x86: Move tsc_calibration to x86_init_ops

TSC calibration is modified by the vmware hypervisor and paravirt by
separate means. Moorestown wants to add its own calibration routine as
well. So make calibrate_tsc a proper x86_init_ops function and
override it by paravirt or by the early setup of the vmware
hypervisor.

Signed-off-by: Thomas Gleixner <tglx@linutronix.de>

9 files changed, 32 insertions, 30 deletions

diff --git a/arch/x86/kernel/cpu/hypervisor.c b/arch/x86/kernel/cpu/hypervisor.c
index 93ba8eeb100a..08be922de33a 100644
--- a/arch/x86/kernel/cpu/hypervisor.c
+++ b/arch/x86/kernel/cpu/hypervisor.c
@@ -34,13 +34,6 @@ detect_hypervisor_vendor(struct cpuinfo_x86 *c)
                 c->x86_hyper_vendor = X86_HYPER_VENDOR_NONE;
 }
 
-unsigned long get_hypervisor_tsc_freq(void)
-{
-        if (boot_cpu_data.x86_hyper_vendor == X86_HYPER_VENDOR_VMWARE)
-                return vmware_get_tsc_khz();
-        return 0;
-}
-
 static inline void __cpuinit
 hypervisor_set_feature_bits(struct cpuinfo_x86 *c)
 {
@@ -55,3 +48,10 @@ void __cpuinit init_hypervisor(struct cpuinfo_x86 *c)
         detect_hypervisor_vendor(c);
         hypervisor_set_feature_bits(c);
 }
+
+void __init init_hypervisor_platform(void)
+{
+        init_hypervisor(&boot_cpu_data);
+        if (boot_cpu_data.x86_hyper_vendor == X86_HYPER_VENDOR_VMWARE)
+                vmware_platform_setup();
+}
diff --git a/arch/x86/kernel/cpu/vmware.c b/arch/x86/kernel/cpu/vmware.c
index bc24f514ec93..0a46b4df5d80 100644
--- a/arch/x86/kernel/cpu/vmware.c
+++ b/arch/x86/kernel/cpu/vmware.c
@@ -24,6 +24,7 @@
 #include <linux/dmi.h>
 #include <asm/div64.h>
 #include <asm/vmware.h>
+#include <asm/x86_init.h>
 
 #define CPUID_VMWARE_INFO_LEAF  0x40000000
 #define VMWARE_HYPERVISOR_MAGIC 0x564D5868
@@ -47,21 +48,29 @@ static inline int __vmware_platform(void)
         return eax != (uint32_t)-1 && ebx == VMWARE_HYPERVISOR_MAGIC;
 }
 
-static unsigned long __vmware_get_tsc_khz(void)
+static unsigned long vmware_get_tsc_khz(void)
 {
         uint64_t tsc_hz;
         uint32_t eax, ebx, ecx, edx;
 
         VMWARE_PORT(GETHZ, eax, ebx, ecx, edx);
 
-        if (ebx == UINT_MAX)
-                return 0;
         tsc_hz = eax | (((uint64_t)ebx) << 32);
         do_div(tsc_hz, 1000);
         BUG_ON(tsc_hz >> 32);
         return tsc_hz;
 }
 
+void __init vmware_platform_setup(void)
+{
+        uint32_t eax, ebx, ecx, edx;
+
+        VMWARE_PORT(GETHZ, eax, ebx, ecx, edx);
+
+        if (ebx != UINT_MAX)
+                x86_platform.calibrate_tsc = vmware_get_tsc_khz;
+}
+
 /*
  * While checking the dmi string infomation, just checking the product
  * serial key should be enough, as this will always have a VMware
@@ -87,12 +96,6 @@ int vmware_platform(void)
         return 0;
 }
 
-unsigned long vmware_get_tsc_khz(void)
-{
-        BUG_ON(!vmware_platform());
-        return __vmware_get_tsc_khz();
-}
-
 /*
  * VMware hypervisor takes care of exporting a reliable TSC to the guest.
  * Still, due to timing difference when running on virtual cpus, the TSC can
diff --git a/arch/x86/kernel/kvmclock.c b/arch/x86/kernel/kvmclock.c
index 64e9b5f59d2d..75a21b61b863 100644
--- a/arch/x86/kernel/kvmclock.c
+++ b/arch/x86/kernel/kvmclock.c
@@ -187,7 +187,7 @@ void __init kvmclock_init(void)
                 pv_time_ops.get_wallclock = kvm_get_wallclock;
                 pv_time_ops.set_wallclock = kvm_set_wallclock;
                 pv_time_ops.sched_clock = kvm_clock_read;
-                pv_time_ops.get_tsc_khz = kvm_get_tsc_khz;
+                x86_platform.calibrate_tsc = kvm_get_tsc_khz;
 #ifdef CONFIG_X86_LOCAL_APIC
                 x86_cpuinit.setup_percpu_clockev =
                         kvm_setup_secondary_clock;
diff --git a/arch/x86/kernel/paravirt.c b/arch/x86/kernel/paravirt.c
index 9c0e644a76dc..7cbf898d839b 100644
--- a/arch/x86/kernel/paravirt.c
+++ b/arch/x86/kernel/paravirt.c
@@ -309,7 +309,6 @@ struct pv_time_ops pv_time_ops = {
         .get_wallclock = native_get_wallclock,
         .set_wallclock = native_set_wallclock,
         .sched_clock = native_sched_clock,
-        .get_tsc_khz = native_calibrate_tsc,
 };
 
