There were a couple problems with error handling in the probe function:
1)  If the "drvdata" allocation failed then it lead to a NULL
    dereference.
2)  On several error paths we decremented "nr_cti_cpu" before it was
    incremented which lead to a reference counting bug.

There were also some parts of the error handling which were not bugs but
were messy.  The error handling was confusing to read.  It printed some
unnecessary error messages.

The simplest way to fix these problems was to create a cti_pm_setup()
function that did all the power management setup in one go.  That way
when we call cti_pm_release() we don't have to deal with the
complications of a partially configured power management config.

I reversed the "if (drvdata->ctidev.cpu >= 0)" condition in
cti_pm_release() so that it mirros the new cti_pm_setup() function.

Fixes: 6a0953ce ("coresight: cti: Add CPU idle pm notifer to CTI devices")
Signed-off-by: Dan Carpenter <dan.carpenter@oracle.com>
Reviewed-by: Mike Leach <mike.leach@linaro.org>
Reviewed-by: Mathieu Poirier <mathieu.poirier@linaro.org>
Link: https://lore.kernel.org/r/20200701160852.2782823-2-mathieu.poirier@linaro.orgSigned-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>