perf_counter: Ammend cleanup in fork() fail

When fork() fails we cannot use perf_counter_exit_task() since that
assumes to operate on current. Write a new helper that cleans up
unused/clean contexts.
Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Mike Galbraith <efault@gmx.de>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Corey Ashford <cjashfor@linux.vnet.ibm.com>
Cc: Marcelo Tosatti <mtosatti@redhat.com>
Cc: Arnaldo Carvalho de Melo <acme@redhat.com>
Cc: John Kacur <jkacur@redhat.com>
LKML-Reference: <new-submission>
Signed-off-by: Ingo Molnar <mingo@elte.hu>

include/linux/perf_counter.h

@@ -579,6 +579,7 @@ extern void perf_counter_task_sched_out(struct task_struct *task,
 extern void perf_counter_task_tick(struct task_struct *task, int cpu);
 extern int perf_counter_init_task(struct task_struct *child);
 extern void perf_counter_exit_task(struct task_struct *child);
+extern void perf_counter_free_task(struct task_struct *task);
 extern void perf_counter_do_pending(void);
 extern void perf_counter_print_debug(void);
 extern void __perf_disable(void);
@@ -644,6 +645,7 @@ static inline void
 perf_counter_task_tick(struct task_struct *task, int cpu)		{ }
 static inline int perf_counter_init_task(struct task_struct *child)	{ return 0; }
 static inline void perf_counter_exit_task(struct task_struct *child)	{ }
+static inline void perf_counter_free_task(struct task_struct *task)	{ }
 static inline void perf_counter_do_pending(void)			{ }
 static inline void perf_counter_print_debug(void)			{ }
 static inline void perf_disable(void)					{ }

kernel/fork.c

@@ -1298,7 +1298,7 @@ static struct task_struct *copy_process(unsigned long clone_flags,
 bad_fork_cleanup_audit:
 	audit_free(p);
 bad_fork_cleanup_policy:
-	perf_counter_exit_task(p);
+	perf_counter_free_task(p);
 #ifdef CONFIG_NUMA
 	mpol_put(p->mempolicy);
 bad_fork_cleanup_cgroup:

kernel/perf_counter.c

@@ -3538,8 +3538,7 @@ static void sync_child_counter(struct perf_counter *child_counter,
 }
 
 static void
-__perf_counter_exit_task(struct task_struct *child,
-			 struct perf_counter *child_counter,
+__perf_counter_exit_task(struct perf_counter *child_counter,
 			 struct perf_counter_context *child_ctx)
 {
 	struct perf_counter *parent_counter;
@@ -3605,7 +3604,7 @@ void perf_counter_exit_task(struct task_struct *child)
 again:
 	list_for_each_entry_safe(child_counter, tmp, &child_ctx->counter_list,
 				 list_entry)
-		__perf_counter_exit_task(child, child_counter, child_ctx);
+		__perf_counter_exit_task(child_counter, child_ctx);
 
 	/*
 	 * If the last counter was a group counter, it will have appended all
@@ -3620,6 +3619,44 @@ void perf_counter_exit_task(struct task_struct *child)
 	put_ctx(child_ctx);
 }
 
+/*
+ * free an unexposed, unused context as created by inheritance by
+ * init_task below, used by fork() in case of fail.
+ */
+void perf_counter_free_task(struct task_struct *task)
+{
+	struct perf_counter_context *ctx = task->perf_counter_ctxp;
+	struct perf_counter *counter, *tmp;
+
+	if (!ctx)
+		return;
+
+	mutex_lock(&ctx->mutex);
+again:
+	list_for_each_entry_safe(counter, tmp, &ctx->counter_list, list_entry) {
+		struct perf_counter *parent = counter->parent;
+
+		if (WARN_ON_ONCE(!parent))
+			continue;
+
+		mutex_lock(&parent->child_mutex);
+		list_del_init(&counter->child_list);
+		mutex_unlock(&parent->child_mutex);
+
+		fput(parent->filp);
+
+		list_del_counter(counter, ctx);
+		free_counter(counter);
+	}
+
+	if (!list_empty(&ctx->counter_list))
+		goto again;
+
+	mutex_unlock(&ctx->mutex);
+
+	put_ctx(ctx);
+}
+
 /*
  * Initialize the perf_counter context in task_struct
  */