Help to get the sensors working for a Gigabyte X670E Aorus Master (AM 5 Board with Ryzen 9) #15
Comments
|
You need to build the it87 module using the code in this repo. See: https://aur.archlinux.org/packages/it87-dkms-git |
|
@f45tb00t sorry about my slow response, I've been tied up with other things. Firstly, As for how to build it. For a simple test you can download this repo and just run "make" and then "sudo make install", which will overwrite the existing kernel module. After that, you may need to reboot for the new module to take effect. However, for ongoing installation, as you upgrade your kernel, you probably want to use the Anyway, see how this goes and let me know if it works for you |
|
Hey @frankcrawford This is the current |
|
@programminghoch10 okay, I can see the issue, I think, the Gigabyte X670E Aorus Master looks to have 2 chips, with the it8689 only providing 5 fans, and the other would provide the other details. However, we need to see what the other chip is, so are you able to provide additional details on what chips are in the system? |
|
@programminghoch10 looking at some other online documentation that board also has an IT8792E (hw ID 0x8733) chip. If you run sensor-detect does it show anything else? Secondly does anything else appear in the boot log for the it87 module? |
|
@frankcrawford Yes it seems there are two ITE chips connected. I don't exactly know what you mean with "boot log", I assume this: |
|
@programminghoch10 how confident are you with git, and with making and installing modules into the kernel? I am thinking of trying a few things in a separate git branch but would need you to compile it and test it, as I don't have that board. |
|
@frankcrawford absolutely no problem. Bring it on! 👍 |
|
@programminghoch10 when you have a chance can you try the dev branch it87922e in the github repo. First checkout the test branch ("git checkout it87922e") and then you will need to compile it ("make"), unload the current one ("modprobe -r it87") and load up the local one ("modprobe hwmon_vid ; insmod ./it87.ko"). If I have it correct, it should detect the second chip as an IT87922E (TBC). Then secondly, see if the values look to be correct. |
|
@frankcrawford still shows only one ITE sensor device
btw thanks for reminding me that kernel modules can be loaded dynamically, I do know me some linux but I totally forgot about that and thought I would have to reboot after |
|
@programminghoch10 well the output from dmesg shows it did detect the other chip, but didn't activate it. It may be an issue that the second chip may need to be configured first, which is a known problem. To correct it, you may need to reboot your system, and ensure that the new module is the one loaded, not the older one. |
|
@frankcrawford |
|
@programminghoch10 okay, I'll have to see what I can find out about why it isn't activating. Thanks. |
|
@programminghoch10 I've just pushed a new version, although all it does is add some more messaging, so I can see what is going on. Also, can you send the output from /sys/class/dmi/id/board_name and /sys/class/dmi/id/board_vendor Thanks. |
|
@frankcrawford i pulled, [14678.472253] it87: it87 driver version v1.0-143-gc51843b.20240213
[14678.472338] it87: Entering config for second chip
[14678.472521] it87: Found IT8689E chip at 0xa40, revision 1
[14678.472656] it87: Beeping is supported
[14678.477131] it87: Device (chip IT87922E (TBC) ioreg 0x4e) not activated, skipping$ cat /sys/class/dmi/id/board_{name,vendor}
X670E AORUS MASTER
Gigabyte Technology Co., Ltd. |
|
Hmm, the values are coming back right, but not activating properly, so I need to look into further. |
|
@programminghoch10 I've done a hack to get past the activation issue. See how it works, but also be aware that this is slighlty dodgy, so be prepared to roll it back. |
|
|
Hi guys, sorry for my late reply... Life... you know... It's awesome that you work on this. I'll test as well. @programminghoch10 Maybe some things to share with you about the board I found out
Beside that I am really happy with the setup. With the Ryzen 7900 (non x version because of 24/7 usage and the power consumption) the pc is a monster :D Also the 6 SATA ports are reliable so far. Running them with a RAID 6 with 6 disks. For the OS and VMs I use two nvme disks |
