1 files changed, 186 insertions, 32 deletions

diff --git a/kernel/events/core.c b/kernel/events/core.c
index e1459df73043..bc9b98a9af9a 100644
--- a/kernel/events/core.c
+++ b/kernel/events/core.c
@@ -28,6 +28,7 @@
 #include <linux/export.h>
 #include <linux/vmalloc.h>
 #include <linux/hardirq.h>
+#include <linux/hugetlb.h>
 #include <linux/rculist.h>
 #include <linux/uaccess.h>
 #include <linux/syscalls.h>
@@ -387,6 +388,7 @@ static atomic_t nr_freq_events __read_mostly;
 static atomic_t nr_switch_events __read_mostly;
 static atomic_t nr_ksymbol_events __read_mostly;
 static atomic_t nr_bpf_events __read_mostly;
+static atomic_t nr_cgroup_events __read_mostly;
 
 static LIST_HEAD(pmus);
 static DEFINE_MUTEX(pmus_lock);
@@ -981,16 +983,10 @@ perf_cgroup_set_shadow_time(struct perf_event *event, u64 now)
 	event->shadow_ctx_time = now - t->timestamp;
 }
 
-/*
- * Update cpuctx->cgrp so that it is set when first cgroup event is added and
- * cleared when last cgroup event is removed.
- */
 static inline void
-list_update_cgroup_event(struct perf_event *event,
-			 struct perf_event_context *ctx, bool add)
+perf_cgroup_event_enable(struct perf_event *event, struct perf_event_context *ctx)
 {
 	struct perf_cpu_context *cpuctx;
-	struct list_head *cpuctx_entry;
 
 	if (!is_cgroup_event(event))
 		return;
@@ -1007,28 +1003,41 @@ list_update_cgroup_event(struct perf_event *event,
 	 * because if the first would mismatch, the second would not try again
 	 * and we would leave cpuctx->cgrp unset.
 	 */
-	if (add && !cpuctx->cgrp) {
+	if (ctx->is_active && !cpuctx->cgrp) {
 		struct perf_cgroup *cgrp = perf_cgroup_from_task(current, ctx);
 
 		if (cgroup_is_descendant(cgrp->css.cgroup, event->cgrp->css.cgroup))
 			cpuctx->cgrp = cgrp;
 	}
 
-	if (add && ctx->nr_cgroups++)
+	if (ctx->nr_cgroups++)
 		return;
-	else if (!add && --ctx->nr_cgroups)
+
+	list_add(&cpuctx->cgrp_cpuctx_entry,
+			per_cpu_ptr(&cgrp_cpuctx_list, event->cpu));
+}
+
+static inline void
+perf_cgroup_event_disable(struct perf_event *event, struct perf_event_context *ctx)
+{
+	struct perf_cpu_context *cpuctx;
+
+	if (!is_cgroup_event(event))
 		return;
 
-	/* no cgroup running */
-	if (!add)
+	/*
+	 * Because cgroup events are always per-cpu events,
+	 * @ctx == &cpuctx->ctx.
+	 */
+	cpuctx = container_of(ctx, struct perf_cpu_context, ctx);
+
+	if (--ctx->nr_cgroups)
+		return;
+
+	if (ctx->is_active && cpuctx->cgrp)
 		cpuctx->cgrp = NULL;
 
-	cpuctx_entry = &cpuctx->cgrp_cpuctx_entry;
-	if (add)
-		list_add(cpuctx_entry,
-			 per_cpu_ptr(&cgrp_cpuctx_list, event->cpu));
-	else
-		list_del(cpuctx_entry);
+	list_del(&cpuctx->cgrp_cpuctx_entry);
 }
 
 #else /* !CONFIG_CGROUP_PERF */
@@ -1094,11 +1103,14 @@ static inline u64 perf_cgroup_event_time(struct perf_event *event)
 }
 
 static inline void
-list_update_cgroup_event(struct perf_event *event,
-			 struct perf_event_context *ctx, bool add)
+perf_cgroup_event_enable(struct perf_event *event, struct perf_event_context *ctx)
 {
 }
 
