[Bug 204807] Hardware monitoring sensor nct6798d doesn't work unless acpi_enforce_resources=lax is enabled

[bugzilla-daemon@bugzilla.kernel.org]

https://bugzilla.kernel.org/show_bug.cgi?id=204807

--- Comment #142 from Andy Shevchenko (andy.shevchenko@gmail.com) ---
(In reply to Feliks from comment #140)
> Can someone add my board please? It is an Asus PRIME X570-P, I guess the
> sensor should be, if not exactly the same, as the one from an Asus PRIME
> X570-Pro.
> I cannot use Linux with my board since the CPU fan is spinning at maximum
> speed making an extreme amount of noise, due to that module's sensor
> readings which are wrong.

You have to test yourself before anybody else will add it.

(In reply to Eugene Shalygin from comment #141)
> I would like to ask for an assistance to understand why reading EC sensors
> takes so much time (1 second). Could, please, users of boards with sensors
> published in EC registers (we currently aware of the following models:  Pro
> WS X570-ACE, ROG Crosshair VIII Hero, ROG Crosshair VIII Dark Hero, ROG
> Crosshair VIII Formula, G STRIX B550-E GAMING, ROG STRIX X570-E GAMING)
> measure how long does it take to read all 256 EC registers in their system
> and report back the time and the board name?
> 
> # modprobe ec_sys
> # time cat /sys/kernel/debug/ec/ec0/io > /dev/null

It won't mean anything. The each register read separately may take a long time
since EC is a separate uController that may be interrupted at any time by any
higher priority task (to be sure you have to have a look into the firmware
source code). So, I'll be not surprised if 1s in some cases is not enough. Not
I'm against the shrtening the timeouts, but somebody should really know what
they are about and why.

-- 
You may reply to this email to add a comment.

You are receiving this mail because:
You are watching the assignee of the bug.