sparc64: Fix pcr_ops initialization and usage bugs.
authorDavid S. Miller <davem@davemloft.net>
Mon, 11 Aug 2014 22:38:46 +0000 (15:38 -0700)
committerDavid S. Miller <davem@davemloft.net>
Tue, 12 Aug 2014 03:45:12 +0000 (20:45 -0700)
commit8bccf5b313180faefce38e0d1140f76e0f327d28
treeed8d8f8c5696526f1ecf6fccd94680fe166427f4
parent58556104e9cd0107a7a8d2692cf04ef31669f6e4
sparc64: Fix pcr_ops initialization and usage bugs.

Christopher reports that perf_event_print_debug() can crash in uniprocessor
builds.  The crash is due to pcr_ops being NULL.

This happens because pcr_arch_init() is only invoked by smp_cpus_done() which
only executes in SMP builds.

init_hw_perf_events() is closely intertwined with pcr_ops being setup properly,
therefore:

1) Call pcr_arch_init() early on from init_hw_perf_events(), instead of
   from smp_cpus_done().

2) Do not hook up a PMU type if pcr_ops is NULL after pcr_arch_init().

3) Move init_hw_perf_events to a later initcall so that it we will be
   sure to invoke pcr_arch_init() after all cpus are brought up.

Finally, guard the one naked sequence of pcr_ops dereferences in
__global_pmu_self() with an appropriate NULL check.

Reported-by: Christopher Alexander Tobias Schulze <cat.schulze@alice-dsl.net>
Signed-off-by: David S. Miller <davem@davemloft.net>
arch/sparc/kernel/perf_event.c
arch/sparc/kernel/process_64.c
arch/sparc/kernel/smp_64.c