[<prev] [next>] [<thread-prev] [day] [month] [year] [list]
Message-ID: <20210426131538.0b69c69b@redhat.com>
Date: Mon, 26 Apr 2021 13:15:38 -0600
From: Alex Williamson <alex.williamson@...hat.com>
To: Christoph Hellwig <hch@...radead.org>
Cc: Shanker R Donthineni <sdonthineni@...dia.com>,
Sinan Kaya <okaya@...nel.org>,
Bjorn Helgaas <bhelgaas@...gle.com>, linux-pci@...r.kernel.org,
linux-kernel@...r.kernel.org, Vikram Sethi <vsethi@...dia.com>
Subject: Re: [PATCH 1/1] PCI: Add pci reset quirk for Nvidia GPUs
On Mon, 26 Apr 2021 19:19:43 +0100
Christoph Hellwig <hch@...radead.org> wrote:
> On Fri, Apr 23, 2021 at 04:45:15PM -0500, Shanker R Donthineni wrote:
> > > specific platforms (embedded device?), and the failure mode of the SBR.
> > These are not plug-in PCIe GPU cards, will exist on upcoming
> > server baseboards. Triggering SBR without firmware notification
>
> Please submit the quirks together with the actual support for the GPUs
> in the nouveau driver, as they are completely useless without that.
My default assumption would be that this resolves an issue with
assigning this device to a userspace or VM driver through vfio-pci, as
most in-kernel drivers don't make use of this interface themselves;
they often know more device specific ways to re-initialize hardware.
This reset path is also trivially accessible through pci-sysfs. I
don't expect nouveau would have much use for this even if it did
include support for these devices. Thanks,
Alex
Powered by blists - more mailing lists