[PATCH 0/7] updates for oprofile

[PATCH 0/7] updates for oprofile

[Robert Richter]

This patch set contains updates for oprofile, mainly with some reworks
of x86 code that improves the error handling.

-Robert

[PATCH 1/7] oprofile: update file list in MAINTAINERS file

[Robert Richter]

File list now catches:

 $ xargs | eval ls -d $(cat) | sort -u
 arch/*/include/asm/oprofile*.h
 arch/*/oprofile/
 drivers/oprofile/
 include/linux/oprofile.h

 arch/alpha/oprofile/
 arch/arm/oprofile/
 arch/avr32/oprofile/
 arch/blackfin/oprofile/
 arch/ia64/oprofile/
 arch/m32r/oprofile/
 arch/microblaze/oprofile/
 arch/mips/oprofile/
 arch/mn10300/oprofile/
 arch/parisc/oprofile/
 arch/powerpc/include/asm/oprofile_impl.h
 arch/powerpc/oprofile/
 arch/s390/oprofile/
 arch/sh/oprofile/
 arch/sparc/oprofile/
 arch/x86/oprofile/
 drivers/oprofile/
 include/linux/oprofile.h

Signed-off-by: Robert Richter <robert.richter@amd.com>
---
 MAINTAINERS |    1 +
 1 files changed, 1 insertions(+), 0 deletions(-)

diff --git a/MAINTAINERS b/MAINTAINERS
index a0e3c3a..31e8586 100644
--- a/MAINTAINERS
+++ b/MAINTAINERS
@@ -4165,6 +4165,7 @@ OPROFILE
 M:	Robert Richter <robert.richter@amd.com>
 L:	oprofile-list@lists.sf.net
 S:	Maintained
+F:	arch/*/include/asm/oprofile*.h
 F:	arch/*/oprofile/
 F:	drivers/oprofile/
 F:	include/linux/oprofile.h
-- 
1.7.0.3

[PATCH 2/7] oprofile/x86: rework error handler in nmi_setup()

[Robert Richter]

This patch improves the error handler in nmi_setup(). Most parts of
the code are moved to allocate_msrs(). In case of an error
allocate_msrs() also frees already allocated memory. nmi_setup()
becomes easier and better extendable.

Signed-off-by: Robert Richter <robert.richter@amd.com>
---
 arch/x86/oprofile/nmi_int.c |   33 +++++++++++++++++++--------------
 1 files changed, 19 insertions(+), 14 deletions(-)

diff --git a/arch/x86/oprofile/nmi_int.c b/arch/x86/oprofile/nmi_int.c
index 2c505ee..c0c21f2 100644
--- a/arch/x86/oprofile/nmi_int.c
+++ b/arch/x86/oprofile/nmi_int.c
@@ -295,6 +295,7 @@ static void free_msrs(void)
 		kfree(per_cpu(cpu_msrs, i).controls);
 		per_cpu(cpu_msrs, i).controls = NULL;
 	}
+	nmi_shutdown_mux();
 }
 
 static int allocate_msrs(void)
@@ -307,14 +308,21 @@ static int allocate_msrs(void)
 		per_cpu(cpu_msrs, i).counters = kzalloc(counters_size,
 							GFP_KERNEL);
 		if (!per_cpu(cpu_msrs, i).counters)
-			return 0;
+			goto fail;
 		per_cpu(cpu_msrs, i).controls = kzalloc(controls_size,
 							GFP_KERNEL);
 		if (!per_cpu(cpu_msrs, i).controls)
-			return 0;
+			goto fail;
 	}
 
+	if (!nmi_setup_mux())
+		goto fail;
+
 	return 1;
+
+fail:
+	free_msrs();
+	return 0;
 }
 
 static void nmi_cpu_setup(void *dummy)
@@ -342,17 +350,7 @@ static int nmi_setup(void)
 	int cpu;
 
 	if (!allocate_msrs())
-		err = -ENOMEM;
-	else if (!nmi_setup_mux())
-		err = -ENOMEM;
-	else
-		err = register_die_notifier(&profile_exceptions_nb);
-
-	if (err) {
-		free_msrs();
-		nmi_shutdown_mux();
-		return err;
-	}
+		return -ENOMEM;
 
 	/* We need to serialize save and setup for HT because the subset
 	 * of msrs are distinct for save and setup operations
@@ -374,9 +372,17 @@ static int nmi_setup(void)
 
 		mux_clone(cpu);
 	}
+
+	err = register_die_notifier(&profile_exceptions_nb);
+	if (err)
+		goto fail;
+
 	on_each_cpu(nmi_cpu_setup, NULL, 1);
 	nmi_enabled = 1;
 	return 0;
+fail:
+	free_msrs();
+	return err;
 }
 
 static void nmi_cpu_restore_registers(struct op_msrs *msrs)
@@ -421,7 +427,6 @@ static void nmi_shutdown(void)
 	nmi_enabled = 0;
 	on_each_cpu(nmi_cpu_shutdown, NULL, 1);
 	unregister_die_notifier(&profile_exceptions_nb);
-	nmi_shutdown_mux();
 	msrs = &get_cpu_var(cpu_msrs);
 	model->shutdown(msrs);
 	free_msrs();
-- 
1.7.0.3

[PATCH 3/7] oprofile/x86: reserve counter msrs pairwise

[Robert Richter]

For AMD's and Intel's P6 generic performance counters have pairwise
counter and control msrs. This patch changes the counter reservation
in a way that both msrs must be registered. It joins some counter
loops and also removes the unnecessary NUM_CONTROLS macro in the AMD
implementation.

