Porting this solution to Linux

[LightTreasure]

Hi, you are one of the few people who know this space well enough to have solved this issue - to the point that it was merged into LibreHardwareMonitor.

I wanted to ask if you'd be interested in making a similar solution for Linux? Check out this discussion talking about the same issue on Linux: frankcrawford/it87#11

Naively, it seems to me that the core idea of your solution (overriding the Gigabyte custom fan control code on the 8792 chip so that it goes back to listening to standard fan control commands) would be OS-independent. The difference in approach would be to use the appropriate OS-specific API to access the 8792 chip's memory.

Of course, that is just based on my naive understanding of how this all works. What do you think?

[javs]

Hey! Yeah, it's been something in my list for a while. I've certainly read frank's code and issue #11 in there, but I do appreciate the ping. I've recently migrated a system with a GB motherboard to linux, so I may try to look at this sometime soon.

Regarding the technical details, I think hardware wise you are correct, it's definitely doable. There is nothing we do in LHM that would be unique to windows, it would only change the way we do it. The question is how to do it.

The method used in LHM touches the PCI and LPC buses in order to access the chip's ram. One can probably just raw write to the PCI related registers within it87 and it would work as a proof of concept, but in linux this may not be ideal since there are other kernel mods that likely handle that. Plus it's still an AMD specific solution, and no one has written the Intel side.

I wanted to take a look at the ACPI tables to see if there is any chance GB has defined a WMI object for this. This would certainly make things easier, as we could just use https://github.com/torvalds/linux/blob/master/drivers/platform/x86/gigabyte-wmi.c, but last I checked some years ago I couldn't find anything (and also GB's own software doesn't have any indication there is something).

[LightTreasure]

Thanks for replying! I wonder if I could help with this - I have some systems programming experience and I'd love to dig in deeper. Is the code in this repo doing similar things as the LHM implementation?

Also if you'd like to close this issue, please feel free to do so!

[javs]

Hey, sure! whatever you can do.

Some info:

1. Recap comment Problem with ITE IT8792E/IT8795E - values are ignored LibreHardwareMonitor/LibreHardwareMonitor#251 (comment)
   • The code is here: https://github.com/LibreHardwareMonitor/LibreHardwareMonitor/blob/master/LibreHardwareMonitorLib/Hardware/Motherboard/Lpc/IsaBridgeGigabyteController.cs
   • The detection portion is here https://github.com/LibreHardwareMonitor/LibreHardwareMonitor/blob/master/LibreHardwareMonitorLib/Hardware/Motherboard/Lpc/LpcIO.cs#L647
2. Later in the same thread, someone found a new IO port in their newer AMD motherboards
   • That simplifies the problem a lot since you don't need to do this via ISA+LPC (and thus works for both Intel and AMD presumably).
   • However, it's not clear if this works on older motherboards too (maybe a bios update?). They said "ECIO is only available on AMD motherboards with IT8791E/IT8792E/IT8795E.", but it's not clear if that is still true.
   • I was just rereading LHM and it seems they have implemented it here: https://github.com/LibreHardwareMonitor/LibreHardwareMonitor/blob/master/LibreHardwareMonitorLib/Hardware/Motherboard/Lpc/EcioPortGigabyteController.cs

Most of this was found by observing what GB services do on windows.

My discord is in my github profile if there is something I can help with.