[PATCH v4 bpf-next 0/3] bpf: sharing bpf runtime stats with

[PATCH v4 bpf-next 0/3] bpf: sharing bpf runtime stats with

[Song Liu]

run_time_ns is a useful stats for BPF programs. However, it is gated by
sysctl kernel.bpf_stats_enabled. When multiple user space tools are
toggling kernl.bpf_stats_enabled at the same time, they may confuse each
other.

Solve this problem with a new BPF command BPF_ENABLE_STATS.

Changes v3 => v4:
  1. Add libbpf support and selftest;
  2. Avoid cleaning trailing space.

Changes v2 => v3:
  1. Rename the command to BPF_ENABLE_STATS, and make it extendible.
  2. fix commit log;
  3. remove unnecessary headers.

Changes RFC => v2:
  1. Add a new bpf command instead of /dev/bpf_stats;
  2. Remove the jump_label patch, which is no longer needed;
  3. Add a static variable to save previous value of the sysctl.

Song Liu (3):
  bpf: sharing bpf runtime stats with BPF_ENABLE_STATS
  libbpf: add support for command BPF_ENABLE_STATS
  bpf: add selftest for BPF_ENABLE_STATS

 include/linux/bpf.h                           |  1 +
 include/uapi/linux/bpf.h                      | 11 ++++
 kernel/bpf/syscall.c                          | 50 +++++++++++++++++++
 kernel/sysctl.c                               | 36 ++++++++++++-
 tools/include/uapi/linux/bpf.h                | 11 ++++
 tools/lib/bpf/bpf.c                           |  9 ++++
 tools/lib/bpf/bpf.h                           |  1 +
 tools/lib/bpf/libbpf.map                      |  5 ++
 .../selftests/bpf/prog_tests/enable_stats.c   | 45 +++++++++++++++++
 .../selftests/bpf/progs/test_enable_stats.c   | 28 +++++++++++
 10 files changed, 196 insertions(+), 1 deletion(-)
 create mode 100644 tools/testing/selftests/bpf/prog_tests/enable_stats.c
 create mode 100644 tools/testing/selftests/bpf/progs/test_enable_stats.c

--
2.24.1

[PATCH v4 bpf-next 1/3] bpf: sharing bpf runtime stats with BPF_ENABLE_STATS

[Song Liu]

Currently, sysctl kernel.bpf_stats_enabled controls BPF runtime stats.
Typical userspace tools use kernel.bpf_stats_enabled as follows:

  1. Enable kernel.bpf_stats_enabled;
  2. Check program run_time_ns;
  3. Sleep for the monitoring period;
  4. Check program run_time_ns again, calculate the difference;
  5. Disable kernel.bpf_stats_enabled.

The problem with this approach is that only one userspace tool can toggle
this sysctl. If multiple tools toggle the sysctl at the same time, the
measurement may be inaccurate.

To fix this problem while keep backward compatibility, introduce a new
bpf command BPF_ENABLE_STATS. On success, this command enables stats and
returns a valid fd. BPF_ENABLE_STATS takes argument "type". Currently,
only one type, BPF_STATS_RUNTIME_CNT, is supported. We can extend the
command to support other types of stats in the future.

With BPF_ENABLE_STATS, user space tool would have the following flow:

  1. Get a fd with BPF_ENABLE_STATS, and make sure it is valid;
  2. Check program run_time_ns;
  3. Sleep for the monitoring period;
  4. Check program run_time_ns again, calculate the difference;
  5. Close the fd.

Signed-off-by: Song Liu <songliubraving@fb.com>
---
 include/linux/bpf.h            |  1 +
 include/uapi/linux/bpf.h       | 11 ++++++++
 kernel/bpf/syscall.c           | 50 ++++++++++++++++++++++++++++++++++
 kernel/sysctl.c                | 36 +++++++++++++++++++++++-
 tools/include/uapi/linux/bpf.h | 11 ++++++++
 5 files changed, 108 insertions(+), 1 deletion(-)

diff --git a/include/linux/bpf.h b/include/linux/bpf.h
index 25da6ff2a880..5b840e7c69ce 100644
--- a/include/linux/bpf.h
+++ b/include/linux/bpf.h
@@ -987,6 +987,7 @@ _out:							\
 
 #ifdef CONFIG_BPF_SYSCALL
 DECLARE_PER_CPU(int, bpf_prog_active);
+extern struct mutex bpf_stats_enabled_mutex;
 
