cpufreq: fix bad unlock balance on !CONFIG_SMP

This patch tries to fix lockdep complaint attached below.

It seems that we should always read acquire the cpufreq_rwsem,
whether CONFIG_SMP is enabled or not.  And CONFIG_HOTPLUG_CPU
depends on CONFIG_SMP, so it seems we don't need CONFIG_SMP for the
code enabled by CONFIG_HOTPLUG_CPU.

[    0.504191] =====================================
[    0.504627] [ BUG: bad unlock balance detected! ]
[    0.504627] 3.11.0-rc6-next-20130819 #1 Not tainted
[    0.504627] -------------------------------------
[    0.504627] swapper/1 is trying to release lock (cpufreq_rwsem) at:
[    0.504627] [<ffffffff813d927a>] cpufreq_add_dev+0x13a/0x3e0
[    0.504627] but there are no more locks to release!
[    0.504627]
[    0.504627] other info that might help us debug this:
[    0.504627] 1 lock held by swapper/1:
[    0.504627]  #0:  (subsys mutex#4){+.+.+.}, at: [<ffffffff8134a7bf>] subsys_interface_register+0x4f/0xe0
[    0.504627]
[    0.504627] stack backtrace:
[    0.504627] CPU: 0 PID: 1 Comm: swapper Not tainted 3.11.0-rc6-next-20130819 #1
[    0.504627] Hardware name: Bochs Bochs, BIOS Bochs 01/01/2007
[    0.504627]  ffffffff813d927a ffff88007f847c98 ffffffff814c062b ffff88007f847cc8
[    0.504627]  ffffffff81098bce ffff88007f847cf8 ffffffff81aadc30 ffffffff813d927a
[    0.504627]  00000000ffffffff ffff88007f847d68 ffffffff8109d0be 0000000000000006
[    0.504627] Call Trace:
[    0.504627]  [<ffffffff813d927a>] ? cpufreq_add_dev+0x13a/0x3e0
[    0.504627]  [<ffffffff814c062b>] dump_stack+0x19/0x1b
[    0.504627]  [<ffffffff81098bce>] print_unlock_imbalance_bug+0xfe/0x110
[    0.504627]  [<ffffffff813d927a>] ? cpufreq_add_dev+0x13a/0x3e0
[    0.504627]  [<ffffffff8109d0be>] lock_release_non_nested+0x1ee/0x310
[    0.504627]  [<ffffffff81099d0e>] ? mark_held_locks+0xae/0x120
[    0.504627]  [<ffffffff811510cb>] ? kfree+0xcb/0x1d0
[    0.504627]  [<ffffffff813d77ea>] ? cpufreq_policy_free+0x4a/0x60
[    0.504627]  [<ffffffff813d927a>] ? cpufreq_add_dev+0x13a/0x3e0
[    0.504627]  [<ffffffff8109d2a4>] lock_release+0xc4/0x250
[    0.504627]  [<ffffffff8106c9f3>] up_read+0x23/0x40
[    0.504627]  [<ffffffff813d927a>] cpufreq_add_dev+0x13a/0x3e0
[    0.504627]  [<ffffffff8134a809>] subsys_interface_register+0x99/0xe0
[    0.504627]  [<ffffffff81b19f3b>] ? cpufreq_gov_dbs_init+0x12/0x12
[    0.504627]  [<ffffffff813d7f0d>] cpufreq_register_driver+0x9d/0x1d0
[    0.504627]  [<ffffffff81b19f3b>] ? cpufreq_gov_dbs_init+0x12/0x12
[    0.504627]  [<ffffffff81b1a039>] acpi_cpufreq_init+0xfe/0x1f8
[    0.504627]  [<ffffffff810002ba>] do_one_initcall+0xda/0x180
[    0.504627]  [<ffffffff81ae301e>] kernel_init_freeable+0x12c/0x1bb
[    0.504627]  [<ffffffff81ae2841>] ? do_early_param+0x8c/0x8c
[    0.504627]  [<ffffffff814b4dd0>] ? rest_init+0x140/0x140
[    0.504627]  [<ffffffff814b4dde>] kernel_init+0xe/0xf0
[    0.504627]  [<ffffffff814d029a>] ret_from_fork+0x7a/0xb0
[    0.504627]  [<ffffffff814b4dd0>] ? rest_init+0x140/0x140

Signed-off-by: Li Zhong <zhong@linux.vnet.ibm.com>
Acked-and-tested-by: Viresh Kumar <viresh.kumar@linaro.org>
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>

drivers/cpufreq/cpufreq.c

@@ -978,6 +978,7 @@ static int __cpufreq_add_dev(struct device *dev, struct subsys_interface *sif,
 		cpufreq_cpu_put(policy);
 		return 0;
 	}
+#endif
 
 	if (!down_read_trylock(&cpufreq_rwsem))
 		return 0;
@@ -994,7 +995,6 @@ static int __cpufreq_add_dev(struct device *dev, struct subsys_interface *sif,
 		}
 	}
 	read_unlock_irqrestore(&cpufreq_driver_lock, flags);
-#endif
 #endif
 
 	if (frozen)