[PATCH 0/2] nvme-cli: add support for an abort command with documentation

[Minwoo Im]

Hi James,

On 01/25/2018 02:26 AM, James Smart wrote:
> On 1/24/2018 5:15 AM, Minwoo Im wrote:
>> Or.. Is there any cases that breaks behavior of kernel by an abort
>> command from user space?
>>
> 
> Is it possible yes. Is it simple, no.   Depends on what the command is 
> and when.  You may confuse a transport which uses aborts - and you're 
> also mandating the transport peek into the pass-through command to check 
> for abort so that is can honor the # of outstanding aborts supported by 
> the controller.
> 
> I agree with Johannes. this is an ugly thing to do . The first question 
> I had is - how do you even know what the command and its CID value is ? 
> Also, many devices won't actually even do aborts.
> 
> -- james
> 

I totally agree with you. I really appreciate your comments.
I thought abort command has nothing to do with behavior of kernel, but
CID is assigned by kernel when issuing an command to device.

Thanks again.