Simple fan speed control for the Firebox X750e

[stephenw10]

Indeed this was a problem with mbmon too. Interesting that it has taken this long for the error to occur. Any idea how long the script had been running before it defaulted to 255C? It may be possible to reset the SuperIO chip using some set of registers.

An alternative is the new BIOS from the SSL box has correct DSDT tables that pass the CPU temp via ACPI. If you're running B8.X then the temperature is reported by:

sysctl dev.cpu.0.temperature

I think. Something close to that anyway.  ;) Of course that is also just reading the superio chip so I don't know if that might be any better.

Steve

[Steve Evans]

Before the recent 2.1.1/2.1.2 upgrades my box had been up for 82 days and the fan control was still working OK.

Steve

[power_matz]

@stephenw10:

An alternative is the new BIOS from the SSL box has correct DSDT tables that pass the CPU temp via ACPI. If you're running B8.X then the temperature is reported by:

sysctl dev.cpu.0.temperature

This sysctl is not listed on my system…

I am running the B8 version. My uptime is at this moment only 6 days.

Matthias

[stephenw10]

Hmm. Perhaps ACPI is not enabled in the BIOS by default or the setting carried over from a previous bios version. Check your bios setting.
I'll have to check the default values but that's become a lot more difficult now that I no longer have a Win XP box to run modbin6 on.  ::)

Steve

[Steve Evans]

Virtual machine!

Steve

[power_matz]

I flashed Version 8.1 of the XEOS BIOS.
Still no entry in sysctl for this.

While booting pressing the TAB key doesn't bring me in the BIOS setup anymore.

Should I reset BIOS setting (jumper)?

Matthias

[stephenw10]

If you flashed it using flashrom from within pfSense then the CMOS will not have cleared which could be causing an issue.
I'm going to have review the changes I made. Fortunately it occurred to me the modbin6 is actually a DOS program so I should be able to fire up FreeDOS and run it.  ::)

Steve

[power_matz]

Hi,

after flashing BIOS 8.1 XEOS and restting the CMOS I could go into the appropriate section to enable ACPI.
Now the system is running fine. No issue with the temperature display so far.

Even the powerd stuff is working without adding the specific sysctl.

But the best is: my big problem is solved also. I had some weired crashes while connecting the WAN port (I write this in some other topics).
The reason was an entry in the loader.conf.local to support my processor 2.13 Ghz Pentium M with powerd.

But after the CMOS reset this is not necessary anymore.

So now my systems is running with HD and faster CPU and fan control and powerd. Fine!

Thanx so far to all who helped me.

Matthias

[stephenw10]

That's great news.
I tested my X5500e box that has the 2GHz Pentium-M in it (533MHz FSB) and still exhibited weird symptoms. I was hoping the fixed ACPI might send the correct valued to powerd. My box could have some other problem though I did get it very cheaply!
Out of interest what loader.conf value were you passing to make powerd work with your CPU?

Steve

[power_matz]

I don't have any settings now for using powerd. It is just working…
No config under Advance (sysctl) and no loader.conf stuff.

But there are only two speeds: 2.13 Ghz and 1003 Mhz in display.
Maybe the Pentium M have more steps!?

Matthias

[stephenw10]

Hmm, yes there should be more steps if the ACPI tables are sending the correct data. Since Watchguard never fitted that CPU it's impossible to say what support there might be.

With ACPI enabled the kernel timecounter uses the ACPI-fast counter as it's the most accurate. That means you don't have to use a sysctl to switch away from TSC (which changes with CPU frequency).

I was just interested to know what loader.conf value you were using previously since my CPU does not appear to be supported.

Steve

[power_matz]

Hi,

I added the following:

cpufreq_load="YES"

Matthias

[browntown]

After 3 months of happy uptime, I upgraded from 2.1.1 to 2.1.3.  The fan no longer throttles down to quiet.  Are there cliff notes on how to get it back running again?  Do I need to reload WGXepc?

