[PATCH] i386: vMI timer patches

VMI timer code.  It works by taking over the local APIC clock when APIC is
configured, which requires a couple hooks into the APIC code.  The backend
timer code could be commonized into the timer infrastructure, but there are
some pieces missing (stolen time, in particular), and the exact semantics of
when to do accounting for NO_IDLE need to be shared between different
hypervisors as well.  So for now, VMI timer is a separate module.

[Adrian Bunk: cleanups]

Subject: VMI timer patches
Signed-off-by: Zachary Amsden <zach@vmware.com>
Signed-off-by: Andi Kleen <ak@suse.de>
Cc: Andi Kleen <ak@suse.de>
Cc: Jeremy Fitzhardinge <jeremy@xensource.com>
Cc: Rusty Russell <rusty@rustcorp.com.au>
Cc: Chris Wright <chrisw@sous-sol.org>
Signed-off-by: Andrew Morton <akpm@osdl.org>

arch/i386/Kconfig

@@ -1272,3 +1272,12 @@ config X86_TRAMPOLINE
 config KTIME_SCALAR
 	bool
 	default y
+
+config NO_IDLE_HZ
+	bool
+	depends on PARAVIRT
+	default y
+	help
+	  Switches the regular HZ timer off when the system is going idle.
+	  This helps a hypervisor detect that the Linux system is idle,
+	  reducing the overhead of idle systems.

arch/i386/kernel/Makefile

@@ -40,7 +40,7 @@ obj-$(CONFIG_EARLY_PRINTK)	+= early_printk.o
 obj-$(CONFIG_HPET_TIMER) 	+= hpet.o
 obj-$(CONFIG_K8_NB)		+= k8.o
 
-obj-$(CONFIG_VMI)		+= vmi.o
+obj-$(CONFIG_VMI)		+= vmi.o vmitime.o
 
 # Make sure this is linked after any other paravirt_ops structs: see head.S
 obj-$(CONFIG_PARAVIRT)		+= paravirt.o

arch/i386/kernel/apic.c

@@ -1395,7 +1395,7 @@ int __init APIC_init_uniprocessor (void)
 		if (!skip_ioapic_setup && nr_ioapics)
 			setup_IO_APIC();
 #endif
-	setup_boot_APIC_clock();
+	setup_boot_clock();
 
 	return 0;
 }

arch/i386/kernel/entry.S

@@ -626,6 +626,11 @@ ENTRY(name)				\
 /* The include is where all of the SMP etc. interrupts come from */
 #include "entry_arch.h"
 
+/* This alternate entry is needed because we hijack the apic LVTT */
+#if defined(CONFIG_VMI) && defined(CONFIG_X86_LOCAL_APIC)
+BUILD_INTERRUPT(apic_vmi_timer_interrupt,LOCAL_TIMER_VECTOR)
+#endif
+
 KPROBE_ENTRY(page_fault)
 	RING0_EC_FRAME
 	pushl $do_page_fault

arch/i386/kernel/paravirt.c

@@ -544,6 +544,8 @@ struct paravirt_ops paravirt_ops = {
 	.apic_write = native_apic_write,
 	.apic_write_atomic = native_apic_write_atomic,
 	.apic_read = native_apic_read,
+	.setup_boot_clock = setup_boot_APIC_clock,
+	.setup_secondary_clock = setup_secondary_APIC_clock,
 #endif
 	.set_lazy_mode = (void *)native_nop,
 

arch/i386/kernel/smpboot.c

@@ -554,7 +554,7 @@ static void __cpuinit start_secondary(void *unused)
 	smp_callin();
 	while (!cpu_isset(smp_processor_id(), smp_commenced_mask))
 		rep_nop();
-	setup_secondary_APIC_clock();
+	setup_secondary_clock();
 	if (nmi_watchdog == NMI_IO_APIC) {
 		disable_8259A_irq(0);
 		enable_NMI_through_LVT0(NULL);
@@ -1331,7 +1331,7 @@ static void __init smp_boot_cpus(unsigned int max_cpus)
 
 	smpboot_setup_io_apic();
 
-	setup_boot_APIC_clock();
+	setup_boot_clock();
 
 	/*
 	 * Synchronize the TSC with the AP

arch/i386/kernel/time.c

@@ -232,6 +232,7 @@ EXPORT_SYMBOL(get_cmos_time);
 static void sync_cmos_clock(unsigned long dummy);
 
 static DEFINE_TIMER(sync_cmos_timer, sync_cmos_clock, 0, 0);
+int no_sync_cmos_clock;
 
 static void sync_cmos_clock(unsigned long dummy)
 {
@@ -275,7 +276,8 @@ static void sync_cmos_clock(unsigned long dummy)
 
 void notify_arch_cmos_timer(void)
 {
-	mod_timer(&sync_cmos_timer, jiffies + 1);
+	if (!no_sync_cmos_clock)
+		mod_timer(&sync_cmos_timer, jiffies + 1);
 }
 
 static long clock_cmos_diff;

arch/i386/kernel/tsc.c

@@ -23,6 +23,7 @@
  * an extra value to store the TSC freq
  */
 unsigned int tsc_khz;
+unsigned long long (*custom_sched_clock)(void);
 
 int tsc_disable;
 
