linux-kernel - Re: [PATCH v2] PCI: pciehp: Don't enable PME on runtime suspend

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Message-ID: <20170208180458.GC27298@bhelgaas-glaptop.roam.corp.google.com>
Date:   Wed, 8 Feb 2017 12:04:58 -0600
From:   Bjorn Helgaas <helgaas@...nel.org>
To:     Lukas Wunner <lukas@...ner.de>
Cc:     "Rafael J. Wysocki" <rjw@...ysocki.net>,
        Yinghai Lu <yinghai@...nel.org>,
        Mika Westerberg <mika.westerberg@...ux.intel.com>,
        linux-pci@...r.kernel.org, linux-kernel@...r.kernel.org
Subject: Re: [PATCH v2] PCI: pciehp: Don't enable PME on runtime suspend

On Wed, Feb 08, 2017 at 05:23:54AM +0100, Lukas Wunner wrote:
> On Tue, Feb 07, 2017 at 05:04:45PM +0100, Rafael J. Wysocki wrote:

> > The bottom line, in any case, is that the current code causes problems to
> > happen somewhere and as a rule we don't release code that is known to
> > cause problems to happen to anyone.  This means something needs to be done
> > about that and the choice at this point is pretty much between reverting and
> > quirking the affected system(s).
> 
> Quirking is not an option in this case because the PCI device IDs of the
> affected hotplug ports as well as DMI data are unknown.  Yinghai Lu is
> refusing to publish that, for both affected systems.

A quirk is appropriate if the hardware isn't working per spec.  I
don't think we have convincing evidence of that yet, so I'm more
interested in investigating that than I am in getting the device IDs
to add a quirk.

I suspect there's useful information to be had by disabling CONFIG_PM
and CONFIG_HOTPLUG_PCI_PCIE and poking things with setpci, which could
be done regardless of any NDAs.

Bjorn