Steve mentioned once that Shellcmd would be better as to survive an update, but I lost my bookmarks on how to do that. Searching these giant threads with all the quotes is a headache. Forgive my impatience.

edit:

I guess I am doing shellcmd, here's a screenie.

But it doesn't seem to work.

Edit a little later:
Guess the upgrade wiped wgxepc from the local folder. Re-downloaded into conf using the command function in the webscreen and the fetch commands from the wiki. and changed shellcmd to point to /conf/wgxepc .  All working now, thanks for your indulgence. I'm keeping this here so I can find it next update.

Note to self:
Go to Diagnostics -> nanobsd  –  enable read/write filesystem
Go to Diagnostics -> command  -- fetch -o /conf https://sites.google.com/site/pfsensefirebox/home/WGXepc
*****Can then test settings by running command --  /conf/WGXepc -l green or /conf/WGXepc -f 30
Go to Diagnostics -> nanobsd  --  Switch back to read only filesystem
Go to Services -> Shellcmd -- Add entry type shellcmd (or early) and your command  /conf/WGXepc -f 32

reboot and enjoy the quiet

Or just build a rasberry pi box and go fanless.

[stephenw10]

Yep, like you found only the /conf (actually a link to the config slice) is shared between slices and not overwritten.
I take it you're not actually using the fan speed control program then?

Steve

[power_matz]

Hi,

after enabling the ACPI settings the system runs good with the script. But now it is screwed up after some time with the display attached. The fan is running at ff then.

Any idea what is happening here?

Matthias


[stephenw10]

So that dashboard shot is with the temperature sensor set to ACPI?
This seems to be the same issue that has been seen throughout this thread. When ACPI is reporting 0 what does WGXepc or mbmon report? I expect to see it's gone to the maximum value.
It's hard to know quite what can be done about this. It might be feasible to somehow reset the SuperIO chip but it's hard to test anything because there's no reliable way to trigger the issue.

Steve

[power_matz]

ok. I wait until the error occurs. Then I will check what WGXexp reports

[Guest]

@browntown:

Or just build a rasberry pi box and go fanless.

pfSense doesn't run on the raspberry pi.

I've considered a port to the beaglebone black though.

[chpalmer]

@power_matz:

Hi,

after flashing BIOS 8.1 XEOS and restting the CMOS I could go into the appropriate section to enable ACPI.
Now the system is running fine. No issue with the temperature display so far.

Even the powerd stuff is working without adding the specific sysctl.

But the best is: my big problem is solved also. I had some weired crashes while connecting the WAN port (I write this in some other topics).
The reason was an entry in the loader.conf.local to support my processor 2.13 Ghz Pentium M with powerd.

But after the CMOS reset this is not necessary anymore.

So now my systems is running with HD and faster CPU and fan control and powerd. Fine!

Thanx so far to all who helped me.

Matthias

Thanks-  you got me doing a little more experimentation tonight.  ;D

CPU: Intel(R) Pentium(R) M processor 2.26GHz (2266.76-MHz 686-class CPU)
cpu0: ACPI CPU on acpi0
est0: Enhanced SpeedStep Frequency Control on cpu0
est: CPU supports Enhanced Speedstep, but is not recognized. Rats!  ::)
est: cpu_vendor GenuineIntel, msr 612112b0600112b My LCD display is finally showing the CPU speed.

Edit-  However…

Intel(R) Pentium(R) M processor 2.26GHz Current: 1133 MHz, Max: 2266 MHz

;D

[chpalmer]

Interesting-  after several hours of being out I came home to see pfmbmon now reporting-

| | CPU: Intel(R) Pentium(R) M processor 2.26GHz (2266.76-MHz 686-class CPU)
cpu0: ACPI CPU on acpi0
acpi_throttle0: ACPI CPU Throttling on cpu0 |
| dev.cpu.0.freq | 2266 |