perf: add hotplug support

The change is a squash of the following four commits taken
from msm-4.14:

1) commit <b02d7648fbd1> ("enable perf to continue across
hotplug")
Currently perf hardware, software and tracepoint events are
deleted when a cpu is hotplugged out. This change restarts
the events after hotplug. In arm_pmu.c most of the code
for handline power collapse is reused for hotplug.
This change supercedes commit 1f0f95c5fe9e ("perf: add hotplug
support so that perf continues after hotplug") and uses the
new hotplug notification method.

2) commit <9768e7af40d6> ("disable perf_event_read during hotplug")
core.c should not allow perf_event_read access during hotplug.
DCVS may try to read events during hotplug startup or
shutdown. Set a flag to not allow access during hotplug.

3) commit <1fc690b7b8c6> ("perf: core: Avoid race condition when
releasing perf-events")
Before calling perf_event_release_kernel(), capture
the perf's pmus_mutex lock to prevent the CPU from going offline
during the operation.

4) commit <77257e46efea> ("perf: Manage CPU hotplug events at
core level")
the approach here is to achieve the hotplug management
at the perf-core level. The idea is to detach the event from the
context (perf_remove_from_context()), when the CPU is about to come
down and re-attach it back (perf_install_in_context()), when the CPU
comes back online. The approach involves removing the logic for
maintaining zombie events (PERF_EVENT_STATE_ZOMBIE) and let the
dormant list itself carry the events whose CPUs are offline.

Change-Id: I3738bdcaae2d2199ba3fb68ce77f637d867c5c14
Signed-off-by: Raghavendra Rao Ananta <rananta@codeaurora.org>
Signed-off-by: Rishabh Bhatnagar <rishabhb@codeaurora.org>