ACK / APPLIED[artful]: [PATCH 0/3] [SRU Y/Z/A] APST gets enabled against explicit kernel option

Seth Forshee seth.forshee at canonical.com
Wed Jul 5 15:49:35 UTC 2017


On Thu, Jun 29, 2017 at 05:59:18PM +0800, Kai-Heng Feng wrote:
> BugLink: https://bugs.launchpad.net/bugs/1699004
> 
> [Impact]
> The NVMe driver doesn't set APSTE on APST quirked devices at initialization.
> If the BIOS or NVMe enables APST before driver loading, APST will never be
> disabled, it also ignores explicit kernel option as result.
> So the faulty NVMe may not work as intended.
> 
> [Test Case]
> $ sudo nvme get-feature -f 0x0c -H /dev/nvme0
> ...will show APST is "Disabled" instead of "Enabled"
> 
> [Fix]
> Originally, APSTA is used to determined whether APST should be enabled or not,
> but it actually means hardware support instead of APST enabled/disabled.
> Separate the logic "apst is supported" and "to enable apst", so we can
> use the latter one to explicitly disable APST at initialiaztion.
> 
> [Regression Potential]
> Very low.
> This SRU didn't change anything really - it just explicitly set APSTE at
> initialization.
> 
> Andy Lutomirski (2):
>   nvme: Display raw APST configuration via DYNAMIC_DEBUG
>   nvme: Add nvme_core.force_apst to ignore the NO_APST quirk
> 
> Kai-Heng Feng (1):
>   nvme: explicitly disable APST on quirked devices

Note that this is now commit 76a5af841755a0427229a6a77ca83781d61e5b2a
upstream.

Acked-by: Seth Forshee <seth.forshee at canonical.com>

Applied to artful/master-next and unstable/master, thanks.




More information about the kernel-team mailing list