From: 'Aristeu Rozanski' <aris@redhat.com>
To: Borislav Petkov <bp@alien8.de>
Cc: "Luck, Tony" <tony.luck@intel.com>,
"'linux-edac@vger.kernel.org'" <linux-edac@vger.kernel.org>,
'Mauro Carvalho Chehab' <mchehab@kernel.org>
Subject: Re: [PATCH] EDAC: skx_common: downgrade message importance on missing PCI device
Date: Mon, 6 Jan 2020 10:12:42 -0500 [thread overview]
Message-ID: <20200106151242.vkdiiwhubmkx5osh@redhat.com> (raw)
In-Reply-To: <20191210090013.GA9395@zn.tnic>
On Tue, Dec 10, 2019 at 10:00:13AM +0100, Borislav Petkov wrote:
> On Tue, Dec 10, 2019 at 12:02:45AM +0000, Luck, Tony wrote:
> > > This looks like we call skx_init() once per core. Do we keep calling it because
> > > the calls are failing? Or do we do that even when calls succeed?
> > >
> > > I was only really expecting that skx_init() would be called once.
> >
> > So (by experimentation) it seems that if the module load fails it
> > will be retried num_online_cpus times (though not bound to each
> > CPU in turn ... it will maybe try the init call on the same CPU multiple
> > times, but miss running on some CPUs).
> >
> > If the load succeeds, then whoever is repeating the load decides
> > to stop.
>
> That's the result of our conversion to MODULE_DEVICE_TABLE to match CPU
> models. So it tries once on each CPU:
>
> https://lkml.kernel.org/r/20191107103857.GC19501@zn.tnic
>
> I have no clean solution for this except maybe remembering the return
> value of the first instance probing in the edac core module and then
> asking it... it ain't pretty though.
The other option I can think about is just allowing the module to load
in this situation: the CPU is present but you have no memory controller
PCI devices present. Some drivers will allow loading without having a
device present without errors. It's not clean as having to come up with
something modutils can pick up as hint to not try to load more than once.
Or could just downgrade this specific message since it's a very common
case and let the more unusual situations report more than once.
--
Aristeu
next prev parent reply other threads:[~2020-01-06 15:12 UTC|newest]
Thread overview: 13+ messages / expand[flat|nested] mbox.gz Atom feed top
2019-12-04 21:23 [PATCH] EDAC: skx_common: downgrade message importance on missing PCI device Aristeu Rozanski
2019-12-09 21:52 ` Luck, Tony
2019-12-10 0:02 ` Luck, Tony
2019-12-10 9:00 ` Borislav Petkov
2020-01-06 15:12 ` 'Aristeu Rozanski' [this message]
2020-01-06 16:17 ` Borislav Petkov
2020-01-06 16:20 ` 'Aristeu Rozanski'
2020-01-06 16:23 ` Borislav Petkov
2020-01-07 15:51 ` 'Aristeu Rozanski'
2020-01-07 16:45 ` Borislav Petkov
2020-01-07 21:43 ` Luck, Tony
2020-01-08 14:54 ` 'Aristeu Rozanski'
2019-12-10 14:18 ` 'Aristeu Rozanski'
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=20200106151242.vkdiiwhubmkx5osh@redhat.com \
--to=aris@redhat.com \
--cc=bp@alien8.de \
--cc=linux-edac@vger.kernel.org \
--cc=mchehab@kernel.org \
--cc=tony.luck@intel.com \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).