Subject: cell: update latency time of Cell BE processor to the correct value

The original method of measuring the latency time for frequency switches was
inaccurate. This fixes the method, and the measured latency time reported by
the backend cpufreq driver.

Signed-off-by: Christian Krafft <krafft@de.ibm.com>

Index: linux-2.6/arch/powerpc/platforms/cell/cbe_cpufreq.c
===================================================================
--- linux-2.6.orig/arch/powerpc/platforms/cell/cbe_cpufreq.c
+++ linux-2.6/arch/powerpc/platforms/cell/cbe_cpufreq.c
@@ -124,8 +124,8 @@ static int set_pmode(int cpu, unsigned i
 
 #ifdef DEBUG
 	time = (u64) get_cycles() - time; /* actual cycles (not cpu cycles!) */
-	time = 1000000000 * time / CLOCK_TICK_RATE; /* time in ns (10^-9) */
-	pr_debug("had to wait %lu ns for a transition\n", time);
+	time = time / tb_ticks_per_usec; /* time in usec (10^-6) */
+	pr_debug("had to wait %lu usec for a transition\n", time);
 #endif
 
 	rc = 0;
@@ -171,7 +171,7 @@ static int cbe_cpufreq_cpu_init (struct 
 
 	policy->governor = CPUFREQ_DEFAULT_GOVERNOR;
 	/* if DEBUG is enabled set_pmode() measures the correct latency of a transition */
-	policy->cpuinfo.transition_latency = 430000;
+	policy->cpuinfo.transition_latency = 25000;
 
 	cur_pmode = get_pmode(policy->cpu);
 	pr_debug("current pmode is at %d\n",cur_pmode);