udevd warning during boot on -current

saxa · 10-18-2015, 01:43 PM

Hi I just noticed few days ago that I get this kind of strange warning during boot time:

Code:

udevd[397]: Failed to execute '/usr/bin/xcmddc' '/usr/bin/xcmddc --i2c /dev/i2c-1 --identify': No such file or directory

udevd[398]: Failed to execute '/usr/bin/xcmddc' '/usr/bin/xcmddc --i2c /dev/i2c-1 --identify': No such file or directory

udevd[399]: Failed to execute '/usr/bin/xcmddc' '/usr/bin/xcmddc --i2c /dev/i2c-1 --identify': No such file or directory

Any ideas on why of this ?

Rgds
Saxa

genss · 10-18-2015, 02:08 PM

it is trying to execute some binary that does something (identifies?) on a device node that represents an i2c interface
google does not mention that binary

to find out what udev rule tells it to do such a thing do a `grep -r xcmddmc /lib/udev/`
if it is a rule, that is

to find out what device that is one would
ls -l /dev/i2c-1
then look at /sys/dev/*/ for that node

for example on my computer

Code:

bash-4.3# ls -l /dev/i2c-1 
crw------- 1 root root 89, 1 Lis 18 17:43 /dev/i2c-1
# the "c" says it is a character device, so /sys/dev/char/
# node is 89 1
bash-4.3# cd /sys/dev/char/
bash-4.3# ls
bunch of nodes
bash-4.3# cd 89\:1/
bash-4.3# ls
dev  device  name  power  subsystem  uevent
bash-4.3# cat name 
NVIDIA i2c adapter 2 at 1:00.0

maybe someone else knows what xcmddmc is

saxa · 10-18-2015, 02:21 PM

Thats what I get:

Code:

root@rocker:~# ls -l /dev/i2c-1 
crw------- 1 root root 89, 1 Oct 18 15:23 /dev/i2c-1

root@rocker:/sys/dev/char/89:1# cat name 
nvkm-0000:02:00.0-0

ljb643 · 10-18-2015, 02:34 PM

xcmddc is included in Slackware 14.1 (I know, you are running -current) in xcm-0.5.2. The description of xcm is "X color management tools", but xcmddc is a program to get EDID information (which describes the capabilities of a video display) over the I2C bus. The udev rule file 90-xcm-i2c.rules file references this program, but that file is also in the xcm package. So you shouldn't have that udev rule unless you also have the program it runs.

ReaperX7 · 10-18-2015, 06:28 PM

So, the $64,000 question is:

Did you perform a complete, full installation of Slackware?

saxa · 10-18-2015, 06:32 PM

ok, need to check that more in detail. Yes, the install is fresh and complete, I did that because my old install on a hdd suffered a failure of the hdd so I inserted a spare SSD I had and made a clean fresh install.

ReaperX7 · 10-18-2015, 06:39 PM

It shouldn't complain then, especially with Slackware's udev-classic package, but there is always the chance something either didn't install right, or there might be a rule issue. Track down the package that binary came in using the Slackware website package tracker, and then try re-downloading and re-installing the package see if that clears it up first, if not, let us know and we can think of something else. There is always the chance it's one of the udev rules having a case of the hiccups.

Just use "upgradepkg --reinstall *.t?z" on the package in a directory you download it to.

saxa · 10-18-2015, 06:51 PM

Ok, will try to look into it, if I find it out i will report here. Thing is that I noticed that only few days ago, so I'm not sure if this was happening from the beginning or after I did some updates with slackpkg to the latest -current state. I nearly sure it appeared after the latest updates. But nearly means nearly, so I wont affirm that.
What I'm sure is that except alienbobs libreoffice, and Virtualbox I have nothing else installed what would be out of the slackware current tree. This is 64bit install.

chris.willing · 10-19-2015, 07:34 PM

Quote:

Originally Posted by saxa

Hi I just noticed few days ago that I get this kind of strange warning during boot time:

Code:

udevd[397]: Failed to execute '/usr/bin/xcmddmc' '/usr/bin/xcmddc --i2c /dev/i2c-1 --identify': No such file or directory

udevd[398]: Failed to execute '/usr/bin/xcmddmc' '/usr/bin/xcmddc --i2c /dev/i2c-1 --identify': No such file or directory

udevd[399]: Failed to execute '/usr/bin/xcmddmc' '/usr/bin/xcmddc --i2c /dev/i2c-1 --identify': No such file or directory

Presumably you cut and pasted that output? Notice that there are two similar looking but different binaries referenced here - xcmddmc and xcmddc. As already mentioned, the xcmddc comes from the xcm package in both 14.1 and -current (which also includes the 90-xcm-i2c.rules file installed in /lib/udev/rules.d). However the xcmddmc binary isn't part of 14.1 or -current that I can see. Maybe it's being called via some script which has a typo ...

chris

saxa · 10-19-2015, 09:39 PM

No I have not pasted it out, I could not do that as this error does not appear in the /var/log/messages or dmesg files. But I noticed there is an "m" more in the first binary. And this is my typo.

saxa · 10-19-2015, 09:40 PM

So to avoid confusion I will edit the initial post and correct it.

---------- Post added 10-20-15 at 02:41 AM ----------

Done.

Philip Lacroix · 11-19-2015, 02:42 PM

Quote:

Originally Posted by saxa

Code:

udevd[397]: Failed to execute '/usr/bin/xcmddc' '/usr/bin/xcmddc --i2c /dev/i2c-1 --identify': No such file or directory
(...)

Any ideas on why of this ?

Do you have /usr and /var on separate partitions? I see exactly the same messages on a machine with 32bit -current (full fresh reinstall, nothing missing) with a split filesystem setup: /, /usr, /var, /tmp and /home. When the messages are shown, the /usr filesystem is not yet mounted, neither is /var, which apparently should explain the error messages themselves and why they are not being logged. IIRC this is a known issue with udev on similar setups.

I commented out the related line in udev rule "/lib/udev/rules.d/90-xcm-i2c.rules" and as expected the errors disappeared. Everything seem to be fine so far.

Code:

# load the device node kernel module
SUBSYSTEM=="i2c-adapter", ACTION=="add", RUN+="/sbin/modprobe i2c-dev"

# openSUSE_11.3 preferes:
SUBSYSTEM=="i2c", RUN+="/sbin/modprobe -bv i2c-dev"

# match only DDC devices
#KERNEL=="i2c-[0-9]*", ACTION=="add", PROGRAM=="/usr/bin/xcmddc --i2c /dev/%k --identify", RESULT=="0", GROUP="video", MODE="0660"

Philip

saxa · 11-19-2015, 05:40 PM

Hi Philip, yes I have it on separate partitions.

Code:

bash-4.3$ cat /etc/fstab
/dev/sda2        /                btrfs       defaults,noatime,discard         1   1
/dev/sda1        /boot            btrfs       defaults,noatime,discard         1   2
/dev/sda3        /usr             btrfs       defaults,noatime,discard         1   2
/dev/sda4        /home            btrfs       defaults,noatime,discard         1   2
#/dev/cdrom      /mnt/cdrom       auto        noauto,owner,ro,comment=x-gvfs-show 0   0
/dev/fd0         /mnt/floppy      auto        noauto,owner     0   0
devpts           /dev/pts         devpts      gid=5,mode=620   0   0
proc             /proc            proc        defaults         0   0
tmpfs            /dev/shm         tmpfs       defaults         0   0
bash-4.3$

as you see /usr is on a separate partition from /var which is on the same partition as / , but will try to use your solution of commenting it out.

Rgds
Saxa

Philip Lacroix · 11-20-2015, 06:23 PM

Hi Saxa. The switch from udev to eudev on -current fixed the mentioned issue. I'm not sure to what extent this is related to eudev's "--enable-split-usr" build option.

eudev.SlackBuild

Best
Philip

saxa · 11-20-2015, 07:20 PM

Ok, thanks for the info. I will probably stay with the default install.