warning: ‘x86_get_event_constraints’ defined but not used

warning: ‘x86_get_event_constraints’ defined but not used

[Kevin Winchester]

Hi,

From:

    gcc (GCC) 4.6.1

I get the following warning:

    arch/x86/kernel/cpu/perf_event.c:1444:1: warning:
‘x86_get_event_constraints’ defined but not used [-Wunused-function]

It appears that function is only called in some of the C files that
are included into perf_event.c, namely perf_event_intel.c,
perf_event_p4.c and perf_event_p6.c.  Each of those files is
completely dependent on CONFIG_CPU_SUP_INTEL, which I obviously do not
have enabled.

At the moment, I have eliminated the warning with the patch below, but
I do not consider that to be optimal.  Is there are reason why those C
files are included directly in another C file rather than being linked
together?  Or is there some other way the code could be improved and
also remove the warning at the same time?

I do not mind doing some coding here to make this happen.  I am very
close to a warning-free build, so I would definitely like to take care
of this one.

diff --git a/arch/x86/kernel/cpu/perf_event.c b/arch/x86/kernel/cpu/perf_event.c
index 4ee3abf..a37e7e9 100644
--- a/arch/x86/kernel/cpu/perf_event.c
+++ b/arch/x86/kernel/cpu/perf_event.c
@@ -1440,6 +1440,7 @@ static __read_mostly struct notifier_block
perf_event_nmi_notifier = {
 static struct event_constraint unconstrained;
 static struct event_constraint emptyconstraint;

+#ifdef CONFIG_CPU_SUP_INTEL
 static struct event_constraint *
 x86_get_event_constraints(struct cpu_hw_events *cpuc, struct perf_event *event)
 {
@@ -1454,6 +1455,7 @@ x86_get_event_constraints(struct cpu_hw_events
*cpuc, struct perf_event *event)

        return &unconstrained;
 }
+#endif

 #include "perf_event_amd.c"
 #include "perf_event_p6.c"


Thanks,

-- 
Kevin Winchester

Re: warning: ‘x86_get_event_constraints’ defined but not used

[Peter Zijlstra]

On Mon, 2011-08-15 at 20:40 -0300, Kevin Winchester wrote:
> Is there are reason why those C
> files are included directly in another C file rather than being linked
> together?  

Sheer suckyness on my part.. we needed a 'quick' fix to make these
CPU_SUP_* things build at all and never got around to cleaning it up
afterwards.

> Or is there some other way the code could be improved and
> also remove the warning at the same time?

creation of a arch/x86/kernel/cpu/perf_event.h and enough code massage
to make all these .c files compile independently would be appreciated,
not sure that'll immediately solve your issue though.

> I do not mind doing some coding here to make this happen.  I am very
> close to a warning-free build, so I would definitely like to take care
> of this one.

You could slip something like it into the 'cleanup' if nothing else
presents itself :-)

Re: warning: ‘x86_get_event_constraints’ defined but not used

[Kevin Winchester]

On 23 August 2011 10:23, Peter Zijlstra <a.p.zijlstra@chello.nl> wrote:
> On Mon, 2011-08-15 at 20:40 -0300, Kevin Winchester wrote:
>> Is there are reason why those C
>> files are included directly in another C file rather than being linked
>> together?
>
> Sheer suckyness on my part.. we needed a 'quick' fix to make these
> CPU_SUP_* things build at all and never got around to cleaning it up
> afterwards.
>
>> Or is there some other way the code could be improved and
>> also remove the warning at the same time?
>
> creation of a arch/x86/kernel/cpu/perf_event.h and enough code massage
> to make all these .c files compile independently would be appreciated,
> not sure that'll immediately solve your issue though.
>
>> I do not mind doing some coding here to make this happen.  I am very
>> close to a warning-free build, so I would definitely like to take care
>> of this one.
>
> You could slip something like it into the 'cleanup' if nothing else
> presents itself :-)
>

Sure, I'll give the above suggestion a try and see if I can ensure
that the function is not compiled unless it is needed.

Kevin

[RFC PATCH] x86: perf: Clean up perf_event cpu code

[Kevin Winchester]

The CPU support for perf events on x86 was implemented via included C files
with #ifdefs.  Clean this up by creating a new header file and compiling
the vendor-specific files as needed.

Signed-off-by: Kevin Winchester <kjwinchester@gmail.com>
---

This is the first time I've attempted a patch this large.  Should I break it up
into smaller pieces?  Or does that requirement not really apply since it's just
a cleanup patch?

Also, it is based on Linus' tree, should I base it on tip instead?



 arch/x86/kernel/cpu/Makefile               |    5 +
 arch/x86/kernel/cpu/perf_event.c           |  367 +--------------------
 arch/x86/kernel/cpu/perf_event.h           |  491 ++++++++++++++++++++++++++++
 arch/x86/kernel/cpu/perf_event_amd.c       |   18 +-
 arch/x86/kernel/cpu/perf_event_intel.c     |   52 ++--
 arch/x86/kernel/cpu/perf_event_intel_ds.c  |   79 ++---
 arch/x86/kernel/cpu/perf_event_intel_lbr.c |   28 +-
 arch/x86/kernel/cpu/perf_event_p4.c        |    9 +-
 arch/x86/kernel/cpu/perf_event_p6.c        |    9 +-
 9 files changed, 600 insertions(+), 458 deletions(-)

diff --git a/arch/x86/kernel/cpu/Makefile b/arch/x86/kernel/cpu/Makefile
index 6042981..1044fd7 100644
--- a/arch/x86/kernel/cpu/Makefile
+++ b/arch/x86/kernel/cpu/Makefile
@@ -28,6 +28,11 @@ obj-$(CONFIG_CPU_SUP_UMC_32)		+= umc.o
 
 obj-$(CONFIG_PERF_EVENTS)		+= perf_event.o
 
+ifdef CONFIG_PERF_EVENTS
+obj-$(CONFIG_CPU_SUP_AMD)		+= perf_event_amd.o
+obj-$(CONFIG_CPU_SUP_INTEL)		+= perf_event_p6.o perf_event_p4.o perf_event_intel_lbr.o perf_event_intel_ds.o perf_event_intel.o
+endif
+
 obj-$(CONFIG_X86_MCE)			+= mcheck/
 obj-$(CONFIG_MTRR)			+= mtrr/
 
diff --git a/arch/x86/kernel/cpu/perf_event.c b/arch/x86/kernel/cpu/perf_event.c
index 4ee3abf..44574ae 100644
--- a/arch/x86/kernel/cpu/perf_event.c
+++ b/arch/x86/kernel/cpu/perf_event.c
@@ -32,6 +32,8 @@
 #include <asm/smp.h>
 #include <asm/alternative.h>
 
+#include "perf_event.h"
+
 #if 0
 #undef wrmsrl
 #define wrmsrl(msr, val) 					\
@@ -43,283 +45,17 @@ do {								\
 } while (0)
 #endif
 
-/*
- *          |   NHM/WSM    |      SNB     |
- * register -------------------------------
- *          |  HT  | no HT |  HT  | no HT |
- *-----------------------------------------
- * offcore  | core | core  | cpu  | core  |
- * lbr_sel  | core | core  | cpu  | core  |
- * ld_lat   | cpu  | core  | cpu  | core  |
- *-----------------------------------------
- *
- * Given that there is a small number of shared regs,
- * we can pre-allocate their slot in the per-cpu
- * per-core reg tables.
- */
-enum extra_reg_type {
-	EXTRA_REG_NONE  = -1,	/* not used */
-
-	EXTRA_REG_RSP_0 = 0,	/* offcore_response_0 */
-	EXTRA_REG_RSP_1 = 1,	/* offcore_response_1 */
-
-	EXTRA_REG_MAX		/* number of entries needed */
-};
-
-struct event_constraint {
-	union {
-		unsigned long	idxmsk[BITS_TO_LONGS(X86_PMC_IDX_MAX)];
-		u64		idxmsk64;
-	};
-	u64	code;
-	u64	cmask;
-	int	weight;
-};
-
-struct amd_nb {
-	int nb_id;  /* NorthBridge id */
-	int refcnt; /* reference count */
-	struct perf_event *owners[X86_PMC_IDX_MAX];
-	struct event_constraint event_constraints[X86_PMC_IDX_MAX];
-};
-
-struct intel_percore;
-
-#define MAX_LBR_ENTRIES		16
-
-struct cpu_hw_events {
-	/*
-	 * Generic x86 PMC bits
-	 */
-	struct perf_event	*events[X86_PMC_IDX_MAX]; /* in counter order */
-	unsigned long		active_mask[BITS_TO_LONGS(X86_PMC_IDX_MAX)];
-	unsigned long		running[BITS_TO_LONGS(X86_PMC_IDX_MAX)];
-	int			enabled;
-
-	int			n_events;
-	int			n_added;
-	int			n_txn;
-	int			assign[X86_PMC_IDX_MAX]; /* event to counter assignment */
-	u64			tags[X86_PMC_IDX_MAX];
-	struct perf_event	*event_list[X86_PMC_IDX_MAX]; /* in enabled order */
-
-	unsigned int		group_flag;
-
-	/*
-	 * Intel DebugStore bits
-	 */
-	struct debug_store	*ds;
-	u64			pebs_enabled;
-
-	/*
-	 * Intel LBR bits
-	 */
-	int				lbr_users;
-	void				*lbr_context;
-	struct perf_branch_stack	lbr_stack;
-	struct perf_branch_entry	lbr_entries[MAX_LBR_ENTRIES];
-
-	/*
-	 * manage shared (per-core, per-cpu) registers
-	 * used on Intel NHM/WSM/SNB
-	 */
-	struct intel_shared_regs	*shared_regs;
-
-	/*
-	 * AMD specific bits
-	 */
-	struct amd_nb		*amd_nb;
-};
-
-#define __EVENT_CONSTRAINT(c, n, m, w) {\
-	{ .idxmsk64 = (n) },		\
-	.code = (c),			\
-	.cmask = (m),			\
-	.weight = (w),			\
-}
-
-#define EVENT_CONSTRAINT(c, n, m)	\
-	__EVENT_CONSTRAINT(c, n, m, HWEIGHT(n))
-
-/*
- * Constraint on the Event code.
- */
-#define INTEL_EVENT_CONSTRAINT(c, n)	\
-	EVENT_CONSTRAINT(c, n, ARCH_PERFMON_EVENTSEL_EVENT)
-
-/*
- * Constraint on the Event code + UMask + fixed-mask
- *
- * filter mask to validate fixed counter events.
- * the following filters disqualify for fixed counters:
- *  - inv
- *  - edge
- *  - cnt-mask
- *  The other filters are supported by fixed counters.
- *  The any-thread option is supported starting with v3.
- */
-#define FIXED_EVENT_CONSTRAINT(c, n)	\
-	EVENT_CONSTRAINT(c, (1ULL << (32+n)), X86_RAW_EVENT_MASK)
-
-/*
- * Constraint on the Event code + UMask
- */
-#define INTEL_UEVENT_CONSTRAINT(c, n)	\
-	EVENT_CONSTRAINT(c, n, INTEL_ARCH_EVENT_MASK)
-
-#define EVENT_CONSTRAINT_END		\
-	EVENT_CONSTRAINT(0, 0, 0)
-
-#define for_each_event_constraint(e, c)	\
-	for ((e) = (c); (e)->weight; (e)++)
-
-/*
- * Per register state.
- */
-struct er_account {
-	raw_spinlock_t		lock;	/* per-core: protect structure */
-	u64			config;	/* extra MSR config */
-	u64			reg;	/* extra MSR number */
-	atomic_t		ref;	/* reference count */
-};
-
-/*
- * Extra registers for specific events.
- *
- * Some events need large masks and require external MSRs.
- * Those extra MSRs end up being shared for all events on
- * a PMU and sometimes between PMU of sibling HT threads.
- * In either case, the kernel needs to handle conflicting
- * accesses to those extra, shared, regs. The data structure
- * to manage those registers is stored in cpu_hw_event.
- */
-struct extra_reg {
-	unsigned int		event;
-	unsigned int		msr;
-	u64			config_mask;
-	u64			valid_mask;
-	int			idx;  /* per_xxx->regs[] reg index */
-};
-
-#define EVENT_EXTRA_REG(e, ms, m, vm, i) {	\
-	.event = (e),		\
-	.msr = (ms),		\
-	.config_mask = (m),	\
-	.valid_mask = (vm),	\
-	.idx = EXTRA_REG_##i	\
-	}
-
-#define INTEL_EVENT_EXTRA_REG(event, msr, vm, idx)	\
-	EVENT_EXTRA_REG(event, msr, ARCH_PERFMON_EVENTSEL_EVENT, vm, idx)
-
-#define EVENT_EXTRA_END EVENT_EXTRA_REG(0, 0, 0, 0, RSP_0)
+struct x86_pmu x86_pmu __read_mostly;
 
-union perf_capabilities {
-	struct {
-		u64	lbr_format    : 6;
-		u64	pebs_trap     : 1;
-		u64	pebs_arch_reg : 1;
-		u64	pebs_format   : 4;
-		u64	smm_freeze    : 1;
-	};
-	u64	capabilities;
-};
-
-/*
- * struct x86_pmu - generic x86 pmu
- */
-struct x86_pmu {
-	/*
-	 * Generic x86 PMC bits
-	 */
-	const char	*name;
-	int		version;
-	int		(*handle_irq)(struct pt_regs *);
-	void		(*disable_all)(void);
-	void		(*enable_all)(int added);
-	void		(*enable)(struct perf_event *);
-	void		(*disable)(struct perf_event *);
-	int		(*hw_config)(struct perf_event *event);
-	int		(*schedule_events)(struct cpu_hw_events *cpuc, int n, int *assign);
-	unsigned	eventsel;
-	unsigned	perfctr;
-	u64		(*event_map)(int);
-	int		max_events;
-	int		num_counters;
-	int		num_counters_fixed;
-	int		cntval_bits;
-	u64		cntval_mask;
-	int		apic;
-	u64		max_period;
-	struct event_constraint *
-			(*get_event_constraints)(struct cpu_hw_events *cpuc,
-						 struct perf_event *event);
-
-	void		(*put_event_constraints)(struct cpu_hw_events *cpuc,
-						 struct perf_event *event);
-	struct event_constraint *event_constraints;
-	void		(*quirks)(void);
-	int		perfctr_second_write;
-
-	int		(*cpu_prepare)(int cpu);
-	void		(*cpu_starting)(int cpu);
-	void		(*cpu_dying)(int cpu);
-	void		(*cpu_dead)(int cpu);
-
-	/*
-	 * Intel Arch Perfmon v2+
-	 */
-	u64			intel_ctrl;
-	union perf_capabilities intel_cap;
-
-	/*
-	 * Intel DebugStore bits
-	 */
-	int		bts, pebs;
-	int		bts_active, pebs_active;
-	int		pebs_record_size;
-	void		(*drain_pebs)(struct pt_regs *regs);
-	struct event_constraint *pebs_constraints;
-
-	/*
-	 * Intel LBR
-	 */
-	unsigned long	lbr_tos, lbr_from, lbr_to; /* MSR base regs       */
-	int		lbr_nr;			   /* hardware stack size */
-
-	/*
-	 * Extra registers for events
-	 */
-	struct extra_reg *extra_regs;
-	unsigned int er_flags;
-};
-
-#define ERF_NO_HT_SHARING	1
-#define ERF_HAS_RSP_1		2
-
-static struct x86_pmu x86_pmu __read_mostly;
-
-static DEFINE_PER_CPU(struct cpu_hw_events, cpu_hw_events) = {
+DEFINE_PER_CPU(struct cpu_hw_events, cpu_hw_events) = {
 	.enabled = 1,
 };
 
-static int x86_perf_event_set_period(struct perf_event *event);
-
-/*
- * Generalized hw caching related hw_event table, filled
- * in on a per model basis. A value of 0 means
- * 'not supported', -1 means 'hw_event makes no sense on
- * this CPU', any other value means the raw hw_event
- * ID.
- */
-
-#define C(x) PERF_COUNT_HW_CACHE_##x
-
-static u64 __read_mostly hw_cache_event_ids
+u64 __read_mostly hw_cache_event_ids
 				[PERF_COUNT_HW_CACHE_MAX]
 				[PERF_COUNT_HW_CACHE_OP_MAX]
 				[PERF_COUNT_HW_CACHE_RESULT_MAX];
-static u64 __read_mostly hw_cache_extra_regs
+u64 __read_mostly hw_cache_extra_regs
 				[PERF_COUNT_HW_CACHE_MAX]
 				[PERF_COUNT_HW_CACHE_OP_MAX]
 				[PERF_COUNT_HW_CACHE_RESULT_MAX];
@@ -329,8 +65,7 @@ static u64 __read_mostly hw_cache_extra_regs
  * Can only be executed on the CPU where the event is active.
  * Returns the delta events processed.
  */
-static u64
-x86_perf_event_update(struct perf_event *event)
+u64 x86_perf_event_update(struct perf_event *event)
 {
 	struct hw_perf_event *hwc = &event->hw;
 	int shift = 64 - x86_pmu.cntval_bits;
@@ -373,30 +108,6 @@ again:
 	return new_raw_count;
 }
 
-static inline int x86_pmu_addr_offset(int index)
-{
-	int offset;
-
-	/* offset = X86_FEATURE_PERFCTR_CORE ? index << 1 : index */
-	alternative_io(ASM_NOP2,
-		       "shll $1, %%eax",
-		       X86_FEATURE_PERFCTR_CORE,
-		       "=a" (offset),
-		       "a"  (index));
-
-	return offset;
-}
-
-static inline unsigned int x86_pmu_config_addr(int index)
-{
-	return x86_pmu.eventsel + x86_pmu_addr_offset(index);
-}
-
-static inline unsigned int x86_pmu_event_addr(int index)
-{
-	return x86_pmu.perfctr + x86_pmu_addr_offset(index);
-}
-
 /*
  * Find and validate any extra registers to set up.
  */
@@ -532,9 +243,6 @@ msr_fail:
 	return false;
 }
 
-static void reserve_ds_buffers(void);
-static void release_ds_buffers(void);
-
 static void hw_perf_event_destroy(struct perf_event *event)
 {
 	if (atomic_dec_and_mutex_lock(&active_events, &pmc_reserve_mutex)) {
@@ -583,7 +291,7 @@ set_ext_hw_attr(struct hw_perf_event *hwc, struct perf_event *event)
 	return x86_pmu_extra_regs(val, event);
 }
 
-static int x86_setup_perfctr(struct perf_event *event)
+int x86_setup_perfctr(struct perf_event *event)
 {
 	struct perf_event_attr *attr = &event->attr;
 	struct hw_perf_event *hwc = &event->hw;
@@ -647,7 +355,7 @@ static int x86_setup_perfctr(struct perf_event *event)
 	return 0;
 }
 
-static int x86_pmu_hw_config(struct perf_event *event)
+int x86_pmu_hw_config(struct perf_event *event)
 {
 	if (event->attr.precise_ip) {
 		int precise = 0;
@@ -723,7 +431,7 @@ static int __x86_pmu_event_init(struct perf_event *event)
 	return x86_pmu.hw_config(event);
 }
 
-static void x86_pmu_disable_all(void)
+void x86_pmu_disable_all(void)
 {
 	struct cpu_hw_events *cpuc = &__get_cpu_var(cpu_hw_events);
 	int idx;
@@ -758,15 +466,7 @@ static void x86_pmu_disable(struct pmu *pmu)
 	x86_pmu.disable_all();
 }
 
-static inline void __x86_pmu_enable_event(struct hw_perf_event *hwc,
-					  u64 enable_mask)
-{
-	if (hwc->extra_reg.reg)
-		wrmsrl(hwc->extra_reg.reg, hwc->extra_reg.config);
-	wrmsrl(hwc->config_base, hwc->config | enable_mask);
-}
-
-static void x86_pmu_enable_all(int added)
+void x86_pmu_enable_all(int added)
 {
 	struct cpu_hw_events *cpuc = &__get_cpu_var(cpu_hw_events);
 	int idx;
@@ -788,7 +488,7 @@ static inline int is_x86_event(struct perf_event *event)
 	return event->pmu == &pmu;
 }
 
-static int x86_schedule_events(struct cpu_hw_events *cpuc, int n, int *assign)
+int x86_schedule_events(struct cpu_hw_events *cpuc, int n, int *assign)
 {
 	struct event_constraint *c, *constraints[X86_PMC_IDX_MAX];
 	unsigned long used_mask[BITS_TO_LONGS(X86_PMC_IDX_MAX)];
@@ -959,7 +659,6 @@ static inline int match_prev_assignment(struct hw_perf_event *hwc,
 }
 
 static void x86_pmu_start(struct perf_event *event, int flags);
-static void x86_pmu_stop(struct perf_event *event, int flags);
 
 static void x86_pmu_enable(struct pmu *pmu)
 {
@@ -1031,21 +730,13 @@ static void x86_pmu_enable(struct pmu *pmu)
 	x86_pmu.enable_all(added);
 }
 
-static inline void x86_pmu_disable_event(struct perf_event *event)
-{
-	struct hw_perf_event *hwc = &event->hw;
-
-	wrmsrl(hwc->config_base, hwc->config);
-}
-
 static DEFINE_PER_CPU(u64 [X86_PMC_IDX_MAX], pmc_prev_left);
 
 /*
  * Set the next IRQ period, based on the hwc->period_left value.
  * To be called with the event disabled in hw:
  */
-static int
-x86_perf_event_set_period(struct perf_event *event)
+int x86_perf_event_set_period(struct perf_event *event)
 {
 	struct hw_perf_event *hwc = &event->hw;
 	s64 left = local64_read(&hwc->period_left);
@@ -1105,7 +796,7 @@ x86_perf_event_set_period(struct perf_event *event)
 	return ret;
 }
 
-static void x86_pmu_enable_event(struct perf_event *event)
+void x86_pmu_enable_event(struct perf_event *event)
 {
 	if (__this_cpu_read(cpu_hw_events.enabled))
 		__x86_pmu_enable_event(&event->hw,
@@ -1244,7 +935,7 @@ void perf_event_print_debug(void)
 	local_irq_restore(flags);
 }
 
-static void x86_pmu_stop(struct perf_event *event, int flags)
+void x86_pmu_stop(struct perf_event *event, int flags)
 {
 	struct cpu_hw_events *cpuc = &__get_cpu_var(cpu_hw_events);
 	struct hw_perf_event *hwc = &event->hw;
@@ -1297,7 +988,7 @@ static void x86_pmu_del(struct perf_event *event, int flags)
 	perf_event_update_userpage(event);
 }
 
-static int x86_pmu_handle_irq(struct pt_regs *regs)
+int x86_pmu_handle_irq(struct pt_regs *regs)
 {
 	struct perf_sample_data data;
 	struct cpu_hw_events *cpuc;
@@ -1437,30 +1128,8 @@ static __read_mostly struct notifier_block perf_event_nmi_notifier = {
 	.priority		= NMI_LOCAL_LOW_PRIOR,
 };
 
-static struct event_constraint unconstrained;
-static struct event_constraint emptyconstraint;
-
-static struct event_constraint *
-x86_get_event_constraints(struct cpu_hw_events *cpuc, struct perf_event *event)
-{
-	struct event_constraint *c;
-
-	if (x86_pmu.event_constraints) {
-		for_each_event_constraint(c, x86_pmu.event_constraints) {
-			if ((event->hw.config & c->cmask) == c->code)
-				return c;
-		}
-	}
-
-	return &unconstrained;
-}
-
-#include "perf_event_amd.c"
-#include "perf_event_p6.c"
-#include "perf_event_p4.c"
-#include "perf_event_intel_lbr.c"
-#include "perf_event_intel_ds.c"
-#include "perf_event_intel.c"
+struct event_constraint emptyconstraint;
+struct event_constraint unconstrained;
 