Signed-off-by: Robert Richter <robert.richter@amd.com>
---
 arch/x86/oprofile/op_model_amd.c  |   43 ++++++++++++++++--------------------
 arch/x86/oprofile/op_model_ppro.c |   36 ++++++++++++++----------------
 2 files changed, 36 insertions(+), 43 deletions(-)

diff --git a/arch/x86/oprofile/op_model_amd.c b/arch/x86/oprofile/op_model_amd.c
index 090cbbe..2e2bc90 100644
--- a/arch/x86/oprofile/op_model_amd.c
+++ b/arch/x86/oprofile/op_model_amd.c
@@ -30,13 +30,10 @@
 #include "op_counter.h"
 
 #define NUM_COUNTERS 4
-#define NUM_CONTROLS 4
 #ifdef CONFIG_OPROFILE_EVENT_MULTIPLEX
 #define NUM_VIRT_COUNTERS 32
-#define NUM_VIRT_CONTROLS 32
 #else
 #define NUM_VIRT_COUNTERS NUM_COUNTERS
-#define NUM_VIRT_CONTROLS NUM_CONTROLS
 #endif
 
 #define OP_EVENT_MASK			0x0FFF
@@ -134,13 +131,15 @@ static void op_amd_fill_in_addresses(struct op_msrs * const msrs)
 	int i;
 
 	for (i = 0; i < NUM_COUNTERS; i++) {
-		if (reserve_perfctr_nmi(MSR_K7_PERFCTR0 + i))
-			msrs->counters[i].addr = MSR_K7_PERFCTR0 + i;
-	}
-
-	for (i = 0; i < NUM_CONTROLS; i++) {
-		if (reserve_evntsel_nmi(MSR_K7_EVNTSEL0 + i))
-			msrs->controls[i].addr = MSR_K7_EVNTSEL0 + i;
+		if (!reserve_perfctr_nmi(MSR_K7_PERFCTR0 + i))
+			continue;
+		if (!reserve_evntsel_nmi(MSR_K7_EVNTSEL0 + i)) {
+			release_perfctr_nmi(MSR_K7_PERFCTR0 + i);
+			continue;
+		}
+		/* both registers must be reserved */
+		msrs->counters[i].addr = MSR_K7_PERFCTR0 + i;
+		msrs->controls[i].addr = MSR_K7_EVNTSEL0 + i;
 	}
 }
 
@@ -160,7 +159,7 @@ static void op_amd_setup_ctrs(struct op_x86_model_spec const *model,
 	}
 
 	/* clear all counters */
-	for (i = 0; i < NUM_CONTROLS; ++i) {
+	for (i = 0; i < NUM_COUNTERS; ++i) {
 		if (unlikely(!msrs->controls[i].addr)) {
 			if (counter_config[i].enabled && !smp_processor_id())
 				/*
@@ -175,12 +174,10 @@ static void op_amd_setup_ctrs(struct op_x86_model_spec const *model,
 			op_x86_warn_in_use(i);
 		val &= model->reserved;
 		wrmsrl(msrs->controls[i].addr, val);
-	}
-
-	/* avoid a false detection of ctr overflows in NMI handler */
-	for (i = 0; i < NUM_COUNTERS; ++i) {
-		if (unlikely(!msrs->counters[i].addr))
-			continue;
+		/*
+		 * avoid a false detection of ctr overflows in NMI
+		 * handler
+		 */
 		wrmsrl(msrs->counters[i].addr, -1LL);
 	}
 
@@ -430,12 +427,10 @@ static void op_amd_shutdown(struct op_msrs const * const msrs)
 	int i;
 
 	for (i = 0; i < NUM_COUNTERS; ++i) {
-		if (msrs->counters[i].addr)
-			release_perfctr_nmi(MSR_K7_PERFCTR0 + i);
-	}
-	for (i = 0; i < NUM_CONTROLS; ++i) {
-		if (msrs->controls[i].addr)
-			release_evntsel_nmi(MSR_K7_EVNTSEL0 + i);
+		if (!msrs->counters[i].addr)
+			continue;
+		release_perfctr_nmi(MSR_K7_PERFCTR0 + i);
+		release_evntsel_nmi(MSR_K7_EVNTSEL0 + i);
 	}
 }
 
