[PATCH] nvmet-rdma: Suppress a class of lockdep complaints

[Max Gurtovoy]

On 10/05/2023 2:42, Bart Van Assche wrote:
> On 5/9/23 16:24, Max Gurtovoy wrote:
>> But the concept of "dirtying" the base code only for the sake of the 
>> above reason sounds like a bad practice in general.
> 
> The use of the words "dirtying" and "bad practice" in the above sentence 
> seems completely inappropriate to me.

I'm sorry, this was not my intention. This was the reason I put dirtying 
under " ". To mention explicitly that it's not really dirtying.

> 
>> Maybe for that we can have some mechanism of side patches that one 
>> apply before running the blktests ?
>> This will leave the base code clean but will need to maintain these 
>> patches to be applied cleanly on the base code.
>> But I guess we can have some automation for that..
> 
> So you are proposing to maintain this patch as an out-of-tree patch? 
> That sounds like a bad idea to me. Anyone who modifies the NVMe code 
> wouldn't see the out-of-tree patch so the out-of-tree patch would get 
> broken quickly.

In tree patches that can be applied before building the environment for 
testing.
This way the base code stays clean and you also fix the blktests 
false-positive issues.

> 
> Bart.
>