 static int __cpuinit
 x86_pmu_notifier(struct notifier_block *self, unsigned long action, void *hcpu)
diff --git a/arch/x86/kernel/cpu/perf_event.h b/arch/x86/kernel/cpu/perf_event.h
new file mode 100644
index 0000000..9381f32
--- /dev/null
+++ b/arch/x86/kernel/cpu/perf_event.h
@@ -0,0 +1,491 @@
+/*
+ * Performance events x86 architecture header
+ *
+ *  Copyright (C) 2008 Thomas Gleixner <tglx@linutronix.de>
+ *  Copyright (C) 2008-2009 Red Hat, Inc., Ingo Molnar
+ *  Copyright (C) 2009 Jaswinder Singh Rajput
+ *  Copyright (C) 2009 Advanced Micro Devices, Inc., Robert Richter
+ *  Copyright (C) 2008-2009 Red Hat, Inc., Peter Zijlstra <pzijlstr@redhat.com>
+ *  Copyright (C) 2009 Intel Corporation, <markus.t.metzger@intel.com>
+ *  Copyright (C) 2009 Google, Inc., Stephane Eranian
+ *
+ *  For licencing details see kernel-base/COPYING
+ */
+
+#include <linux/perf_event.h>
+
+/*
+ *          |   NHM/WSM    |      SNB     |
+ * register -------------------------------
+ *          |  HT  | no HT |  HT  | no HT |
+ *-----------------------------------------
+ * offcore  | core | core  | cpu  | core  |
+ * lbr_sel  | core | core  | cpu  | core  |
+ * ld_lat   | cpu  | core  | cpu  | core  |
+ *-----------------------------------------
+ *
+ * Given that there is a small number of shared regs,
+ * we can pre-allocate their slot in the per-cpu
+ * per-core reg tables.
+ */
+enum extra_reg_type {
+	EXTRA_REG_NONE  = -1,	/* not used */
+
+	EXTRA_REG_RSP_0 = 0,	/* offcore_response_0 */
+	EXTRA_REG_RSP_1 = 1,	/* offcore_response_1 */
+
+	EXTRA_REG_MAX		/* number of entries needed */
+};
+
+struct event_constraint {
+	union {
+		unsigned long	idxmsk[BITS_TO_LONGS(X86_PMC_IDX_MAX)];
+		u64		idxmsk64;
+	};
+	u64	code;
+	u64	cmask;
+	int	weight;
+};
+
+struct amd_nb {
+	int nb_id;  /* NorthBridge id */
+	int refcnt; /* reference count */
+	struct perf_event *owners[X86_PMC_IDX_MAX];
+	struct event_constraint event_constraints[X86_PMC_IDX_MAX];
+};
+
+/* The maximal number of PEBS events: */
+#define MAX_PEBS_EVENTS		4
+
+/*
+ * A debug store configuration.
+ *
+ * We only support architectures that use 64bit fields.
+ */
+struct debug_store {
+	u64	bts_buffer_base;
+	u64	bts_index;
+	u64	bts_absolute_maximum;
+	u64	bts_interrupt_threshold;
+	u64	pebs_buffer_base;
+	u64	pebs_index;
+	u64	pebs_absolute_maximum;
+	u64	pebs_interrupt_threshold;
+	u64	pebs_event_reset[MAX_PEBS_EVENTS];
+};
+
+/*
+ * Per register state.
+ */
+struct er_account {
+	raw_spinlock_t		lock;	/* per-core: protect structure */
+	u64                 config;	/* extra MSR config */
+	u64                 reg;	/* extra MSR number */
+	atomic_t            ref;	/* reference count */
+};
+
+/*
+ * Per core/cpu state
+ *
+ * Used to coordinate shared registers between HT threads or
+ * among events on a single PMU.
+ */
+struct intel_shared_regs {
+	struct er_account       regs[EXTRA_REG_MAX];
+	int                     refcnt;		/* per-core: #HT threads */
+	unsigned                core_id;	/* per-core: core id */
+};
+
+#define MAX_LBR_ENTRIES		16
+
+struct cpu_hw_events {
+	/*
+	 * Generic x86 PMC bits
+	 */
+	struct perf_event	*events[X86_PMC_IDX_MAX]; /* in counter order */
+	unsigned long		active_mask[BITS_TO_LONGS(X86_PMC_IDX_MAX)];
+	unsigned long		running[BITS_TO_LONGS(X86_PMC_IDX_MAX)];
+	int			enabled;
+
+	int			n_events;
+	int			n_added;
+	int			n_txn;
+	int			assign[X86_PMC_IDX_MAX]; /* event to counter assignment */
+	u64			tags[X86_PMC_IDX_MAX];
+	struct perf_event	*event_list[X86_PMC_IDX_MAX]; /* in enabled order */
+
+	unsigned int		group_flag;
+
+	/*
+	 * Intel DebugStore bits
+	 */
+	struct debug_store	*ds;
+	u64			pebs_enabled;
+
+	/*
+	 * Intel LBR bits
+	 */
+	int				lbr_users;
+	void				*lbr_context;
+	struct perf_branch_stack	lbr_stack;
+	struct perf_branch_entry	lbr_entries[MAX_LBR_ENTRIES];
+
+	/*
+	 * manage shared (per-core, per-cpu) registers
+	 * used on Intel NHM/WSM/SNB
+	 */
+	struct intel_shared_regs	*shared_regs;
+
+	/*
+	 * AMD specific bits
+	 */
+	struct amd_nb		*amd_nb;
+};
+
+#define __EVENT_CONSTRAINT(c, n, m, w) {\
+	{ .idxmsk64 = (n) },		\
+	.code = (c),			\
+	.cmask = (m),			\
+	.weight = (w),			\
+}
+
+#define EVENT_CONSTRAINT(c, n, m)	\
+	__EVENT_CONSTRAINT(c, n, m, HWEIGHT(n))
+
+/*
+ * Constraint on the Event code.
+ */
+#define INTEL_EVENT_CONSTRAINT(c, n)	\
+	EVENT_CONSTRAINT(c, n, ARCH_PERFMON_EVENTSEL_EVENT)
+
+/*
+ * Constraint on the Event code + UMask + fixed-mask
+ *
+ * filter mask to validate fixed counter events.
+ * the following filters disqualify for fixed counters:
+ *  - inv
+ *  - edge
+ *  - cnt-mask
+ *  The other filters are supported by fixed counters.
+ *  The any-thread option is supported starting with v3.
+ */
+#define FIXED_EVENT_CONSTRAINT(c, n)	\
+	EVENT_CONSTRAINT(c, (1ULL << (32+n)), X86_RAW_EVENT_MASK)
+
+/*
+ * Constraint on the Event code + UMask
+ */
+#define INTEL_UEVENT_CONSTRAINT(c, n)	\
+	EVENT_CONSTRAINT(c, n, INTEL_ARCH_EVENT_MASK)
+
+#define EVENT_CONSTRAINT_END		\
+	EVENT_CONSTRAINT(0, 0, 0)
+
+#define for_each_event_constraint(e, c)	\
+	for ((e) = (c); (e)->weight; (e)++)
+
+/*
+ * Extra registers for specific events.
+ *
+ * Some events need large masks and require external MSRs.
+ * Those extra MSRs end up being shared for all events on
+ * a PMU and sometimes between PMU of sibling HT threads.
+ * In either case, the kernel needs to handle conflicting
+ * accesses to those extra, shared, regs. The data structure
+ * to manage those registers is stored in cpu_hw_event.
+ */
+struct extra_reg {
+	unsigned int		event;
+	unsigned int		msr;
+	u64			config_mask;
+	u64			valid_mask;
+	int			idx;  /* per_xxx->regs[] reg index */
+};
+
+#define EVENT_EXTRA_REG(e, ms, m, vm, i) {	\
+	.event = (e),		\
+	.msr = (ms),		\
+	.config_mask = (m),	\
+	.valid_mask = (vm),	\
+	.idx = EXTRA_REG_##i	\
+	}
+
+#define INTEL_EVENT_EXTRA_REG(event, msr, vm, idx)	\
+	EVENT_EXTRA_REG(event, msr, ARCH_PERFMON_EVENTSEL_EVENT, vm, idx)
+
+#define EVENT_EXTRA_END EVENT_EXTRA_REG(0, 0, 0, 0, RSP_0)
+
+union perf_capabilities {
+	struct {
+		u64	lbr_format:6;
+		u64	pebs_trap:1;
+		u64	pebs_arch_reg:1;
+		u64	pebs_format:4;
+		u64	smm_freeze:1;
+	};
+	u64	capabilities;
+};
+
+/*
+ * struct x86_pmu - generic x86 pmu
+ */
+struct x86_pmu {
+	/*
+	 * Generic x86 PMC bits
+	 */
+	const char	*name;
+	int		version;
+	int		(*handle_irq)(struct pt_regs *);
+	void		(*disable_all)(void);
+	void		(*enable_all)(int added);
+	void		(*enable)(struct perf_event *);
+	void		(*disable)(struct perf_event *);
+	int		(*hw_config)(struct perf_event *event);
+	int		(*schedule_events)(struct cpu_hw_events *cpuc, int n, int *assign);
+	unsigned	eventsel;
+	unsigned	perfctr;
+	u64		(*event_map)(int);
+	int		max_events;
+	int		num_counters;
+	int		num_counters_fixed;
+	int		cntval_bits;
+	u64		cntval_mask;
+	int		apic;
+	u64		max_period;
+	struct event_constraint *
+			(*get_event_constraints)(struct cpu_hw_events *cpuc,
+						 struct perf_event *event);
+
+	void		(*put_event_constraints)(struct cpu_hw_events *cpuc,
+						 struct perf_event *event);
+	struct event_constraint *event_constraints;
+	void		(*quirks)(void);
+	int		perfctr_second_write;
+
+	int		(*cpu_prepare)(int cpu);
+	void		(*cpu_starting)(int cpu);
+	void		(*cpu_dying)(int cpu);
+	void		(*cpu_dead)(int cpu);
+
+	/*
+	 * Intel Arch Perfmon v2+
+	 */
+	u64			intel_ctrl;
+	union perf_capabilities intel_cap;
+
+	/*
+	 * Intel DebugStore bits
+	 */
+	int		bts, pebs;
+	int		bts_active, pebs_active;
+	int		pebs_record_size;
+	void		(*drain_pebs)(struct pt_regs *regs);
+	struct event_constraint *pebs_constraints;
+
+	/*
+	 * Intel LBR
+	 */
+	unsigned long	lbr_tos, lbr_from, lbr_to; /* MSR base regs       */
+	int		lbr_nr;			   /* hardware stack size */
+
+	/*
+	 * Extra registers for events
+	 */
+	struct extra_reg *extra_regs;
+	unsigned int er_flags;
+};
+
+#define ERF_NO_HT_SHARING	1
+#define ERF_HAS_RSP_1		2
+
+extern struct x86_pmu x86_pmu __read_mostly;
+
+DECLARE_PER_CPU(struct cpu_hw_events, cpu_hw_events);
+
+int x86_perf_event_set_period(struct perf_event *event);
+
+/*
+ * Generalized hw caching related hw_event table, filled
+ * in on a per model basis. A value of 0 means
+ * 'not supported', -1 means 'hw_event makes no sense on
+ * this CPU', any other value means the raw hw_event
+ * ID.
+ */
+
+#define C(x) PERF_COUNT_HW_CACHE_##x
+
+extern u64 __read_mostly hw_cache_event_ids
+				[PERF_COUNT_HW_CACHE_MAX]
+				[PERF_COUNT_HW_CACHE_OP_MAX]
+				[PERF_COUNT_HW_CACHE_RESULT_MAX];
+extern u64 __read_mostly hw_cache_extra_regs
+				[PERF_COUNT_HW_CACHE_MAX]
+				[PERF_COUNT_HW_CACHE_OP_MAX]
+				[PERF_COUNT_HW_CACHE_RESULT_MAX];
+
+u64 x86_perf_event_update(struct perf_event *event);
+
+static inline int x86_pmu_addr_offset(int index)
+{
+	int offset;
+
+	/* offset = X86_FEATURE_PERFCTR_CORE ? index << 1 : index */
+	alternative_io(ASM_NOP2,
+		       "shll $1, %%eax",
+		       X86_FEATURE_PERFCTR_CORE,
+		       "=a" (offset),
+		       "a"  (index));
+
+	return offset;
+}
+
+static inline unsigned int x86_pmu_config_addr(int index)
+{
+	return x86_pmu.eventsel + x86_pmu_addr_offset(index);
+}
+
+static inline unsigned int x86_pmu_event_addr(int index)
+{
+	return x86_pmu.perfctr + x86_pmu_addr_offset(index);
+}
+
+int x86_setup_perfctr(struct perf_event *event);
+
+int x86_pmu_hw_config(struct perf_event *event);
+
+void x86_pmu_disable_all(void);
+
+static inline void __x86_pmu_enable_event(struct hw_perf_event *hwc,
+					  u64 enable_mask)
+{
+	if (hwc->extra_reg.reg)
+		wrmsrl(hwc->extra_reg.reg, hwc->extra_reg.config);
+	wrmsrl(hwc->config_base, hwc->config | enable_mask);
+}
+
+void x86_pmu_enable_all(int added);
+
+int x86_schedule_events(struct cpu_hw_events *cpuc, int n, int *assign);
+
+void x86_pmu_stop(struct perf_event *event, int flags);
+
+static inline void x86_pmu_disable_event(struct perf_event *event)
+{
+	struct hw_perf_event *hwc = &event->hw;
+
+	wrmsrl(hwc->config_base, hwc->config);
+}
+
+void x86_pmu_enable_event(struct perf_event *event);
+
+int x86_pmu_handle_irq(struct pt_regs *regs);
+
+extern struct event_constraint emptyconstraint;
+
+extern struct event_constraint unconstrained;
+
+#ifdef CONFIG_CPU_SUP_AMD
+
+int amd_pmu_init(void);
+
+#else /* CONFIG_CPU_SUP_AMD */
+
+static inline int amd_pmu_init(void)
+{
+	return 0;
+}
+
+#endif /* CONFIG_CPU_SUP_AMD */
+
+#ifdef CONFIG_CPU_SUP_INTEL
+
+int intel_pmu_save_and_restart(struct perf_event *event);
+
+struct event_constraint *
+x86_get_event_constraints(struct cpu_hw_events *cpuc, struct perf_event *event);
+
+struct intel_shared_regs *allocate_shared_regs(int cpu);
+
+int intel_pmu_init(void);
+
+void init_debug_store_on_cpu(int cpu);
+
+void fini_debug_store_on_cpu(int cpu);
+
+void release_ds_buffers(void);
+
+void reserve_ds_buffers(void);
+
+extern struct event_constraint bts_constraint;
+
+void intel_pmu_enable_bts(u64 config);
+
+void intel_pmu_disable_bts(void);
+
+int intel_pmu_drain_bts_buffer(void);
+
+extern struct event_constraint intel_core2_pebs_event_constraints[];
+
+extern struct event_constraint intel_atom_pebs_event_constraints[];
+
+extern struct event_constraint intel_nehalem_pebs_event_constraints[];
+
+extern struct event_constraint intel_westmere_pebs_event_constraints[];
+
+extern struct event_constraint intel_snb_pebs_event_constraints[];
+
+struct event_constraint *intel_pebs_constraints(struct perf_event *event);
+
+void intel_pmu_pebs_enable(struct perf_event *event);
+
+void intel_pmu_pebs_disable(struct perf_event *event);
+
+void intel_pmu_pebs_enable_all(void);
+
+void intel_pmu_pebs_disable_all(void);
+
+void intel_ds_init(void);
+
+void intel_pmu_lbr_reset(void);
+
+void intel_pmu_lbr_enable(struct perf_event *event);
+
+void intel_pmu_lbr_disable(struct perf_event *event);
+
+void intel_pmu_lbr_enable_all(void);
+
+void intel_pmu_lbr_disable_all(void);
+
+void intel_pmu_lbr_read(void);
+
+void intel_pmu_lbr_init_core(void);
+
+void intel_pmu_lbr_init_nhm(void);
+
+void intel_pmu_lbr_init_atom(void);
+
+int p4_pmu_init(void);
+
+int p6_pmu_init(void);
+
+#else /* CONFIG_CPU_SUP_INTEL */
+
+static inline void reserve_ds_buffers(void)
+{
+}
+
+static inline void release_ds_buffers(void)
+{
+}
+
+static inline int intel_pmu_init(void)
+{
+	return 0;
+}
+
+static inline struct intel_shared_regs *allocate_shared_regs(int cpu)
+{
+	return NULL;
+}
+
+#endif /* CONFIG_CPU_SUP_INTEL */
diff --git a/arch/x86/kernel/cpu/perf_event_amd.c b/arch/x86/kernel/cpu/perf_event_amd.c
index 941caa2..9a526ee 100644
--- a/arch/x86/kernel/cpu/perf_event_amd.c
+++ b/arch/x86/kernel/cpu/perf_event_amd.c
@@ -1,4 +1,9 @@
-#ifdef CONFIG_CPU_SUP_AMD
+#include <linux/perf_event.h>
+#include <linux/types.h>
+#include <linux/init.h>
+#include <linux/slab.h>
+
+#include "perf_event.h"
 
 static __initconst const u64 amd_hw_cache_event_ids
 				[PERF_COUNT_HW_CACHE_MAX]
@@ -573,7 +578,7 @@ static __initconst const struct x86_pmu amd_pmu_f15h = {
 #endif
 };
 
-static __init int amd_pmu_init(void)
+__init int amd_pmu_init(void)
 {
 	/* Performance-monitoring supported from K7 and later: */
 	if (boot_cpu_data.x86 < 6)
@@ -602,12 +607,3 @@ static __init int amd_pmu_init(void)
 
 	return 0;
 }
-
-#else /* CONFIG_CPU_SUP_AMD */
-
-static int amd_pmu_init(void)
-{
-	return 0;
-}
-
-#endif
diff --git a/arch/x86/kernel/cpu/perf_event_intel.c b/arch/x86/kernel/cpu/perf_event_intel.c
index f88af2c..277891d 100644
--- a/arch/x86/kernel/cpu/perf_event_intel.c
+++ b/arch/x86/kernel/cpu/perf_event_intel.c
@@ -1,16 +1,18 @@
-#ifdef CONFIG_CPU_SUP_INTEL
-
 /*
  * Per core/cpu state
  *
  * Used to coordinate shared registers between HT threads or
  * among events on a single PMU.
  */
-struct intel_shared_regs {
-	struct er_account       regs[EXTRA_REG_MAX];
-	int                     refcnt;		/* per-core: #HT threads */
-	unsigned                core_id;	/* per-core: core id */
-};
+
+#include <linux/stddef.h>
+#include <linux/types.h>
+#include <linux/init.h>
+#include <linux/slab.h>
+
+#include <asm/hardirq.h>
+
+#include "perf_event.h"
 
 /*
  * Intel PerfMon, used on Core and later.
@@ -945,7 +947,7 @@ static void intel_pmu_enable_event(struct perf_event *event)
  * Save and restart an expired event. Called by NMI contexts,
  * so it has to be careful about preempting normal event ops:
  */
-static int intel_pmu_save_and_restart(struct perf_event *event)
+int intel_pmu_save_and_restart(struct perf_event *event)
 {
 	x86_perf_event_update(event);
 	return x86_perf_event_set_period(event);
@@ -1197,6 +1199,21 @@ intel_shared_regs_constraints(struct cpu_hw_events *cpuc,
 	return c;
 }
 
+struct event_constraint *
+x86_get_event_constraints(struct cpu_hw_events *cpuc, struct perf_event *event)
+{
+	struct event_constraint *c;
+
+	if (x86_pmu.event_constraints) {
+		for_each_event_constraint(c, x86_pmu.event_constraints) {
+			if ((event->hw.config & c->cmask) == c->code)
+				return c;
+		}
+	}
+
+	return &unconstrained;
+}
+
 static struct event_constraint *
 intel_get_event_constraints(struct cpu_hw_events *cpuc, struct perf_event *event)
 {
@@ -1309,7 +1326,7 @@ static __initconst const struct x86_pmu core_pmu = {
 	.event_constraints	= intel_core_event_constraints,
 };
 
-static struct intel_shared_regs *allocate_shared_regs(int cpu)
+struct intel_shared_regs *allocate_shared_regs(int cpu)
 {
 	struct intel_shared_regs *regs;
 	int i;
@@ -1441,7 +1458,7 @@ static void intel_clovertown_quirks(void)
 	x86_pmu.pebs_constraints = NULL;
 }
 
-static __init int intel_pmu_init(void)
+__init int intel_pmu_init(void)
 {
 	union cpuid10_edx edx;
 	union cpuid10_eax eax;
@@ -1597,7 +1614,7 @@ static __init int intel_pmu_init(void)
 		intel_pmu_lbr_init_nhm();
 
 		x86_pmu.event_constraints = intel_snb_event_constraints;
-		x86_pmu.pebs_constraints = intel_snb_pebs_events;
+		x86_pmu.pebs_constraints = intel_snb_pebs_event_constraints;
 		x86_pmu.extra_regs = intel_snb_extra_regs;
 		/* all extra regs are per-cpu when HT is on */
 		x86_pmu.er_flags |= ERF_HAS_RSP_1;
@@ -1628,16 +1645,3 @@ static __init int intel_pmu_init(void)
 	}
 	return 0;
 }
-
-#else /* CONFIG_CPU_SUP_INTEL */
-
-static int intel_pmu_init(void)
-{
-	return 0;
-}
-
-static struct intel_shared_regs *allocate_shared_regs(int cpu)
-{
-	return NULL;
-}
-#endif /* CONFIG_CPU_SUP_INTEL */
diff --git a/arch/x86/kernel/cpu/perf_event_intel_ds.c b/arch/x86/kernel/cpu/perf_event_intel_ds.c
index 1b1ef3a..c0d238f 100644
--- a/arch/x86/kernel/cpu/perf_event_intel_ds.c
+++ b/arch/x86/kernel/cpu/perf_event_intel_ds.c
@@ -1,7 +1,10 @@
-#ifdef CONFIG_CPU_SUP_INTEL
+#include <linux/bitops.h>
+#include <linux/types.h>
+#include <linux/slab.h>
 
-/* The maximal number of PEBS events: */
-#define MAX_PEBS_EVENTS		4
+#include <asm/perf_event.h>
+
+#include "perf_event.h"
 
 /* The size of a BTS record in bytes: */
 #define BTS_RECORD_SIZE		24
@@ -37,24 +40,7 @@ struct pebs_record_nhm {
 	u64 status, dla, dse, lat;
 };
 
-/*
- * A debug store configuration.
- *
- * We only support architectures that use 64bit fields.
- */
-struct debug_store {
-	u64	bts_buffer_base;
-	u64	bts_index;
-	u64	bts_absolute_maximum;
-	u64	bts_interrupt_threshold;
-	u64	pebs_buffer_base;
-	u64	pebs_index;
-	u64	pebs_absolute_maximum;
-	u64	pebs_interrupt_threshold;
-	u64	pebs_event_reset[MAX_PEBS_EVENTS];
-};
-
-static void init_debug_store_on_cpu(int cpu)
+void init_debug_store_on_cpu(int cpu)
 {
 	struct debug_store *ds = per_cpu(cpu_hw_events, cpu).ds;
 
@@ -66,7 +52,7 @@ static void init_debug_store_on_cpu(int cpu)
 		     (u32)((u64)(unsigned long)ds >> 32));
 }
 
-static void fini_debug_store_on_cpu(int cpu)
+void fini_debug_store_on_cpu(int cpu)
 {
 	if (!per_cpu(cpu_hw_events, cpu).ds)
 		return;
@@ -175,7 +161,7 @@ static void release_ds_buffer(int cpu)
 	kfree(ds);
 }
 
-static void release_ds_buffers(void)
+void release_ds_buffers(void)
 {
 	int cpu;
 
@@ -194,7 +180,7 @@ static void release_ds_buffers(void)
 	put_online_cpus();
 }
 
-static void reserve_ds_buffers(void)
+void reserve_ds_buffers(void)
 {
 	int bts_err = 0, pebs_err = 0;
 	int cpu;
@@ -260,10 +246,10 @@ static void reserve_ds_buffers(void)
  * BTS
  */
 
-static struct event_constraint bts_constraint =
+struct event_constraint bts_constraint =
 	EVENT_CONSTRAINT(0, 1ULL << X86_PMC_IDX_FIXED_BTS, 0);
 
-static void intel_pmu_enable_bts(u64 config)
+void intel_pmu_enable_bts(u64 config)
 {
 	unsigned long debugctlmsr;
 
@@ -282,7 +268,7 @@ static void intel_pmu_enable_bts(u64 config)
 	update_debugctlmsr(debugctlmsr);
 }
 
-static void intel_pmu_disable_bts(void)
+void intel_pmu_disable_bts(void)
 {
 	struct cpu_hw_events *cpuc = &__get_cpu_var(cpu_hw_events);
 	unsigned long debugctlmsr;
@@ -299,7 +285,7 @@ static void intel_pmu_disable_bts(void)
 	update_debugctlmsr(debugctlmsr);
 }
 
-static int intel_pmu_drain_bts_buffer(void)
+int intel_pmu_drain_bts_buffer(void)
 {
 	struct cpu_hw_events *cpuc = &__get_cpu_var(cpu_hw_events);
 	struct debug_store *ds = cpuc->ds;
@@ -361,7 +347,7 @@ static int intel_pmu_drain_bts_buffer(void)
 /*
  * PEBS
  */
-static struct event_constraint intel_core2_pebs_event_constraints[] = {
+struct event_constraint intel_core2_pebs_event_constraints[] = {
 	INTEL_UEVENT_CONSTRAINT(0x00c0, 0x1), /* INST_RETIRED.ANY */
 	INTEL_UEVENT_CONSTRAINT(0xfec1, 0x1), /* X87_OPS_RETIRED.ANY */
 	INTEL_UEVENT_CONSTRAINT(0x00c5, 0x1), /* BR_INST_RETIRED.MISPRED */
@@ -370,14 +356,14 @@ static struct event_constraint intel_core2_pebs_event_constraints[] = {
 	EVENT_CONSTRAINT_END
 };
 
-static struct event_constraint intel_atom_pebs_event_constraints[] = {
+struct event_constraint intel_atom_pebs_event_constraints[] = {
 	INTEL_UEVENT_CONSTRAINT(0x00c0, 0x1), /* INST_RETIRED.ANY */
 	INTEL_UEVENT_CONSTRAINT(0x00c5, 0x1), /* MISPREDICTED_BRANCH_RETIRED */
 	INTEL_EVENT_CONSTRAINT(0xcb, 0x1),    /* MEM_LOAD_RETIRED.* */
 	EVENT_CONSTRAINT_END
 };
 
-static struct event_constraint intel_nehalem_pebs_event_constraints[] = {
+struct event_constraint intel_nehalem_pebs_event_constraints[] = {
 	INTEL_EVENT_CONSTRAINT(0x0b, 0xf),    /* MEM_INST_RETIRED.* */
 	INTEL_EVENT_CONSTRAINT(0x0f, 0xf),    /* MEM_UNCORE_RETIRED.* */
 	INTEL_UEVENT_CONSTRAINT(0x010c, 0xf), /* MEM_STORE_RETIRED.DTLB_MISS */
@@ -392,7 +378,7 @@ static struct event_constraint intel_nehalem_pebs_event_constraints[] = {
 	EVENT_CONSTRAINT_END
 };
 
-static struct event_constraint intel_westmere_pebs_event_constraints[] = {
+struct event_constraint intel_westmere_pebs_event_constraints[] = {
 	INTEL_EVENT_CONSTRAINT(0x0b, 0xf),    /* MEM_INST_RETIRED.* */
 	INTEL_EVENT_CONSTRAINT(0x0f, 0xf),    /* MEM_UNCORE_RETIRED.* */
 	INTEL_UEVENT_CONSTRAINT(0x010c, 0xf), /* MEM_STORE_RETIRED.DTLB_MISS */
@@ -407,7 +393,7 @@ static struct event_constraint intel_westmere_pebs_event_constraints[] = {
 	EVENT_CONSTRAINT_END
 };
 
-static struct event_constraint intel_snb_pebs_events[] = {
+struct event_constraint intel_snb_pebs_event_constraints[] = {
 	INTEL_UEVENT_CONSTRAINT(0x01c0, 0x2), /* INST_RETIRED.PRECDIST */
 	INTEL_UEVENT_CONSTRAINT(0x01c2, 0xf), /* UOPS_RETIRED.ALL */
 	INTEL_UEVENT_CONSTRAINT(0x02c2, 0xf), /* UOPS_RETIRED.RETIRE_SLOTS */
@@ -428,8 +414,7 @@ static struct event_constraint intel_snb_pebs_events[] = {
 	EVENT_CONSTRAINT_END
 };
 
-static struct event_constraint *
-intel_pebs_constraints(struct perf_event *event)
+struct event_constraint *intel_pebs_constraints(struct perf_event *event)
 {
 	struct event_constraint *c;
 
@@ -446,7 +431,7 @@ intel_pebs_constraints(struct perf_event *event)
 	return &emptyconstraint;
 }
 
-static void intel_pmu_pebs_enable(struct perf_event *event)
+void intel_pmu_pebs_enable(struct perf_event *event)
 {
 	struct cpu_hw_events *cpuc = &__get_cpu_var(cpu_hw_events);
 	struct hw_perf_event *hwc = &event->hw;
@@ -460,7 +445,7 @@ static void intel_pmu_pebs_enable(struct perf_event *event)
 		intel_pmu_lbr_enable(event);
 }
 
-static void intel_pmu_pebs_disable(struct perf_event *event)
+void intel_pmu_pebs_disable(struct perf_event *event)
 {
 	struct cpu_hw_events *cpuc = &__get_cpu_var(cpu_hw_events);
 	struct hw_perf_event *hwc = &event->hw;
@@ -475,7 +460,7 @@ static void intel_pmu_pebs_disable(struct perf_event *event)
 		intel_pmu_lbr_disable(event);
 }
 
-static void intel_pmu_pebs_enable_all(void)
+void intel_pmu_pebs_enable_all(void)
 {
 	struct cpu_hw_events *cpuc = &__get_cpu_var(cpu_hw_events);
 
@@ -483,7 +468,7 @@ static void intel_pmu_pebs_enable_all(void)
 		wrmsrl(MSR_IA32_PEBS_ENABLE, cpuc->pebs_enabled);
 }
 
-static void intel_pmu_pebs_disable_all(void)
+void intel_pmu_pebs_disable_all(void)
 {
 	struct cpu_hw_events *cpuc = &__get_cpu_var(cpu_hw_events);
 
@@ -576,8 +561,6 @@ static int intel_pmu_pebs_fixup_ip(struct pt_regs *regs)
 	return 0;
 }
 
-static int intel_pmu_save_and_restart(struct perf_event *event);
-
 static void __intel_pmu_pebs_event(struct perf_event *event,
 				   struct pt_regs *iregs, void *__pebs)
 {
@@ -716,7 +699,7 @@ static void intel_pmu_drain_pebs_nhm(struct pt_regs *iregs)
  * BTS, PEBS probe and setup
  */
 
-static void intel_ds_init(void)
+void intel_ds_init(void)
 {
 	/*
 	 * No support for 32bit formats
@@ -749,15 +732,3 @@ static void intel_ds_init(void)
 		}
 	}
 }
-
-#else /* CONFIG_CPU_SUP_INTEL */
-
-static void reserve_ds_buffers(void)
-{
-}
-
-static void release_ds_buffers(void)
-{
-}
-
-#endif /* CONFIG_CPU_SUP_INTEL */
diff --git a/arch/x86/kernel/cpu/perf_event_intel_lbr.c b/arch/x86/kernel/cpu/perf_event_intel_lbr.c
index d202c1b..3fab3de 100644
--- a/arch/x86/kernel/cpu/perf_event_intel_lbr.c
+++ b/arch/x86/kernel/cpu/perf_event_intel_lbr.c
@@ -1,4 +1,10 @@
-#ifdef CONFIG_CPU_SUP_INTEL
+#include <linux/perf_event.h>
+#include <linux/types.h>
+
+#include <asm/perf_event.h>
+#include <asm/msr.h>
+
+#include "perf_event.h"
 
 enum {
 	LBR_FORMAT_32		= 0x00,
@@ -48,7 +54,7 @@ static void intel_pmu_lbr_reset_64(void)
 	}
 }
 
-static void intel_pmu_lbr_reset(void)
+void intel_pmu_lbr_reset(void)
 {
 	if (!x86_pmu.lbr_nr)
 		return;
@@ -59,7 +65,7 @@ static void intel_pmu_lbr_reset(void)
 		intel_pmu_lbr_reset_64();
 }
 
-static void intel_pmu_lbr_enable(struct perf_event *event)
+void intel_pmu_lbr_enable(struct perf_event *event)
 {
 	struct cpu_hw_events *cpuc = &__get_cpu_var(cpu_hw_events);
 
@@ -81,7 +87,7 @@ static void intel_pmu_lbr_enable(struct perf_event *event)
 	cpuc->lbr_users++;
 }
 
-static void intel_pmu_lbr_disable(struct perf_event *event)
+void intel_pmu_lbr_disable(struct perf_event *event)
 {
 	struct cpu_hw_events *cpuc = &__get_cpu_var(cpu_hw_events);
 
@@ -95,7 +101,7 @@ static void intel_pmu_lbr_disable(struct perf_event *event)
 		__intel_pmu_lbr_disable();
 }
 
-static void intel_pmu_lbr_enable_all(void)
+void intel_pmu_lbr_enable_all(void)
 {
 	struct cpu_hw_events *cpuc = &__get_cpu_var(cpu_hw_events);
 
@@ -103,7 +109,7 @@ static void intel_pmu_lbr_enable_all(void)
 		__intel_pmu_lbr_enable();
 }
 
-static void intel_pmu_lbr_disable_all(void)
+void intel_pmu_lbr_disable_all(void)
 {
 	struct cpu_hw_events *cpuc = &__get_cpu_var(cpu_hw_events);
 
@@ -178,7 +184,7 @@ static void intel_pmu_lbr_read_64(struct cpu_hw_events *cpuc)
 	cpuc->lbr_stack.nr = i;
 }
 
-static void intel_pmu_lbr_read(void)
+void intel_pmu_lbr_read(void)
 {
 	struct cpu_hw_events *cpuc = &__get_cpu_var(cpu_hw_events);
 
@@ -191,7 +197,7 @@ static void intel_pmu_lbr_read(void)
 		intel_pmu_lbr_read_64(cpuc);
 }
 
-static void intel_pmu_lbr_init_core(void)
+void intel_pmu_lbr_init_core(void)
 {
 	x86_pmu.lbr_nr     = 4;
 	x86_pmu.lbr_tos    = 0x01c9;
@@ -199,7 +205,7 @@ static void intel_pmu_lbr_init_core(void)
 	x86_pmu.lbr_to     = 0x60;
 }
 
-static void intel_pmu_lbr_init_nhm(void)
+void intel_pmu_lbr_init_nhm(void)
 {
 	x86_pmu.lbr_nr     = 16;
 	x86_pmu.lbr_tos    = 0x01c9;
@@ -207,12 +213,10 @@ static void intel_pmu_lbr_init_nhm(void)
 	x86_pmu.lbr_to     = 0x6c0;
 }
 
-static void intel_pmu_lbr_init_atom(void)
+void intel_pmu_lbr_init_atom(void)
 {
 	x86_pmu.lbr_nr	   = 8;
 	x86_pmu.lbr_tos    = 0x01c9;
 	x86_pmu.lbr_from   = 0x40;
 	x86_pmu.lbr_to     = 0x60;
 }
-
-#endif /* CONFIG_CPU_SUP_INTEL */
diff --git a/arch/x86/kernel/cpu/perf_event_p4.c b/arch/x86/kernel/cpu/perf_event_p4.c
index 7809d2bc..5edb266 100644
--- a/arch/x86/kernel/cpu/perf_event_p4.c
+++ b/arch/x86/kernel/cpu/perf_event_p4.c
@@ -7,9 +7,12 @@
  *  For licencing details see kernel-base/COPYING
  */
 
-#ifdef CONFIG_CPU_SUP_INTEL
+#include <linux/perf_event.h>
 
 #include <asm/perf_event_p4.h>
+#include <asm/hardirq.h>
+
+#include "perf_event.h"
 
 #define P4_CNTR_LIMIT 3
 /*
@@ -1303,7 +1306,7 @@ static __initconst const struct x86_pmu p4_pmu = {
 	.perfctr_second_write	= 1,
 };
 
-static __init int p4_pmu_init(void)
+__init int p4_pmu_init(void)
 {
 	unsigned int low, high;
 
@@ -1326,5 +1329,3 @@ static __init int p4_pmu_init(void)
 
 	return 0;
 }
-
-#endif /* CONFIG_CPU_SUP_INTEL */
diff --git a/arch/x86/kernel/cpu/perf_event_p6.c b/arch/x86/kernel/cpu/perf_event_p6.c
index 20c097e..c7181be 100644
--- a/arch/x86/kernel/cpu/perf_event_p6.c
+++ b/arch/x86/kernel/cpu/perf_event_p6.c
@@ -1,4 +1,7 @@
-#ifdef CONFIG_CPU_SUP_INTEL
+#include <linux/perf_event.h>
+#include <linux/types.h>
+
+#include "perf_event.h"
 
 /*
  * Not sure about some of these
@@ -114,7 +117,7 @@ static __initconst const struct x86_pmu p6_pmu = {
 	.event_constraints	= p6_event_constraints,
 };
 
-static __init int p6_pmu_init(void)
+__init int p6_pmu_init(void)
 {
 	switch (boot_cpu_data.x86_model) {
 	case 1:
@@ -138,5 +141,3 @@ static __init int p6_pmu_init(void)
 
 	return 0;
 }
-
-#endif /* CONFIG_CPU_SUP_INTEL */
-- 
1.7.6.1

Re: [RFC PATCH] x86: perf: Clean up perf_event cpu code

[Peter Zijlstra]

On Tue, 2011-08-30 at 08:21 -0300, Kevin Winchester wrote:
> 
> This is the first time I've attempted a patch this large.  Should I break it up
> into smaller pieces?  

Only if it makes sens, intermediate stages should still compile. I could
imagine you creating the header file and moving stuff in there one thing
at a time while keeping the whole thing compiling. Eventually resulting
in the state where the #include hackery is superfluous and can be
removed.

If its worth spending a lot of time making that happen is another thing.

> Or does that requirement not really apply since it's just
> a cleanup patch?

Cleanup isn't the criteria. The individual steps have to make sense.

Take for instance the uprobe patch set as an example of how not to split
up things. The individual patches don't really stand on their own and
you're constantly forced to switch back and forth while reviewing, a
total pain.

> Also, it is based on Linus' tree, should I base it on tip instead?

Yes please. I don't think they differ a lot but tip is where it'll end
up being applied to, so it helps if it does indeed apply ;-)

Otherwise looks good, thanks a bunch!

[PATCH] x86: perf: Clean up perf_event cpu code

[Kevin Winchester]

The CPU support for perf events on x86 was implemented via included C files
with #ifdefs.  Clean this up by creating a new header file and compiling
the vendor-specific files as needed.

Signed-off-by: Kevin Winchester <kjwinchester@gmail.com>
---

This is now against the tip tree.  I left it all as one big commit for now,
although I can break it up if necessary.  I'm just not sure it would gain much.

Again, this is by far the largest out of the handful of patches I have ever
submitted, so please let me know if I am doing anything wrong.


 arch/x86/kernel/cpu/Makefile               |    5 +
 arch/x86/kernel/cpu/perf_event.c           |  369 +--------------------
 arch/x86/kernel/cpu/perf_event.h           |  493 ++++++++++++++++++++++++++++
 arch/x86/kernel/cpu/perf_event_amd.c       |   18 +-
 arch/x86/kernel/cpu/perf_event_intel.c     |   52 ++--
 arch/x86/kernel/cpu/perf_event_intel_ds.c  |   79 ++---
 arch/x86/kernel/cpu/perf_event_intel_lbr.c |   28 +-
 arch/x86/kernel/cpu/perf_event_p4.c        |    9 +-
 arch/x86/kernel/cpu/perf_event_p6.c        |    9 +-
 9 files changed, 602 insertions(+), 460 deletions(-)

diff --git a/arch/x86/kernel/cpu/Makefile b/arch/x86/kernel/cpu/Makefile
index 0e3a82a..55622b9 100644
--- a/arch/x86/kernel/cpu/Makefile
+++ b/arch/x86/kernel/cpu/Makefile
@@ -29,6 +29,11 @@ obj-$(CONFIG_CPU_SUP_UMC_32)		+= umc.o
 
 obj-$(CONFIG_PERF_EVENTS)		+= perf_event.o
 
+ifdef CONFIG_PERF_EVENTS
+obj-$(CONFIG_CPU_SUP_AMD)		+= perf_event_amd.o
+obj-$(CONFIG_CPU_SUP_INTEL)		+= perf_event_p6.o perf_event_p4.o perf_event_intel_lbr.o perf_event_intel_ds.o perf_event_intel.o
+endif
+
 obj-$(CONFIG_X86_MCE)			+= mcheck/
 obj-$(CONFIG_MTRR)			+= mtrr/
 
diff --git a/arch/x86/kernel/cpu/perf_event.c b/arch/x86/kernel/cpu/perf_event.c
index 594d425..eb8fc15 100644
--- a/arch/x86/kernel/cpu/perf_event.c
+++ b/arch/x86/kernel/cpu/perf_event.c
@@ -32,6 +32,8 @@
 #include <asm/smp.h>
 #include <asm/alternative.h>
 
+#include "perf_event.h"
+
 #if 0
 #undef wrmsrl
 #define wrmsrl(msr, val) 					\
@@ -43,285 +45,17 @@ do {								\
 } while (0)
 #endif
 
-/*
- *          |   NHM/WSM    |      SNB     |
- * register -------------------------------
- *          |  HT  | no HT |  HT  | no HT |
- *-----------------------------------------
- * offcore  | core | core  | cpu  | core  |
- * lbr_sel  | core | core  | cpu  | core  |
- * ld_lat   | cpu  | core  | cpu  | core  |
- *-----------------------------------------
- *
- * Given that there is a small number of shared regs,
- * we can pre-allocate their slot in the per-cpu
- * per-core reg tables.
- */
-enum extra_reg_type {
-	EXTRA_REG_NONE  = -1,	/* not used */
-
-	EXTRA_REG_RSP_0 = 0,	/* offcore_response_0 */
-	EXTRA_REG_RSP_1 = 1,	/* offcore_response_1 */
-
-	EXTRA_REG_MAX		/* number of entries needed */
-};
-
-struct event_constraint {
-	union {
-		unsigned long	idxmsk[BITS_TO_LONGS(X86_PMC_IDX_MAX)];
-		u64		idxmsk64;
-	};
-	u64	code;
-	u64	cmask;
-	int	weight;
-};
-
-struct amd_nb {
-	int nb_id;  /* NorthBridge id */
-	int refcnt; /* reference count */
-	struct perf_event *owners[X86_PMC_IDX_MAX];
-	struct event_constraint event_constraints[X86_PMC_IDX_MAX];
-};
-
-struct intel_percore;
-
-#define MAX_LBR_ENTRIES		16
-
-struct cpu_hw_events {
-	/*
-	 * Generic x86 PMC bits
-	 */
-	struct perf_event	*events[X86_PMC_IDX_MAX]; /* in counter order */
-	unsigned long		active_mask[BITS_TO_LONGS(X86_PMC_IDX_MAX)];
-	unsigned long		running[BITS_TO_LONGS(X86_PMC_IDX_MAX)];
-	int			enabled;
-
-	int			n_events;
-	int			n_added;
-	int			n_txn;
-	int			assign[X86_PMC_IDX_MAX]; /* event to counter assignment */
-	u64			tags[X86_PMC_IDX_MAX];
-	struct perf_event	*event_list[X86_PMC_IDX_MAX]; /* in enabled order */
-
-	unsigned int		group_flag;
-
-	/*
-	 * Intel DebugStore bits
-	 */
-	struct debug_store	*ds;
-	u64			pebs_enabled;
-
-	/*
-	 * Intel LBR bits
-	 */
-	int				lbr_users;
-	void				*lbr_context;
-	struct perf_branch_stack	lbr_stack;
-	struct perf_branch_entry	lbr_entries[MAX_LBR_ENTRIES];
-
-	/*
-	 * manage shared (per-core, per-cpu) registers
-	 * used on Intel NHM/WSM/SNB
-	 */
-	struct intel_shared_regs	*shared_regs;
-
-	/*
-	 * AMD specific bits
-	 */
-	struct amd_nb		*amd_nb;
-
-	void			*kfree_on_online;
-};
-
-#define __EVENT_CONSTRAINT(c, n, m, w) {\
-	{ .idxmsk64 = (n) },		\
-	.code = (c),			\
-	.cmask = (m),			\
-	.weight = (w),			\
-}
-
-#define EVENT_CONSTRAINT(c, n, m)	\
-	__EVENT_CONSTRAINT(c, n, m, HWEIGHT(n))
-
-/*
- * Constraint on the Event code.
- */
-#define INTEL_EVENT_CONSTRAINT(c, n)	\
-	EVENT_CONSTRAINT(c, n, ARCH_PERFMON_EVENTSEL_EVENT)
-
-/*
- * Constraint on the Event code + UMask + fixed-mask
- *
- * filter mask to validate fixed counter events.
- * the following filters disqualify for fixed counters:
- *  - inv
- *  - edge
- *  - cnt-mask
- *  The other filters are supported by fixed counters.
- *  The any-thread option is supported starting with v3.
- */
-#define FIXED_EVENT_CONSTRAINT(c, n)	\
-	EVENT_CONSTRAINT(c, (1ULL << (32+n)), X86_RAW_EVENT_MASK)
-
-/*
- * Constraint on the Event code + UMask
- */
-#define INTEL_UEVENT_CONSTRAINT(c, n)	\
-	EVENT_CONSTRAINT(c, n, INTEL_ARCH_EVENT_MASK)
-
-#define EVENT_CONSTRAINT_END		\
-	EVENT_CONSTRAINT(0, 0, 0)
-
-#define for_each_event_constraint(e, c)	\
-	for ((e) = (c); (e)->weight; (e)++)
-
-/*
- * Per register state.
- */
-struct er_account {
-	raw_spinlock_t		lock;	/* per-core: protect structure */
-	u64			config;	/* extra MSR config */
-	u64			reg;	/* extra MSR number */
-	atomic_t		ref;	/* reference count */
-};
-
-/*
- * Extra registers for specific events.
- *
- * Some events need large masks and require external MSRs.
- * Those extra MSRs end up being shared for all events on
- * a PMU and sometimes between PMU of sibling HT threads.
- * In either case, the kernel needs to handle conflicting
- * accesses to those extra, shared, regs. The data structure
- * to manage those registers is stored in cpu_hw_event.
- */
-struct extra_reg {
-	unsigned int		event;
-	unsigned int		msr;
-	u64			config_mask;
-	u64			valid_mask;
-	int			idx;  /* per_xxx->regs[] reg index */
-};
-
-#define EVENT_EXTRA_REG(e, ms, m, vm, i) {	\
-	.event = (e),		\
-	.msr = (ms),		\
-	.config_mask = (m),	\
-	.valid_mask = (vm),	\
-	.idx = EXTRA_REG_##i	\
-	}
-
-#define INTEL_EVENT_EXTRA_REG(event, msr, vm, idx)	\
-	EVENT_EXTRA_REG(event, msr, ARCH_PERFMON_EVENTSEL_EVENT, vm, idx)
+struct x86_pmu x86_pmu __read_mostly;
 
-#define EVENT_EXTRA_END EVENT_EXTRA_REG(0, 0, 0, 0, RSP_0)
-
-union perf_capabilities {
-	struct {
-		u64	lbr_format    : 6;
-		u64	pebs_trap     : 1;
-		u64	pebs_arch_reg : 1;
-		u64	pebs_format   : 4;
-		u64	smm_freeze    : 1;
-	};
-	u64	capabilities;
-};
-
-/*
- * struct x86_pmu - generic x86 pmu
- */
-struct x86_pmu {
-	/*
-	 * Generic x86 PMC bits
-	 */
-	const char	*name;
-	int		version;
-	int		(*handle_irq)(struct pt_regs *);
-	void		(*disable_all)(void);
-	void		(*enable_all)(int added);
-	void		(*enable)(struct perf_event *);
-	void		(*disable)(struct perf_event *);
-	int		(*hw_config)(struct perf_event *event);
-	int		(*schedule_events)(struct cpu_hw_events *cpuc, int n, int *assign);
-	unsigned	eventsel;
-	unsigned	perfctr;
-	u64		(*event_map)(int);
-	int		max_events;
-	int		num_counters;
-	int		num_counters_fixed;
-	int		cntval_bits;
-	u64		cntval_mask;
-	int		apic;
-	u64		max_period;
-	struct event_constraint *
-			(*get_event_constraints)(struct cpu_hw_events *cpuc,
-						 struct perf_event *event);
-
-	void		(*put_event_constraints)(struct cpu_hw_events *cpuc,
-						 struct perf_event *event);
-	struct event_constraint *event_constraints;
-	void		(*quirks)(void);
-	int		perfctr_second_write;
-
-	int		(*cpu_prepare)(int cpu);
-	void		(*cpu_starting)(int cpu);
-	void		(*cpu_dying)(int cpu);
-	void		(*cpu_dead)(int cpu);
-
-	/*
-	 * Intel Arch Perfmon v2+
-	 */
-	u64			intel_ctrl;
-	union perf_capabilities intel_cap;
-
-	/*
-	 * Intel DebugStore bits
-	 */
-	int		bts, pebs;
-	int		bts_active, pebs_active;
-	int		pebs_record_size;
-	void		(*drain_pebs)(struct pt_regs *regs);
-	struct event_constraint *pebs_constraints;
-
-	/*
-	 * Intel LBR
-	 */
-	unsigned long	lbr_tos, lbr_from, lbr_to; /* MSR base regs       */
-	int		lbr_nr;			   /* hardware stack size */
-
-	/*
-	 * Extra registers for events
-	 */
-	struct extra_reg *extra_regs;
-	unsigned int er_flags;
-};
-
-#define ERF_NO_HT_SHARING	1
-#define ERF_HAS_RSP_1		2
-
-static struct x86_pmu x86_pmu __read_mostly;
-
-static DEFINE_PER_CPU(struct cpu_hw_events, cpu_hw_events) = {
+DEFINE_PER_CPU(struct cpu_hw_events, cpu_hw_events) = {
 	.enabled = 1,
 };
 
-static int x86_perf_event_set_period(struct perf_event *event);
-
-/*
- * Generalized hw caching related hw_event table, filled
- * in on a per model basis. A value of 0 means
- * 'not supported', -1 means 'hw_event makes no sense on
- * this CPU', any other value means the raw hw_event
- * ID.
- */
-
-#define C(x) PERF_COUNT_HW_CACHE_##x
-
-static u64 __read_mostly hw_cache_event_ids
+u64 __read_mostly hw_cache_event_ids
 				[PERF_COUNT_HW_CACHE_MAX]
 				[PERF_COUNT_HW_CACHE_OP_MAX]
 				[PERF_COUNT_HW_CACHE_RESULT_MAX];
-static u64 __read_mostly hw_cache_extra_regs
+u64 __read_mostly hw_cache_extra_regs
 				[PERF_COUNT_HW_CACHE_MAX]
 				[PERF_COUNT_HW_CACHE_OP_MAX]
 				[PERF_COUNT_HW_CACHE_RESULT_MAX];
@@ -331,8 +65,7 @@ static u64 __read_mostly hw_cache_extra_regs
  * Can only be executed on the CPU where the event is active.
  * Returns the delta events processed.
  */
-static u64
-x86_perf_event_update(struct perf_event *event)
+u64 x86_perf_event_update(struct perf_event *event)
 {
 	struct hw_perf_event *hwc = &event->hw;
 	int shift = 64 - x86_pmu.cntval_bits;
@@ -375,30 +108,6 @@ again:
 	return new_raw_count;
 }
 
-static inline int x86_pmu_addr_offset(int index)
-{
-	int offset;
-
-	/* offset = X86_FEATURE_PERFCTR_CORE ? index << 1 : index */
-	alternative_io(ASM_NOP2,
-		       "shll $1, %%eax",
-		       X86_FEATURE_PERFCTR_CORE,
-		       "=a" (offset),
-		       "a"  (index));
-
-	return offset;
-}
-
-static inline unsigned int x86_pmu_config_addr(int index)
-{
-	return x86_pmu.eventsel + x86_pmu_addr_offset(index);
-}
-
-static inline unsigned int x86_pmu_event_addr(int index)
-{
-	return x86_pmu.perfctr + x86_pmu_addr_offset(index);
-}
-
 /*
  * Find and validate any extra registers to set up.
  */
@@ -534,9 +243,6 @@ msr_fail:
 	return false;
 }
 
-static void reserve_ds_buffers(void);
-static void release_ds_buffers(void);
-
 static void hw_perf_event_destroy(struct perf_event *event)
 {
 	if (atomic_dec_and_mutex_lock(&active_events, &pmc_reserve_mutex)) {
@@ -585,7 +291,7 @@ set_ext_hw_attr(struct hw_perf_event *hwc, struct perf_event *event)
 	return x86_pmu_extra_regs(val, event);
 }
 