@@ -583,7 +578,7 @@ static void op_amd_exit(void)
 
 struct op_x86_model_spec op_amd_spec = {
 	.num_counters		= NUM_COUNTERS,
-	.num_controls		= NUM_CONTROLS,
+	.num_controls		= NUM_COUNTERS,
 	.num_virt_counters	= NUM_VIRT_COUNTERS,
 	.reserved		= MSR_AMD_EVENTSEL_RESERVED,
 	.event_mask		= OP_EVENT_MASK,
diff --git a/arch/x86/oprofile/op_model_ppro.c b/arch/x86/oprofile/op_model_ppro.c
index 2bf90fa..f8e268e 100644
--- a/arch/x86/oprofile/op_model_ppro.c
+++ b/arch/x86/oprofile/op_model_ppro.c
@@ -35,13 +35,15 @@ static void ppro_fill_in_addresses(struct op_msrs * const msrs)
 	int i;
 
 	for (i = 0; i < num_counters; i++) {
-		if (reserve_perfctr_nmi(MSR_P6_PERFCTR0 + i))
-			msrs->counters[i].addr = MSR_P6_PERFCTR0 + i;
-	}
-
-	for (i = 0; i < num_counters; i++) {
-		if (reserve_evntsel_nmi(MSR_P6_EVNTSEL0 + i))
-			msrs->controls[i].addr = MSR_P6_EVNTSEL0 + i;
+		if (!reserve_perfctr_nmi(MSR_P6_PERFCTR0 + i))
+			continue;
+		if (!reserve_evntsel_nmi(MSR_P6_EVNTSEL0 + i)) {
+			release_perfctr_nmi(MSR_P6_PERFCTR0 + i);
+			continue;
+		}
+		/* both registers must be reserved */
+		msrs->counters[i].addr = MSR_P6_PERFCTR0 + i;
+		msrs->controls[i].addr = MSR_P6_EVNTSEL0 + i;
 	}
 }
 
@@ -92,12 +94,10 @@ static void ppro_setup_ctrs(struct op_x86_model_spec const *model,
 			op_x86_warn_in_use(i);
 		val &= model->reserved;
 		wrmsrl(msrs->controls[i].addr, val);
-	}
-
-	/* avoid a false detection of ctr overflows in NMI handler */
-	for (i = 0; i < num_counters; ++i) {
-		if (unlikely(!msrs->counters[i].addr))
-			continue;
+		/*
+		 * avoid a false detection of ctr overflows in NMI *
+		 * handler
+		 */
 		wrmsrl(msrs->counters[i].addr, -1LL);
 	}
 
@@ -194,12 +194,10 @@ static void ppro_shutdown(struct op_msrs const * const msrs)
 	int i;
 
 	for (i = 0; i < num_counters; ++i) {
-		if (msrs->counters[i].addr)
-			release_perfctr_nmi(MSR_P6_PERFCTR0 + i);
-	}
-	for (i = 0; i < num_counters; ++i) {
-		if (msrs->controls[i].addr)
-			release_evntsel_nmi(MSR_P6_EVNTSEL0 + i);
+		if (!msrs->counters[i].addr)
+			continue;
+		release_perfctr_nmi(MSR_P6_PERFCTR0 + i);
+		release_evntsel_nmi(MSR_P6_EVNTSEL0 + i);
 	}
 	if (reset_value) {
 		kfree(reset_value);
-- 
1.7.0.3

[PATCH 4/7] oprofile/x86: moving shutdown functions

[Robert Richter]

Moving some code in preparation of the next patch.

Signed-off-by: Robert Richter <robert.richter@amd.com>
---
 arch/x86/oprofile/op_model_amd.c  |   24 +++++++++++-----------
 arch/x86/oprofile/op_model_p4.c   |   38 +++++++++++++++++-------------------
 arch/x86/oprofile/op_model_ppro.c |   33 +++++++++++++++----------------
 3 files changed, 46 insertions(+), 49 deletions(-)

diff --git a/arch/x86/oprofile/op_model_amd.c b/arch/x86/oprofile/op_model_amd.c
index 2e2bc90..7e5886d 100644
--- a/arch/x86/oprofile/op_model_amd.c
+++ b/arch/x86/oprofile/op_model_amd.c
@@ -126,6 +126,18 @@ static void op_mux_switch_ctrl(struct op_x86_model_spec const *model,
 
 /* functions for op_amd_spec */
 
+static void op_amd_shutdown(struct op_msrs const * const msrs)
+{
+	int i;
+
+	for (i = 0; i < NUM_COUNTERS; ++i) {
+		if (!msrs->counters[i].addr)
+			continue;
+		release_perfctr_nmi(MSR_K7_PERFCTR0 + i);
+		release_evntsel_nmi(MSR_K7_EVNTSEL0 + i);
+	}
+}
+
 static void op_amd_fill_in_addresses(struct op_msrs * const msrs)
 {
 	int i;
@@ -422,18 +434,6 @@ static void op_amd_stop(struct op_msrs const * const msrs)
 	op_amd_stop_ibs();
 }
 
