Linux Omap]     [Fedora ARM]     [IETF Annouce]     [Security]     [Bugtraq]     [Linux OMAP]     [Linux MIPS]     [eCos]     [Asterisk Internet PBX]     [Linux API]     [Monitors]

  Powered by Linux