unable to set speeds git version | octo #671
Comments
nyllewin
changed the title
|
Can you try to bypass liquidctl and manually set a value in hwmon? Locate Quadro in echo 255 > pwm1(pwm1 is fan1, etc. 255 PWM = 100%) Then reboot, and try the same thing again with a different value, and report back if it works. This is what liquidctl does internally now, but I'd like to see what happens at the source.
That's normal, it detects a hwmon driver and warns that you're bypassing it. Evidently, that's necessary sometimes. |
|
Also, is anything else configuring the device? Are you maybe booting into windows with aquasuite after poweroff or similar? |
|
I got in to this and this looks familiar to the liquidctl output. the last step worked. I have no Idea why it isn't working via sudo.
Rebooted now, same behavior. none of these work until after I talked to that specific fan via direct-access flag then and only then the pwm is accepted.
I tried installing the rules for not needing sudo but after installing them and reloading I still needed to authorize, not sure if the way I did it was wrong but the reloads didn't trigger any errors when executed with sudo. I went in to the directory for user rules, sudo nano filename 71xxxx then copied via ctrl c from github raw the text and pasted, reloaded. So after that didn't work i renamed them to 71xx.rules.backup in both directories where they were present. the distro one and the user one.
Controlling not so much. There is coolercontrol where I am also not able to set speeds with on either device, the hwmon one and the liquidctl one so I wouldn't call that " controlling " .. its reading the sensors and so on but that's about it. I can uninstall. I don't have a windows install anymore so that's out. I wish I could have brought better news but yea. let me know, I'm down for anything basically :D |
|
Yeah, you'll need
That seems weird, looks like something else is controlling it? Can you please try the github version of the driver (instructions)? Then go to hwmon again, and do: cat pwm1_enableAnd see what the value is. Should be
So maybe it's something else before a "direct access" setting. If it is, set it to I am not familiar with coolercontrol as I don't use it, but let's try to rule it out, so kill the process or something. And yes, Octo or Quadro should make no difference. |
|
I stopped the coolercontrol daemon service did this after:
this succeeded I think? I did this, off / on cycle and did this again. I'm sorry I'm really new lmao checked for coolercontrol service or if its working doing anything and we got nothing, good! Now I did this: So I went back and did make dev in the d5 next driver again and then they showed up. Also assuming that's normal? because i reboot and then it loads the old kernel version and so on.. surely. anyways. this is from fresh boot, device running at fallback speeds as far as I can tell because this stupid 80mm fan is driving me nuts with its noise and some others also just run rather high rpm. Ok this kicked the fan from 25% (fallback speed) to 100%
is actually 1. sorry for not using pwm1 as you told me but its the pump and I'd like to keep it running so I'm using another channel :D |
Yeah, on boot it loads the built-in one.
That was just an example, use whichever channel you want. I just got tired of typing "pwmX, where X is ...". pwm_enable sheds some light and proves what I thought: So it's using PID mode, and the driver in the kernel is (right now) only setting the PWM value and not setting the internal PWM controller value to the equivalent of That's a fix for the kernel, then, as I suspected. Thanks for going through this with me, much appreciated. |
|
Thank you! Can I somehow just .. run the github version of the kernel driver for the time being until things are sorted or would it be more wise to just to default to direct access for now? scrap that I just realized that its the same problem because we did this with the git version, right. ok Nevermind! It would honestly just be fine to default to direct access I think, I dont use much.. maybe I can make something up with delta from the sensors and set speeds automatically accordingly with direct-access flag once every 30 seconds or so. |
|
Yes, you can run the github version just fine. Or just use direct access, it doesn't matter to the device. The outcome is the same.
Nope. I told you to try the github version so we get access to pwm_enable to read it out; if you use it, liquidctl (and everything else) will work just fine because it knows how to set both pwmX_enable and pwmX. (And the driver kind of forces you to do the same, when done manually.) So long story short, either use the github driver until a future kernel update arrives or use direct access. |
|
@aleksamagicka, should we maybe revert1 #670? From what I understood of the above discussion, it seems that Footnotes
|
|
I've just sent a fix for the in-kernel driver to LKML, so that should end up in stable trees. That fixes it up so that just setting pwmX does the work of setting both pwmX_enable and pwmX values, since the in-kernel one only has pwmX. The github one has both, which makes it obvious that you have to set them both to see an effect.
As for adding a comment, I'm all for it. Drivers develop over time and differences should be noted. |
|
Looks like I commented a minute too soon. Guenter rejected the patch, so that leaves us with indeed reverting #670. I'll see if I can create pwm_enable patches for the in-tree driver tomorrow. |
Describe the bug
Hey, I switched to the git version due to recent pwm issue and fix and just noticed now that I'm unable to set speeds for fans after a power off and power on.
Things were fine with the version that came via pacman install if not counting for the pwm issue.
Should I install a version before the pwm fix (2 days ago) to see if it's also the case before that? (how do I do that :D)
edit: It works with direct-access.
But I get this warning:
WARNING: directly writing fixed speed despite aquacomputer_d5next kernel driver having supportAlso it's requiring sudo now, is that normal?
Thank you!
Commands executed
sudo liquidctl set fan4 speed 0
sudo liquidctl set fan4 speed 100
sudo liquidctl set fan4 speed 0 --direct-access
Output of all relevant commands with
--debugflagDirect Access debug output:
uname -r
6.7.4-arch1-1
non direct access output:
sudo liquidctl status
sudo liquidctl set fan2 speed 0 --debug
sudo liquidctl status
Affected device
aquacomputer octo
Does your version of liquidctl support the device in question?
Yes, my version supports it
Operating system and version
arch linux
Installation method
aur
Version of liquidctl
liquidctl v1.14.0.dev37+g14dd33a (Linux-6.7.4-arch1-1-x86_64-with-glibc2.39)
The text was updated successfully, but these errors were encountered: