[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <aCNiiHzOksQFrPe1@black.fi.intel.com>
Date: Tue, 13 May 2025 18:17:28 +0300
From: Raag Jadav <raag.jadav@...el.com>
To: Lukas Wunner <lukas@...ner.de>
Cc: rafael@...nel.org, mahesh@...ux.ibm.com, oohall@...il.com,
bhelgaas@...gle.com, linux-pci@...r.kernel.org,
linux-kernel@...r.kernel.org, ilpo.jarvinen@...ux.intel.com,
aravind.iddamsetty@...ux.intel.com
Subject: Re: [PATCH v3] PCI: Prevent power state transition of erroneous
device
On Tue, May 13, 2025 at 06:41:31AM +0200, Lukas Wunner wrote:
> On Sun, May 04, 2025 at 02:34:44PM +0530, Raag Jadav wrote:
> > If error flags are set on an AER capable device, most likely either the
> > device recovery is in progress or has already failed. Neither of the
> > cases are well suited for power state transition of the device, since
> > this can lead to unpredictable consequences like resume failure, or in
> > worst case the device is lost because of it. Leave the device in its
> > existing power state to avoid such issues.
>
> Have you witnessed this on a particular platform / hardware combination?
> If so, it would be good to mention it. If I'd happen to find this
> commit in the future through "git blame", that's the first question
> that would come to mind: How and on what hardware was this actually
> triggered, how can I reproduce it.
We have a few issues[1] reported which are similar in nature. But since
they are not easily reproducible and still under investigation, I'm
a bit hesitant to explicitly reference it.
[1] https://gitlab.freedesktop.org/drm/xe/kernel/-/issues/4395
> > + /*
> > + * If error flags are set on an AER capable device, most likely either
> > + * the device recovery is in progress or has already failed. Neither of
> > + * the cases are well suited for power state transition of the device,
> > + * since this can lead to unpredictable consequences like resume
> > + * failure, or in worst case the device is lost because of it. Leave the
> > + * device in its existing power state to avoid such issues.
> > + */
>
> That's quite verbose and merely a 1:1 repetition of the commit message.
> I'd recommend a more condensed code comment and anyone interested in
> further details may look them up in the commit message.
Sure, will update.
Raag
Powered by blists - more mailing lists