-static void op_amd_shutdown(struct op_msrs const * const msrs)
-{
-	int i;
-
-	for (i = 0; i < NUM_COUNTERS; ++i) {
-		if (!msrs->counters[i].addr)
-			continue;
-		release_perfctr_nmi(MSR_K7_PERFCTR0 + i);
-		release_evntsel_nmi(MSR_K7_EVNTSEL0 + i);
-	}
-}
-
 static u8 ibs_eilvt_off;
 
 static inline void apic_init_ibs_nmi_per_cpu(void *arg)
diff --git a/arch/x86/oprofile/op_model_p4.c b/arch/x86/oprofile/op_model_p4.c
index e6a160a..7cc80df 100644
--- a/arch/x86/oprofile/op_model_p4.c
+++ b/arch/x86/oprofile/op_model_p4.c
@@ -385,6 +385,24 @@ static unsigned int get_stagger(void)
 
 static unsigned long reset_value[NUM_COUNTERS_NON_HT];
 
+static void p4_shutdown(struct op_msrs const * const msrs)
+{
+	int i;
+
+	for (i = 0; i < num_counters; ++i) {
+		if (msrs->counters[i].addr)
+			release_perfctr_nmi(msrs->counters[i].addr);
+	}
+	/*
+	 * some of the control registers are specially reserved in
+	 * conjunction with the counter registers (hence the starting offset).
+	 * This saves a few bits.
+	 */
+	for (i = num_counters; i < num_controls; ++i) {
+		if (msrs->controls[i].addr)
+			release_evntsel_nmi(msrs->controls[i].addr);
+	}
+}
 
 static void p4_fill_in_addresses(struct op_msrs * const msrs)
 {
@@ -668,26 +686,6 @@ static void p4_stop(struct op_msrs const * const msrs)
 	}
 }
 
-static void p4_shutdown(struct op_msrs const * const msrs)
-{
-	int i;
-
-	for (i = 0; i < num_counters; ++i) {
-		if (msrs->counters[i].addr)
-			release_perfctr_nmi(msrs->counters[i].addr);
-	}
-	/*
-	 * some of the control registers are specially reserved in
-	 * conjunction with the counter registers (hence the starting offset).
-	 * This saves a few bits.
-	 */
-	for (i = num_counters; i < num_controls; ++i) {
-		if (msrs->controls[i].addr)
-			release_evntsel_nmi(msrs->controls[i].addr);
-	}
-}
-
-
 #ifdef CONFIG_SMP
 struct op_x86_model_spec op_p4_ht2_spec = {
 	.num_counters		= NUM_COUNTERS_HT2,
diff --git a/arch/x86/oprofile/op_model_ppro.c b/arch/x86/oprofile/op_model_ppro.c
index f8e268e..b07d25a 100644
--- a/arch/x86/oprofile/op_model_ppro.c
+++ b/arch/x86/oprofile/op_model_ppro.c
@@ -30,6 +30,22 @@ static int counter_width = 32;
 
 static u64 *reset_value;
 
+static void ppro_shutdown(struct op_msrs const * const msrs)
+{
+	int i;
+
+	for (i = 0; i < num_counters; ++i) {
+		if (!msrs->counters[i].addr)
+			continue;
+		release_perfctr_nmi(MSR_P6_PERFCTR0 + i);
+		release_evntsel_nmi(MSR_P6_EVNTSEL0 + i);
+	}
+	if (reset_value) {
+		kfree(reset_value);
+		reset_value = NULL;
+	}
+}
+
 static void ppro_fill_in_addresses(struct op_msrs * const msrs)
 {
 	int i;
@@ -189,23 +205,6 @@ static void ppro_stop(struct op_msrs const * const msrs)
 	}
 }
 
-static void ppro_shutdown(struct op_msrs const * const msrs)
-{
-	int i;
-
-	for (i = 0; i < num_counters; ++i) {
-		if (!msrs->counters[i].addr)
-			continue;
-		release_perfctr_nmi(MSR_P6_PERFCTR0 + i);
-		release_evntsel_nmi(MSR_P6_EVNTSEL0 + i);
-	}
-	if (reset_value) {
-		kfree(reset_value);
-		reset_value = NULL;
-	}
-}
-
-
 struct op_x86_model_spec op_ppro_spec = {
 	.num_counters		= 2,
 	.num_controls		= 2,
-- 
1.7.0.3

[PATCH 5/7] oprofile/x86: return -EBUSY if counters are already reserved

[Robert Richter]

In case a counter is already reserved by the watchdog or perf_event
subsystem, oprofile ignored this counters silently. This case is
handled now and oprofile_setup() now reports an error.

