Re: [PATCH] block: protect debugfs attributes using q->elevator_lock

[Nilay Shroff <nilay@linux.ibm.com>]

On 3/13/25 1:24 PM, Christoph Hellwig wrote:
> On Wed, Mar 12, 2025 at 08:49:33PM +0530, Nilay Shroff wrote:
>>> really want this read system call interrupted by random signals, do
>>> we?  I guess this should be mutex_lock_killable.
>>>
>>> (and the same for the existing methods this is copy and pasted from).
>>>
>> I thought we wanted to interrupt using SIGINT (CTRL+C) in case user opens 
>> this file using cat. Maybe that's more convenient than sending SIGKILL. 
>> And yes I used mutex_lock_interruptible because for other attributes we are 
>> already using it. But if mutex_lock_killable is preferred then I'd change it
>> for all methods.
> 
> Let's leave it alone for this series.  While I think it's the wrong
> choice it's been there for a long time, so we might as well not change
> it now for unrelated reasons.
> 
Alright, then I will not replace mutex_lock_interruptible with 
mutex_lock_killable for this series. A next version of this 
series is on the way... 

Thanks,
--Nilay