+static inline void
+perf_cgroup_event_disable(struct perf_event *event, struct perf_event_context *ctx)
+{
+}
 #endif
 
 /*
@@ -1789,13 +1801,14 @@ list_add_event(struct perf_event *event, struct perf_event_context *ctx)
 		add_event_to_groups(event, ctx);
 	}
 
-	list_update_cgroup_event(event, ctx, true);
-
 	list_add_rcu(&event->event_entry, &ctx->event_list);
 	ctx->nr_events++;
 	if (event->attr.inherit_stat)
 		ctx->nr_stat++;
 
+	if (event->state > PERF_EVENT_STATE_OFF)
+		perf_cgroup_event_enable(event, ctx);
+
 	ctx->generation++;
 }
 
@@ -1861,6 +1874,9 @@ static void __perf_event_header_size(struct perf_event *event, u64 sample_type)
 	if (sample_type & PERF_SAMPLE_PHYS_ADDR)
 		size += sizeof(data->phys_addr);
 
+	if (sample_type & PERF_SAMPLE_CGROUP)
+		size += sizeof(data->cgroup);
+
 	event->header_size = size;
 }
 
@@ -1971,8 +1987,6 @@ list_del_event(struct perf_event *event, struct perf_event_context *ctx)
 
 	event->attach_state &= ~PERF_ATTACH_CONTEXT;
 
-	list_update_cgroup_event(event, ctx, false);
-
 	ctx->nr_events--;
 	if (event->attr.inherit_stat)
 		ctx->nr_stat--;
@@ -1989,8 +2003,10 @@ list_del_event(struct perf_event *event, struct perf_event_context *ctx)
 	 * of error state is by explicit re-enabling
 	 * of the event
 	 */
-	if (event->state > PERF_EVENT_STATE_OFF)
+	if (event->state > PERF_EVENT_STATE_OFF) {
+		perf_cgroup_event_disable(event, ctx);
 		perf_event_set_state(event, PERF_EVENT_STATE_OFF);
+	}
 
 	ctx->generation++;
 }
@@ -2221,6 +2237,7 @@ event_sched_out(struct perf_event *event,
 
 	if (READ_ONCE(event->pending_disable) >= 0) {
 		WRITE_ONCE(event->pending_disable, -1);
+		perf_cgroup_event_disable(event, ctx);
 		state = PERF_EVENT_STATE_OFF;
 	}
 	perf_event_set_state(event, state);
@@ -2358,6 +2375,7 @@ static void __perf_event_disable(struct perf_event *event,
 		event_sched_out(event, cpuctx, ctx);
 
 	perf_event_set_state(event, PERF_EVENT_STATE_OFF);
+	perf_cgroup_event_disable(event, ctx);
 }
 
 /*
@@ -2741,7 +2759,7 @@ static int  __perf_install_in_context(void *info)
 	}
 
 #ifdef CONFIG_CGROUP_PERF
-	if (is_cgroup_event(event)) {
+	if (event->state > PERF_EVENT_STATE_OFF && is_cgroup_event(event)) {
 		/*
 		 * If the current cgroup doesn't match the event's
 		 * cgroup, we should not try to schedule it.
@@ -2901,6 +2919,7 @@ static void __perf_event_enable(struct perf_event *event,
 		ctx_sched_out(ctx, cpuctx, EVENT_TIME);
 
 	perf_event_set_state(event, PERF_EVENT_STATE_INACTIVE);
+	perf_cgroup_event_enable(event, ctx);
 
 	if (!ctx->is_active)
 		return;
@@ -3503,7 +3522,8 @@ static void cpu_ctx_sched_out(struct perf_cpu_context *cpuctx,
 
 static bool perf_less_group_idx(const void *l, const void *r)
 {
-	const struct perf_event *le = l, *re = r;
+	const struct perf_event *le = *(const struct perf_event **)l;
+	const struct perf_event *re = *(const struct perf_event **)r;
 
 	return le->group_index < re->group_index;
 }
@@ -3611,8 +3631,10 @@ static int merge_sched_in(struct perf_event *event, void *data)
 	}
 
 	if (event->state == PERF_EVENT_STATE_INACTIVE) {
-		if (event->attr.pinned)
+		if (event->attr.pinned) {
+			perf_cgroup_event_disable(event, ctx);
 			perf_event_set_state(event, PERF_EVENT_STATE_ERROR);
+		}
 
