Re: [tip:perf/x86] perf/x86/intel: Support task events with Intel CQM

[Peter Zijlstra]

On Thu, Mar 05, 2015 at 01:55:16AM +0100, Ingo Molnar wrote:
> That's a fair point. Peter?

Indeed, sorry. And its because of a stupid reason.

---
Subject: perf: Remove type specific target pointers

The only reason CQM had to use a hard-coded pmu type was so it could use
cqm_target in hw_perf_event.

Do away with the {tp,bp,cqm}_target pointers and provide a non type
specific one.

This allows us to do away with that silly pmu type as well.

Signed-off-by: Peter Zijlstra (Intel) <peterz@xxxxxxxxxxxxx>
---
 arch/arm/kernel/hw_breakpoint.c            |  2 +-
 arch/arm64/kernel/hw_breakpoint.c          |  2 +-
 arch/x86/kernel/cpu/perf_event_intel_cqm.c |  7 +++----
 include/linux/perf_event.h                 |  4 +---
 include/uapi/linux/perf_event.h            |  1 -
 kernel/events/core.c                       | 14 ++++----------
 kernel/events/hw_breakpoint.c              |  8 ++++----
 kernel/trace/trace_uprobe.c                | 10 +++++-----
 8 files changed, 19 insertions(+), 29 deletions(-)

diff --git a/arch/arm/kernel/hw_breakpoint.c b/arch/arm/kernel/hw_breakpoint.c
index 7fc70ae21185..dc7d0a95bd36 100644
--- a/arch/arm/kernel/hw_breakpoint.c
+++ b/arch/arm/kernel/hw_breakpoint.c
@@ -648,7 +648,7 @@ int arch_validate_hwbkpt_settings(struct perf_event *bp)
 		 * Per-cpu breakpoints are not supported by our stepping
 		 * mechanism.
 		 */
-		if (!bp->hw.bp_target)
+		if (!bp->hw.target)
 			return -EINVAL;
 
 		/*
diff --git a/arch/arm64/kernel/hw_breakpoint.c b/arch/arm64/kernel/hw_breakpoint.c
index 98bbe06e469c..e7d934d3afe0 100644
--- a/arch/arm64/kernel/hw_breakpoint.c
+++ b/arch/arm64/kernel/hw_breakpoint.c
@@ -527,7 +527,7 @@ int arch_validate_hwbkpt_settings(struct perf_event *bp)
 	 * Disallow per-task kernel breakpoints since these would
 	 * complicate the stepping code.
 	 */
-	if (info->ctrl.privilege == AARCH64_BREAKPOINT_EL1 && bp->hw.bp_target)
+	if (info->ctrl.privilege == AARCH64_BREAKPOINT_EL1 && bp->hw.target)
 		return -EINVAL;
 
 	return 0;
diff --git a/arch/x86/kernel/cpu/perf_event_intel_cqm.c b/arch/x86/kernel/cpu/perf_event_intel_cqm.c
index 9a8ef8376fcd..e4d1b8b738fa 100644
--- a/arch/x86/kernel/cpu/perf_event_intel_cqm.c
+++ b/arch/x86/kernel/cpu/perf_event_intel_cqm.c
@@ -263,7 +263,7 @@ static bool __match_event(struct perf_event *a, struct perf_event *b)
 	/*
 	 * Events that target same task are placed into the same cache group.
 	 */
-	if (a->hw.cqm_target == b->hw.cqm_target)
+	if (a->hw.target == b->hw.target)
 		return true;
 
 	/*
@@ -279,7 +279,7 @@ static bool __match_event(struct perf_event *a, struct perf_event *b)
 static inline struct perf_cgroup *event_to_cgroup(struct perf_event *event)
 {
 	if (event->attach_state & PERF_ATTACH_TASK)
-		return perf_cgroup_from_task(event->hw.cqm_target);
+		return perf_cgroup_from_task(event->hw.target);
 
 	return event->cgrp;
 }
@@ -1365,8 +1365,7 @@ static int __init intel_cqm_init(void)
 
 	__perf_cpu_notifier(intel_cqm_cpu_notifier);
 
-	ret = perf_pmu_register(&intel_cqm_pmu, "intel_cqm",
-				PERF_TYPE_INTEL_CQM);
+	ret = perf_pmu_register(&intel_cqm_pmu, "intel_cqm", -1);
 	if (ret)
 		pr_err("Intel CQM perf registration failed: %d\n", ret);
 	else
diff --git a/include/linux/perf_event.h b/include/linux/perf_event.h
index b8f69d35b5c7..b16eac5f54ce 100644
--- a/include/linux/perf_event.h
+++ b/include/linux/perf_event.h
@@ -119,7 +119,6 @@ struct hw_perf_event {
 			struct hrtimer	hrtimer;
 		};
 		struct { /* tracepoint */
-			struct task_struct	*tp_target;
 			/* for tp_event->class */
 			struct list_head	tp_list;
 		};
@@ -129,7 +128,6 @@ struct hw_perf_event {
 			struct list_head	cqm_events_entry;
 			struct list_head	cqm_groups_entry;
 			struct list_head	cqm_group_entry;
-			struct task_struct	*cqm_target;
 		};
 #ifdef CONFIG_HAVE_HW_BREAKPOINT
 		struct { /* breakpoint */
@@ -138,12 +136,12 @@ struct hw_perf_event {
 			 * problem hw_breakpoint has with context
 			 * creation and event initalization.
 			 */
-			struct task_struct		*bp_target;
 			struct arch_hw_breakpoint	info;
 			struct list_head		bp_list;
 		};
 #endif
 	};
+	struct task_struct		*target;
 	int				state;
 	local64_t			prev_count;
 	u64				sample_period;
diff --git a/include/uapi/linux/perf_event.h b/include/uapi/linux/perf_event.h
index 3c8b45de57ec..1e3cd07cf76e 100644
--- a/include/uapi/linux/perf_event.h
+++ b/include/uapi/linux/perf_event.h
@@ -32,7 +32,6 @@ enum perf_type_id {
 	PERF_TYPE_HW_CACHE			= 3,
 	PERF_TYPE_RAW				= 4,
 	PERF_TYPE_BREAKPOINT			= 5,
-	PERF_TYPE_INTEL_CQM			= 6,
 