Signed-off-by: Robert Richter <robert.richter@amd.com>
---
 arch/x86/oprofile/nmi_int.c       |    5 ++++-
 arch/x86/oprofile/op_model_amd.c  |   24 +++++++++++++-----------
 arch/x86/oprofile/op_model_p4.c   |   14 +++++++++++++-
 arch/x86/oprofile/op_model_ppro.c |   24 +++++++++++++-----------
 arch/x86/oprofile/op_x86_model.h  |    2 +-
 5 files changed, 44 insertions(+), 25 deletions(-)

diff --git a/arch/x86/oprofile/nmi_int.c b/arch/x86/oprofile/nmi_int.c
index c0c21f2..9f001d9 100644
--- a/arch/x86/oprofile/nmi_int.c
+++ b/arch/x86/oprofile/nmi_int.c
@@ -357,7 +357,10 @@ static int nmi_setup(void)
 	 */
 
 	/* Assume saved/restored counters are the same on all CPUs */
-	model->fill_in_addresses(&per_cpu(cpu_msrs, 0));
+	err = model->fill_in_addresses(&per_cpu(cpu_msrs, 0));
+	if (err)
+		goto fail;
+
 	for_each_possible_cpu(cpu) {
 		if (!cpu)
 			continue;
diff --git a/arch/x86/oprofile/op_model_amd.c b/arch/x86/oprofile/op_model_amd.c
index 7e5886d..536d0b0 100644
--- a/arch/x86/oprofile/op_model_amd.c
+++ b/arch/x86/oprofile/op_model_amd.c
@@ -138,21 +138,30 @@ static void op_amd_shutdown(struct op_msrs const * const msrs)
 	}
 }
 
-static void op_amd_fill_in_addresses(struct op_msrs * const msrs)
+static int op_amd_fill_in_addresses(struct op_msrs * const msrs)
 {
 	int i;
 
 	for (i = 0; i < NUM_COUNTERS; i++) {
 		if (!reserve_perfctr_nmi(MSR_K7_PERFCTR0 + i))
-			continue;
+			goto fail;
 		if (!reserve_evntsel_nmi(MSR_K7_EVNTSEL0 + i)) {
 			release_perfctr_nmi(MSR_K7_PERFCTR0 + i);
-			continue;
+			goto fail;
 		}
 		/* both registers must be reserved */
 		msrs->counters[i].addr = MSR_K7_PERFCTR0 + i;
 		msrs->controls[i].addr = MSR_K7_EVNTSEL0 + i;
+		continue;
+	fail:
+		if (!counter_config[i].enabled)
+			continue;
+		op_x86_warn_reserved(i);
+		op_amd_shutdown(msrs);
+		return -EBUSY;
 	}
+
+	return 0;
 }
 
 static void op_amd_setup_ctrs(struct op_x86_model_spec const *model,
@@ -172,15 +181,8 @@ static void op_amd_setup_ctrs(struct op_x86_model_spec const *model,
 
 	/* clear all counters */
 	for (i = 0; i < NUM_COUNTERS; ++i) {
-		if (unlikely(!msrs->controls[i].addr)) {
-			if (counter_config[i].enabled && !smp_processor_id())
-				/*
-				 * counter is reserved, this is on all
-				 * cpus, so report only for cpu #0
-				 */
-				op_x86_warn_reserved(i);
+		if (!msrs->controls[i].addr)
 			continue;
-		}
 		rdmsrl(msrs->controls[i].addr, val);
 		if (val & ARCH_PERFMON_EVENTSEL_ENABLE)
 			op_x86_warn_in_use(i);
diff --git a/arch/x86/oprofile/op_model_p4.c b/arch/x86/oprofile/op_model_p4.c
index 7cc80df..182558d 100644
--- a/arch/x86/oprofile/op_model_p4.c
+++ b/arch/x86/oprofile/op_model_p4.c
@@ -404,7 +404,7 @@ static void p4_shutdown(struct op_msrs const * const msrs)
 	}
 }
 
-static void p4_fill_in_addresses(struct op_msrs * const msrs)
+static int p4_fill_in_addresses(struct op_msrs * const msrs)
 {
 	unsigned int i;
 	unsigned int addr, cccraddr, stag;
@@ -486,6 +486,18 @@ static void p4_fill_in_addresses(struct op_msrs * const msrs)
 			msrs->controls[i++].addr = MSR_P4_CRU_ESCR5;
 		}
 	}
+
+	for (i = 0; i < num_counters; ++i) {
+		if (!counter_config[i].enabled)
+			continue;
+		if (msrs->controls[i].addr)
+			continue;
+		op_x86_warn_reserved(i);
+		p4_shutdown(msrs);
+		return -EBUSY;
+	}
+
+	return 0;
 }
 
 
diff --git a/arch/x86/oprofile/op_model_ppro.c b/arch/x86/oprofile/op_model_ppro.c
index b07d25a..1fd17cf 100644
--- a/arch/x86/oprofile/op_model_ppro.c
+++ b/arch/x86/oprofile/op_model_ppro.c
@@ -46,21 +46,30 @@ static void ppro_shutdown(struct op_msrs const * const msrs)
 	}
 }
 
-static void ppro_fill_in_addresses(struct op_msrs * const msrs)
+static int ppro_fill_in_addresses(struct op_msrs * const msrs)
 {
 	int i;
 
 	for (i = 0; i < num_counters; i++) {
 		if (!reserve_perfctr_nmi(MSR_P6_PERFCTR0 + i))
-			continue;
+			goto fail;
 		if (!reserve_evntsel_nmi(MSR_P6_EVNTSEL0 + i)) {
 			release_perfctr_nmi(MSR_P6_PERFCTR0 + i);
-			continue;
+			goto fail;
 		}
 		/* both registers must be reserved */
 		msrs->counters[i].addr = MSR_P6_PERFCTR0 + i;
 		msrs->controls[i].addr = MSR_P6_EVNTSEL0 + i;
+		continue;
+	fail:
+		if (!counter_config[i].enabled)
+			continue;
+		op_x86_warn_reserved(i);
+		ppro_shutdown(msrs);
+		return -EBUSY;
 	}
+
+	return 0;
 }
 