|
@programminghoch10 Another attempt, based on some comments I've recently been given. |
|
@f45tb00t that is an interesting item about 4X RAM causing a problem with EXPO profiles. I think I've seen that previously on other boards. |
It seems this is a common problem. Unfortunately. Ref: https://www.amd.com/de/product/12746 To be honest I was not aware about that. I think that would have changed my "buying-mind" a little. Because why should I buy then faster RAM for much more money, if they just do not run as expected. I did not try to get the maximum out by manually tweaking, as I need a reliable and stable production system. Furthermore I am not in that tweaking anymore. I don't play games with this pc, so therefore I don't care about 0,00001% more speed. But surely the EXPO was an interesting thing. My last PC was like 10+++ years old, still running, but got very slow over the last years due to the CPU vulns and the applied fixes. So I decided to create a new one without any compromises this time. Once I get the temperatures of the Cores and fix the 5.1 sound issues, I'll be totally happy with this setup. Hope it lasts also for 10+++ years. :) |
|
@programminghoch10 so, talking with someone who has some knowledge of the chips involved, it turns out we have this all wrong. What he said is 0x8883 is IT8883FN which is just a plain eSPI to LPC Bridge, more-less transparent on the bus. Behind it usually lies the above chip. The bug he is describing is that the chip locks up if not accessed correctly, which I thought we did, but may not be done correctly. Anyway, firstly try the version I pushed recently, but expect more changes, to try and get it working properly. |
@f45tb00t thanks for sharing your experience with the mainboard |
|
@programminghoch10 rev 1.0. As far as I can see there is no 1.1 yet? |
|
@programminghoch10 when you do have some time, let me know and I'll get you to try something else more to see the results. |
|
@f45tb00t there is and i have one after quite some issues with rev 1.0 @frankcrawford |
|
@programminghoch10 I am going to keep working on it, but will do a bit of restructuring, which will affect not just you but other chips as well. However, what I would like you to try, when you have a chance is to: The reason for this is I want to test if it is the it87 module that is crashing the second chip or something else. If it is the it87 module, you should no longer see the chip ID 0x8887, but a different chip ID (I think). |
|
Oh, rats, I just checked the sensor-detect module and it may be running the bad code as well. Anyway, we will see. |
|
Just changing my request please do the following: a) blacklist the it87 module (depending on system is something like adding "blacklist it87" into /etc/modprobe.d/blacklist) Thanks. |
|
@programminghoch10 After you try the above tests see how the new code I've just pushed works. |
|
@frankcrawford
then i inserted the module (without rebooting), still the same |
|
@programminghoch10 that result is both good and bad. Firstly, the isadump at offset 20 shows the chip ID and is is 8733, which is a valid IT87 chip, not 8883, which is causing us trouble, so it does look like it is getting broken by the initialisation somehow. So, try the following: Also, make sure to grab the latest git version, I did late last night. Thanks |
|
@f45tb00t I got this mainboard when it was fairly new, around autumn 2022. However mid 2023 a weird issue appeared which I couldn't figure out: So me now being on the new revision I retested all the stuffs. Here are my results mixed with what I found out on the previous board too.
Regarding fan control: I also had some issues installing to NVMe but since I didn't feel like re-installing the whole OS anyways, I've been using my old SATA SSD install for over a year, so I can't say much about that. But I'll definitely switch to NVMe at some point, currently the NVMe SSD is just sitting around and collecting "Power-On Hours" lol. I also did a quick reboot while writing to document my BIOS settings for you (the ones I changed from default setting):
I also completely turn of my I currently have the BIOS version Hope you found it interesting :) |
|
@frankcrawford
I only did the sensors detect until this part showed up, then i left with ctrl-c.
