Hi Frank, > I was able to sneak some time in today between builds and look at this > driver some more. I was able to get your driver to work with a simple > modification. Let me describe my setup first. Kernel is 2.6.9 except > when I started having problems with this driver I brought adm1031.c from > 2.6.10 and put the normal_i2c stuff back in. It appears that in > adm1031_attach_adapter() adapter->class is zero and HWMON is 1 so I get > the return 0 instead of calling i2c_detect. I am using an adm1030 part. > When I remove the adapter->class test then the driver is up and running > fine. adapter->class is provided by the driver of the bus your ADM1030 chip sits on. If it is zero this means that the driver did not properly declare its class. This is a problem with the bus driver, not the adm1031 driver. So both your claim that the adm1031 driver was broken and your fix are incorrect. > I see some other drivers have removed this sort of test but I have > not looked into the consequences of doing this. Which drivers please? We are rather currently *adding* the checks, not removing them. > Any ideas or clues on what I might need to fix to get the class set > properly? Which bus driver are you using? It obviously lacks class declaration, so the correct fix is to add the class there. Thanks, -- Jean Delvare