 
@@ -96,15 +105,8 @@ static void ppro_setup_ctrs(struct op_x86_model_spec const *model,
 
 	/* clear all counters */
 	for (i = 0; i < num_counters; ++i) {
-		if (unlikely(!msrs->controls[i].addr)) {
-			if (counter_config[i].enabled && !smp_processor_id())
-				/*
-				 * counter is reserved, this is on all
-				 * cpus, so report only for cpu #0
-				 */
-				op_x86_warn_reserved(i);
+		if (!msrs->controls[i].addr)
 			continue;
-		}
 		rdmsrl(msrs->controls[i].addr, val);
 		if (val & ARCH_PERFMON_EVENTSEL_ENABLE)
 			op_x86_warn_in_use(i);
diff --git a/arch/x86/oprofile/op_x86_model.h b/arch/x86/oprofile/op_x86_model.h
index ff82a75..5514013 100644
--- a/arch/x86/oprofile/op_x86_model.h
+++ b/arch/x86/oprofile/op_x86_model.h
@@ -41,7 +41,7 @@ struct op_x86_model_spec {
 	u16		event_mask;
 	int		(*init)(struct oprofile_operations *ops);
 	void		(*exit)(void);
-	void		(*fill_in_addresses)(struct op_msrs * const msrs);
+	int		(*fill_in_addresses)(struct op_msrs * const msrs);
 	void		(*setup_ctrs)(struct op_x86_model_spec const *model,
 				      struct op_msrs const * const msrs);
 	int		(*check_ctrs)(struct pt_regs * const regs,
-- 
1.7.0.3

[PATCH 6/7] oprofile/x86: move IBS code

[Robert Richter]

Moving code to make future changes easier. This groups all IBS code
together.

Signed-off-by: Robert Richter <robert.richter@amd.com>
---
 arch/x86/oprofile/op_model_amd.c |  220 +++++++++++++++++++-------------------
 1 files changed, 110 insertions(+), 110 deletions(-)

diff --git a/arch/x86/oprofile/op_model_amd.c b/arch/x86/oprofile/op_model_amd.c
index 536d0b0..e159254 100644
--- a/arch/x86/oprofile/op_model_amd.c
+++ b/arch/x86/oprofile/op_model_amd.c
@@ -102,116 +102,6 @@ static u32 get_ibs_caps(void)
 	return ibs_caps;
 }
 
-#ifdef CONFIG_OPROFILE_EVENT_MULTIPLEX
-
-static void op_mux_switch_ctrl(struct op_x86_model_spec const *model,
-			       struct op_msrs const * const msrs)
-{
-	u64 val;
-	int i;
-
-	/* enable active counters */
-	for (i = 0; i < NUM_COUNTERS; ++i) {
-		int virt = op_x86_phys_to_virt(i);
-		if (!reset_value[virt])
-			continue;
-		rdmsrl(msrs->controls[i].addr, val);
-		val &= model->reserved;
-		val |= op_x86_get_ctrl(model, &counter_config[virt]);
-		wrmsrl(msrs->controls[i].addr, val);
-	}
-}
-
-#endif
-
-/* functions for op_amd_spec */
-
-static void op_amd_shutdown(struct op_msrs const * const msrs)
-{
-	int i;
-
-	for (i = 0; i < NUM_COUNTERS; ++i) {
-		if (!msrs->counters[i].addr)
-			continue;
-		release_perfctr_nmi(MSR_K7_PERFCTR0 + i);
-		release_evntsel_nmi(MSR_K7_EVNTSEL0 + i);
-	}
-}
-
-static int op_amd_fill_in_addresses(struct op_msrs * const msrs)
-{
-	int i;
-
-	for (i = 0; i < NUM_COUNTERS; i++) {
-		if (!reserve_perfctr_nmi(MSR_K7_PERFCTR0 + i))
-			goto fail;
-		if (!reserve_evntsel_nmi(MSR_K7_EVNTSEL0 + i)) {
-			release_perfctr_nmi(MSR_K7_PERFCTR0 + i);
-			goto fail;
-		}
-		/* both registers must be reserved */
-		msrs->counters[i].addr = MSR_K7_PERFCTR0 + i;
-		msrs->controls[i].addr = MSR_K7_EVNTSEL0 + i;
-		continue;
-	fail:
-		if (!counter_config[i].enabled)
-			continue;
-		op_x86_warn_reserved(i);
-		op_amd_shutdown(msrs);
-		return -EBUSY;
-	}
-
-	return 0;
-}
-
-static void op_amd_setup_ctrs(struct op_x86_model_spec const *model,
-			      struct op_msrs const * const msrs)
-{
-	u64 val;
-	int i;
-
-	/* setup reset_value */
-	for (i = 0; i < NUM_VIRT_COUNTERS; ++i) {
-		if (counter_config[i].enabled
-		    && msrs->counters[op_x86_virt_to_phys(i)].addr)
-			reset_value[i] = counter_config[i].count;
-		else
-			reset_value[i] = 0;
-	}
-
-	/* clear all counters */
-	for (i = 0; i < NUM_COUNTERS; ++i) {
-		if (!msrs->controls[i].addr)
-			continue;
-		rdmsrl(msrs->controls[i].addr, val);
-		if (val & ARCH_PERFMON_EVENTSEL_ENABLE)
-			op_x86_warn_in_use(i);
-		val &= model->reserved;
-		wrmsrl(msrs->controls[i].addr, val);
-		/*
-		 * avoid a false detection of ctr overflows in NMI
-		 * handler
-		 */
-		wrmsrl(msrs->counters[i].addr, -1LL);
-	}
-
-	/* enable active counters */
-	for (i = 0; i < NUM_COUNTERS; ++i) {
-		int virt = op_x86_phys_to_virt(i);
-		if (!reset_value[virt])
-			continue;
-
-		/* setup counter registers */
-		wrmsrl(msrs->counters[i].addr, -(u64)reset_value[virt]);
-
-		/* setup control registers */
-		rdmsrl(msrs->controls[i].addr, val);
-		val &= model->reserved;
-		val |= op_x86_get_ctrl(model, &counter_config[virt]);
-		wrmsrl(msrs->controls[i].addr, val);
-	}
-}
-
 /*
  * 16-bit Linear Feedback Shift Register (LFSR)
  *
@@ -376,6 +266,116 @@ static void op_amd_stop_ibs(void)
 		wrmsrl(MSR_AMD64_IBSOPCTL, 0);
 }
 