Definitelfy something fishy there. I kept the module blacklisted, then inserted the module after booting: btw I always pull and build the
Wow two chips are present! Nice! |
|
also quick and dirty test, I did a full load stress test on the CPU and I can see 4 fans spinning in the case. so were still losing a fan sensor somewhere, but I guess this will need some more extensive testing |
|
@programminghoch10 that is great, although missing a fan is a bit annoying. However, it is easier to find one, if it is missing than a whole sensor. It is likely that it is a different offset for that fans than I currently have, and if you are sure you have values, it should be easy to find it. I'll send some details a bit later. I am also interested in the temperature and voltages, to make sure we have them correct too, if you can send them over some time. It may be worth looking at the values in the sensor conf files in "Sensor Configs", as it is probably close to the fan values you would see on your board. |
|
@programminghoch10 when you have a chance, we will see if we can find your missing fan. To do this, send through two more dumps: isadump 0xa45 0xa46 Note, this doesn't need a reboot or anything. It should just be looking up the same registers as the driver does. Thank. |
i don't know what you mean by
How am I supposed to know whether the shown values are correct?
And I also don't know which sensor conf files you mean. |
|
@programminghoch10 thanks for those dumps, I'll look at them a bit later. As for the comments, for the voltage and temp, it is just to see if they look reasonable, by comparing what you see in BIOS, or on Windows with things like hwinfo. It is only to see if they look right, not if t hey are 100% accurate, as I don't know how you can do that either. Note, especially for the voltages, the mappings are important, as they scale the voltages depending on what type they are. The temperatures don't normally need any scaling. As for the sensor scalings, in the folder "Sensor configs" as part of the git repo. If you look in any of the con files, they list the chips on that server and the equivalent names of the sensor items. While they don't need to be exactly right, Gigabyte are fairly consistent, so on one board for a certain chip they will tend to match another board too. Note that the names are based on the motherboard, but sensors doesn't care about that, it only cares about the chips listed. Look for the entries like chip "it8792-" or chip "it8689-", but also chip "it8792-isa-0a60" is fine for your board. Something like GA-Z690-AORUS-PRO.conf is pretty comprehensive, but has some slightly different chips as well. Feel free to copy one of those files (or generate a whole new one) for your board and load it in /etc/sensors.d, and it will then be used automatically by sensors. |
|
@programminghoch10 I've tried a small update to the fan measurements, so, please give it a go. Also, what connectors do you have the fans connected to? What sort of fans are they? |
|
@frankcrawford ive tested it, still the same, 4 fans spinning, 3 PWM sensors showing activity. @f45tb00t |
|
@programminghoch10 since this seems to be working correctly and not breaking other chips, I've now rolled this into the mainline. Going forward I probably won't keep updating the it87922e branch unless something dramatic happens. |
|
Hey ho @programminghoch10 @frankcrawford thank you very much for working on this issue, I really appreciate your work! I compiled the kernel by archlinux AUR (dkms is used), rebooted and some fans are found, but not every.
Please do not get irritated by sensors output, for getting the mapping table I used a lose fan on free ports ... Is there any chance to give you more information to find out which address to read to get other pins working as well?
EDIT: I fixed this now by switching the ports on mainboard... |
|
@dr3st Thanks for the report, and even more interesting the comment that the maybe they are using USB for the other fans. However, I am seeing what I can find with the current chipsets. The item about RGB LED Controller is more related to the ability to flash lights on the fan, etc, not directly to the management of the fan itself. There is a package that allows you to manage it, https://github.com/liquidctl/liquidctl which seems to work well. |
Hello Frank,
over the last weeks since I have a new PC built I did quite a lot of research how to get the temperature sensors to work. Just with the stock lm-sensors, nothing is really working.
I also tried to get to know how I could build a driver, but actually I am a little lost what I need to do.
Maybe you can help, as I would like to provide also to others a working solution. As we are kinda blind when it's about CPU temperatures with this new boards.
Some Questions:
With sensors-detect I found this:
Where can I find the the correct drivers for these chips?
How do I build the drivers and make them available as you did?
Thanks a lot for the good work! :)
regards
FB
The text was updated successfully, but these errors were encountered: