[Bug 218685] asus-nb-wmi fails to load due to conflict with amd-pmf

[bugzilla-daemon@kernel.org]

https://bugzilla.kernel.org/show_bug.cgi?id=218685

--- Comment #13 from Mario Limonciello (AMD) (mario.limonciello@amd.com) ---
OK, from that log and your acpidump it looks to me that it supports these PMF
functions:

APMF_FUNC_GET_SYS_PARAMS
APMF_FUNC_SBIOS_REQUESTS
APMF_FUNC_OS_POWER_SLIDER_UPDATE


The one that is specifically causing the conflict the ASUS driver is
APMF_FUNC_OS_POWER_SLIDER_UPDATE.  This is what causes the ACPI power profile
to get registered:

[   71.242252] amd-pmf AMDI0102:00: SPS enabled and Platform Profiles
registered

When the profile is moved it calls APMF() with an arg of 8, which calls PMF8():

                    Case (0x08)
                    {
                        PMF8 (Arg1)
                    }

PMF8 outputs some debug using M460 but eventually calls APX8():

            Method (PMF8, 1, Serialized)
            {
                M460 ("FEA-ASL-\\_SB.PMF.APMF Function 8 call PMF8\n", Zero,
Zero, Zero, Zero, Zero, Zero)
                CreateByteField (Arg0, 0x02, M490)
                M460 ("  Slider Event Notification: 0x%x\n", M490, Zero, Zero,
Zero, Zero, Zero)
                If (CondRefOf (\_SB.APX8))
                {
                    M460 ("  Call OEM ACPI APX8\n", Zero, Zero, Zero, Zero,
Zero, Zero)
                    \_SB.APX8 (M490)
                }
            }

This updates a "PMTP" field:

        Method (APX8, 1, Serialized)
        {
            PMTP = (0xD6080000 | Arg0)
        }

That PTMP field is part of an Operation Region used for Port I/O:

        OperationRegion (TP80, SystemIO, 0x80, 0x04)
        Field (TP80, DWordAcc, NoLock, Preserve)
        {
            PMTP,   32
        }

To me this looks like the system should probably be responding to power slider
events from the amd-pmf driver.  Do you find that various CPU coefficients
don't get updated even when you've changed the power profile when amd-pmf is
bound?

If they do get updated, then I think the bug should be in asus-wmi.c that it
shouldn't be loading a platform profile because amd-pmf is taking this role. 
Perhaps the DSTS method has an unknown bit to indicate this?

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

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