 	PERF_TYPE_MAX,				/* non-ABI */
 };
diff --git a/kernel/events/core.c b/kernel/events/core.c
index 89f0f16d55f9..7a0816ef7d18 100644
--- a/kernel/events/core.c
+++ b/kernel/events/core.c
@@ -7171,18 +7171,12 @@ perf_event_alloc(struct perf_event_attr *attr, int cpu,
 
 	if (task) {
 		event->attach_state = PERF_ATTACH_TASK;
-
-		if (attr->type == PERF_TYPE_TRACEPOINT)
-			event->hw.tp_target = task;
-#ifdef CONFIG_HAVE_HW_BREAKPOINT
 		/*
-		 * hw_breakpoint is a bit difficult here..
+		 * XXX pmu::event_init needs to know what task to account to
+		 * and we cannot use the ctx information because we need the
+		 * pmu before we get a ctx.
 		 */
-		else if (attr->type == PERF_TYPE_BREAKPOINT)
-			event->hw.bp_target = task;
-#endif
-		else if (attr->type == PERF_TYPE_INTEL_CQM)
-			event->hw.cqm_target = task;
+		event->hw.target = task;
 	}
 
 	if (!overflow_handler && parent_event) {
diff --git a/kernel/events/hw_breakpoint.c b/kernel/events/hw_breakpoint.c
index 9803a6600d49..92ce5f4ccc26 100644
--- a/kernel/events/hw_breakpoint.c
+++ b/kernel/events/hw_breakpoint.c
@@ -116,12 +116,12 @@ static unsigned int max_task_bp_pinned(int cpu, enum bp_type_idx type)
  */
 static int task_bp_pinned(int cpu, struct perf_event *bp, enum bp_type_idx type)
 {
-	struct task_struct *tsk = bp->hw.bp_target;
+	struct task_struct *tsk = bp->hw.target;
 	struct perf_event *iter;
 	int count = 0;
 
 	list_for_each_entry(iter, &bp_task_head, hw.bp_list) {
-		if (iter->hw.bp_target == tsk &&
+		if (iter->hw.target == tsk &&
 		    find_slot_idx(iter) == type &&
 		    (iter->cpu < 0 || cpu == iter->cpu))
 			count += hw_breakpoint_weight(iter);
@@ -153,7 +153,7 @@ fetch_bp_busy_slots(struct bp_busy_slots *slots, struct perf_event *bp,
 		int nr;
 
 		nr = info->cpu_pinned;
-		if (!bp->hw.bp_target)
+		if (!bp->hw.target)
 			nr += max_task_bp_pinned(cpu, type);
 		else
 			nr += task_bp_pinned(cpu, bp, type);
@@ -210,7 +210,7 @@ toggle_bp_slot(struct perf_event *bp, bool enable, enum bp_type_idx type,
 		weight = -weight;
 
 	/* Pinned counter cpu profiling */
-	if (!bp->hw.bp_target) {
+	if (!bp->hw.target) {
 		get_bp_info(bp->cpu, type)->cpu_pinned += weight;
 		return;
 	}
diff --git a/kernel/trace/trace_uprobe.c b/kernel/trace/trace_uprobe.c
index 7dc1c8abecd6..996e452e1eb3 100644
--- a/kernel/trace/trace_uprobe.c
+++ b/kernel/trace/trace_uprobe.c
@@ -1005,7 +1005,7 @@ __uprobe_perf_filter(struct trace_uprobe_filter *filter, struct mm_struct *mm)
 		return true;
 
 	list_for_each_entry(event, &filter->perf_events, hw.tp_list) {
-		if (event->hw.tp_target->mm == mm)
+		if (event->hw.target->mm == mm)
 			return true;
 	}
 
@@ -1015,7 +1015,7 @@ __uprobe_perf_filter(struct trace_uprobe_filter *filter, struct mm_struct *mm)
 static inline bool
 uprobe_filter_event(struct trace_uprobe *tu, struct perf_event *event)
 {
-	return __uprobe_perf_filter(&tu->filter, event->hw.tp_target->mm);
+	return __uprobe_perf_filter(&tu->filter, event->hw.target->mm);
 }
 
 static int uprobe_perf_close(struct trace_uprobe *tu, struct perf_event *event)
@@ -1023,10 +1023,10 @@ static int uprobe_perf_close(struct trace_uprobe *tu, struct perf_event *event)
 	bool done;
 
 	write_lock(&tu->filter.rwlock);
-	if (event->hw.tp_target) {
+	if (event->hw.target) {
 		list_del(&event->hw.tp_list);
 		done = tu->filter.nr_systemwide ||
-			(event->hw.tp_target->flags & PF_EXITING) ||
+			(event->hw.target->flags & PF_EXITING) ||
 			uprobe_filter_event(tu, event);
 	} else {
 		tu->filter.nr_systemwide--;
@@ -1046,7 +1046,7 @@ static int uprobe_perf_open(struct trace_uprobe *tu, struct perf_event *event)
 	int err;
 
 	write_lock(&tu->filter.rwlock);
-	if (event->hw.tp_target) {
+	if (event->hw.target) {
 		/*
 		 * event->parent != NULL means copy_process(), we can avoid
 		 * uprobe_apply(). current->mm must be probed and we can rely

--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/