+#ifdef CONFIG_OPROFILE_EVENT_MULTIPLEX
+
+static void op_mux_switch_ctrl(struct op_x86_model_spec const *model,
+			       struct op_msrs const * const msrs)
+{
+	u64 val;
+	int i;
+
+	/* enable active counters */
+	for (i = 0; i < NUM_COUNTERS; ++i) {
+		int virt = op_x86_phys_to_virt(i);
+		if (!reset_value[virt])
+			continue;
+		rdmsrl(msrs->controls[i].addr, val);
+		val &= model->reserved;
+		val |= op_x86_get_ctrl(model, &counter_config[virt]);
+		wrmsrl(msrs->controls[i].addr, val);
+	}
+}
+
+#endif
+
+/* functions for op_amd_spec */
+
+static void op_amd_shutdown(struct op_msrs const * const msrs)
+{
+	int i;
+
+	for (i = 0; i < NUM_COUNTERS; ++i) {
+		if (!msrs->counters[i].addr)
+			continue;
+		release_perfctr_nmi(MSR_K7_PERFCTR0 + i);
+		release_evntsel_nmi(MSR_K7_EVNTSEL0 + i);
+	}
+}
+
+static int op_amd_fill_in_addresses(struct op_msrs * const msrs)
+{
+	int i;
+
+	for (i = 0; i < NUM_COUNTERS; i++) {
+		if (!reserve_perfctr_nmi(MSR_K7_PERFCTR0 + i))
+			goto fail;
+		if (!reserve_evntsel_nmi(MSR_K7_EVNTSEL0 + i)) {
+			release_perfctr_nmi(MSR_K7_PERFCTR0 + i);
+			goto fail;
+		}
+		/* both registers must be reserved */
+		msrs->counters[i].addr = MSR_K7_PERFCTR0 + i;
+		msrs->controls[i].addr = MSR_K7_EVNTSEL0 + i;
+		continue;
+	fail:
+		if (!counter_config[i].enabled)
+			continue;
+		op_x86_warn_reserved(i);
+		op_amd_shutdown(msrs);
+		return -EBUSY;
+	}
+
+	return 0;
+}
+
+static void op_amd_setup_ctrs(struct op_x86_model_spec const *model,
+			      struct op_msrs const * const msrs)
+{
+	u64 val;
+	int i;
+
+	/* setup reset_value */
+	for (i = 0; i < NUM_VIRT_COUNTERS; ++i) {
+		if (counter_config[i].enabled
+		    && msrs->counters[op_x86_virt_to_phys(i)].addr)
+			reset_value[i] = counter_config[i].count;
+		else
+			reset_value[i] = 0;
+	}
+
+	/* clear all counters */
+	for (i = 0; i < NUM_COUNTERS; ++i) {
+		if (!msrs->controls[i].addr)
+			continue;
+		rdmsrl(msrs->controls[i].addr, val);
+		if (val & ARCH_PERFMON_EVENTSEL_ENABLE)
+			op_x86_warn_in_use(i);
+		val &= model->reserved;
+		wrmsrl(msrs->controls[i].addr, val);
+		/*
+		 * avoid a false detection of ctr overflows in NMI
+		 * handler
+		 */
+		wrmsrl(msrs->counters[i].addr, -1LL);
+	}
+
+	/* enable active counters */
+	for (i = 0; i < NUM_COUNTERS; ++i) {
+		int virt = op_x86_phys_to_virt(i);
+		if (!reset_value[virt])
+			continue;
+
+		/* setup counter registers */
+		wrmsrl(msrs->counters[i].addr, -(u64)reset_value[virt]);
+
+		/* setup control registers */
+		rdmsrl(msrs->controls[i].addr, val);
+		val &= model->reserved;
+		val |= op_x86_get_ctrl(model, &counter_config[virt]);
+		wrmsrl(msrs->controls[i].addr, val);
+	}
+}
+
 static int op_amd_check_ctrs(struct pt_regs * const regs,
 			     struct op_msrs const * const msrs)
 {
-- 
1.7.0.3

