Merge branch 'x86-vdso-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/linux-2.6-tip

* 'x86-vdso-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/linux-2.6-tip:
  x86: vdso: Remove unused variable
  x86-64: Optimize vDSO time()
  x86-64: Add time to vDSO
  x86-64: Turn off -pg and turn on -foptimize-sibling-calls for vDSO
  x86-64: Move vread_tsc into a new file with sensible options
  x86-64: Vclock_gettime(CLOCK_MONOTONIC) can't ever see nsec < 0
  x86-64: Don't generate cmov in vread_tsc
  x86-64: Remove unnecessary barrier in vread_tsc
  x86-64: Clean up vdso/kernel shared variables

6 files changed, 73 insertions, 72 deletions

diff --git a/arch/x86/kernel/Makefile b/arch/x86/kernel/Makefile
index 250806472a7..f5abe3a245b 100644
--- a/arch/x86/kernel/Makefile
+++ b/arch/x86/kernel/Makefile
@@ -8,7 +8,6 @@ CPPFLAGS_vmlinux.lds += -U$(UTS_MACHINE)
 
 ifdef CONFIG_FUNCTION_TRACER
 # Do not profile debug and lowlevel utilities
-CFLAGS_REMOVE_tsc.o = -pg
 CFLAGS_REMOVE_rtc.o = -pg
 CFLAGS_REMOVE_paravirt-spinlocks.o = -pg
 CFLAGS_REMOVE_pvclock.o = -pg
@@ -24,13 +23,16 @@ endif
 nostackp := $(call cc-option, -fno-stack-protector)
 CFLAGS_vsyscall_64.o	:= $(PROFILING) -g0 $(nostackp)
 CFLAGS_hpet.o		:= $(nostackp)
-CFLAGS_tsc.o		:= $(nostackp)
+CFLAGS_vread_tsc_64.o	:= $(nostackp)
 CFLAGS_paravirt.o	:= $(nostackp)
 GCOV_PROFILE_vsyscall_64.o	:= n
 GCOV_PROFILE_hpet.o		:= n
 GCOV_PROFILE_tsc.o		:= n
 GCOV_PROFILE_paravirt.o		:= n
 
+# vread_tsc_64 is hot and should be fully optimized:
+CFLAGS_REMOVE_vread_tsc_64.o = -pg -fno-optimize-sibling-calls
+
 obj-y			:= process_$(BITS).o signal.o entry_$(BITS).o
 obj-y			+= traps.o irq.o irq_$(BITS).o dumpstack_$(BITS).o
 obj-y			+= time.o ioport.o ldt.o dumpstack.o
@@ -39,7 +41,7 @@ obj-$(CONFIG_IRQ_WORK)  += irq_work.o
 obj-y			+= probe_roms.o
 obj-$(CONFIG_X86_32)	+= sys_i386_32.o i386_ksyms_32.o
 obj-$(CONFIG_X86_64)	+= sys_x86_64.o x8664_ksyms_64.o
-obj-$(CONFIG_X86_64)	+= syscall_64.o vsyscall_64.o
+obj-$(CONFIG_X86_64)	+= syscall_64.o vsyscall_64.o vread_tsc_64.o
 obj-y			+= bootflag.o e820.o
 obj-y			+= pci-dma.o quirks.o topology.o kdebugfs.o
 obj-y			+= alternative.o i8253.o pci-nommu.o hw_breakpoint.o
diff --git a/arch/x86/kernel/time.c b/arch/x86/kernel/time.c
index 25a28a24593..00cbb272627 100644
--- a/arch/x86/kernel/time.c
+++ b/arch/x86/kernel/time.c
@@ -23,7 +23,7 @@
 #include <asm/time.h>
 
 #ifdef CONFIG_X86_64
-volatile unsigned long __jiffies __section_jiffies = INITIAL_JIFFIES;
+DEFINE_VVAR(volatile unsigned long, jiffies) = INITIAL_JIFFIES;
 #endif
 
 unsigned long profile_pc(struct pt_regs *regs)
diff --git a/arch/x86/kernel/tsc.c b/arch/x86/kernel/tsc.c
index 9335bf7dd2e..6cc6922262a 100644
--- a/arch/x86/kernel/tsc.c
+++ b/arch/x86/kernel/tsc.c
@@ -763,25 +763,6 @@ static cycle_t read_tsc(struct clocksource *cs)
 		ret : clocksource_tsc.cycle_last;
 }
 
-#ifdef CONFIG_X86_64
-static cycle_t __vsyscall_fn vread_tsc(void)
-{
-	cycle_t ret;
-
-	/*
-	 * Surround the RDTSC by barriers, to make sure it's not
-	 * speculated to outside the seqlock critical section and
-	 * does not cause time warps:
-	 */
-	rdtsc_barrier();
-	ret = (cycle_t)vget_cycles();
-	rdtsc_barrier();
-
-	return ret >= __vsyscall_gtod_data.clock.cycle_last ?
-		ret : __vsyscall_gtod_data.clock.cycle_last;
-}
-#endif
-
 static void resume_tsc(struct clocksource *cs)
 {
 	clocksource_tsc.cycle_last = 0;
diff --git a/arch/x86/kernel/vmlinux.lds.S b/arch/x86/kernel/vmlinux.lds.S
index 61682f0ac26..89aed99aafc 100644
--- a/arch/x86/kernel/vmlinux.lds.S
+++ b/arch/x86/kernel/vmlinux.lds.S
@@ -161,6 +161,12 @@ SECTIONS
 
