Re: [PATCH 0/3] cxl/mbox: support background operation abort requests

[Ravis OpenSrc <Ravis.OpenSrc@micron.com>]

>>On Wed, 23 Oct 2024, Ravis OpenSrc wrote:\n
>>Hi Davidlohr,
>>
>>   We have recently submitted an RFC to implement Request Background Abort
>>in case a timeout is encountered while executing a background command.
>>The patch series disables those background commands which do not support abort
>>based on their individual CEL properties.
>>
>>https://lore.kernel.org/linux-cxl/20241017163215.00000547@Huawei.com/T/#ma6710971908b85a5f8c5da2e23b8102b5a6e277c
>
>*sigh* it would have been nice to be Cc'ed.

Noted. Will take care of this in future.

>
>>
>>This implementation is based on Dan's suggestion in a previous thread.
>>https://lore.kernel.org/linux-cxl/66035c2e8ba17_770232948b@dwillia2-xfh.jf.intel.com.notmuch/
>>
>>We can discuss more on how we can potentially merge and arrive at a common ground.
>
>I think that my approach is way more robust then defining a timeout as in that series.
>It also aligns more with Jonathan's comments about the abort being on demand.

The one major functionality in our original proposal apart from implementing abort is

Allowing background commands to be invoked from user space through the primary mailbox
by ensuring only those background commands are enabled which also support request abort operation
by checking their individual CEL details.

Few questions on abort implementation are:
1. Would abort be required only when a new bg command is issued?
2. What is the path for supporting background commands from primary mailbox if not done through
CEL check?
3. Should all future background commands follow sysfs path?

--Ravi.