Re: [PATCH 0/2] power: supply: Fix AXP288 fallback when not needed

From: Hans de Goede
Date: Thu Feb 15 2018 - 11:15:45 EST


Hi Carlo,

On 14-02-18 20:41, Carlo Caione wrote:
From: Carlo Caione <carlo@xxxxxxxxxxxx>

With commits af3ec837 and dccfae6d a blacklist was introduced to avoid
using the ACPI drivers for AC and battery when a native PMIC driver was
already present. While this is in general a good idea (because of broken
DSDT or proprietary and undocumented ACPI opregions for the ACPI
AC/battery devices) we have come across at least one CherryTrail laptop
(ECS EF20EA) shipping the AXP288 together with a separate FG controller
(a MAX17047) instead of the one embedded in the AXP288.

Thank you for the patches, the code looks good, but I think you need
to split them up a bit differently, the patches under drivers/acpi
need to be merged into a different subsys tree as those under
drivers/power/supply, so the "power: supply: ACPI/AXP288: Add quirks
for ECS EF20EA" needs to be split I believe.

Also the patches to files under drivers/acpi should have an
"ACPI: " prefix for their subject AFAIK.

Regards,

Hans




This is the interesting analisys done by Hans de Goede (thank you):

Looking at the _BIX method of the BATC/PNP0C0A device, we see it referencing
FG10:

Method (_BIX, 0, NotSerialized) // _BIX: Battery Information Extend
{
If (AVBL == One)
{
BUF2 = FG10 /* \_SB_.PCI0.I2C1.FG10 */

And FG10 is defined as:

Field (DVFG, BufferAcc, NoLock, Preserve)
{
Connection (SMFG),
Offset (0x10),
AccessAs (BufferAcc, AttribBytes (0x02)),
FG10, 8
}

With SMFG being defined as:

Name (SMFG, ResourceTemplate ()
{
I2cSerialBusV2 (0x0036, ControllerInitiated, 0x000186A0,
AddressingMode7Bit, "\\_SB.PCI0.I2C1",
0x00, ResourceConsumer, , Exclusive,
)
})

Looking for I2C1 address 0x0036 we find:

Device (ANFG)
{
Name (_HID, "MAX17047" /* Fuel Gauge Controller */) // _HID: Hardwa
Name (_CID, "MAX17047" /* Fuel Gauge Controller */) // _CID: Compat
Name (_DDN, "Fuel Gauge Controller") // _DDN: DOS Device Name
Name (RBUF, ResourceTemplate ()
{
I2cSerialBusV2 (0x0036, ControllerInitiated, 0x00061A80,
AddressingMode7Bit, "\\_SB.PCI0.I2C1",
0x00, ResourceConsumer, , Exclusive,
)
GpioInt (Level, ActiveLow, ExclusiveAndWake, PullNone, 0x0000,
"\\_SB.GPO3", 0x00, ResourceConsumer, ,
)
{ // Pin list
0x0001
}
})

Where as the AXP288 PMIC is I2C7 address 0x034:

Device (PMI1)
{
Name (_ADR, Zero) // _ADR: Address
Name (_HID, "INT33F4" /* XPOWER PMIC Controller */) // _HID: Ha
Name (_CID, "INT33F4" /* XPOWER PMIC Controller */) // _CID: Co
Name (_DDN, "XPOWER PMIC Controller") // _DDN: DOS Device Name
Name (_HRV, 0x03) // _HRV: Hardware Revision
Name (_UID, One) // _UID: Unique ID

Method (_CRS, 0, NotSerialized) // _CRS: Current Resource Setti
{
Name (SBUF, ResourceTemplate ()
{
I2cSerialBusV2 (0x0034, ControllerInitiated, 0x000F4240,
AddressingMode7Bit, "\\_SB.PCI0.I2C7",
0x00, ResourceConsumer, , Exclusive,
)

So basically this laptopt is using a separate FG chip instead of the one
embedded in the AXP288.

To have this correctly working we need basically to avoid the fallback on the
AXP288 driver enabling again the ACPI AC/battery drivers and at the same time
avoiding that the AXP288 FG driver is probed at all.

I'm still not fully convinced that having two different quirks (one to disable
the blacklist and another to disable the AXP288 FG probing) is the right way to
fix this. So any comment is welcome.

Carlo Caione (2):
power: supply: ACPI/AXP288: Add quirk to avoid using PMIC
power: supply: ACPI/AXP288: Add quirks for ECS EF20EA

drivers/acpi/ac.c | 33 ++++++++++++++++++++++++--------
drivers/acpi/battery.c | 33 ++++++++++++++++++++++++--------
drivers/power/supply/axp288_fuel_gauge.c | 6 ++++++
3 files changed, 56 insertions(+), 16 deletions(-)