2008-07-01 18:44:18

by Alok Kataria

[permalink] [raw]
Subject: [PATCH 03/05] x86: Merge cpu-freq

[VMware, Inc. is providing the following patch to you under the terms of the
GPL version 2 and no later version. This patch is provided as is, with no
warranties or support. VMware disclaims all liability in connection with
the use/inability to use this patch. Any use of the attached is considered
acceptance of the above.]

From: Alok N Kataria <[email protected]>

Unify the cpufreq code.

Signed-off-by: Alok N Kataria <[email protected]>
Signed-off-by: Dan Hecht <[email protected]>

---

arch/x86/kernel/tsc.c | 114 ++++++++++++++++++++++++++++++++++++++++++++++
arch/x86/kernel/tsc_32.c | 113 ----------------------------------------------
arch/x86/kernel/tsc_64.c | 114 ----------------------------------------------
3 files changed, 114 insertions(+), 227 deletions(-)


diff --git a/arch/x86/kernel/tsc.c b/arch/x86/kernel/tsc.c
index e6ee145..595f78a 100644
--- a/arch/x86/kernel/tsc.c
+++ b/arch/x86/kernel/tsc.c
@@ -4,6 +4,7 @@
#include <linux/module.h>
#include <linux/timer.h>
#include <linux/acpi_pmtmr.h>
+#include <linux/cpufreq.h>

#include <asm/hpet.h>

@@ -215,3 +216,116 @@ int recalibrate_cpu_khz(void)
EXPORT_SYMBOL(recalibrate_cpu_khz);

#endif /* CONFIG_X86_32 */
+
+/* Accelerators for sched_clock()
+ * convert from cycles(64bits) => nanoseconds (64bits)
+ * basic equation:
+ * ns = cycles / (freq / ns_per_sec)
+ * ns = cycles * (ns_per_sec / freq)
+ * ns = cycles * (10^9 / (cpu_khz * 10^3))
+ * ns = cycles * (10^6 / cpu_khz)
+ *
+ * Then we use scaling math (suggested by [email protected]) to get:
+ * ns = cycles * (10^6 * SC / cpu_khz) / SC
+ * ns = cycles * cyc2ns_scale / SC
+ *
+ * And since SC is a constant power of two, we can convert the div
+ * into a shift.
+ *
+ * We can use khz divisor instead of mhz to keep a better precision, since
+ * cyc2ns_scale is limited to 10^6 * 2^10, which fits in 32 bits.
+ * ([email protected])
+ *
+ * [email protected] "math is hard, lets go shopping!"
+ */
+
+DEFINE_PER_CPU(unsigned long, cyc2ns);
+
+void set_cyc2ns_scale(unsigned long cpu_khz, int cpu)
+{
+ unsigned long long tsc_now, ns_now;
+ unsigned long flags, *scale;
+
+ local_irq_save(flags);
+ sched_clock_idle_sleep_event();
+
+ scale = &per_cpu(cyc2ns, cpu);
+
+ rdtscll(tsc_now);
+ ns_now = __cycles_2_ns(tsc_now);
+
+ if (cpu_khz)
+ *scale = (NSEC_PER_MSEC << CYC2NS_SCALE_FACTOR)/cpu_khz;
+
+ sched_clock_idle_wakeup_event(0);
+ local_irq_restore(flags);
+}
+
+#ifdef CONFIG_CPU_FREQ
+
+/* Frequency scaling support. Adjust the TSC based timer when the cpu frequency
+ * changes.
+ *
+ * RED-PEN: On SMP we assume all CPUs run with the same frequency. It's
+ * not that important because current Opteron setups do not support
+ * scaling on SMP anyroads.
+ *
+ * Should fix up last_tsc too. Currently gettimeofday in the
+ * first tick after the change will be slightly wrong.
+ */
+
+static unsigned int ref_freq;
+static unsigned long loops_per_jiffy_ref;
+static unsigned long tsc_khz_ref;
+
+static int time_cpufreq_notifier(struct notifier_block *nb, unsigned long val,
+ void *data)
+{
+ struct cpufreq_freqs *freq = data;
+ unsigned long *lpj, dummy;
+
+ if (cpu_has(&cpu_data(freq->cpu), X86_FEATURE_CONSTANT_TSC))
+ return 0;
+
+ lpj = &dummy;
+ if (!(freq->flags & CPUFREQ_CONST_LOOPS))
+#ifdef CONFIG_SMP
+ lpj = &cpu_data(freq->cpu).loops_per_jiffy;
+#else
+ lpj = &boot_cpu_data.loops_per_jiffy;
+#endif
+
+ if (!ref_freq) {
+ ref_freq = freq->old;
+ loops_per_jiffy_ref = *lpj;
+ tsc_khz_ref = tsc_khz;
+ }
+ if ((val == CPUFREQ_PRECHANGE && freq->old < freq->new) ||
+ (val == CPUFREQ_POSTCHANGE && freq->old > freq->new) ||
+ (val == CPUFREQ_RESUMECHANGE)) {
+ *lpj = cpufreq_scale(loops_per_jiffy_ref, ref_freq, freq->new);
+
+ tsc_khz = cpufreq_scale(tsc_khz_ref, ref_freq, freq->new);
+ if (!(freq->flags & CPUFREQ_CONST_LOOPS))
+ mark_tsc_unstable("cpufreq changes");
+ }
+
+ set_cyc2ns_scale(tsc_khz_ref, freq->cpu);
+
+ return 0;
+}
+
+static struct notifier_block time_cpufreq_notifier_block = {
+ .notifier_call = time_cpufreq_notifier
+};
+
+static int __init cpufreq_tsc(void)
+{
+ cpufreq_register_notifier(&time_cpufreq_notifier_block,
+ CPUFREQ_TRANSITION_NOTIFIER);
+ return 0;
+}
+
+core_initcall(cpufreq_tsc);
+
+#endif /* CONFIG_CPU_FREQ */
diff --git a/arch/x86/kernel/tsc_32.c b/arch/x86/kernel/tsc_32.c
index 40c0aaf..bbc153d 100644
--- a/arch/x86/kernel/tsc_32.c
+++ b/arch/x86/kernel/tsc_32.c
@@ -18,119 +18,6 @@
extern int tsc_unstable;
extern int tsc_disabled;

