[PATCH linux dev-4.10 0/6] Hardware work-arounds for max31785

Andrew Jeffery andrew at aj.id.au
Sat Jul 29 02:22:43 AEST 2017


Two issues have reared up regarding the recently applied MAX31785 driver:

The first issue, regarding unexpected NACKs from the device, was discovered in
earlier testing but its impact wasn't clear and its cause was not nailed down
at the time. The thought was it could be solved in slow-time, but this turned
out not to be the case. A work-around is presented in the last patch.

The second issue stems from a difference in hardware behaviour between
Witherspoon hardware levels. Integration testing machines do not have the
"installed" bit programmed into the max31785 chip out of the box, whilst it
appears that functional testing level machines do. The consequence is the
driver emitted some warnings and exposed none of the fans to userspace on ITC
level machines. A new devicetree property is added to make the driver configure
the installed bits in the max31785 chip during initialisation.

Please review.


Andrew Jeffery (6):
  dt-bindings: max31785: Add mark-fan-installed property
  ARM: dts: aspeed: Configure Firestone to mark its fans as installed
  ARM: dts: aspeed: Configure Romulus to mark its fans as installed
  ARM: dts: aspeed: Configure Witherspoon to mark its fans as installed
  max31785: Support mark-fan-installed devicetree property
  pmbus: max31785: Work around back-to-back writes with FAN_CONFIG_1_2

 .../devicetree/bindings/hwmon/pmbus/max31785.txt   |   2 +
 arch/arm/boot/dts/aspeed-bmc-opp-firestone.dts     |   6 ++
 arch/arm/boot/dts/aspeed-bmc-opp-romulus.dts       |   6 ++
 arch/arm/boot/dts/aspeed-bmc-opp-witherspoon.dts   |   4 +
 drivers/hwmon/pmbus/max31785.c                     | 116 +++++++++++++++++++--
 5 files changed, 123 insertions(+), 11 deletions(-)


More information about the openbmc mailing list