Re: [PATCH for-5.13/block] blk-iocost: don't ignore vrate_min on QD contention

[Jens Axboe]

On 4/22/21 7:54 PM, Tejun Heo wrote:
> ioc_adjust_base_vrate() ignored vrate_min when rq_wait_pct indicates that
> there is QD contention. The reasoning was that QD depletion always reliably
> indicates device saturation and thus it's safe to override user specified
> vrate_min. However, this sometimes leads to unnecessary throttling,
> especially on really fast devices, because vrate adjustments have delays and
> inertia. It also confuses users because the behavior violates the explicitly
> specified configuration.
> 
> This patch drops the special case handling so that vrate_min is always
> applied.

Applied, thanks.

-- 
Jens Axboe