[PATCH 7/7] oprofile/x86: remove duplicate IBS capability check

[Robert Richter]

The check is already done in ibs_exit().

Signed-off-by: Robert Richter <robert.richter@amd.com>
---
 arch/x86/oprofile/op_model_amd.c |    3 +--
 1 files changed, 1 insertions(+), 2 deletions(-)

diff --git a/arch/x86/oprofile/op_model_amd.c b/arch/x86/oprofile/op_model_amd.c
index e159254..384c524 100644
--- a/arch/x86/oprofile/op_model_amd.c
+++ b/arch/x86/oprofile/op_model_amd.c
@@ -490,8 +490,7 @@ static int init_ibs_nmi(void)
 /* uninitialize the APIC for the IBS interrupts if needed */
 static void clear_ibs_nmi(void)
 {
-	if (ibs_caps)
-		on_each_cpu(apic_clear_ibs_nmi_per_cpu, NULL, 1);
+	on_each_cpu(apic_clear_ibs_nmi_per_cpu, NULL, 1);
 }
 
 /* initialize the APIC for the IBS interrupts if available */
-- 
1.7.0.3

[GIT PULL] updates for oprofile

[Robert Richter]

Ingo,

please pull oprofile updates for tip from

 git://git.kernel.org/pub/scm/linux/kernel/git/rric/oprofile.git core

Thanks,

-Robert

-- 

The following changes since commit 9414e99672271adcc661f3c160a30b374179b92f:
  Phil Carmody (1):
        oprofile: protect from not being in an IRQ context

are available in the git repository at:

  git://git.kernel.org/pub/scm/linux/kernel/git/rric/oprofile.git core

Robert Richter (7):
      oprofile: update file list in MAINTAINERS file
      oprofile/x86: rework error handler in nmi_setup()
      oprofile/x86: reserve counter msrs pairwise
      oprofile/x86: moving shutdown functions
      oprofile/x86: return -EBUSY if counters are already reserved
      oprofile/x86: move IBS code
      oprofile/x86: remove duplicate IBS capability check

 MAINTAINERS                       |    1 +
 arch/x86/oprofile/nmi_int.c       |   38 ++++---
 arch/x86/oprofile/op_model_amd.c  |  228 ++++++++++++++++++-------------------
 arch/x86/oprofile/op_model_p4.c   |   52 +++++----
 arch/x86/oprofile/op_model_ppro.c |   77 ++++++-------
 arch/x86/oprofile/op_x86_model.h  |    2 +-
 6 files changed, 206 insertions(+), 192 deletions(-)

-- 
Advanced Micro Devices, Inc.
Operating System Research Center
email: robert.richter@amd.com

Re: [GIT PULL] updates for oprofile

[Ingo Molnar]

* Robert Richter <robert.richter@amd.com> wrote:

> Ingo,
> 
> please pull oprofile updates for tip from
> 
>  git://git.kernel.org/pub/scm/linux/kernel/git/rric/oprofile.git core
> 
> Thanks,
> 
> -Robert
> 
> -- 
> 
> The following changes since commit 9414e99672271adcc661f3c160a30b374179b92f:
>   Phil Carmody (1):
>         oprofile: protect from not being in an IRQ context
> 
> are available in the git repository at:
> 
>   git://git.kernel.org/pub/scm/linux/kernel/git/rric/oprofile.git core
> 
> Robert Richter (7):
>       oprofile: update file list in MAINTAINERS file
>       oprofile/x86: rework error handler in nmi_setup()
>       oprofile/x86: reserve counter msrs pairwise
>       oprofile/x86: moving shutdown functions
>       oprofile/x86: return -EBUSY if counters are already reserved
>       oprofile/x86: move IBS code
>       oprofile/x86: remove duplicate IBS capability check
> 
>  MAINTAINERS                       |    1 +
>  arch/x86/oprofile/nmi_int.c       |   38 ++++---
>  arch/x86/oprofile/op_model_amd.c  |  228 ++++++++++++++++++-------------------
>  arch/x86/oprofile/op_model_p4.c   |   52 +++++----
>  arch/x86/oprofile/op_model_ppro.c |   77 ++++++-------
>  arch/x86/oprofile/op_x86_model.h  |    2 +-
>  6 files changed, 206 insertions(+), 192 deletions(-)

Pulled, thanks Robert!

	Ingo