flightlessmango
Wrong path for some hwmon sensors
One of my friend(has amdgpu) had that issue MangoHud showed 0 gpu power usage however days ago it worked. From just static review I think this function https://github.com/flightlessmango/MangoHud/blob/41f923b0b2be07f035df221d01625e11f1305fde/src/overlay.cpp#L735 needs to be reviewed by somebody related to how it uses it's local variables. And also when to break or continue from the loops.( I've nothing to do with hwmon interfaces nor mangohud, it's a pure guess...)
So I think after this commit https://github.com/flightlessmango/MangoHud/commit/dd5795146258704bec66472e24d9fb0f2edee633 dirs only gets new value in scope of the if statement, but I assume it should have the same value here https://github.com/flightlessmango/MangoHud/blob/41f923b0b2be07f035df221d01625e11f1305fde/src/overlay.cpp#L906C33-L906C37 too(as how it was before). Don't know why shadowing needed here instead different name anyway.
Some other what maybe could be a problem too at some places/in the future but not related to this issue(not so important as I said I have nothing to do with these things nor can test this dynamically): Here https://github.com/flightlessmango/MangoHud/blob/41f923b0b2be07f035df221d01625e11f1305fde/src/overlay.cpp#L896 and https://github.com/flightlessmango/MangoHud/blob/41f923b0b2be07f035df221d01625e11f1305fde/src/overlay.cpp#L902 Are you sure you use these statements right...?
- Like fore sure you read all metrics data(what possibly opened after line 896) from gpu_metrics, and do not read ones with metrics what opened before(like near line 882)?
- Also when continue at line 902 it will try to open for ex.: a hwinfo file for a cardY at next iteration. And I think this can be a problem because at current iteration we possibly already opened some hwinfo interfaces from cardX. => amdgpu_files structure will contain file pointers related to different cards??
Not really a github user so just writing like this https://github.com/flightlessmango/MangoHud/issues/1099 probably the same
#1098 does not resolve this for me, I still need to revert that commit.
no, I meant https://github.com/flightlessmango/MangoHud/pull/1098 is CPU related which is unrelated to our GPU isssues... (I have GPU issue too)
For me it's GPU ;) Think that is the difference in our issues.
One of my friend(has amdgpu) had that issue MangoHud showed 0 gpu power usage however days ago it worked.
Does this https://github.com/flightlessmango/MangoHud/commit/c33ce2b734856934f53721eeffe421b1da0fb935 commit fix the issue for your friend? Apparently it's related to the changes in kernel, not mangohud