@@ -107,6 +108,9 @@ unsigned long long sched_clock(void)
 {
 	unsigned long long this_offset;
 
+	if (unlikely(custom_sched_clock))
+		return (*custom_sched_clock)();
+
 	/*
 	 * in the NUMA case we dont use the TSC as they are not
 	 * synchronized across all CPUs.

arch/i386/kernel/vmi.c

@@ -34,6 +34,7 @@
 #include <asm/apic.h>
 #include <asm/processor.h>
 #include <asm/timer.h>
+#include <asm/vmi_time.h>
 
 /* Convenient for calling VMI functions indirectly in the ROM */
 typedef u32 __attribute__((regparm(1))) (VROMFUNC)(void);
@@ -67,6 +68,7 @@ struct {
 	void (*set_linear_mapping)(int, u32, u32, u32);
 	void (*flush_tlb)(int);
 	void (*set_initial_ap_state)(int, int);
+	void (*halt)(void);
 } vmi_ops;
 
 /* XXX move this to alternative.h */
@@ -252,6 +254,19 @@ static void vmi_nop(void)
 {
 }
 
+/* For NO_IDLE_HZ, we stop the clock when halting the kernel */
+#ifdef CONFIG_NO_IDLE_HZ
+static fastcall void vmi_safe_halt(void)
+{
+	int idle = vmi_stop_hz_timer();
+	vmi_ops.halt();
+	if (idle) {
+		local_irq_disable();
+		vmi_account_time_restart_hz_timer();
+		local_irq_enable();
+	}
+}
+#endif
 
 #ifdef CONFIG_DEBUG_PAGE_TYPE
 
@@ -727,7 +742,12 @@ static inline int __init activate_vmi(void)
 		     (char *)paravirt_ops.save_fl);
 	patch_offset(&irq_save_disable_callout[IRQ_PATCH_DISABLE],
 		     (char *)paravirt_ops.irq_disable);
+#ifndef CONFIG_NO_IDLE_HZ
 	para_fill(safe_halt, Halt);
+#else
+	vmi_ops.halt = vmi_get_function(VMI_CALL_Halt);
+	paravirt_ops.safe_halt = vmi_safe_halt;
+#endif
 	para_fill(wbinvd, WBINVD);
 	/* paravirt_ops.read_msr = vmi_rdmsr */
 	/* paravirt_ops.write_msr = vmi_wrmsr */
@@ -837,6 +857,31 @@ static inline int __init activate_vmi(void)
 	paravirt_ops.apic_write_atomic = vmi_get_function(VMI_CALL_APICWrite);
 #endif
 
+	/*
+	 * Check for VMI timer functionality by probing for a cycle frequency method
+	 */
+	reloc = call_vrom_long_func(vmi_rom, get_reloc, VMI_CALL_GetCycleFrequency);
+	if (rel->type != VMI_RELOCATION_NONE) {
+		vmi_timer_ops.get_cycle_frequency = (void *)rel->eip;
+		vmi_timer_ops.get_cycle_counter =
+			vmi_get_function(VMI_CALL_GetCycleCounter);
+		vmi_timer_ops.get_wallclock =
+			vmi_get_function(VMI_CALL_GetWallclockTime);
+		vmi_timer_ops.wallclock_updated =
+			vmi_get_function(VMI_CALL_WallclockUpdated);
+		vmi_timer_ops.set_alarm = vmi_get_function(VMI_CALL_SetAlarm);
+		vmi_timer_ops.cancel_alarm =
+			 vmi_get_function(VMI_CALL_CancelAlarm);
+		paravirt_ops.time_init = vmi_time_init;
+		paravirt_ops.get_wallclock = vmi_get_wallclock;
+		paravirt_ops.set_wallclock = vmi_set_wallclock;
+#ifdef CONFIG_X86_LOCAL_APIC
+		paravirt_ops.setup_boot_clock = vmi_timer_setup_boot_alarm;
+		paravirt_ops.setup_secondary_clock = vmi_timer_setup_secondary_alarm;
+#endif
+		custom_sched_clock = vmi_sched_clock;
+	}
+
 	/*
 	 * Alternative instruction rewriting doesn't happen soon enough
 	 * to convert VMI_IRET to a call instead of a jump; so we have