-static int x86_setup_perfctr(struct perf_event *event)
+int x86_setup_perfctr(struct perf_event *event)
 {
 	struct perf_event_attr *attr = &event->attr;
 	struct hw_perf_event *hwc = &event->hw;
@@ -649,7 +355,7 @@ static int x86_setup_perfctr(struct perf_event *event)
 	return 0;
 }
 
-static int x86_pmu_hw_config(struct perf_event *event)
+int x86_pmu_hw_config(struct perf_event *event)
 {
 	if (event->attr.precise_ip) {
 		int precise = 0;
@@ -725,7 +431,7 @@ static int __x86_pmu_event_init(struct perf_event *event)
 	return x86_pmu.hw_config(event);
 }
 
-static void x86_pmu_disable_all(void)
+void x86_pmu_disable_all(void)
 {
 	struct cpu_hw_events *cpuc = &__get_cpu_var(cpu_hw_events);
 	int idx;
@@ -760,15 +466,7 @@ static void x86_pmu_disable(struct pmu *pmu)
 	x86_pmu.disable_all();
 }
 
-static inline void __x86_pmu_enable_event(struct hw_perf_event *hwc,
-					  u64 enable_mask)
-{
-	if (hwc->extra_reg.reg)
-		wrmsrl(hwc->extra_reg.reg, hwc->extra_reg.config);
-	wrmsrl(hwc->config_base, hwc->config | enable_mask);
-}
-
-static void x86_pmu_enable_all(int added)
+void x86_pmu_enable_all(int added)
 {
 	struct cpu_hw_events *cpuc = &__get_cpu_var(cpu_hw_events);
 	int idx;
@@ -790,7 +488,7 @@ static inline int is_x86_event(struct perf_event *event)
 	return event->pmu == &pmu;
 }
 
-static int x86_schedule_events(struct cpu_hw_events *cpuc, int n, int *assign)
+int x86_schedule_events(struct cpu_hw_events *cpuc, int n, int *assign)
 {
 	struct event_constraint *c, *constraints[X86_PMC_IDX_MAX];
 	unsigned long used_mask[BITS_TO_LONGS(X86_PMC_IDX_MAX)];
@@ -961,7 +659,6 @@ static inline int match_prev_assignment(struct hw_perf_event *hwc,
 }
 
 static void x86_pmu_start(struct perf_event *event, int flags);
-static void x86_pmu_stop(struct perf_event *event, int flags);
 
 static void x86_pmu_enable(struct pmu *pmu)
 {
@@ -1033,21 +730,13 @@ static void x86_pmu_enable(struct pmu *pmu)
 	x86_pmu.enable_all(added);
 }
 
-static inline void x86_pmu_disable_event(struct perf_event *event)
-{
-	struct hw_perf_event *hwc = &event->hw;
-
-	wrmsrl(hwc->config_base, hwc->config);
-}
-
 static DEFINE_PER_CPU(u64 [X86_PMC_IDX_MAX], pmc_prev_left);
 
 /*
  * Set the next IRQ period, based on the hwc->period_left value.
  * To be called with the event disabled in hw:
  */
-static int
-x86_perf_event_set_period(struct perf_event *event)
+int x86_perf_event_set_period(struct perf_event *event)
 {
 	struct hw_perf_event *hwc = &event->hw;
 	s64 left = local64_read(&hwc->period_left);
@@ -1107,7 +796,7 @@ x86_perf_event_set_period(struct perf_event *event)
 	return ret;
 }
 
-static void x86_pmu_enable_event(struct perf_event *event)
+void x86_pmu_enable_event(struct perf_event *event)
 {
 	if (__this_cpu_read(cpu_hw_events.enabled))
 		__x86_pmu_enable_event(&event->hw,
@@ -1246,7 +935,7 @@ void perf_event_print_debug(void)
 	local_irq_restore(flags);
 }
 
-static void x86_pmu_stop(struct perf_event *event, int flags)
+void x86_pmu_stop(struct perf_event *event, int flags)
 {
 	struct cpu_hw_events *cpuc = &__get_cpu_var(cpu_hw_events);
 	struct hw_perf_event *hwc = &event->hw;
@@ -1299,7 +988,7 @@ static void x86_pmu_del(struct perf_event *event, int flags)
 	perf_event_update_userpage(event);
 }
 
-static int x86_pmu_handle_irq(struct pt_regs *regs)
+int x86_pmu_handle_irq(struct pt_regs *regs)
 {
 	struct perf_sample_data data;
 	struct cpu_hw_events *cpuc;
@@ -1439,30 +1128,8 @@ static __read_mostly struct notifier_block perf_event_nmi_notifier = {
 	.priority		= NMI_LOCAL_LOW_PRIOR,
 };
 
-static struct event_constraint unconstrained;
-static struct event_constraint emptyconstraint;
-
-static struct event_constraint *
-x86_get_event_constraints(struct cpu_hw_events *cpuc, struct perf_event *event)
-{
-	struct event_constraint *c;
-
-	if (x86_pmu.event_constraints) {
-		for_each_event_constraint(c, x86_pmu.event_constraints) {
-			if ((event->hw.config & c->cmask) == c->code)
-				return c;
-		}
-	}
-
-	return &unconstrained;
-}
-
-#include "perf_event_amd.c"
-#include "perf_event_p6.c"
-#include "perf_event_p4.c"
-#include "perf_event_intel_lbr.c"
-#include "perf_event_intel_ds.c"
-#include "perf_event_intel.c"
+struct event_constraint emptyconstraint;
+struct event_constraint unconstrained;
 
 static int __cpuinit
 x86_pmu_notifier(struct notifier_block *self, unsigned long action, void *hcpu)
diff --git a/arch/x86/kernel/cpu/perf_event.h b/arch/x86/kernel/cpu/perf_event.h
new file mode 100644
index 0000000..fb330b0
--- /dev/null
+++ b/arch/x86/kernel/cpu/perf_event.h
@@ -0,0 +1,493 @@
+/*
+ * Performance events x86 architecture header
+ *
+ *  Copyright (C) 2008 Thomas Gleixner <tglx@linutronix.de>
+ *  Copyright (C) 2008-2009 Red Hat, Inc., Ingo Molnar
+ *  Copyright (C) 2009 Jaswinder Singh Rajput
+ *  Copyright (C) 2009 Advanced Micro Devices, Inc., Robert Richter
+ *  Copyright (C) 2008-2009 Red Hat, Inc., Peter Zijlstra <pzijlstr@redhat.com>
+ *  Copyright (C) 2009 Intel Corporation, <markus.t.metzger@intel.com>
+ *  Copyright (C) 2009 Google, Inc., Stephane Eranian
+ *
+ *  For licencing details see kernel-base/COPYING
+ */
+
+#include <linux/perf_event.h>
+
+/*
+ *          |   NHM/WSM    |      SNB     |
+ * register -------------------------------
+ *          |  HT  | no HT |  HT  | no HT |
+ *-----------------------------------------
+ * offcore  | core | core  | cpu  | core  |
+ * lbr_sel  | core | core  | cpu  | core  |
+ * ld_lat   | cpu  | core  | cpu  | core  |
+ *-----------------------------------------
+ *
+ * Given that there is a small number of shared regs,
+ * we can pre-allocate their slot in the per-cpu
+ * per-core reg tables.
+ */
+enum extra_reg_type {
+	EXTRA_REG_NONE  = -1,	/* not used */
+
+	EXTRA_REG_RSP_0 = 0,	/* offcore_response_0 */
+	EXTRA_REG_RSP_1 = 1,	/* offcore_response_1 */
+
+	EXTRA_REG_MAX		/* number of entries needed */
+};
+
+struct event_constraint {
+	union {
+		unsigned long	idxmsk[BITS_TO_LONGS(X86_PMC_IDX_MAX)];
+		u64		idxmsk64;
+	};
+	u64	code;
+	u64	cmask;
+	int	weight;
+};
+
+struct amd_nb {
+	int nb_id;  /* NorthBridge id */
+	int refcnt; /* reference count */
+	struct perf_event *owners[X86_PMC_IDX_MAX];
+	struct event_constraint event_constraints[X86_PMC_IDX_MAX];
+};
+
+/* The maximal number of PEBS events: */
+#define MAX_PEBS_EVENTS		4
+
+/*
+ * A debug store configuration.
+ *
+ * We only support architectures that use 64bit fields.
+ */
+struct debug_store {
+	u64	bts_buffer_base;
+	u64	bts_index;
+	u64	bts_absolute_maximum;
+	u64	bts_interrupt_threshold;
+	u64	pebs_buffer_base;
+	u64	pebs_index;
+	u64	pebs_absolute_maximum;
+	u64	pebs_interrupt_threshold;
+	u64	pebs_event_reset[MAX_PEBS_EVENTS];
+};
+
+/*
+ * Per register state.
+ */
+struct er_account {
+	raw_spinlock_t		lock;	/* per-core: protect structure */
+	u64                 config;	/* extra MSR config */
+	u64                 reg;	/* extra MSR number */
+	atomic_t            ref;	/* reference count */
+};
+
+/*
+ * Per core/cpu state
+ *
+ * Used to coordinate shared registers between HT threads or
+ * among events on a single PMU.
+ */
+struct intel_shared_regs {
+	struct er_account       regs[EXTRA_REG_MAX];
+	int                     refcnt;		/* per-core: #HT threads */
+	unsigned                core_id;	/* per-core: core id */
+};
+
+#define MAX_LBR_ENTRIES		16
+
+struct cpu_hw_events {
+	/*
+	 * Generic x86 PMC bits
+	 */
+	struct perf_event	*events[X86_PMC_IDX_MAX]; /* in counter order */
+	unsigned long		active_mask[BITS_TO_LONGS(X86_PMC_IDX_MAX)];
+	unsigned long		running[BITS_TO_LONGS(X86_PMC_IDX_MAX)];
+	int			enabled;
+
+	int			n_events;
+	int			n_added;
+	int			n_txn;
+	int			assign[X86_PMC_IDX_MAX]; /* event to counter assignment */
+	u64			tags[X86_PMC_IDX_MAX];
+	struct perf_event	*event_list[X86_PMC_IDX_MAX]; /* in enabled order */
+
+	unsigned int		group_flag;
+
+	/*
+	 * Intel DebugStore bits
+	 */
+	struct debug_store	*ds;
+	u64			pebs_enabled;
+
+	/*
+	 * Intel LBR bits
+	 */
+	int				lbr_users;
+	void				*lbr_context;
+	struct perf_branch_stack	lbr_stack;
+	struct perf_branch_entry	lbr_entries[MAX_LBR_ENTRIES];
+
+	/*
+	 * manage shared (per-core, per-cpu) registers
+	 * used on Intel NHM/WSM/SNB
+	 */
+	struct intel_shared_regs	*shared_regs;
+
+	/*
+	 * AMD specific bits
+	 */
+	struct amd_nb		*amd_nb;
+
+	void				*kfree_on_online;
+};
+
+#define __EVENT_CONSTRAINT(c, n, m, w) {\
+	{ .idxmsk64 = (n) },		\
+	.code = (c),			\
+	.cmask = (m),			\
+	.weight = (w),			\
+}
+
+#define EVENT_CONSTRAINT(c, n, m)	\
+	__EVENT_CONSTRAINT(c, n, m, HWEIGHT(n))
+
+/*
+ * Constraint on the Event code.
+ */
+#define INTEL_EVENT_CONSTRAINT(c, n)	\
+	EVENT_CONSTRAINT(c, n, ARCH_PERFMON_EVENTSEL_EVENT)
+
+/*
+ * Constraint on the Event code + UMask + fixed-mask
+ *
+ * filter mask to validate fixed counter events.
+ * the following filters disqualify for fixed counters:
+ *  - inv
+ *  - edge
+ *  - cnt-mask
+ *  The other filters are supported by fixed counters.
+ *  The any-thread option is supported starting with v3.
+ */
+#define FIXED_EVENT_CONSTRAINT(c, n)	\
+	EVENT_CONSTRAINT(c, (1ULL << (32+n)), X86_RAW_EVENT_MASK)
+
+/*
+ * Constraint on the Event code + UMask
+ */
+#define INTEL_UEVENT_CONSTRAINT(c, n)	\
+	EVENT_CONSTRAINT(c, n, INTEL_ARCH_EVENT_MASK)
+
+#define EVENT_CONSTRAINT_END		\
+	EVENT_CONSTRAINT(0, 0, 0)
+
+#define for_each_event_constraint(e, c)	\
+	for ((e) = (c); (e)->weight; (e)++)
+
+/*
+ * Extra registers for specific events.
+ *
+ * Some events need large masks and require external MSRs.
+ * Those extra MSRs end up being shared for all events on
+ * a PMU and sometimes between PMU of sibling HT threads.
+ * In either case, the kernel needs to handle conflicting
+ * accesses to those extra, shared, regs. The data structure
+ * to manage those registers is stored in cpu_hw_event.
+ */
+struct extra_reg {
+	unsigned int		event;
+	unsigned int		msr;
+	u64			config_mask;
+	u64			valid_mask;
+	int			idx;  /* per_xxx->regs[] reg index */
+};
+
+#define EVENT_EXTRA_REG(e, ms, m, vm, i) {	\
+	.event = (e),		\
+	.msr = (ms),		\
+	.config_mask = (m),	\
+	.valid_mask = (vm),	\
+	.idx = EXTRA_REG_##i	\
+	}
+
+#define INTEL_EVENT_EXTRA_REG(event, msr, vm, idx)	\
+	EVENT_EXTRA_REG(event, msr, ARCH_PERFMON_EVENTSEL_EVENT, vm, idx)
+
+#define EVENT_EXTRA_END EVENT_EXTRA_REG(0, 0, 0, 0, RSP_0)
+
+union perf_capabilities {
+	struct {
+		u64	lbr_format:6;
+		u64	pebs_trap:1;
+		u64	pebs_arch_reg:1;
+		u64	pebs_format:4;
+		u64	smm_freeze:1;
+	};
+	u64	capabilities;
+};
+
+/*
+ * struct x86_pmu - generic x86 pmu
+ */
+struct x86_pmu {
+	/*
+	 * Generic x86 PMC bits
+	 */
+	const char	*name;
+	int		version;
+	int		(*handle_irq)(struct pt_regs *);
+	void		(*disable_all)(void);
+	void		(*enable_all)(int added);
+	void		(*enable)(struct perf_event *);
+	void		(*disable)(struct perf_event *);
+	int		(*hw_config)(struct perf_event *event);
+	int		(*schedule_events)(struct cpu_hw_events *cpuc, int n, int *assign);
+	unsigned	eventsel;
+	unsigned	perfctr;
+	u64		(*event_map)(int);
+	int		max_events;
+	int		num_counters;
+	int		num_counters_fixed;
+	int		cntval_bits;
+	u64		cntval_mask;
+	int		apic;
+	u64		max_period;
+	struct event_constraint *
+			(*get_event_constraints)(struct cpu_hw_events *cpuc,
+						 struct perf_event *event);
+
+	void		(*put_event_constraints)(struct cpu_hw_events *cpuc,
+						 struct perf_event *event);
+	struct event_constraint *event_constraints;
+	void		(*quirks)(void);
+	int		perfctr_second_write;
+
+	int		(*cpu_prepare)(int cpu);
+	void		(*cpu_starting)(int cpu);
+	void		(*cpu_dying)(int cpu);
+	void		(*cpu_dead)(int cpu);
+
+	/*
+	 * Intel Arch Perfmon v2+
+	 */
+	u64			intel_ctrl;
+	union perf_capabilities intel_cap;
+
+	/*
+	 * Intel DebugStore bits
+	 */
+	int		bts, pebs;
+	int		bts_active, pebs_active;
+	int		pebs_record_size;
+	void		(*drain_pebs)(struct pt_regs *regs);
+	struct event_constraint *pebs_constraints;
+
+	/*
+	 * Intel LBR
+	 */
+	unsigned long	lbr_tos, lbr_from, lbr_to; /* MSR base regs       */
+	int		lbr_nr;			   /* hardware stack size */
+
+	/*
+	 * Extra registers for events
+	 */
+	struct extra_reg *extra_regs;
+	unsigned int er_flags;
+};
+
+#define ERF_NO_HT_SHARING	1
+#define ERF_HAS_RSP_1		2
+
+extern struct x86_pmu x86_pmu __read_mostly;
+
+DECLARE_PER_CPU(struct cpu_hw_events, cpu_hw_events);
+
+int x86_perf_event_set_period(struct perf_event *event);
+
+/*
+ * Generalized hw caching related hw_event table, filled
+ * in on a per model basis. A value of 0 means
+ * 'not supported', -1 means 'hw_event makes no sense on
+ * this CPU', any other value means the raw hw_event
+ * ID.
+ */
+
+#define C(x) PERF_COUNT_HW_CACHE_##x
+
+extern u64 __read_mostly hw_cache_event_ids
+				[PERF_COUNT_HW_CACHE_MAX]
+				[PERF_COUNT_HW_CACHE_OP_MAX]
+				[PERF_COUNT_HW_CACHE_RESULT_MAX];
+extern u64 __read_mostly hw_cache_extra_regs
+				[PERF_COUNT_HW_CACHE_MAX]
+				[PERF_COUNT_HW_CACHE_OP_MAX]
+				[PERF_COUNT_HW_CACHE_RESULT_MAX];
+
+u64 x86_perf_event_update(struct perf_event *event);
+
+static inline int x86_pmu_addr_offset(int index)
+{
+	int offset;
+
+	/* offset = X86_FEATURE_PERFCTR_CORE ? index << 1 : index */
+	alternative_io(ASM_NOP2,
+		       "shll $1, %%eax",
+		       X86_FEATURE_PERFCTR_CORE,
+		       "=a" (offset),
+		       "a"  (index));
+
+	return offset;
+}
+
+static inline unsigned int x86_pmu_config_addr(int index)
+{
+	return x86_pmu.eventsel + x86_pmu_addr_offset(index);
+}
+
+static inline unsigned int x86_pmu_event_addr(int index)
+{
+	return x86_pmu.perfctr + x86_pmu_addr_offset(index);
+}
+
+int x86_setup_perfctr(struct perf_event *event);
+
+int x86_pmu_hw_config(struct perf_event *event);
+
+void x86_pmu_disable_all(void);
+
+static inline void __x86_pmu_enable_event(struct hw_perf_event *hwc,
+					  u64 enable_mask)
+{
+	if (hwc->extra_reg.reg)
+		wrmsrl(hwc->extra_reg.reg, hwc->extra_reg.config);
+	wrmsrl(hwc->config_base, hwc->config | enable_mask);
+}
+
+void x86_pmu_enable_all(int added);
+
+int x86_schedule_events(struct cpu_hw_events *cpuc, int n, int *assign);
+
+void x86_pmu_stop(struct perf_event *event, int flags);
+
+static inline void x86_pmu_disable_event(struct perf_event *event)
+{
+	struct hw_perf_event *hwc = &event->hw;
+
+	wrmsrl(hwc->config_base, hwc->config);
+}
+
+void x86_pmu_enable_event(struct perf_event *event);
+
+int x86_pmu_handle_irq(struct pt_regs *regs);
+
+extern struct event_constraint emptyconstraint;
+
+extern struct event_constraint unconstrained;
+
+#ifdef CONFIG_CPU_SUP_AMD
+
+int amd_pmu_init(void);
+
+#else /* CONFIG_CPU_SUP_AMD */
+
+static inline int amd_pmu_init(void)
+{
+	return 0;
+}
+
+#endif /* CONFIG_CPU_SUP_AMD */
+
+#ifdef CONFIG_CPU_SUP_INTEL
+
+int intel_pmu_save_and_restart(struct perf_event *event);
+
+struct event_constraint *
+x86_get_event_constraints(struct cpu_hw_events *cpuc, struct perf_event *event);
+
+struct intel_shared_regs *allocate_shared_regs(int cpu);
+
+int intel_pmu_init(void);
+
+void init_debug_store_on_cpu(int cpu);
+
+void fini_debug_store_on_cpu(int cpu);
+
+void release_ds_buffers(void);
+
+void reserve_ds_buffers(void);
+
+extern struct event_constraint bts_constraint;
+
+void intel_pmu_enable_bts(u64 config);
+
+void intel_pmu_disable_bts(void);
+
+int intel_pmu_drain_bts_buffer(void);
+
+extern struct event_constraint intel_core2_pebs_event_constraints[];
+
+extern struct event_constraint intel_atom_pebs_event_constraints[];
+
+extern struct event_constraint intel_nehalem_pebs_event_constraints[];
+
+extern struct event_constraint intel_westmere_pebs_event_constraints[];
+
+extern struct event_constraint intel_snb_pebs_event_constraints[];
+
+struct event_constraint *intel_pebs_constraints(struct perf_event *event);
+
+void intel_pmu_pebs_enable(struct perf_event *event);
+
+void intel_pmu_pebs_disable(struct perf_event *event);
+
+void intel_pmu_pebs_enable_all(void);
+
+void intel_pmu_pebs_disable_all(void);
+
+void intel_ds_init(void);
+
+void intel_pmu_lbr_reset(void);
+
+void intel_pmu_lbr_enable(struct perf_event *event);
+
+void intel_pmu_lbr_disable(struct perf_event *event);
+
+void intel_pmu_lbr_enable_all(void);
+
+void intel_pmu_lbr_disable_all(void);
+
+void intel_pmu_lbr_read(void);
+
+void intel_pmu_lbr_init_core(void);
+
+void intel_pmu_lbr_init_nhm(void);
+
+void intel_pmu_lbr_init_atom(void);
+
+int p4_pmu_init(void);
+
+int p6_pmu_init(void);
+
+#else /* CONFIG_CPU_SUP_INTEL */
+
+static inline void reserve_ds_buffers(void)
+{
+}
+
+static inline void release_ds_buffers(void)
+{
+}
+
+static inline int intel_pmu_init(void)
+{
+	return 0;
+}
+
+static inline struct intel_shared_regs *allocate_shared_regs(int cpu)
+{
+	return NULL;
+}
+
+#endif /* CONFIG_CPU_SUP_INTEL */
diff --git a/arch/x86/kernel/cpu/perf_event_amd.c b/arch/x86/kernel/cpu/perf_event_amd.c
index ee9436c..ed334c8 100644
--- a/arch/x86/kernel/cpu/perf_event_amd.c
+++ b/arch/x86/kernel/cpu/perf_event_amd.c
@@ -1,4 +1,9 @@
-#ifdef CONFIG_CPU_SUP_AMD
+#include <linux/perf_event.h>
+#include <linux/types.h>
+#include <linux/init.h>
+#include <linux/slab.h>
+
+#include "perf_event.h"
 
