From mboxrd@z Thu Jan  1 00:00:00 1970
From: keith.busch@intel.com (Keith Busch)
Date: Mon, 9 Oct 2017 16:22:15 -0600
Subject: Fwd: [Bug 196907] [Regression] New default s2idle does not work
 on Dell XPS 13 9360 with Ubuntu 16.04.3 LTS (Xenial Xerus)
In-Reply-To: <5479293.9hsiVanX3S@aspire.rjw.lan>
References: <1636342.6qveo3Fvtu@aspire.rjw.lan>
 <20171009221413.GE19329@localhost.localdomain>
 <5479293.9hsiVanX3S@aspire.rjw.lan>
Message-ID: <20171009222215.GF19329@localhost.localdomain>

On Tue, Oct 10, 2017@12:01:33AM +0200, Rafael J. Wysocki wrote:
> On Tuesday, October 10, 2017 12:14:14 AM CEST Keith Busch wrote:
> > On Mon, Oct 09, 2017@11:52:26PM +0200, Rafael J. Wysocki wrote:
> > > Hi,
> > > 
> > > We are seeing a resume issue with a Hynix SSD (PC300 NVMe SK hynix 512GB)
> > > with respect to suspend-to-idle (S3 resume works for the reporter,
> > > interestingly enough).  It appears to be limited to Hynix and to that
> > > specific SSD ATM.
> > > 
> > > Also it is reported to work in 4.11-rc1 and it already is failing in 4.13,
> > > so it looks like something between the two broke it.
> > > 
> > > Any ideas of what that might be?
> > > 
> > > Or anything we should try/test to narrow it down?
> > > 
> > > Any help will be appreciated.
> > 
> > I'd start with the APST feature and see if disabling that helps.
> > That can be done with the nvme_core module param:
> > 
> >   default_ps_max_latency_us=0
> > 
> 
> We've tried that and it doesn't help.

But it wasn't done correctly according to the notes on the
bugzilla. May be just a typo, but want make sure: it says they used
"nvme.default_ps_max_latency_us=0" on the kernel's command line, which
doesn't mean anything. It's "nvme_core" that provides that module
parameter.