Groups | Search | Server Info | Keyboard shortcuts | Login | Register [http] [https] [nntp] [nntps]

Groups > linux.debian.kernel > #73447 > unrolled thread

Bug#996672: linux-image-5.10.0-9-amd64: ite_cir floods syslog with receive overflow warning messages

Back to article view | Back to linux.debian.kernel

This discussion starts older than the indexed window; earlier articles aren't shown. The article labeled Started by below is the oldest one visible, not the original post.

Contents

  Bug#996672: linux-image-5.10.0-9-amd64: ite_cir floods syslog with receive overflow warning messages Salvatore Bonaccorso <carnil@debian.org> - 2021-10-17 14:50 +0200
    Bug#996672: SV: Bug#996672: linux-image-5.10.0-9-amd64: ite_cir floods syslog with receive overflow warning messages Salvatore Bonaccorso <carnil@debian.org> - 2021-10-17 21:40 +0200
      Bug#996672: SV: Bug#996672: linux-image-5.10.0-9-amd64: ite_cir floods syslog with receive overflow warning messages Salvatore Bonaccorso <carnil@debian.org> - 2021-10-19 07:10 +0200
      Bug#996672: SV: Bug#996672: linux-image-5.10.0-9-amd64: ite_cir floods syslog with receive overflow warning messages Salvatore Bonaccorso <carnil@debian.org> - 2021-10-21 08:00 +0200
        Bug#996672: Sv: Bug#996672: SV: Bug#996672: linux-image-5.10.0-9-amd64: ite_cir floods syslog with receive overflow warning messages Salvatore Bonaccorso <carnil@debian.org> - 2021-10-23 14:20 +0200
          Processed: Re: Sv: Bug#996672: SV: Bug#996672: linux-image-5.10.0-9-amd64:  ite_cir floods syslog with receive overflow warning messages "Debian Bug Tracking System" <owner@bugs.debian.org> - 2021-10-23 14:20 +0200

#73447 — Bug#996672: linux-image-5.10.0-9-amd64: ite_cir floods syslog with receive overflow warning messages

Hi Björn,

On Sun, Oct 17, 2021 at 08:51:02AM +0200, Björn Wiberg wrote:
> Package: src:linux
> Version: 5.10.70-1
> Severity: minor
> 
> Starting with inux-image-5.10.0-9-amd64, the ite_cir kernel module appears to massively flood (approximately 80 log entries per second) syslog with messages stating that a receive overflow has occurred:
> 
> /../
> Oct 10 19:31:39 glimmer kernel: [    3.758499] rc rc0: ITE8708 CIR transceiver as /devices/virtual/rc/rc0
> Oct 10 19:31:39 glimmer kernel: [    3.758557] rc rc0: lirc_dev: driver ite-cir registered at minor = 0, raw IR receiver, raw IR transmitter
> Oct 10 19:31:39 glimmer kernel: [    3.758619] input: ITE8708 CIR transceiver as /devices/virtual/rc/rc0/input3
> Oct 10 19:31:39 glimmer kernel: [    3.760191] ite_cir: driver has been successfully loaded
> /../
> Oct 10 20:42:51 glimmer kernel: [ 4276.187105] rc rc0: receive overflow
> Oct 10 20:42:51 glimmer kernel: [ 4276.195016] rc rc0: receive overflow
> /../ (repeated receive overflow messages)
> 
> This always occurs first after the system has been running for some time (approximately 1½ hours).
> 
> The logging of this condition appears to have been added to the source code earlier this year:
> https://lore.kernel.org/lkml/20210510102009.237644288@linuxfoundation.org/
> 
> It appears from the source code that some kind of reset is attempted, however the outcome of this is unknown (and apparently does not succeed and/or solve the receive overflow condition).
> 
> The machine in question is located in a closet with no known IR sending from other devices.
> There are no known applications running that are using the input device (/devices/virtual/rc/rc0/input3).
> 
> Output from lsmod shows the following:
> 
> root@glimmer:~# lsmod | grep -Ei '(ir_|_ir|rc_|_rc|ite)' | grep -v crc
> ir_rc6_decoder         20480  0
> rc_rc6_mce             16384  0
> ite_cir                32768  0
> rc_core                61440  4 ite_cir,ir_rc6_decoder,rc_rc6_mce
> 
> Possible work-arounds include blacklisting the ite_cir module (e.g. in /etc/modprobe.d/local.conf):
> blacklist ite_cir
> 
> ...which would then disable the IR functionality altogether, or to temporarily discard the log messages (e.g. in /etc/rsyslog.d/local.conf):
> if $syslogfacility-text == 'kern' and $programname == 'kernel' and re_match($msg, '\\] rc rc[^:]+: receive overflow$') and $syslogseverity >= 4 then stop
> 
> However this does not solve the underlying problem.
> 
> I understand that the problem could potentially be because of "iffy" hardware, but still wanted to report this, e.g. in case many others are seeing the same thing on their systems.

I assume you get the same logged as well using newer kernel from
unstable (or bullseye-backports, 5.14.9-2~bpo11+1, there are though
not yet signed packages)?

If so it might be best to ask upstream to rule out the iffy hardware
part.

Regards,
Salvatore

[toc] | [next] | [standalone]

#73454 — Bug#996672: SV: Bug#996672: linux-image-5.10.0-9-amd64: ite_cir floods syslog with receive overflow warning messages

Hi Björn,

On Sun, Oct 17, 2021 at 05:40:43PM +0000, Björn Wiberg wrote:
> Hello Salvatore!
> 
> Many thanks for your reply!
> 
> I’m afraid I cannot switch to unstable (or ”mix”), and if I
> understand it correctly, the unsigned kernel from backports won’t
> boot if Secure Boot is being used (which is the case on this
> server)?

