perf core: Allow setting up max frame stack depth via sysctl

The default remains 127, which is good for most cases, and not even hit
most of the time, but then for some cases, as reported by Brendan, 1024+
deep frames are appearing on the radar for things like groovy, ruby.

And in some workloads putting a _lower_ cap on this may make sense. One
that is per event still needs to be put in place tho.

The new file is:

  # cat /proc/sys/kernel/perf_event_max_stack
  127

Chaging it:

  # echo 256 > /proc/sys/kernel/perf_event_max_stack
  # cat /proc/sys/kernel/perf_event_max_stack
  256

But as soon as there is some event using callchains we get:

  # echo 512 > /proc/sys/kernel/perf_event_max_stack
  -bash: echo: write error: Device or resource busy
  #

Because we only allocate the callchain percpu data structures when there
is a user, which allows for changing the max easily, its just a matter
of having no callchain users at that point.

Reported-and-Tested-by: Brendan Gregg <[email protected]>
Reviewed-by: Frederic Weisbecker <[email protected]>
Acked-by: Alexei Starovoitov <[email protected]>
Acked-by: David Ahern <[email protected]>
Cc: Adrian Hunter <[email protected]>
Cc: Alexander Shishkin <[email protected]>
Cc: He Kuang <[email protected]>
Cc: Jiri Olsa <[email protected]>
Cc: Linus Torvalds <[email protected]>
Cc: Masami Hiramatsu <[email protected]>
Cc: Milian Wolff <[email protected]>
Cc: Namhyung Kim <[email protected]>
Cc: Peter Zijlstra <[email protected]>
Cc: Stephane Eranian <[email protected]>
Cc: Thomas Gleixner <[email protected]>
Cc: Vince Weaver <[email protected]>
Cc: Wang Nan <[email protected]>
Cc: Zefan Li <[email protected]>
Link: http://lkml.kernel.org/r/[email protected]
Signed-off-by: Arnaldo Carvalho de Melo <[email protected]>

3 files changed, 49 insertions, 6 deletions

diff --git a/kernel/bpf/stackmap.c b/kernel/bpf/stackmap.c
index 499d9e933f8e..f5a19548be12 100644
--- a/kernel/bpf/stackmap.c
+++ b/kernel/bpf/stackmap.c
@@ -66,7 +66,7 @@ static struct bpf_map *stack_map_alloc(union bpf_attr *attr)
 	/* check sanity of attributes */
 	if (attr->max_entries == 0 || attr->key_size != 4 ||
 	    value_size < 8 || value_size % 8 ||
-	    value_size / 8 > PERF_MAX_STACK_DEPTH)
+	    value_size / 8 > sysctl_perf_event_max_stack)
 		return ERR_PTR(-EINVAL);
 
 	/* hash table size must be power of 2 */
@@ -124,8 +124,8 @@ static u64 bpf_get_stackid(u64 r1, u64 r2, u64 flags, u64 r4, u64 r5)
 	struct perf_callchain_entry *trace;
 	struct stack_map_bucket *bucket, *new_bucket, *old_bucket;
 	u32 max_depth = map->value_size / 8;
-	/* stack_map_alloc() checks that max_depth <= PERF_MAX_STACK_DEPTH */
-	u32 init_nr = PERF_MAX_STACK_DEPTH - max_depth;
+	/* stack_map_alloc() checks that max_depth <= sysctl_perf_event_max_stack */
+	u32 init_nr = sysctl_perf_event_max_stack - max_depth;
 	u32 skip = flags & BPF_F_SKIP_FIELD_MASK;
 	u32 hash, id, trace_nr, trace_len;
 	bool user = flags & BPF_F_USER_STACK;
@@ -143,7 +143,7 @@ static u64 bpf_get_stackid(u64 r1, u64 r2, u64 flags, u64 r4, u64 r5)
 		return -EFAULT;
 
 	/* get_perf_callchain() guarantees that trace->nr >= init_nr
-	 * and trace-nr <= PERF_MAX_STACK_DEPTH, so trace_nr <= max_depth
+	 * and trace-nr <= sysctl_perf_event_max_stack, so trace_nr <= max_depth
 	 */
 	trace_nr = trace->nr - init_nr;
 
diff --git a/kernel/events/callchain.c b/kernel/events/callchain.c
index 343c22f5e867..b9325e7dcba1 100644
--- a/kernel/events/callchain.c
+++ b/kernel/events/callchain.c
@@ -18,6 +18,14 @@ struct callchain_cpus_entries {
 	struct perf_callchain_entry	*cpu_entries[0];
 };
 
+int sysctl_perf_event_max_stack __read_mostly = PERF_MAX_STACK_DEPTH;
+
+static inline size_t perf_callchain_entry__sizeof(void)
+{
+	return (sizeof(struct perf_callchain_entry) +
+		sizeof(__u64) * sysctl_perf_event_max_stack);
+}
+
 static DEFINE_PER_CPU(int, callchain_recursion[PERF_NR_CONTEXTS]);
 static atomic_t nr_callchain_events;
 static DEFINE_MUTEX(callchain_mutex);
@@ -73,7 +81,7 @@ static int alloc_callchain_buffers(void)
 	if (!entries)
 		return -ENOMEM;
 
-	size = sizeof(struct perf_callchain_entry) * PERF_NR_CONTEXTS;
+	size = perf_callchain_entry__sizeof() * PERF_NR_CONTEXTS;
 
 	for_each_possible_cpu(cpu) {
 		entries->cpu_entries[cpu] = kmalloc_node(size, GFP_KERNEL,
@@ -147,7 +155,8 @@ static struct perf_callchain_entry *get_callchain_entry(int *rctx)
 
 	cpu = smp_processor_id();
 
-	return &entries->cpu_entries[cpu][*rctx];
+	return (((void *)entries->cpu_entries[cpu]) +
+		(*rctx * perf_callchain_entry__sizeof()));
 }
 
 static void
@@ -215,3 +224,25 @@ exit_put:
 
 	return entry;
 }
+
+int perf_event_max_stack_handler(struct ctl_table *table, int write,
+				 void __user *buffer, size_t *lenp, loff_t *ppos)
+{
+	int new_value = sysctl_perf_event_max_stack, ret;
+	struct ctl_table new_table = *table;
+
+	new_table.data = &new_value;
+	ret = proc_dointvec_minmax(&new_table, write, buffer, lenp, ppos);
+	if (ret || !write)
+		return ret;
+
+	mutex_lock(&callchain_mutex);
+	if (atomic_read(&nr_callchain_events))
+		ret = -EBUSY;
+	else
+		sysctl_perf_event_max_stack = new_value;
+
+	mutex_unlock(&callchain_mutex);
+
+	return ret;
+}
diff --git a/kernel/sysctl.c b/kernel/sysctl.c
index 725587f10667..c8b318663525 100644
--- a/kernel/sysctl.c
+++ b/kernel/sysctl.c
@@ -130,6 +130,9 @@ static int one_thousand = 1000;
 #ifdef CONFIG_PRINTK
 static int ten_thousand = 10000;
 #endif
+#ifdef CONFIG_PERF_EVENTS
+static int six_hundred_forty_kb = 640 * 1024;
+#endif
 
 /* this is needed for the proc_doulongvec_minmax of vm_dirty_bytes */
 static unsigned long dirty_bytes_min = 2 * PAGE_SIZE;
@@ -1144,6 +1147,15 @@ static struct ctl_table kern_table[] = {
 		.extra1		= &zero,
 		.extra2		= &one_hundred,
 	},
+	{
+		.procname	= "perf_event_max_stack",
+		.data		= NULL, /* filled in by handler */
+		.maxlen		= sizeof(sysctl_perf_event_max_stack),
+		.mode		= 0644,
+		.proc_handler	= perf_event_max_stack_handler,
+		.extra1		= &zero,
+		.extra2		= &six_hundred_forty_kb,
+	},
 #endif
 #ifdef CONFIG_KMEMCHECK
 	{