 static __initconst const u64 amd_hw_cache_event_ids
 				[PERF_COUNT_HW_CACHE_MAX]
@@ -573,7 +578,7 @@ static __initconst const struct x86_pmu amd_pmu_f15h = {
 #endif
 };
 
-static __init int amd_pmu_init(void)
+__init int amd_pmu_init(void)
 {
 	/* Performance-monitoring supported from K7 and later: */
 	if (boot_cpu_data.x86 < 6)
@@ -602,12 +607,3 @@ static __init int amd_pmu_init(void)
 
 	return 0;
 }
-
-#else /* CONFIG_CPU_SUP_AMD */
-
-static int amd_pmu_init(void)
-{
-	return 0;
-}
-
-#endif
diff --git a/arch/x86/kernel/cpu/perf_event_intel.c b/arch/x86/kernel/cpu/perf_event_intel.c
index 3751494..6f7626c 100644
--- a/arch/x86/kernel/cpu/perf_event_intel.c
+++ b/arch/x86/kernel/cpu/perf_event_intel.c
@@ -1,16 +1,18 @@
-#ifdef CONFIG_CPU_SUP_INTEL
-
 /*
  * Per core/cpu state
  *
  * Used to coordinate shared registers between HT threads or
  * among events on a single PMU.
  */
-struct intel_shared_regs {
-	struct er_account       regs[EXTRA_REG_MAX];
-	int                     refcnt;		/* per-core: #HT threads */
-	unsigned                core_id;	/* per-core: core id */
-};
+
+#include <linux/stddef.h>
+#include <linux/types.h>
+#include <linux/init.h>
+#include <linux/slab.h>
+
+#include <asm/hardirq.h>
+
+#include "perf_event.h"
 
 /*
  * Intel PerfMon, used on Core and later.
@@ -945,7 +947,7 @@ static void intel_pmu_enable_event(struct perf_event *event)
  * Save and restart an expired event. Called by NMI contexts,
  * so it has to be careful about preempting normal event ops:
  */
-static int intel_pmu_save_and_restart(struct perf_event *event)
+int intel_pmu_save_and_restart(struct perf_event *event)
 {
 	x86_perf_event_update(event);
 	return x86_perf_event_set_period(event);
@@ -1197,6 +1199,21 @@ intel_shared_regs_constraints(struct cpu_hw_events *cpuc,
 	return c;
 }
 
+struct event_constraint *
+x86_get_event_constraints(struct cpu_hw_events *cpuc, struct perf_event *event)
+{
+	struct event_constraint *c;
+
+	if (x86_pmu.event_constraints) {
+		for_each_event_constraint(c, x86_pmu.event_constraints) {
+			if ((event->hw.config & c->cmask) == c->code)
+				return c;
+		}
+	}
+
+	return &unconstrained;
+}
+
 static struct event_constraint *
 intel_get_event_constraints(struct cpu_hw_events *cpuc, struct perf_event *event)
 {
@@ -1309,7 +1326,7 @@ static __initconst const struct x86_pmu core_pmu = {
 	.event_constraints	= intel_core_event_constraints,
 };
 
-static struct intel_shared_regs *allocate_shared_regs(int cpu)
+struct intel_shared_regs *allocate_shared_regs(int cpu)
 {
 	struct intel_shared_regs *regs;
 	int i;
@@ -1441,7 +1458,7 @@ static void intel_clovertown_quirks(void)
 	x86_pmu.pebs_constraints = NULL;
 }
 
-static __init int intel_pmu_init(void)
+__init int intel_pmu_init(void)
 {
 	union cpuid10_edx edx;
 	union cpuid10_eax eax;
@@ -1597,7 +1614,7 @@ static __init int intel_pmu_init(void)
 		intel_pmu_lbr_init_nhm();
 
 		x86_pmu.event_constraints = intel_snb_event_constraints;
-		x86_pmu.pebs_constraints = intel_snb_pebs_events;
+		x86_pmu.pebs_constraints = intel_snb_pebs_event_constraints;
 		x86_pmu.extra_regs = intel_snb_extra_regs;
 		/* all extra regs are per-cpu when HT is on */
 		x86_pmu.er_flags |= ERF_HAS_RSP_1;
@@ -1628,16 +1645,3 @@ static __init int intel_pmu_init(void)
 	}
 	return 0;
 }
-
-#else /* CONFIG_CPU_SUP_INTEL */
-
-static int intel_pmu_init(void)
-{
-	return 0;
-}
-
-static struct intel_shared_regs *allocate_shared_regs(int cpu)
-{
-	return NULL;
-}
-#endif /* CONFIG_CPU_SUP_INTEL */
diff --git a/arch/x86/kernel/cpu/perf_event_intel_ds.c b/arch/x86/kernel/cpu/perf_event_intel_ds.c
index 1b1ef3a..c0d238f 100644
--- a/arch/x86/kernel/cpu/perf_event_intel_ds.c
+++ b/arch/x86/kernel/cpu/perf_event_intel_ds.c
@@ -1,7 +1,10 @@
-#ifdef CONFIG_CPU_SUP_INTEL
+#include <linux/bitops.h>
+#include <linux/types.h>
+#include <linux/slab.h>
 
-/* The maximal number of PEBS events: */
-#define MAX_PEBS_EVENTS		4
+#include <asm/perf_event.h>
+
+#include "perf_event.h"
 
 /* The size of a BTS record in bytes: */
 #define BTS_RECORD_SIZE		24
@@ -37,24 +40,7 @@ struct pebs_record_nhm {
 	u64 status, dla, dse, lat;
 };
 
-/*
- * A debug store configuration.
- *
- * We only support architectures that use 64bit fields.
- */
-struct debug_store {
-	u64	bts_buffer_base;
-	u64	bts_index;
-	u64	bts_absolute_maximum;
-	u64	bts_interrupt_threshold;
-	u64	pebs_buffer_base;
-	u64	pebs_index;
-	u64	pebs_absolute_maximum;
-	u64	pebs_interrupt_threshold;
-	u64	pebs_event_reset[MAX_PEBS_EVENTS];
-};
-
-static void init_debug_store_on_cpu(int cpu)
+void init_debug_store_on_cpu(int cpu)
 {
 	struct debug_store *ds = per_cpu(cpu_hw_events, cpu).ds;
 
@@ -66,7 +52,7 @@ static void init_debug_store_on_cpu(int cpu)
 		     (u32)((u64)(unsigned long)ds >> 32));
 }
 
-static void fini_debug_store_on_cpu(int cpu)
+void fini_debug_store_on_cpu(int cpu)
 {
 	if (!per_cpu(cpu_hw_events, cpu).ds)
 		return;
@@ -175,7 +161,7 @@ static void release_ds_buffer(int cpu)
 	kfree(ds);
 }
 
-static void release_ds_buffers(void)
+void release_ds_buffers(void)
 {
 	int cpu;
 
@@ -194,7 +180,7 @@ static void release_ds_buffers(void)
 	put_online_cpus();
 }
 
-static void reserve_ds_buffers(void)
+void reserve_ds_buffers(void)
 {
 	int bts_err = 0, pebs_err = 0;
 	int cpu;
@@ -260,10 +246,10 @@ static void reserve_ds_buffers(void)
  * BTS
  */
 
-static struct event_constraint bts_constraint =
+struct event_constraint bts_constraint =
 	EVENT_CONSTRAINT(0, 1ULL << X86_PMC_IDX_FIXED_BTS, 0);
 
-static void intel_pmu_enable_bts(u64 config)
+void intel_pmu_enable_bts(u64 config)
 {
 	unsigned long debugctlmsr;
 
@@ -282,7 +268,7 @@ static void intel_pmu_enable_bts(u64 config)
 	update_debugctlmsr(debugctlmsr);
 }
 
-static void intel_pmu_disable_bts(void)
+void intel_pmu_disable_bts(void)
 {
 	struct cpu_hw_events *cpuc = &__get_cpu_var(cpu_hw_events);
 	unsigned long debugctlmsr;
@@ -299,7 +285,7 @@ static void intel_pmu_disable_bts(void)
 	update_debugctlmsr(debugctlmsr);
 }
 