-/* Accelerators for sched_clock()
- * convert from cycles(64bits) => nanoseconds (64bits)
- * basic equation:
- * ns = cycles / (freq / ns_per_sec)
- * ns = cycles * (ns_per_sec / freq)
- * ns = cycles * (10^9 / (cpu_khz * 10^3))
- * ns = cycles * (10^6 / cpu_khz)
- *
- * Then we use scaling math (suggested by [email protected]) to get:
- * ns = cycles * (10^6 * SC / cpu_khz) / SC
- * ns = cycles * cyc2ns_scale / SC
- *
- * And since SC is a constant power of two, we can convert the div
- * into a shift.
- *
- * We can use khz divisor instead of mhz to keep a better precision, since
- * cyc2ns_scale is limited to 10^6 * 2^10, which fits in 32 bits.
- * ([email protected])
- *
- * [email protected] "math is hard, lets go shopping!"
- */
-
-DEFINE_PER_CPU(unsigned long, cyc2ns);
-
-void set_cyc2ns_scale(unsigned long cpu_khz, int cpu)
-{
- unsigned long long tsc_now, ns_now;
- unsigned long flags, *scale;
-
- local_irq_save(flags);
- sched_clock_idle_sleep_event();
-
- scale = &per_cpu(cyc2ns, cpu);
-
- rdtscll(tsc_now);
- ns_now = __cycles_2_ns(tsc_now);
-
- if (cpu_khz)
- *scale = (NSEC_PER_MSEC << CYC2NS_SCALE_FACTOR)/cpu_khz;
-
- /*
- * Start smoothly with the new frequency:
- */
- sched_clock_idle_wakeup_event(0);
- local_irq_restore(flags);
-}
-
-#ifdef CONFIG_CPU_FREQ
-
-/*
- * if the CPU frequency is scaled, TSC-based delays will need a different
- * loops_per_jiffy value to function properly.
- */
-static unsigned int ref_freq;
-static unsigned long loops_per_jiffy_ref;
-static unsigned long cpu_khz_ref;
-
-static int
-time_cpufreq_notifier(struct notifier_block *nb, unsigned long val, void *data)
-{
- struct cpufreq_freqs *freq = data;
-
- if (!ref_freq) {
- if (!freq->old){
- ref_freq = freq->new;
- return 0;
- }
- ref_freq = freq->old;
- loops_per_jiffy_ref = cpu_data(freq->cpu).loops_per_jiffy;
- cpu_khz_ref = cpu_khz;
- }
-
- if ((val == CPUFREQ_PRECHANGE && freq->old < freq->new) ||
- (val == CPUFREQ_POSTCHANGE && freq->old > freq->new) ||
- (val == CPUFREQ_RESUMECHANGE)) {
- if (!(freq->flags & CPUFREQ_CONST_LOOPS))
- cpu_data(freq->cpu).loops_per_jiffy =
- cpufreq_scale(loops_per_jiffy_ref,
- ref_freq, freq->new);
-
- if (cpu_khz) {
-
- if (num_online_cpus() == 1)
- cpu_khz = cpufreq_scale(cpu_khz_ref,
- ref_freq, freq->new);
- if (!(freq->flags & CPUFREQ_CONST_LOOPS)) {
- tsc_khz = cpu_khz;
- set_cyc2ns_scale(cpu_khz, freq->cpu);
- /*
- * TSC based sched_clock turns
- * to junk w/ cpufreq
- */
- mark_tsc_unstable("cpufreq changes");
- }
- }
- }
-
- return 0;
-}
-
-static struct notifier_block time_cpufreq_notifier_block = {
- .notifier_call = time_cpufreq_notifier
-};
-
-static int __init cpufreq_tsc(void)
-{
- return cpufreq_register_notifier(&time_cpufreq_notifier_block,
- CPUFREQ_TRANSITION_NOTIFIER);
-}
-core_initcall(cpufreq_tsc);
-
-#endif
-
/* clock source code */

static struct clocksource clocksource_tsc;
diff --git a/arch/x86/kernel/tsc_64.c b/arch/x86/kernel/tsc_64.c
index c852ff9..80a274b 100644
--- a/arch/x86/kernel/tsc_64.c
+++ b/arch/x86/kernel/tsc_64.c
@@ -16,120 +16,6 @@
extern int tsc_unstable;
extern int tsc_disabled;

