WG x750e - automatic speed adjustment: mbmon going crazy
-
Hello all,
I wrote a script to automatically adjust the fan speed depending on the CPU temperature:
#!/bin/sh # Damien P - 26-May-2013 # Usage auto_speed.sh [interval in seconds] # # Prerequisites: # Install of mbmon package: # pkg_add -r ftp://ftp-archive.freebsd.org/pub/FreeBSD-Archive/ports/i386/packages-8.1-release/Latest/mbmon.tbz # Install of cpuburn to load CPU and test program # pkg_add -r ftp://ftp-archive.freebsd.org/pub/FreeBSD-Archive/ports/i386/packages-8.1-release/Latest/cpuburn.tbz # Validate packages: # rehash # To run the burn test: # burnP6 & # Default check interval in seconds DefaultInterval=10 # First temperature threshold and associated fan speed setting Temp1=35 Fan1=20 # Second temperature threshold and associated fan speed setting Temp2=40 Fan2=28 # Third temperature threshold and associated fan speed setting Temp3=45 Fan3=35 # Last temperature threshold and associated fan speed setting Temp4=60 Fan4=FF # Check interval in seconds CheckInterval=${1:-$DefaultInterval} if ! [ -z "`echo $CheckInterval | tr -d [0-9]`" ]; then i CheckInterval=$DefaultInterval echo "$1 is not a positive integer, using default value [$DefaultInterval]" 1>&2 fi # Enter infinite loop while [ true ]; do # Get CPU temperature (in celsius) CpuTemp=`/usr/local/bin/mbmon -I -i -c1 -T2` if [ $CpuTemp -le $Temp1 ]; then FanSpeed=$Fan1 elif [ $CpuTemp -le $Temp2 ]; then FanSpeed=$Fan2 elif [ $CpuTemp -le $Temp3 ]; then FanSpeed=$Fan3 else FanSpeed=$Fan4 fi /usr/local/bin/WGXepc -f $FanSpeed sleep $CheckInterval doneIt works perfectly well for a couple of days until mbmon (see code for download location) starts showing a CPU temperature of 88 C:
$ /usr/local/bin/mbmon -I -i -c1 -T2
88
From there, the fan speed is set to its maximum (really noisy >:() and though the CPU is almost idle and the air getting out the box really cool, mbmon still indicates 88 C.
A reboot solves the issue but I do not want to have to do it every two days.Did anybody encounter the same issue ?
–
Damien -
Hmm, well I doubt anyone else has experienced that exactly since you're the first to try it! ;)
Anyway looks like interesting stuff. So you're saying that after ~2 days mbmon no longer reports the correct temperature value. Is that 2 days uptime? 2 days running the script? Does mbmon report the other values correctly?Do you end up with a large number of processes running or anything else odd?
Since mbmon is just reading the SuperIO chips values you could try installing the superiotool and reading the values with that instead. The output is not really readable usefully (without the datasheet) but you could see it it's a problem with the sensor rather than mbmon.
Steve
-
Thanks for your quick and thoughtful reply :D
@stephenw10:So you're saying that after ~2 days mbmon no longer reports the correct temperature value. Is that 2 days uptime? 2 days running the script?
It has in fact been running for 6 days now and the auto_speed.sh script starts upon boot time.
Does mbmon report the other values correctly?
I do not know. How does this look like ?
$ /usr/local/bin/mbmon -I -c1
Temp.= 25.0, 88.0, 67.0; Rot.= 0, 0, 0
Vcore = 0.96, 2.26; Volt. = 3.41, 5.16, 12.59, -12.45, -2.18Do you end up with a large number of processes running or anything else odd?
This was my first thought too but the system is running smoothly without any duplicate processes and with a very low CPU load:
last pid: 61328; load averages: 0.01, 0.02, 0.00 up 6+03:56:54 15:41:26
49 processes: 1 running, 48 sleeping
CPU: 0.7% user, 1.1% nice, 2.6% system, 1.5% interrupt, 94.1% idle
Mem: 45M Active, 17M Inact, 76M Wired, 120K Cache, 59M Buf, 1849M FreeSince mbmon is just reading the SuperIO chips values you could try installing the superiotool and reading the values with that instead. The output is not really readable usefully (without the datasheet) but you could see it it's a problem with the sensor rather than mbmon.
