---

yaml
---
r: 98432
b: refs/heads/master
c: 2826f8c
h: refs/heads/master
v: v3

[refs]

@@ -1,2 +1,2 @@
 ---
-refs/heads/master: 035cfc61a523343fe0bee5ec54348e26f330a06c
+refs/heads/master: 2826f8c0f4c97b7db33e2a680f184d828eb7a785

trunk/Documentation/DocBook/kgdb.tmpl

@@ -84,9 +84,10 @@
     runs an instance of gdb against the vmlinux file which contains
     the symbols (not boot image such as bzImage, zImage, uImage...).
     In gdb the developer specifies the connection parameters and
-    connects to kgdb.  The type of connection a developer makes with
-    gdb depends on the availability of kgdb I/O modules compiled as
-    builtin's or kernel modules in the test machine's kernel.
+    connects to kgdb.  Depending on which kgdb I/O modules exist in
+    the kernel for a given architecture, it may be possible to debug
+    the test machine's kernel with the development machine using a
+    rs232 or ethernet connection.
     </para>
   </chapter>
   <chapter id="CompilingAKernel">
@@ -222,7 +223,7 @@
   </para>
   <para>
   IMPORTANT NOTE: Using this option with kgdb over the console
-  (kgdboc) is not supported.
+  (kgdboc) or kgdb over ethernet (kgdboe) is not supported.
   </para>
   </sect1>
   </chapter>
@@ -248,11 +249,18 @@
     (gdb) target remote /dev/ttyS0
     </programlisting>
     <para>
-    Example (kgdb to a terminal server on tcp port 2012):
+    Example (kgdb to a terminal server):
     </para>
     <programlisting>
     % gdb ./vmlinux
-    (gdb) target remote 192.168.2.2:2012
+    (gdb) target remote udp:192.168.2.2:6443
+    </programlisting>
+    <para>
+    Example (kgdb over ethernet):
+    </para>
+    <programlisting>
+    % gdb ./vmlinux
+    (gdb) target remote udp:192.168.2.2:6443
     </programlisting>
     <para>
     Once connected, you can debug a kernel the way you would debug an

trunk/arch/ia64/kernel/setup.c

@@ -578,8 +578,6 @@ setup_arch (char **cmdline_p)
 	cpu_init();	/* initialize the bootstrap CPU */
 	mmu_context_init();	/* initialize context_id bitmap */
 
-	check_sal_cache_flush();
-
 #ifdef CONFIG_ACPI
 	acpi_boot_init();
 #endif
@@ -607,6 +605,7 @@ setup_arch (char **cmdline_p)
 		ia64_mca_init();
 
 	platform_setup(cmdline_p);
+	check_sal_cache_flush();
 	paging_init();
 }
 

trunk/arch/x86/Kconfig

@@ -383,7 +383,6 @@ config VMI
 config KVM_CLOCK
 	bool "KVM paravirtualized clock"
 	select PARAVIRT
-	select PARAVIRT_CLOCK
 	depends on !(X86_VISWS || X86_VOYAGER)
 	help
 	  Turning on this option will allow you to run a paravirtualized clock
@@ -411,10 +410,6 @@ config PARAVIRT
 	  over full virtualization.  However, when run without a hypervisor
 	  the kernel is theoretically slower and slightly larger.
 
-config PARAVIRT_CLOCK
-	bool
-	default n
-
 endif
 
 config MEMTEST_BOOTPARAM

trunk/arch/x86/kernel/Makefile

@@ -82,7 +82,6 @@ obj-$(CONFIG_VMI)		+= vmi_32.o vmiclock_32.o
 obj-$(CONFIG_KVM_GUEST)		+= kvm.o
 obj-$(CONFIG_KVM_CLOCK)		+= kvmclock.o
 obj-$(CONFIG_PARAVIRT)		+= paravirt.o paravirt_patch_$(BITS).o
-obj-$(CONFIG_PARAVIRT_CLOCK)	+= pvclock.o
 
 obj-$(CONFIG_PCSPKR_PLATFORM)	+= pcspeaker.o
 

trunk/arch/x86/kernel/kvmclock.c

@@ -18,7 +18,6 @@
 
 #include <linux/clocksource.h>
 #include <linux/kvm_para.h>
-#include <asm/pvclock.h>
 #include <asm/arch_hooks.h>
 #include <asm/msr.h>
 #include <asm/apic.h>
@@ -37,47 +36,83 @@ static int parse_no_kvmclock(char *arg)
 early_param("no-kvmclock", parse_no_kvmclock);
 
 /* The hypervisor will put information about time periodically here */
-static DEFINE_PER_CPU_SHARED_ALIGNED(struct pvclock_vcpu_time_info, hv_clock);
-static struct pvclock_wall_clock wall_clock;
+static DEFINE_PER_CPU_SHARED_ALIGNED(struct kvm_vcpu_time_info, hv_clock);
+#define get_clock(cpu, field) per_cpu(hv_clock, cpu).field
 