 		*can_add_hw = 0;
 		ctx->rotate_necessary = 1;
@@ -4608,6 +4630,8 @@ static void unaccount_event(struct perf_event *event)
 		atomic_dec(&nr_comm_events);
 	if (event->attr.namespaces)
 		atomic_dec(&nr_namespaces_events);
+	if (event->attr.cgroup)
+		atomic_dec(&nr_cgroup_events);
 	if (event->attr.task)
 		atomic_dec(&nr_task_events);
 	if (event->attr.freq)
@@ -6864,6 +6888,9 @@ void perf_output_sample(struct perf_output_handle *handle,
 	if (sample_type & PERF_SAMPLE_PHYS_ADDR)
 		perf_output_put(handle, data->phys_addr);
 
+	if (sample_type & PERF_SAMPLE_CGROUP)
+		perf_output_put(handle, data->cgroup);
+
 	if (sample_type & PERF_SAMPLE_AUX) {
 		perf_output_put(handle, data->aux_size);
 
@@ -6907,9 +6934,12 @@ static u64 perf_virt_to_phys(u64 virt)
 		 * Try IRQ-safe __get_user_pages_fast first.
 		 * If failed, leave phys_addr as 0.
 		 */
-		if ((current->mm != NULL) &&
-		    (__get_user_pages_fast(virt, 1, 0, &p) == 1))
-			phys_addr = page_to_phys(p) + virt % PAGE_SIZE;
+		if (current->mm != NULL) {
+			pagefault_disable();
+			if (__get_user_pages_fast(virt, 1, 0, &p) == 1)
+				phys_addr = page_to_phys(p) + virt % PAGE_SIZE;
+			pagefault_enable();
+		}
 
 		if (p)
 			put_page(p);
@@ -7063,6 +7093,16 @@ void perf_prepare_sample(struct perf_event_header *header,
 	if (sample_type & PERF_SAMPLE_PHYS_ADDR)
 		data->phys_addr = perf_virt_to_phys(data->addr);
 
+#ifdef CONFIG_CGROUP_PERF
+	if (sample_type & PERF_SAMPLE_CGROUP) {
+		struct cgroup *cgrp;
+
+		/* protected by RCU */
+		cgrp = task_css_check(current, perf_event_cgrp_id, 1)->cgroup;
+		data->cgroup = cgroup_id(cgrp);
+	}
+#endif
+
 	if (sample_type & PERF_SAMPLE_AUX) {
 		u64 size;
 
