From: "Rafael J. Wysocki" <rjw@rjwysocki.net>
To: Alan Stern <stern@rowland.harvard.edu>
Cc: Linux USB <linux-usb@vger.kernel.org>,
        Linux PM <linux-pm@vger.kernel.org>,
        LKML <linux-kernel@vger.kernel.org>,
        Greg Kroah-Hartman <gregkh@linuxfoundation.org>,
        Mathias Nyman <mathias.nyman@linux.intel.com>,
        Felipe Balbi <balbi@kernel.org>
Subject: Re: [PATCH] PM / USB: hcd_pci: Skip secondary root hub check for HCD_DEAD()
Date: Tue, 25 Jul 2017 17:59:04 +0200
Message-ID: <2055160.bTgMse1pkO@aspire.rjw.lan>
User-Agent: KMail/4.14.10 (Linux/4.12.0-rc1+; KDE/4.14.9; x86_64; ; )
In-Reply-To: <Pine.LNX.4.44L0.1707251000470.2256-100000@iolanthe.rowland.org>
References: <Pine.LNX.4.44L0.1707251000470.2256-100000@iolanthe.rowland.org>
MIME-Version: 1.0
Content-Transfer-Encoding: 7Bit
Content-Type: text/plain; charset="us-ascii"
Sender: linux-kernel-owner@vger.kernel.org
Content-Length: 1819
Lines: 54

On Tuesday, July 25, 2017 10:05:03 AM Alan Stern wrote:
> On Tue, 25 Jul 2017, Rafael J. Wysocki wrote:
> 
> > From: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
> > 
> > If HCD_DEAD(hcd) is "true" in check_root_hub_suspended(), it is
> > rather pointless to check the secondary root hub, so return early
> > then.
> > 
> > This actually fixes occasional suspend failures on one of my test
> > machines.
> > 
> > Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
> > ---
> >  drivers/usb/core/hcd-pci.c |    3 +++
> >  1 file changed, 3 insertions(+)
> > 
> > Index: linux-pm/drivers/usb/core/hcd-pci.c
> > ===================================================================
> > --- linux-pm.orig/drivers/usb/core/hcd-pci.c
> > +++ linux-pm/drivers/usb/core/hcd-pci.c
> > @@ -427,6 +427,9 @@ static int check_root_hub_suspended(stru
> >  		dev_warn(dev, "Root hub is not suspended\n");
> >  		return -EBUSY;
> >  	}
> > +	if (HCD_DEAD(hcd))
> > +		return 0;
> > +
> >  	if (hcd->shared_hcd) {
> >  		hcd = hcd->shared_hcd;
> >  		if (HCD_RH_RUNNING(hcd)) {
> 
> While this is an okay solution, IMO it would be more reliable and more 
> general to have usb_hc_died() clear the HCD_FLAG_RH_RUNNING bit and set 
> the HCD_FLAG_DEAD bit in the shared hcd.  Right now it only does these 
> things for the primary.
> 
> Would you like to write and test a patch to do that?

I can do that.

> Incidentally, if this fixes occasional suspend failures on your test 
> machine, does that mean the test machine's host controller occasionally 
> dies?  Maybe that should be fixed too...

Yes, it does sometimes.

What appears to happen is that the platform does not initialize properly
sometimes after a reset and then the HCD dies during suspend.

So reproduction may be somewhat tricky, but oh well.

Thanks,
Rafael