 #define VVIRT_OFFSET (VSYSCALL_ADDR - __vsyscall_0)
 #define VVIRT(x) (ADDR(x) - VVIRT_OFFSET)
+#define EMIT_VVAR(x, offset) .vsyscall_var_ ## x	\
+	ADDR(.vsyscall_0) + offset		 	\
+	: AT(VLOAD(.vsyscall_var_ ## x)) {     		\
+		*(.vsyscall_var_ ## x)			\
+	}						\
+	x = VVIRT(.vsyscall_var_ ## x);
 
 	. = ALIGN(4096);
 	__vsyscall_0 = .;
@@ -175,18 +181,6 @@ SECTIONS
 		*(.vsyscall_fn)
 	}
 
-	. = ALIGN(L1_CACHE_BYTES);
-	.vsyscall_gtod_data : AT(VLOAD(.vsyscall_gtod_data)) {
-		*(.vsyscall_gtod_data)
-	}
-
-	vsyscall_gtod_data = VVIRT(.vsyscall_gtod_data);
-	.vsyscall_clock : AT(VLOAD(.vsyscall_clock)) {
-		*(.vsyscall_clock)
-	}
-	vsyscall_clock = VVIRT(.vsyscall_clock);
-
-
 	.vsyscall_1 ADDR(.vsyscall_0) + 1024: AT(VLOAD(.vsyscall_1)) {
 		*(.vsyscall_1)
 	}
@@ -194,21 +188,14 @@ SECTIONS
 		*(.vsyscall_2)
 	}
 
-	.vgetcpu_mode : AT(VLOAD(.vgetcpu_mode)) {
-		*(.vgetcpu_mode)
-	}
-	vgetcpu_mode = VVIRT(.vgetcpu_mode);
-
-	. = ALIGN(L1_CACHE_BYTES);
-	.jiffies : AT(VLOAD(.jiffies)) {
-		*(.jiffies)
-	}
-	jiffies = VVIRT(.jiffies);
-
 	.vsyscall_3 ADDR(.vsyscall_0) + 3072: AT(VLOAD(.vsyscall_3)) {
 		*(.vsyscall_3)
 	}
 
+#define __VVAR_KERNEL_LDS
+#include <asm/vvar.h>
+#undef __VVAR_KERNEL_LDS
+
 	. = __vsyscall_0 + PAGE_SIZE;
 
 #undef VSYSCALL_ADDR
@@ -216,6 +203,7 @@ SECTIONS
 #undef VLOAD
 #undef VVIRT_OFFSET
 #undef VVIRT
+#undef EMIT_VVAR
 
 #endif /* CONFIG_X86_64 */
 
diff --git a/arch/x86/kernel/vread_tsc_64.c b/arch/x86/kernel/vread_tsc_64.c
new file mode 100644
index 00000000000..a81aa9e9894
--- /dev/null
+++ b/arch/x86/kernel/vread_tsc_64.c
@@ -0,0 +1,36 @@
+/* This code runs in userspace. */
+
+#define DISABLE_BRANCH_PROFILING
+#include <asm/vgtod.h>
+
+notrace cycle_t __vsyscall_fn vread_tsc(void)
+{
+	cycle_t ret;
+	u64 last;
+
+	/*
+	 * Empirically, a fence (of type that depends on the CPU)
+	 * before rdtsc is enough to ensure that rdtsc is ordered
+	 * with respect to loads.  The various CPU manuals are unclear
+	 * as to whether rdtsc can be reordered with later loads,
+	 * but no one has ever seen it happen.
+	 */
+	rdtsc_barrier();
+	ret = (cycle_t)vget_cycles();
+
+	last = VVAR(vsyscall_gtod_data).clock.cycle_last;
+
+	if (likely(ret >= last))
+		return ret;
+
+	/*
+	 * GCC likes to generate cmov here, but this branch is extremely
+	 * predictable (it's just a funciton of time and the likely is
+	 * very likely) and there's a data dependence, so force GCC
+	 * to generate a branch instead.  I don't barrier() because
+	 * we don't actually need a barrier, and if this function
+	 * ever gets inlined it will generate worse code.
+	 */
+	asm volatile ("");
+	return last;
+}
diff --git a/arch/x86/kernel/vsyscall_64.c b/arch/x86/kernel/vsyscall_64.c
index 59be48d0d75..3e682184d76 100644
--- a/arch/x86/kernel/vsyscall_64.c
+++ b/arch/x86/kernel/vsyscall_64.c
@@ -49,15 +49,8 @@
 		__attribute__ ((unused, __section__(".vsyscall_" #nr))) notrace
 #define __syscall_clobber "r11","cx","memory"
 