Thanks for the tip. However I have a hard time understanding the output (mbmon output below but could not get anything matching):
$ superiotool
superiotool r
Found Winbond W83627HF/F/HG/G (id=0x52, rev=0x41) at 0x2e$ ( superiotool -d ; mbmon -I -c1 ) | tail -7
LDN 0x0b (Hardware monitor)
idx 30 60 61 70 f0
val 01 02 90 00 00
def 00 00 00 00 00Temp.= 25.0, 88.0, 67.0; Rot.= 0, 0, 0
Vcore = 0.96, 2.26; Volt. = 3.41, 5.16, 12.59, -12.45, -2.18Here is what I got after a reboot:
$ ( superiotool -d ; mbmon -I -c1 ) | tail -7
LDN 0x0b (Hardware monitor)
idx 30 60 61 70 f0
val 01 02 90 00 00
def 00 00 00 00 00Temp.= 25.0, 30.5, 3.0; Rot.= 675000, 225000, 11842
Vcore = 0.96, 2.16; Volt. = 3.41, 5.16, 12.59, -12.45, -2.18superiotool output is exactely the same, unlike mbmon's.
–
Damien -
Ah, interesting so mbmon, after some time or number of calls, stops reporting both the temperature and the fan speeds. I assume the fans were actually running at that point. You have seen how the fan speed cannot be read correctly by the superio chip at speeds below ~BB. Since at that time they were running at FF they should have been correctly displayed. Potentially I could imagine a scenario where the fan speed read goes into 7 figures and crashed something. Given enough readings this might happen.
I don't have the superIO data sheet here but it's easily available. I remember reading through it for hours to work out what register did what and how to change it. Anyway I believe you need to run the extended dump to see the all the registers including the fan speed and temperature output. One word of caution here as I caught me out. On the XTM5 box, which has a different SuperIO chip, running the extended dump somehow locked the chip such that after running it I was no longer able to change any register. I don't remember that ever happening on the X-e box and since you're only reading values it shouldn't be an issue anyway.
Steve
-
Adding the extended dump on my x750e only outputs one more useless line:
Hardware monitor (0x0295)
I guess I'll have to stick to a fixed fan speed and forget about having it automated.Damien
-
Hmm, that's no good then. Don't give up yet automatic fan speed control would be great. :)
Let me fire up my X750e and see if I can replicate your results.Steve
-
Ok, I checked my box here and yes you're right superiotool does not show the relevant info for some reason. It works fine with some other chips.
However it is the SuperIO chip that mbmon is reading and it is possible to read it directly but it's not that straight forward.To read the values from the chips you set it's index register to required value and then read the value from the data register. In this case (and almost every case) those are I/O space 0x295 and 0x296 respectively. To do this when I was initially looking for the LED control I "wrote" (I use that term very loosely!) two programs, readio and writeio, which can be found at my Google site. These can be used to write to any address in I/O space so can obviously cause problems. There is no sanity checking or anything, you've been warned! ;)
Using those we can read values. So for example the 8bit temperature value from sensor 1 is at control register CR27 (the one value SuperIOtool skips past :-\ ):
[2.0.3-RELEASE][root@testbox.localdomain]/conf(347): ./writeio 0x295 0x27 Setting 295 to 27 [2.0.3-RELEASE][root@testbox.localdomain]/conf(348): ./readio 0x296 Reading 296 :19It's hex value 0x19, 25 degrees C. We know that's correct from mbmon.
Unfortunately sensor 1 is almost certainly not connected. Sensor 2 is the only one returning useful data (though I did play with sensor 3 a bit with interesting results). The data values for Sensor 2 are in 'Bank 1' which can only be accessed by first setting the 'bank select' register, 0x4e, to 1. So:[2.0.3-RELEASE][root@testbox.localdomain]/conf(349): ./writeio 0x295 0x4e Setting 295 to 4e [2.0.3-RELEASE][root@testbox.localdomain]/conf(350): ./writeio 0x296 0x01 Setting 296 to 1 [2.0.3-RELEASE][root@testbox.localdomain]/conf(351): ./writeio 0x295 0x50 Setting 295 to 50 [2.0.3-RELEASE][root@testbox.localdomain]/conf(352): ./readio 0x296 Reading 296 :28We finally get the value 0x28 which is 40C. About what we'd expect. In fact the value is a 9 bit number with the 9th bit offering 0.5C accuracy but these 8MSBs are probably fine here (also it's stored as 2s compliment so if the numbers go negative things get complicated. ;)).
If you don't set the bank select back to 0 mbmon will get confused the first time you run it but runs fine second time around.
It would be interesting to give this a go when mbmon has freaked out to see if the SuperIO chip itself is still giving useful data. It would be easy enough to incorporate this into WGXepc if it does.
Steve
-
I'm not too sure why you set register 0x295 to 0x50 once 0x296 has been properly set to 0x01 ?
Nevertheless, I could confirm that executing these four commands returned the same value than mbmon, both when CPU is idle and supercharged with burnP6.
I restarted my automated script and will patiently wait for it to go crazy again :PThanks for your detailed explanation and time!
Damien
-
I'm not too sure why you set register 0x295 to 0x50 once 0x296 has been properly set to 0x01 ?
