Hi Nicolas,
On Tue, 15 Sep 2009 23:22:14 +0200, Nicolas Krzywinski wrote:
just another desparate user that wrung out lm-sensors.org, Google & Co.
;-)
First of all: I don't really have a problem, because lm-sensors work on
my
baby
But: There is some spamming in the kernel log (see subject and below)
-----
Now lets hear the details:
My system:
1x Intel Xeon DP E5506
..on Tyan S7002WGM2NR-LE (one CPU slot free)
..using KVR1066D3Q8R7SK3/12G as short-term memory
..with several SATA drives attached
..running Debian 2.6.26-17lenny2
As far as I understand aptitude, I installed:
lm-sensors 3.0.2-1+b2
..with libc6 2.7-18
..and libsensors4 3.0.2-1+b2
..and perl 5.10.0-19lenny2
..and sed 4.1.5-6
What about the kernel version? That's all what really matters here.
-----
What I modified in sensors3.conf:
[..]
chip "lm78-*" "lm79-*" "w83781d-*"
[..]
ignore in0
ignore in1
ignore in2
ignore in3
ignore in4
ignore in5
ignore in6
[..]
ignore fan1
ignore fan2
ignore fan3
ignore temp1
ignore cpu0_vid
[..]
chip "w83793-*"
[..]
ignore fan6
ignore fan7
ignore fan8
ignore fan9
ignore fan10
ignore fan11
ignore fan12
# This we have to reactivate when there is a second cpu installed..
ignore temp2
ignore temp5
[..]
Remarks: ignored anything from lm78.. because I got only rubbish from
there - but Winbond works fine, nothing more that I need.
Most probably because you don't have an LM78 and that would be a
misdetection. I remember we have a lot of these some times ago, but
this should be fixed since kernel 2.6.28 and lm-sensors 3.0.3.
Anyway, instead of ignoring everything, you'd rather simply _not_ load
the lm78 kernel module.
-----
Output of "sensors" command gives me pretty good values (except min/max
but I'm not keen on that):
me@myserver:~$ sensors
lm78-i2c-0-2d
Adapter: SMBus I801 adapter at 0400
w83793-i2c-0-2f
Adapter: SMBus I801 adapter at 0400
VCoreA: +1.21 V (min = +0.00 V, max = +2.05 V)
VCoreB: +0.00 V (min = +0.00 V, max = +2.05 V)
Vtt: +1.09 V (min = +0.00 V, max = +2.05 V)
in3: +1.47 V (min = +0.00 V, max = +4.08 V)
in4: +1.07 V (min = +0.00 V, max = +4.08 V)
+3.3V: +3.28 V (min = +0.00 V, max = +4.08 V)
+12V: +11.71 V (min = +0.00 V, max = +24.48 V)
+5V: +4.90 V (min = +0.15 V, max = +6.27 V)
5VSB: +4.88 V (min = +0.15 V, max = +6.27 V)
VBAT: +3.18 V (min = +0.00 V, max = +4.08 V)
fan1: 525 RPM (min = 0 RPM)
fan2: 0 RPM (min = 0 RPM)
fan3: 775 RPM (min = 0 RPM)
fan4: 1131 RPM (min = 0 RPM)
fan5: 703 RPM (min = 0 RPM)
CPU1 Temp: +71.2°C (high = +100.0°C, hyst = +95.0°C) sensor = Intel
PECI
temp6: +49.0°C (high = +100.0°C, hyst = +95.0°C) sensor =
thermistor
beep_enable:disabled
Remarks: don't wonder CPU temp, this is fully on load for more than 2
days
now ;-)
-----
Now, all that I need works thats good ... but lets have a look at my
kernel log:
me@myserver:~$ tail /var/log/kern.log
Sep 15 21:49:59 server7even3 kernel: [2670913.842602] i801_smbus
0000:00:1f.3: Bus collision! SMBus may be locked until next hard reset.
(sorry!)
Sep 15 22:13:48 server7even3 kernel: [2672347.037041] i801_smbus
0000:00:1f.3: Bus collision! SMBus may be locked until next hard reset.
(sorry!)
Sep 15 22:20:55 server7even3 kernel: [2672775.506601] i801_smbus
0000:00:1f.3: Bus collision! SMBus may be locked until next hard reset.
(sorry!)
Sep 15 22:28:04 server7even3 kernel: [2673204.696425] i801_smbus
0000:00:1f.3: Bus collision! SMBus may be locked until next hard reset.
(sorry!)
Sep 15 22:31:39 server7even3 kernel: [2673420.968805] i801_smbus
0000:00:1f.3: Bus collision! SMBus may be locked until next hard reset.
(sorry!)
Sep 15 22:37:15 server7even3 kernel: [2673757.521883] i801_smbus
0000:00:1f.3: Bus collision! SMBus may be locked until next hard reset.
(sorry!)
Sep 15 23:01:02 server7even3 kernel: [2675186.137306] i801_smbus
0000:00:1f.3: Bus collision! SMBus may be locked until next hard reset.
(sorry!)
Sep 15 23:03:58 server7even3 kernel: [2675362.459786] i801_smbus
0000:00:1f.3: Bus collision! SMBus may be locked until next hard reset.
(sorry!)
Sep 15 23:07:30 server7even3 kernel: [2675574.876593] i801_smbus
0000:00:1f.3: Bus collision! SMBus may be locked until next hard reset.
(sorry!)
Sep 15 23:09:01 server7even3 kernel: [2675666.787660] i801_smbus
0000:00:1f.3: Bus collision! SMBus may be locked until next hard reset.
(sorry!)
This message is needlessly alarming. What really happened is that the
Intel 82801 was not able to claim mastership of the SMBus, because
someone else is using it. Maybe you have a BMC module connected to this
motherboard? Are you using IPMI tools somehow? In this case, with older
kernels, the transaction simply fails, so you get transient errors. As
the w83793 driver apparently doesn't handle them properly,
The message was fixed in kernel 2.6.27. Since kernel 2.6.31, the failed
transaction is also retried automatically if the adapter asks for it,
but the i2c-i801 driver doesn't. This should be added. In the meantime
it is possible to set the retry count from user-space using the i2c-dev
driver.
I do not really get to the point on what is causing those messages,
because:
- I got readings (any other occurrence of this message I found got no
sensor readings)
- I added only "ignores" to sensors3.conf (did I confused the kernel
because I ignored anything from lm78??)
No, the kernel messages are unrelated to sensors3.conf.
- There is no time schema I can identify, the messages seem to occur
really arbitrary
What I did not do until now (I apologize, but have reasons):
- Restart of system (I don't want to restart because I think that
message
will come back)
- stopped sensord (because of the need to monitor the cpu temp, you
know..)
Further:
- I installed phpsysinfo as well. Anything evil with that maybe?
No, phpsysinfo merely runs "sensors" and parses its output, it doesn't
so anything you aren't already doing yourself.
Long story, short question: any hints?
Upgrade to kernel >= 2.6.28, stop loading the lm78 kernel module.
Upgrade to lm-sensors >= 3.0.3 and run sensors-detect again, you might
get more sensors values from another monitoring chip.