-/*
- * vsyscall_gtod_data contains data that is :
- * - readonly from vsyscalls
- * - written by timer interrupt or systcl (/proc/sys/kernel/vsyscall64)
- * Try to keep this structure as small as possible to avoid cache line ping pongs
- */
-int __vgetcpu_mode __section_vgetcpu_mode;
-
-struct vsyscall_gtod_data __vsyscall_gtod_data __section_vsyscall_gtod_data =
+DEFINE_VVAR(int, vgetcpu_mode);
+DEFINE_VVAR(struct vsyscall_gtod_data, vsyscall_gtod_data) =
 {
 	.lock = __SEQLOCK_UNLOCKED(__vsyscall_gtod_data.lock),
 	.sysctl_enabled = 1,
@@ -97,7 +90,7 @@ void update_vsyscall(struct timespec *wall_time, struct timespec *wtm,
  */
 static __always_inline void do_get_tz(struct timezone * tz)
 {
-	*tz = __vsyscall_gtod_data.sys_tz;
+	*tz = VVAR(vsyscall_gtod_data).sys_tz;
 }
 
 static __always_inline int gettimeofday(struct timeval *tv, struct timezone *tz)
@@ -126,23 +119,24 @@ static __always_inline void do_vgettimeofday(struct timeval * tv)
 	unsigned long mult, shift, nsec;
 	cycle_t (*vread)(void);
 	do {
-		seq = read_seqbegin(&__vsyscall_gtod_data.lock);
+		seq = read_seqbegin(&VVAR(vsyscall_gtod_data).lock);
 
-		vread = __vsyscall_gtod_data.clock.vread;
-		if (unlikely(!__vsyscall_gtod_data.sysctl_enabled || !vread)) {
+		vread = VVAR(vsyscall_gtod_data).clock.vread;
+		if (unlikely(!VVAR(vsyscall_gtod_data).sysctl_enabled ||
+			     !vread)) {
 			gettimeofday(tv,NULL);
 			return;
 		}
 
 		now = vread();
-		base = __vsyscall_gtod_data.clock.cycle_last;
-		mask = __vsyscall_gtod_data.clock.mask;
-		mult = __vsyscall_gtod_data.clock.mult;
-		shift = __vsyscall_gtod_data.clock.shift;
+		base = VVAR(vsyscall_gtod_data).clock.cycle_last;
+		mask = VVAR(vsyscall_gtod_data).clock.mask;
+		mult = VVAR(vsyscall_gtod_data).clock.mult;
+		shift = VVAR(vsyscall_gtod_data).clock.shift;
 
-		tv->tv_sec = __vsyscall_gtod_data.wall_time_sec;
-		nsec = __vsyscall_gtod_data.wall_time_nsec;
-	} while (read_seqretry(&__vsyscall_gtod_data.lock, seq));
+		tv->tv_sec = VVAR(vsyscall_gtod_data).wall_time_sec;
+		nsec = VVAR(vsyscall_gtod_data).wall_time_nsec;
+	} while (read_seqretry(&VVAR(vsyscall_gtod_data).lock, seq));
 
 	/* calculate interval: */
 	cycle_delta = (now - base) & mask;
@@ -171,15 +165,15 @@ time_t __vsyscall(1) vtime(time_t *t)
 {
 	unsigned seq;
 	time_t result;
-	if (unlikely(!__vsyscall_gtod_data.sysctl_enabled))
+	if (unlikely(!VVAR(vsyscall_gtod_data).sysctl_enabled))
 		return time_syscall(t);
 
 	do {
-		seq = read_seqbegin(&__vsyscall_gtod_data.lock);
+		seq = read_seqbegin(&VVAR(vsyscall_gtod_data).lock);
 
-		result = __vsyscall_gtod_data.wall_time_sec;
+		result = VVAR(vsyscall_gtod_data).wall_time_sec;
 
-	} while (read_seqretry(&__vsyscall_gtod_data.lock, seq));
+	} while (read_seqretry(&VVAR(vsyscall_gtod_data).lock, seq));
 
 	if (t)
 		*t = result;
@@ -208,9 +202,9 @@ vgetcpu(unsigned *cpu, unsigned *node, struct getcpu_cache *tcache)
 	   We do this here because otherwise user space would do it on
 	   its own in a likely inferior way (no access to jiffies).
 	   If you don't like it pass NULL. */
-	if (tcache && tcache->blob[0] == (j = __jiffies)) {
+	if (tcache && tcache->blob[0] == (j = VVAR(jiffies))) {
 		p = tcache->blob[1];
-	} else if (__vgetcpu_mode == VGETCPU_RDTSCP) {
+	} else if (VVAR(vgetcpu_mode) == VGETCPU_RDTSCP) {
 		/* Load per CPU data from RDTSCP */
 		native_read_tscp(&p);
 	} else {