[<prev] [next>] [<thread-prev] [day] [month] [year] [list]
Message-ID: <20180219204331.GB48133@bhelgaas-glaptop.roam.corp.google.com>
Date: Mon, 19 Feb 2018 14:43:31 -0600
From: Bjorn Helgaas <helgaas@...nel.org>
To: linux-pci@...r.kernel.org
Cc: Jayachandran Nair <Jayachandran.Nair@...ium.com>,
Lorenzo Pieralisi <lorenzo.pieralisi@....com>,
"Rafael J. Wysocki" <rjw@...ysocki.net>,
linux-kernel@...r.kernel.org,
Alex Williamson <alex.williamson@...hat.com>,
Robert Richter <Robert.Richter@...ium.com>,
Tom Vaden <tom.vaden@....com>,
Mika Westerberg <mika.westerberg@...ux.intel.com>,
George Cherian <george.cherian@...ium.com>
Subject: Re: [PATCH v1 0/2] PCI: Probe for reset support earlier
On Fri, Feb 16, 2018 at 04:50:03PM -0600, Bjorn Helgaas wrote:
> The PCI core currently uses a late_initcall to probe each device for
> whether it supports reset. This is dangerous because a driver may have
> already claimed the device by this point, and the PCI core should not
> interfere with the driver by touching the device on its own.
>
> These patches move the probe to be earlier, during enumeration, before a
> driver has a chance to claim the device.
>
> ---
>
> Bjorn Helgaas (2):
> PCI: Probe for device reset support before driver claim
> PCI: Remove redundant probes for device reset support
>
>
> drivers/pci/pci-sysfs.c | 3 +--
> drivers/pci/pci.c | 15 ++++++---------
> drivers/pci/probe.c | 3 +++
> 3 files changed, 10 insertions(+), 11 deletions(-)
Applied both to pci/virtualization with Rafael's reviewed-by (and
update to comment to s/D3hot/D3/) for v4.17.
Powered by blists - more mailing lists