-/* Accelerators for sched_clock()
- * convert from cycles(64bits) => nanoseconds (64bits)
- * basic equation:
- * ns = cycles / (freq / ns_per_sec)
- * ns = cycles * (ns_per_sec / freq)
- * ns = cycles * (10^9 / (cpu_khz * 10^3))
- * ns = cycles * (10^6 / cpu_khz)
- *
- * Then we use scaling math (suggested by [email protected]) to get:
- * ns = cycles * (10^6 * SC / cpu_khz) / SC
- * ns = cycles * cyc2ns_scale / SC
- *
- * And since SC is a constant power of two, we can convert the div
- * into a shift.
- *
- * We can use khz divisor instead of mhz to keep a better precision, since
- * cyc2ns_scale is limited to 10^6 * 2^10, which fits in 32 bits.
- * ([email protected])
- *
- * [email protected] "math is hard, lets go shopping!"
- */
-
-DEFINE_PER_CPU(unsigned long, cyc2ns);
-
-void set_cyc2ns_scale(unsigned long cpu_khz, int cpu)
-{
- unsigned long long tsc_now, ns_now;
- unsigned long flags, *scale;
-
- local_irq_save(flags);
- sched_clock_idle_sleep_event();
-
- scale = &per_cpu(cyc2ns, cpu);
-
- rdtscll(tsc_now);
- ns_now = __cycles_2_ns(tsc_now);
-
- if (cpu_khz)
- *scale = (NSEC_PER_MSEC << CYC2NS_SCALE_FACTOR)/cpu_khz;
-
- sched_clock_idle_wakeup_event(0);
- local_irq_restore(flags);
-}
-
-#ifdef CONFIG_CPU_FREQ
-
-/* Frequency scaling support. Adjust the TSC based timer when the cpu frequency
- * changes.
- *
- * RED-PEN: On SMP we assume all CPUs run with the same frequency. It's
- * not that important because current Opteron setups do not support
- * scaling on SMP anyroads.
- *
- * Should fix up last_tsc too. Currently gettimeofday in the
- * first tick after the change will be slightly wrong.
- */
-
-static unsigned int ref_freq;
-static unsigned long loops_per_jiffy_ref;
-static unsigned long tsc_khz_ref;
-
-static int time_cpufreq_notifier(struct notifier_block *nb, unsigned long val,
- void *data)
-{
- struct cpufreq_freqs *freq = data;
- unsigned long *lpj, dummy;
-
- if (cpu_has(&cpu_data(freq->cpu), X86_FEATURE_CONSTANT_TSC))
- return 0;
-
- lpj = &dummy;
- if (!(freq->flags & CPUFREQ_CONST_LOOPS))
-#ifdef CONFIG_SMP
- lpj = &cpu_data(freq->cpu).loops_per_jiffy;
-#else
- lpj = &boot_cpu_data.loops_per_jiffy;
-#endif
-
- if (!ref_freq) {
- ref_freq = freq->old;
- loops_per_jiffy_ref = *lpj;
- tsc_khz_ref = tsc_khz;
- }
- if ((val == CPUFREQ_PRECHANGE && freq->old < freq->new) ||
- (val == CPUFREQ_POSTCHANGE && freq->old > freq->new) ||
- (val == CPUFREQ_RESUMECHANGE)) {
- *lpj =
- cpufreq_scale(loops_per_jiffy_ref, ref_freq, freq->new);
-
- tsc_khz = cpufreq_scale(tsc_khz_ref, ref_freq, freq->new);
- if (!(freq->flags & CPUFREQ_CONST_LOOPS))
- mark_tsc_unstable("cpufreq changes");
- }
-
- set_cyc2ns_scale(tsc_khz_ref, freq->cpu);
-
- return 0;
-}
-
-static struct notifier_block time_cpufreq_notifier_block = {
- .notifier_call = time_cpufreq_notifier
-};
-
-static int __init cpufreq_tsc(void)
-{
- cpufreq_register_notifier(&time_cpufreq_notifier_block,
- CPUFREQ_TRANSITION_NOTIFIER);
- return 0;
-}
-
-core_initcall(cpufreq_tsc);
-
-#endif
-
/*
* Make an educated guess if the TSC is trustworthy and synchronized
* over all CPUs.