[PATCH v3 0/3] nvme: Add sysfs interface for APST configuration management

[Alexey Bogoslavsky]

Hi Yaxiong Tian,
The ability to configure the APST-related thresholds is not the essence of the commit
you mentioned, but rather an added bonus, which I suppose not many people actually use.
The actual change is about reducing the number of non-operational states in the APST table to just two,
regardless of the number of non-operational states advertised by the device, and, more importantly,
removal of the linear dependency between the ENLAT and EXLAT and the ITPT which, before the change
was made, caused an unreasonably high ITPT in devices with high latencies, which, in turn, prevented
transitions to non-operational states and led to an unjustified waste of power. The modified code
was tested thoroughly and showed significant improvement for some devices and only a minor negative effect 
for others. All results appear in the commit's description.

Thanks,
Alexey

-----Original Message-----
From: Christoph Hellwig <hch at lst.de> 
Sent: Friday, April 4, 2025 11:30 AM
To: Yaxiong Tian <iambestgod at qq.com>
Cc: Christoph Hellwig <hch at lst.de>; kbusch at kernel.org; axboe at kernel.dk; sagi at grimberg.me; chaitanyak at nvidia.com; linux-nvme at lists.infradead.org; linux-kernel at vger.kernel.org; Yaxiong Tian <tianyaxiong at kylinos.cn>; alexey.bogoslavsky at wdc.com
Subject: Re: [PATCH v3 0/3] nvme: Add sysfs interface for APST configuration management

On Thu, Apr 03, 2025 at 03:05:37PM +0800, Yaxiong Tian wrote:
> These two patches don't fundamentally change the APST configuration 
> policy, but rather enable users to configure various APST parameters 
> in real-time across different devices. As mentioned in commit <ebd8a93aa4f5> ("nvme:
> extend and modify the APST configuration algorithm"):

And who are those users?  What tools do they use.  What settings do they set and why?