 struct pv_irq_ops pv_irq_ops = {
diff --git a/arch/x86/kernel/setup.c b/arch/x86/kernel/setup.c
index bb207a47c631..2d93026af7cd 100644
--- a/arch/x86/kernel/setup.c
+++ b/arch/x86/kernel/setup.c
@@ -818,7 +818,7 @@ void __init setup_arch(char **cmdline_p)
          * VMware detection requires dmi to be available, so this
          * needs to be done after dmi_scan_machine, for the BP.
          */
-        init_hypervisor(&boot_cpu_data);
+        init_hypervisor_platform();
 
         x86_init.resources.probe_roms();
 
diff --git a/arch/x86/kernel/tsc.c b/arch/x86/kernel/tsc.c
index 97a0bcbad100..9917632a8b49 100644
--- a/arch/x86/kernel/tsc.c
+++ b/arch/x86/kernel/tsc.c
@@ -18,6 +18,7 @@
 #include <asm/delay.h>
 #include <asm/hypervisor.h>
 #include <asm/nmi.h>
+#include <asm/x86_init.h>
 
 unsigned int __read_mostly cpu_khz;     /* TSC clocks / usec, not used here */
 EXPORT_SYMBOL(cpu_khz);
@@ -401,15 +402,9 @@ unsigned long native_calibrate_tsc(void)
 {
         u64 tsc1, tsc2, delta, ref1, ref2;
         unsigned long tsc_pit_min = ULONG_MAX, tsc_ref_min = ULONG_MAX;
-        unsigned long flags, latch, ms, fast_calibrate, hv_tsc_khz;
+        unsigned long flags, latch, ms, fast_calibrate;
         int hpet = is_hpet_enabled(), i, loopmin;
 
-        hv_tsc_khz = get_hypervisor_tsc_freq();
-        if (hv_tsc_khz) {
-                printk(KERN_INFO "TSC: Frequency read from the hypervisor\n");
-                return hv_tsc_khz;
-        }
-
         local_irq_save(flags);
         fast_calibrate = quick_pit_calibrate();
         local_irq_restore(flags);
@@ -567,7 +562,7 @@ int recalibrate_cpu_khz(void)
         unsigned long cpu_khz_old = cpu_khz;
 
         if (cpu_has_tsc) {
-                tsc_khz = calibrate_tsc();
+                tsc_khz = x86_platform.calibrate_tsc();
                 cpu_khz = tsc_khz;
                 cpu_data(0).loops_per_jiffy =
                         cpufreq_scale(cpu_data(0).loops_per_jiffy,
@@ -917,7 +912,7 @@ void __init tsc_init(void)
         if (!cpu_has_tsc)
                 return;
 
-        tsc_khz = calibrate_tsc();
+        tsc_khz = x86_platform.calibrate_tsc();
         cpu_khz = tsc_khz;
 
         if (!tsc_khz) {
diff --git a/arch/x86/kernel/vmi_32.c b/arch/x86/kernel/vmi_32.c
index cd7d0fbbf66e..052ae81ee08b 100644
--- a/arch/x86/kernel/vmi_32.c
+++ b/arch/x86/kernel/vmi_32.c
@@ -825,7 +825,7 @@ static inline int __init activate_vmi(void)
                 x86_cpuinit.setup_percpu_clockev = vmi_time_ap_init;
 #endif
                 pv_time_ops.sched_clock = vmi_sched_clock;
-                pv_time_ops.get_tsc_khz = vmi_tsc_khz;
+                x86_platform.calibrate_tsc = vmi_tsc_khz;
 
                 /* We have true wallclock functions; disable CMOS clock sync */
                 no_sync_cmos_clock = 1;
diff --git a/arch/x86/kernel/vmiclock_32.c b/arch/x86/kernel/vmiclock_32.c
index 2b3eb82efeeb..611b9e2360d3 100644
--- a/arch/x86/kernel/vmiclock_32.c
+++ b/arch/x86/kernel/vmiclock_32.c
@@ -68,7 +68,7 @@ unsigned long long vmi_sched_clock(void)
         return cycles_2_ns(vmi_timer_ops.get_cycle_counter(VMI_CYCLES_AVAILABLE));
 }
 
-/* paravirt_ops.get_tsc_khz = vmi_tsc_khz */
+/* x86_platform.calibrate_tsc = vmi_tsc_khz */
 unsigned long vmi_tsc_khz(void)
 {
         unsigned long long khz;
diff --git a/arch/x86/kernel/x86_init.c b/arch/x86/kernel/x86_init.c
index 4790b92714a6..13081b921914 100644
--- a/arch/x86/kernel/x86_init.c
+++ b/arch/x86/kernel/x86_init.c
@@ -13,6 +13,7 @@
 #include <asm/e820.h>
 #include <asm/time.h>
 #include <asm/irq.h>
+#include <asm/tsc.h>
 
 void __cpuinit x86_init_noop(void) { }
 void __init x86_init_uint_noop(unsigned int unused) { }
@@ -67,3 +68,7 @@ struct __initdata x86_init_ops x86_init = {
 __cpuinitdata struct x86_cpuinit_ops x86_cpuinit = {
         .setup_percpu_clockev           = setup_secondary_APIC_clock,
 };
+
+struct x86_platform_ops x86_platform = {
+        .calibrate_tsc                  = native_calibrate_tsc,
+};