The offsets values that you see during the calibration are different from what GCS expects from a Pixhawk\APM board. That happens because the internal compass of MPU9250 has a wider range than HMC5883. It shouldn’t be an issue. Here’s some info about it:
Did you disable the APM before running the Barometer example?
Do you have the barometer covered with an open-cell foam?
Where was you drone located when you were running the GPS example?