+static inline u64 kvm_get_delta(u64 last_tsc)
+{
+	int cpu = smp_processor_id();
+	u64 delta = native_read_tsc() - last_tsc;
+	return (delta * get_clock(cpu, tsc_to_system_mul)) >> KVM_SCALE;
+}
+
+static struct kvm_wall_clock wall_clock;
+static cycle_t kvm_clock_read(void);
 /*
  * The wallclock is the time of day when we booted. Since then, some time may
  * have elapsed since the hypervisor wrote the data. So we try to account for
  * that with system time
  */
 static unsigned long kvm_get_wallclock(void)
 {
-	struct pvclock_vcpu_time_info *vcpu_time;
+	u32 wc_sec, wc_nsec;
+	u64 delta;
 	struct timespec ts;
+	int version, nsec;
 	int low, high;
 
 	low = (int)__pa(&wall_clock);
 	high = ((u64)__pa(&wall_clock) >> 32);
-	native_write_msr(MSR_KVM_WALL_CLOCK, low, high);
 
-	vcpu_time = &get_cpu_var(hv_clock);
-	pvclock_read_wallclock(&wall_clock, vcpu_time, &ts);
-	put_cpu_var(hv_clock);
+	delta = kvm_clock_read();
 
-	return ts.tv_sec;
+	native_write_msr(MSR_KVM_WALL_CLOCK, low, high);
+	do {
+		version = wall_clock.wc_version;
+		rmb();
+		wc_sec = wall_clock.wc_sec;
+		wc_nsec = wall_clock.wc_nsec;
+		rmb();
+	} while ((wall_clock.wc_version != version) || (version & 1));
+
+	delta = kvm_clock_read() - delta;
+	delta += wc_nsec;
+	nsec = do_div(delta, NSEC_PER_SEC);
+	set_normalized_timespec(&ts, wc_sec + delta, nsec);
+	/*
+	 * Of all mechanisms of time adjustment I've tested, this one
+	 * was the champion!
+	 */
+	return ts.tv_sec + 1;
 }
 
 static int kvm_set_wallclock(unsigned long now)
 {
-	return -1;
+	return 0;
 }
 
+/*
+ * This is our read_clock function. The host puts an tsc timestamp each time
+ * it updates a new time. Without the tsc adjustment, we can have a situation
+ * in which a vcpu starts to run earlier (smaller system_time), but probes
+ * time later (compared to another vcpu), leading to backwards time
+ */
 static cycle_t kvm_clock_read(void)
 {
-	struct pvclock_vcpu_time_info *src;
-	cycle_t ret;
+	u64 last_tsc, now;
+	int cpu;
 
-	src = &get_cpu_var(hv_clock);
-	ret = pvclock_clocksource_read(src);
-	put_cpu_var(hv_clock);
-	return ret;
-}
+	preempt_disable();
+	cpu = smp_processor_id();
+
+	last_tsc = get_clock(cpu, tsc_timestamp);
+	now = get_clock(cpu, system_time);
+
+	now += kvm_get_delta(last_tsc);
+	preempt_enable();
 
+	return now;
+}
 static struct clocksource kvm_clock = {
 	.name = "kvm-clock",
 	.read = kvm_clock_read,
@@ -88,14 +123,13 @@ static struct clocksource kvm_clock = {
 	.flags = CLOCK_SOURCE_IS_CONTINUOUS,
 };
 
-static int kvm_register_clock(char *txt)
+static int kvm_register_clock(void)
 {
 	int cpu = smp_processor_id();
 	int low, high;
 	low = (int)__pa(&per_cpu(hv_clock, cpu)) | 1;
 	high = ((u64)__pa(&per_cpu(hv_clock, cpu)) >> 32);
-	printk(KERN_INFO "kvm-clock: cpu %d, msr %x:%x, %s\n",
-	       cpu, high, low, txt);
+
 	return native_write_msr_safe(MSR_KVM_SYSTEM_TIME, low, high);
 }
 
@@ -106,20 +140,12 @@ static void kvm_setup_secondary_clock(void)
 	 * Now that the first cpu already had this clocksource initialized,
 	 * we shouldn't fail.
 	 */
-	WARN_ON(kvm_register_clock("secondary cpu clock"));
+	WARN_ON(kvm_register_clock());
 	/* ok, done with our trickery, call native */
 	setup_secondary_APIC_clock();
 }
 #endif
 
-#ifdef CONFIG_SMP
-void __init kvm_smp_prepare_boot_cpu(void)
-{
-	WARN_ON(kvm_register_clock("primary cpu clock"));
-	native_smp_prepare_boot_cpu();
-}
-#endif
-
 /*
  * After the clock is registered, the host will keep writing to the
  * registered memory location. If the guest happens to shutdown, this memory
@@ -148,16 +174,13 @@ void __init kvmclock_init(void)
 		return;
 
 	if (kvmclock && kvm_para_has_feature(KVM_FEATURE_CLOCKSOURCE)) {
-		if (kvm_register_clock("boot clock"))
+		if (kvm_register_clock())
 			return;
 		pv_time_ops.get_wallclock = kvm_get_wallclock;
 		pv_time_ops.set_wallclock = kvm_set_wallclock;
 		pv_time_ops.sched_clock = kvm_clock_read;
 #ifdef CONFIG_X86_LOCAL_APIC
 		pv_apic_ops.setup_secondary_clock = kvm_setup_secondary_clock;
-#endif
-#ifdef CONFIG_SMP
-		smp_ops.smp_prepare_boot_cpu = kvm_smp_prepare_boot_cpu;
 #endif
 		machine_ops.shutdown  = kvm_shutdown;
 #ifdef CONFIG_KEXEC