-static int intel_pmu_drain_bts_buffer(void)
+int intel_pmu_drain_bts_buffer(void)
 {
 	struct cpu_hw_events *cpuc = &__get_cpu_var(cpu_hw_events);
 	struct debug_store *ds = cpuc->ds;
@@ -361,7 +347,7 @@ static int intel_pmu_drain_bts_buffer(void)
 /*
  * PEBS
  */
-static struct event_constraint intel_core2_pebs_event_constraints[] = {
+struct event_constraint intel_core2_pebs_event_constraints[] = {
 	INTEL_UEVENT_CONSTRAINT(0x00c0, 0x1), /* INST_RETIRED.ANY */
 	INTEL_UEVENT_CONSTRAINT(0xfec1, 0x1), /* X87_OPS_RETIRED.ANY */
 	INTEL_UEVENT_CONSTRAINT(0x00c5, 0x1), /* BR_INST_RETIRED.MISPRED */
@@ -370,14 +356,14 @@ static struct event_constraint intel_core2_pebs_event_constraints[] = {
 	EVENT_CONSTRAINT_END
 };
 
-static struct event_constraint intel_atom_pebs_event_constraints[] = {
+struct event_constraint intel_atom_pebs_event_constraints[] = {
 	INTEL_UEVENT_CONSTRAINT(0x00c0, 0x1), /* INST_RETIRED.ANY */
 	INTEL_UEVENT_CONSTRAINT(0x00c5, 0x1), /* MISPREDICTED_BRANCH_RETIRED */
 	INTEL_EVENT_CONSTRAINT(0xcb, 0x1),    /* MEM_LOAD_RETIRED.* */
 	EVENT_CONSTRAINT_END
 };
 
-static struct event_constraint intel_nehalem_pebs_event_constraints[] = {
+struct event_constraint intel_nehalem_pebs_event_constraints[] = {
 	INTEL_EVENT_CONSTRAINT(0x0b, 0xf),    /* MEM_INST_RETIRED.* */
 	INTEL_EVENT_CONSTRAINT(0x0f, 0xf),    /* MEM_UNCORE_RETIRED.* */
 	INTEL_UEVENT_CONSTRAINT(0x010c, 0xf), /* MEM_STORE_RETIRED.DTLB_MISS */
@@ -392,7 +378,7 @@ static struct event_constraint intel_nehalem_pebs_event_constraints[] = {
 	EVENT_CONSTRAINT_END
 };
 
-static struct event_constraint intel_westmere_pebs_event_constraints[] = {
+struct event_constraint intel_westmere_pebs_event_constraints[] = {
 	INTEL_EVENT_CONSTRAINT(0x0b, 0xf),    /* MEM_INST_RETIRED.* */
 	INTEL_EVENT_CONSTRAINT(0x0f, 0xf),    /* MEM_UNCORE_RETIRED.* */
 	INTEL_UEVENT_CONSTRAINT(0x010c, 0xf), /* MEM_STORE_RETIRED.DTLB_MISS */
@@ -407,7 +393,7 @@ static struct event_constraint intel_westmere_pebs_event_constraints[] = {
 	EVENT_CONSTRAINT_END
 };
 
-static struct event_constraint intel_snb_pebs_events[] = {
+struct event_constraint intel_snb_pebs_event_constraints[] = {
 	INTEL_UEVENT_CONSTRAINT(0x01c0, 0x2), /* INST_RETIRED.PRECDIST */
 	INTEL_UEVENT_CONSTRAINT(0x01c2, 0xf), /* UOPS_RETIRED.ALL */
 	INTEL_UEVENT_CONSTRAINT(0x02c2, 0xf), /* UOPS_RETIRED.RETIRE_SLOTS */
@@ -428,8 +414,7 @@ static struct event_constraint intel_snb_pebs_events[] = {
 	EVENT_CONSTRAINT_END
 };
 
-static struct event_constraint *
-intel_pebs_constraints(struct perf_event *event)
+struct event_constraint *intel_pebs_constraints(struct perf_event *event)
 {
 	struct event_constraint *c;
 
@@ -446,7 +431,7 @@ intel_pebs_constraints(struct perf_event *event)
 	return &emptyconstraint;
 }
 
-static void intel_pmu_pebs_enable(struct perf_event *event)
+void intel_pmu_pebs_enable(struct perf_event *event)
 {
 	struct cpu_hw_events *cpuc = &__get_cpu_var(cpu_hw_events);
 	struct hw_perf_event *hwc = &event->hw;
@@ -460,7 +445,7 @@ static void intel_pmu_pebs_enable(struct perf_event *event)
 		intel_pmu_lbr_enable(event);
 }
 
-static void intel_pmu_pebs_disable(struct perf_event *event)
+void intel_pmu_pebs_disable(struct perf_event *event)
 {
 	struct cpu_hw_events *cpuc = &__get_cpu_var(cpu_hw_events);
 	struct hw_perf_event *hwc = &event->hw;
@@ -475,7 +460,7 @@ static void intel_pmu_pebs_disable(struct perf_event *event)
 		intel_pmu_lbr_disable(event);
 }
 
-static void intel_pmu_pebs_enable_all(void)
+void intel_pmu_pebs_enable_all(void)
 {
 	struct cpu_hw_events *cpuc = &__get_cpu_var(cpu_hw_events);
 
@@ -483,7 +468,7 @@ static void intel_pmu_pebs_enable_all(void)
 		wrmsrl(MSR_IA32_PEBS_ENABLE, cpuc->pebs_enabled);
 }
 
-static void intel_pmu_pebs_disable_all(void)
+void intel_pmu_pebs_disable_all(void)
 {
 	struct cpu_hw_events *cpuc = &__get_cpu_var(cpu_hw_events);
 
@@ -576,8 +561,6 @@ static int intel_pmu_pebs_fixup_ip(struct pt_regs *regs)
 	return 0;
 }
 
-static int intel_pmu_save_and_restart(struct perf_event *event);
-
 static void __intel_pmu_pebs_event(struct perf_event *event,
 				   struct pt_regs *iregs, void *__pebs)
 {
@@ -716,7 +699,7 @@ static void intel_pmu_drain_pebs_nhm(struct pt_regs *iregs)
  * BTS, PEBS probe and setup
  */
 
-static void intel_ds_init(void)
+void intel_ds_init(void)
 {
 	/*
 	 * No support for 32bit formats
@@ -749,15 +732,3 @@ static void intel_ds_init(void)
 		}
 	}
 }
-
-#else /* CONFIG_CPU_SUP_INTEL */
-
-static void reserve_ds_buffers(void)
-{
-}
-
-static void release_ds_buffers(void)
-{
-}
-
-#endif /* CONFIG_CPU_SUP_INTEL */
diff --git a/arch/x86/kernel/cpu/perf_event_intel_lbr.c b/arch/x86/kernel/cpu/perf_event_intel_lbr.c
index d202c1b..3fab3de 100644
--- a/arch/x86/kernel/cpu/perf_event_intel_lbr.c
+++ b/arch/x86/kernel/cpu/perf_event_intel_lbr.c
@@ -1,4 +1,10 @@
-#ifdef CONFIG_CPU_SUP_INTEL
+#include <linux/perf_event.h>
+#include <linux/types.h>
+
+#include <asm/perf_event.h>
+#include <asm/msr.h>
+
+#include "perf_event.h"
 
 enum {
 	LBR_FORMAT_32		= 0x00,
@@ -48,7 +54,7 @@ static void intel_pmu_lbr_reset_64(void)
 	}
 }
 
-static void intel_pmu_lbr_reset(void)
+void intel_pmu_lbr_reset(void)
 {
 	if (!x86_pmu.lbr_nr)
 		return;
@@ -59,7 +65,7 @@ static void intel_pmu_lbr_reset(void)
 		intel_pmu_lbr_reset_64();
 }
 
-static void intel_pmu_lbr_enable(struct perf_event *event)
+void intel_pmu_lbr_enable(struct perf_event *event)
 {
 	struct cpu_hw_events *cpuc = &__get_cpu_var(cpu_hw_events);
 
@@ -81,7 +87,7 @@ static void intel_pmu_lbr_enable(struct perf_event *event)
 	cpuc->lbr_users++;
 }
 
-static void intel_pmu_lbr_disable(struct perf_event *event)
+void intel_pmu_lbr_disable(struct perf_event *event)
 {
 	struct cpu_hw_events *cpuc = &__get_cpu_var(cpu_hw_events);
 
@@ -95,7 +101,7 @@ static void intel_pmu_lbr_disable(struct perf_event *event)
 		__intel_pmu_lbr_disable();
 }
 
-static void intel_pmu_lbr_enable_all(void)
+void intel_pmu_lbr_enable_all(void)
 {
 	struct cpu_hw_events *cpuc = &__get_cpu_var(cpu_hw_events);
 
@@ -103,7 +109,7 @@ static void intel_pmu_lbr_enable_all(void)
 		__intel_pmu_lbr_enable();
 }
 
-static void intel_pmu_lbr_disable_all(void)
+void intel_pmu_lbr_disable_all(void)
 {
 	struct cpu_hw_events *cpuc = &__get_cpu_var(cpu_hw_events);
 
@@ -178,7 +184,7 @@ static void intel_pmu_lbr_read_64(struct cpu_hw_events *cpuc)
 	cpuc->lbr_stack.nr = i;
 }
 
-static void intel_pmu_lbr_read(void)
+void intel_pmu_lbr_read(void)
 {
 	struct cpu_hw_events *cpuc = &__get_cpu_var(cpu_hw_events);
 
@@ -191,7 +197,7 @@ static void intel_pmu_lbr_read(void)
 		intel_pmu_lbr_read_64(cpuc);
 }
 
-static void intel_pmu_lbr_init_core(void)
+void intel_pmu_lbr_init_core(void)
 {
 	x86_pmu.lbr_nr     = 4;
 	x86_pmu.lbr_tos    = 0x01c9;
@@ -199,7 +205,7 @@ static void intel_pmu_lbr_init_core(void)
 	x86_pmu.lbr_to     = 0x60;
 }
 
-static void intel_pmu_lbr_init_nhm(void)
+void intel_pmu_lbr_init_nhm(void)
 {
 	x86_pmu.lbr_nr     = 16;
 	x86_pmu.lbr_tos    = 0x01c9;
@@ -207,12 +213,10 @@ static void intel_pmu_lbr_init_nhm(void)
 	x86_pmu.lbr_to     = 0x6c0;
 }
 
-static void intel_pmu_lbr_init_atom(void)
+void intel_pmu_lbr_init_atom(void)
 {
 	x86_pmu.lbr_nr	   = 8;
 	x86_pmu.lbr_tos    = 0x01c9;
 	x86_pmu.lbr_from   = 0x40;
 	x86_pmu.lbr_to     = 0x60;
 }
-
-#endif /* CONFIG_CPU_SUP_INTEL */
diff --git a/arch/x86/kernel/cpu/perf_event_p4.c b/arch/x86/kernel/cpu/perf_event_p4.c
index 7809d2bc..5edb266 100644
--- a/arch/x86/kernel/cpu/perf_event_p4.c
+++ b/arch/x86/kernel/cpu/perf_event_p4.c
@@ -7,9 +7,12 @@
  *  For licencing details see kernel-base/COPYING
  */
 
-#ifdef CONFIG_CPU_SUP_INTEL
+#include <linux/perf_event.h>
 
 #include <asm/perf_event_p4.h>
+#include <asm/hardirq.h>
+
+#include "perf_event.h"
 
 #define P4_CNTR_LIMIT 3
 /*
@@ -1303,7 +1306,7 @@ static __initconst const struct x86_pmu p4_pmu = {
 	.perfctr_second_write	= 1,
 };
 
-static __init int p4_pmu_init(void)
+__init int p4_pmu_init(void)
 {
 	unsigned int low, high;
 
@@ -1326,5 +1329,3 @@ static __init int p4_pmu_init(void)
 
 	return 0;
 }
-
-#endif /* CONFIG_CPU_SUP_INTEL */
diff --git a/arch/x86/kernel/cpu/perf_event_p6.c b/arch/x86/kernel/cpu/perf_event_p6.c
index 20c097e..c7181be 100644
--- a/arch/x86/kernel/cpu/perf_event_p6.c
+++ b/arch/x86/kernel/cpu/perf_event_p6.c
@@ -1,4 +1,7 @@
-#ifdef CONFIG_CPU_SUP_INTEL
+#include <linux/perf_event.h>
+#include <linux/types.h>
+
+#include "perf_event.h"
 
 /*
  * Not sure about some of these
@@ -114,7 +117,7 @@ static __initconst const struct x86_pmu p6_pmu = {
 	.event_constraints	= p6_event_constraints,
 };
 
-static __init int p6_pmu_init(void)
+__init int p6_pmu_init(void)
 {
 	switch (boot_cpu_data.x86_model) {
 	case 1:
@@ -138,5 +141,3 @@ static __init int p6_pmu_init(void)
 
 	return 0;
 }
-
-#endif /* CONFIG_CPU_SUP_INTEL */
-- 
1.7.6.1

Re: [PATCH] x86: perf: Clean up perf_event cpu code

[Peter Zijlstra]

On Tue, 2011-08-30 at 20:41 -0300, Kevin Winchester wrote:
> The CPU support for perf events on x86 was implemented via included C files
> with #ifdefs.  Clean this up by creating a new header file and compiling
> the vendor-specific files as needed.
> 
> Signed-off-by: Kevin Winchester <kjwinchester@gmail.com>
> ---
> 
> This is now against the tip tree.  I left it all as one big commit for now,
> although I can break it up if necessary.  I'm just not sure it would gain much.
> 
> Again, this is by far the largest out of the handful of patches I have ever
> submitted, so please let me know if I am doing anything wrong.
> 

I've stuck it in the queue,. seems to work so far.

If Ingo doesn't complain, it ought to show up for the next release
someplace ;-)

Re: [PATCH] x86: perf: Clean up perf_event cpu code

[Peter Zijlstra]

On Fri, 2011-09-02 at 15:32 +0200, Peter Zijlstra wrote:
> On Tue, 2011-08-30 at 20:41 -0300, Kevin Winchester wrote:
> > The CPU support for perf events on x86 was implemented via included C files
> > with #ifdefs.  Clean this up by creating a new header file and compiling
> > the vendor-specific files as needed.
> > 
> > Signed-off-by: Kevin Winchester <kjwinchester@gmail.com>
> > ---
> > 
> > This is now against the tip tree.  I left it all as one big commit for now,
> > although I can break it up if necessary.  I'm just not sure it would gain much.
> > 
> > Again, this is by far the largest out of the handful of patches I have ever
> > submitted, so please let me know if I am doing anything wrong.
> > 
> 
> I've stuck it in the queue,. seems to work so far.
> 
> If Ingo doesn't complain, it ought to show up for the next release
> someplace ;-)

Ingo found that it failed to build on i386-up, the below delta (already
folded in the patch) cured things:


---
Index: linux-2.6/arch/x86/kernel/cpu/perf_event_intel.c
===================================================================
--- linux-2.6.orig/arch/x86/kernel/cpu/perf_event_intel.c
+++ linux-2.6/arch/x86/kernel/cpu/perf_event_intel.c
@@ -11,6 +11,7 @@
 #include <linux/slab.h>
 
 #include <asm/hardirq.h>
+#include <asm/apic.h>
 
 #include "perf_event.h"
 
Index: linux-2.6/arch/x86/kernel/cpu/perf_event_p4.c
===================================================================
--- linux-2.6.orig/arch/x86/kernel/cpu/perf_event_p4.c
+++ linux-2.6/arch/x86/kernel/cpu/perf_event_p4.c
@@ -11,6 +11,7 @@
 
 #include <asm/perf_event_p4.h>
 #include <asm/hardirq.h>
+#include <asm/apic.h>
 
 #include "perf_event.h"