Because in order to read the actual temperature value, which is at register 50 on Bank1, you need to point the index register, 0x295, at it. It was previously still pointed to 0x4e, the bank select register.
The first time I read through the data sheet and tried to comprehend how this all worked my mind nearly melted. :D It still took a while the second time around. One thing to be aware of is that in order to save processing time/instructions many of the registers automatically increment the index register. This helps if you want to read in a series of values, say 0x20 to 0x2f, you just set the index to 0x20 and then read the data register. That automatically increments the index by 1 so now you just read the data register again and it will be the value of 0x21 etc. This means that if you set the Bank to 1, set the index to 0x50 and read the value at the data register you cannot just read the data again to see it has changed, the index register will now be pointing at 0x51. You have to set the index to 0x50 every time.
It's interesting that mbmon expects to find the bank select register at 0 and when it's not it has a hard time, perhaps a clue? Something else intereacting with the SuperIO chip is leaving it in a state that confuses mbmon perhaps.
As an aside I notice your values for sensor 3 are different from mine. is 67C a common reading? My own shows mostly 127 with occasional 0 and even more occasional 2.0. Have you changed the CPU? Interestingly reconfiguring it as a thermistor instead of a thermal diode produced much more reasonable values though still clearly not real temperatures.
Steve
-
Ah, it's (somewhat) clearer now :)
I did not have a look at the mbmon code yet so I do not know why it behaves like this.
I can't wait though to see whether your trick with the registers will still work when mbmon doesn't anymore !I indeed invested
massively($10 including shipping) to replace the original Celeron 1.3GHz:
Intel(R) Pentium(R) M processor 2.00GHz
Current: 225 MHz, Max: 1500 MHzMore horsepower and less electricity/heat ;D
Damien
-
I added some detailed logging to understand what is going one and after 11 days, mbmon was only sporadically returning incoherent values (around once a day whereas measures are made every 10 seconds).
Maybe the multiple register manipulations helped ???At this stage, I'd say the automatic fan speed adjustment is working !
Thanks Steve for your input ;D
Damien
-
Maybe the multiple register manipulations helped ???
Plausible but sort of disappointing. Nothing to say it won't happen again or start doing something different.
Steve
-
Well, time will tell. At some point, I will reboot the x750e and with the level of logging I have in place now, it won't escape me.
Damien
-
I added the temperature reading to WGXepc so you can try that if mbmon starts acting up again. Though you'd have to parse it's output for the value in your script somehow.
http://forum.pfsense.org/index.php/topic,32013.msg346092.html#msg346092Steve
-
This time, mbmon started producing bogus data (CPU temp around 9°C) after 24 days of uptime.
Unfortunately, the new version of WGXepc provides the same measure.Let me know if you want me to try something else, otherwise, I'll just reboot the machine.
Damien
-
So, is auto fan speed control daemon showing up in the next release as an option click-to-enable?
-
This script is specifically for the Watchguard firebox X-e boxes so it's very unlikely to be in a pfSense realease. Even if it were more generic it probably wouldn't ever be included as more than a package. Any script that can control the fan speeds in a box has the potential to cause damage by overheating the CPU. Usually in something with thermal fan control, like a laptop, the control is handled directly by the SuperIO chip such that it will continue to cool correctly even if the OS crashes. Unfortunately the superio chip in the X-e box doesn't support this.
@bigramon
So reading the temperature with WGXepc gives the same value as mbmon. That implies the SuperIO chip is actually reporting the wrong value. Why could that be? It could be the temperature offset register has been set some how or that it's reading the wrong register for some reason. We could try investigating that but it will be quite involved. You could try using WGXepc in your script instead. Since it only reads one register (where as mbmon reads all registers everytime it's run) it may make a difference.Steve
-
Unfortunately, it did not work much better. After 2 weeks, I started getting temperatures of 255°C.
When I tried mbmon from the command line, it reported an error stating it could not access the hardware.
Everything went back to normal after a reboot.Damien
-
Ah, so what were you using in the script that didn't work? I forget where we left off. ::)
Steve
-
Something gets wrong after a short while:
- WGXepc does not properly set the fan speed anymore
- WGXepc does not report the proper temperature anymore
- mbmon does not execute properly anymore.
$ /usr/local/bin/WGXepc -f 50 Found Firebox X-E Fanspeed set to 50 $ /usr/local/bin/WGXepc -f Found Firebox X-E Fanspeed is ff $ /usr/local/bin/WGXepc -t Found Firebox X-E SuperIO sensor 2 reads: 255 $ /usr/local/bin/mbmon -I -i -c1 -T2 No ISA-IO HWM available!! InitMBInfo: Unknown error: 0At this stage, I guess the best option would be to modify the WGXepc source code to allow it to keep executing in the background and set the fan speed depending on the temperature that is read.
Damien