Right but i expect the signed packages to appear "soonish" as well.
The signing machinery needs to be triggered once the respective
linux-signed-{amd64,i386,arm64} are there. So we can afford to wait a
couple of days yet if you are open to test the bpo kernel version.
> 
> Not sure about the best way to go forward – should I attempt to
> report it upstream, or will you be doing so?

It would be best if you can report it directly, otherwise we the
kernel maintainers are the bottleneck, help from people actually able
to do tests is more than welcome :). The reason I asked to test as
well the bpo version is that this brings us nearer to mainline.

I checked possible commits in the driver, and there is not much
so I wanted to check if you can reproduce the issue still as well with
newer stable version or ideally even better with mainline directly.

https://git.kernel.org/linus/49e851de7e573529885fd1df4365e2459c6030ee
is already for fixing an issue at probe time, so not sure.

> Another option would of course be to wait and see if anything
> changes in future versions.  This bug report will still be
> searchable for others, and can be resurrected if needed.

From my point of view keeping the list of bugs
https://bugs.debian.org/src:linux shorter than longer (with bugs where
we cannot really do something -- not to hide problems, but to have
them reported where it's more sensible, in many cases its just better
to handle upstream) is better.

So with the report downstream now with Debian, there should go along
at least a reporting upstream to narrow it down at the right place.

Hope this makes sense.

When you report it upstream, let us know as well in case we miss it.

Regards,
Salvatore

[toc] | [prev] | [next] | [standalone]

#73468 — Bug#996672: SV: Bug#996672: linux-image-5.10.0-9-amd64: ite_cir floods syslog with receive overflow warning messages

Hi,

This bugfix might be related:
https://lore.kernel.org/linux-media/20211017144453.9795-1-sean@mess.org/

Regards,
Salvatore

[toc] | [prev] | [next] | [standalone]

#73483 — Bug#996672: SV: Bug#996672: linux-image-5.10.0-9-amd64: ite_cir floods syslog with receive overflow warning messages

Hi Björn,

On Sun, Oct 17, 2021 at 09:26:39PM +0200, Salvatore Bonaccorso wrote:
> Hi Björn,
> 
> On Sun, Oct 17, 2021 at 05:40:43PM +0000, Björn Wiberg wrote:
> > Hello Salvatore!
> > 
> > Many thanks for your reply!
> > 
> > I’m afraid I cannot switch to unstable (or ”mix”), and if I
> > understand it correctly, the unsigned kernel from backports won’t
> > boot if Secure Boot is being used (which is the case on this
> > server)?
> 
> Right but i expect the signed packages to appear "soonish" as well.
> The signing machinery needs to be triggered once the respective
> linux-signed-{amd64,i386,arm64} are there. So we can afford to wait a
> couple of days yet if you are open to test the bpo kernel version.

FTR, bullseye-backports packages are now as well available for the
signed packages for linux-signed-amd64.

Regards,
Salvatore

[toc] | [prev] | [next] | [standalone]

#73512 — Bug#996672: Sv: Bug#996672: SV: Bug#996672: linux-image-5.10.0-9-amd64: ite_cir floods syslog with receive overflow warning messages

Control: tags -1 + upstream confirmed

Hi,

On Sat, Oct 23, 2021 at 10:50:43AM +0000, Björn Wiberg wrote:
> Hello again Salvatore!
> 
> I’ve been in touch with Sean Young regarding ite_cir and the
> ITE87080, and it appears that the problem might be caused by cheap
> hardware, exhibiting the problem of not recovering from a receive
> overflow unless reading out the FIFO buffer.
> 
> This was introduced in
> https://github.com/torvalds/linux/commit/28c7afb07ccfc0a939bb06ac1e7afe669901c65a#diff-469248a6db13b2310197808cca4dd41a36cc818a57fe15da07700e200c5d31c2
> where the buffer readout was skipped, and they will probably be
> pushing a fix
> (https://git.linuxtv.org/media_stage.git/commit/?id=fdc881783099c6343921ff017450831c8766d12a)
> upstream in due time.
> 
> Most probably, this will fix the issue for both NUC6s and NUC7s as
> well as newer versions.
> They all appear to be using ITE CIR transceivers (models ITE8704,
> ITE8708 and ITE8713 are listed in INF files inside Intel’s driver
> downloads for Windows).
> 
> So, I’ll just wait for now. 😊
> 
> Thanks again for your help!

Thanks for following up with the results. So this seems the same as in
https://bugs.debian.org/996672#27 . I can look if I can cherry-pick
this as well in the next upload.

Regards,
Salvatore

[toc] | [prev] | [next] | [standalone]

#73513 — Processed: Re: Sv: Bug#996672: SV: Bug#996672: linux-image-5.10.0-9-amd64: ite_cir floods syslog with receive overflow warning messages

Processing control commands:

> tags -1 + upstream confirmed
Bug #996672 [src:linux] linux-image-5.10.0-9-amd64: ite_cir floods syslog with receive overflow warning messages
Added tag(s) confirmed and upstream.

-- 
996672: https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=996672
Debian Bug Tracking System
Contact owner@bugs.debian.org with problems

[toc] | [prev] | [standalone]

Back to top | Article view | linux.debian.kernel

csiph-web