@@ -7736,6 +7776,105 @@ void perf_event_namespaces(struct task_struct *task)
 }
 
 /*
+ * cgroup tracking
+ */
+#ifdef CONFIG_CGROUP_PERF
+
+struct perf_cgroup_event {
+	char				*path;
+	int				path_size;
+	struct {
+		struct perf_event_header	header;
+		u64				id;
+		char				path[];
+	} event_id;
+};
+
+static int perf_event_cgroup_match(struct perf_event *event)
+{
+	return event->attr.cgroup;
+}
+
+static void perf_event_cgroup_output(struct perf_event *event, void *data)
+{
+	struct perf_cgroup_event *cgroup_event = data;
+	struct perf_output_handle handle;
+	struct perf_sample_data sample;
+	u16 header_size = cgroup_event->event_id.header.size;
+	int ret;
+
+	if (!perf_event_cgroup_match(event))
+		return;
+
+	perf_event_header__init_id(&cgroup_event->event_id.header,
+				   &sample, event);
+	ret = perf_output_begin(&handle, event,
+				cgroup_event->event_id.header.size);
+	if (ret)
+		goto out;
+
+	perf_output_put(&handle, cgroup_event->event_id);
+	__output_copy(&handle, cgroup_event->path, cgroup_event->path_size);
+
+	perf_event__output_id_sample(event, &handle, &sample);
+
+	perf_output_end(&handle);
+out:
+	cgroup_event->event_id.header.size = header_size;
+}
+
+static void perf_event_cgroup(struct cgroup *cgrp)
+{
+	struct perf_cgroup_event cgroup_event;
+	char path_enomem[16] = "//enomem";
+	char *pathname;
+	size_t size;
+
+	if (!atomic_read(&nr_cgroup_events))
+		return;
+
+	cgroup_event = (struct perf_cgroup_event){
+		.event_id  = {
+			.header = {
+				.type = PERF_RECORD_CGROUP,
+				.misc = 0,
+				.size = sizeof(cgroup_event.event_id),
+			},
+			.id = cgroup_id(cgrp),
+		},
+	};
+
+	pathname = kmalloc(PATH_MAX, GFP_KERNEL);
+	if (pathname == NULL) {
+		cgroup_event.path = path_enomem;
+	} else {
+		/* just to be sure to have enough space for alignment */
+		cgroup_path(cgrp, pathname, PATH_MAX - sizeof(u64));
+		cgroup_event.path = pathname;
+	}
+
+	/*
+	 * Since our buffer works in 8 byte units we need to align our string
+	 * size to a multiple of 8. However, we must guarantee the tail end is
+	 * zero'd out to avoid leaking random bits to userspace.
+	 */
+	size = strlen(cgroup_event.path) + 1;
+	while (!IS_ALIGNED(size, sizeof(u64)))
+		cgroup_event.path[size++] = '\0';
+
+	cgroup_event.event_id.header.size += size;
+	cgroup_event.path_size = size;
+
+	perf_iterate_sb(perf_event_cgroup_output,
+			&cgroup_event,
+			NULL);
+
+	kfree(pathname);
+}
+
+#endif
+
+/*
  * mmap tracking
  */
 
@@ -7855,7 +7994,7 @@ static void perf_event_mmap_event(struct perf_mmap_event *mmap_event)
 		flags |= MAP_EXECUTABLE;
 	if (vma->vm_flags & VM_LOCKED)
 		flags |= MAP_LOCKED;
-	if (vma->vm_flags & VM_HUGETLB)
+	if (is_vm_hugetlb_page(vma))
 		flags |= MAP_HUGETLB;
 
 	if (file) {
@@ -10778,6 +10917,8 @@ static void account_event(struct perf_event *event)
 		atomic_inc(&nr_comm_events);
 	if (event->attr.namespaces)
 		atomic_inc(&nr_namespaces_events);
+	if (event->attr.cgroup)
+		atomic_inc(&nr_cgroup_events);
 	if (event->attr.task)
 		atomic_inc(&nr_task_events);
 	if (event->attr.freq)
@@ -11157,6 +11298,12 @@ static int perf_copy_attr(struct perf_event_attr __user *uattr,
 
 	if (attr->sample_type & PERF_SAMPLE_REGS_INTR)
 		ret = perf_reg_validate(attr->sample_regs_intr);
+
+#ifndef CONFIG_CGROUP_PERF
+	if (attr->sample_type & PERF_SAMPLE_CGROUP)
+		return -EINVAL;
+#endif
+
 out:
 	return ret;
 
@@ -12754,6 +12901,12 @@ static void perf_cgroup_css_free(struct cgroup_subsys_state *css)
 	kfree(jc);
 }
 
+static int perf_cgroup_css_online(struct cgroup_subsys_state *css)
+{
+	perf_event_cgroup(css->cgroup);
+	return 0;
+}
+
 static int __perf_cgroup_move(void *info)
 {
 	struct task_struct *task = info;
@@ -12775,6 +12928,7 @@ static void perf_cgroup_attach(struct cgroup_taskset *tset)
 struct cgroup_subsys perf_event_cgrp_subsys = {
 	.css_alloc	= perf_cgroup_css_alloc,
 	.css_free	= perf_cgroup_css_free,
+	.css_online	= perf_cgroup_css_online,
 	.attach		= perf_cgroup_attach,
 	/*
 	 * Implicitly enable on dfl hierarchy so that perf events can