 /*
  * Block execution of BPF programs attached to instrumentation (perf,
diff --git a/include/uapi/linux/bpf.h b/include/uapi/linux/bpf.h
index 2e29a671d67e..e16dba8d466a 100644
--- a/include/uapi/linux/bpf.h
+++ b/include/uapi/linux/bpf.h
@@ -113,6 +113,7 @@ enum bpf_cmd {
 	BPF_MAP_DELETE_BATCH,
 	BPF_LINK_CREATE,
 	BPF_LINK_UPDATE,
+	BPF_ENABLE_STATS,
 };
 
 enum bpf_map_type {
@@ -379,6 +380,12 @@ enum {
  */
 #define BPF_F_QUERY_EFFECTIVE	(1U << 0)
 
+/* type for BPF_ENABLE_STATS */
+enum bpf_stats_type {
+	/* enabled run_time_ns and run_cnt */
+	BPF_STATS_RUNTIME_CNT = 0,
+};
+
 enum bpf_stack_build_id_status {
 	/* user space need an empty entry to identify end of a trace */
 	BPF_STACK_BUILD_ID_EMPTY = 0,
@@ -589,6 +596,10 @@ union bpf_attr {
 		__u32		old_prog_fd;
 	} link_update;
 
+	struct { /* struct used by BPF_ENABLE_STATS command */
+		__u32		type;
+	} enable_stats;
+
 } __attribute__((aligned(8)));
 
 /* The description below is an attempt at providing documentation to eBPF
diff --git a/kernel/bpf/syscall.c b/kernel/bpf/syscall.c
index d85f37239540..766778368c5f 100644
--- a/kernel/bpf/syscall.c
+++ b/kernel/bpf/syscall.c
@@ -3656,6 +3656,53 @@ static int link_update(union bpf_attr *attr)
 	return ret;
 }
 
+DEFINE_MUTEX(bpf_stats_enabled_mutex);
+
+static int bpf_stats_release(struct inode *inode, struct file *file)
+{
+	mutex_lock(&bpf_stats_enabled_mutex);
+	static_key_slow_dec(&bpf_stats_enabled_key.key);
+	mutex_unlock(&bpf_stats_enabled_mutex);
+	return 0;
+}
+
+static const struct file_operations bpf_stats_fops = {
+	.release = bpf_stats_release,
+};
+
+static int bpf_enable_runtime_stats(void)
+{
+	int fd;
+
+	if (!capable(CAP_SYS_ADMIN))
+		return -EPERM;
+
+	mutex_lock(&bpf_stats_enabled_mutex);
+	/* Set a very high limit to avoid overflow */
+	if (static_key_count(&bpf_stats_enabled_key.key) > INT_MAX / 2) {
+		mutex_unlock(&bpf_stats_enabled_mutex);
+		return -EBUSY;
+	}
+
+	fd = anon_inode_getfd("bpf-stats", &bpf_stats_fops, NULL, 0);
+	if (fd >= 0)
+		static_key_slow_inc(&bpf_stats_enabled_key.key);
+
+	mutex_unlock(&bpf_stats_enabled_mutex);
+	return fd;
+}
+
+static int bpf_enable_stats(union bpf_attr *attr)
+{
+	switch (attr->enable_stats.type) {
+	case BPF_STATS_RUNTIME_CNT:
+		return bpf_enable_runtime_stats();
+	default:
+		break;
+	}
+	return -EINVAL;
+}
+
 SYSCALL_DEFINE3(bpf, int, cmd, union bpf_attr __user *, uattr, unsigned int, size)
 {
 	union bpf_attr attr;
@@ -3773,6 +3820,9 @@ SYSCALL_DEFINE3(bpf, int, cmd, union bpf_attr __user *, uattr, unsigned int, siz
 	case BPF_LINK_UPDATE:
 		err = link_update(&attr);
 		break;
+	case BPF_ENABLE_STATS:
+		err = bpf_enable_stats(&attr);
+		break;
 	default:
 		err = -EINVAL;
 		break;
diff --git a/kernel/sysctl.c b/kernel/sysctl.c
index 8a176d8727a3..2d72f4ad3063 100644
--- a/kernel/sysctl.c
+++ b/kernel/sysctl.c
@@ -304,6 +304,40 @@ static int min_extfrag_threshold;
 static int max_extfrag_threshold = 1000;
 #endif
 
+#ifdef CONFIG_BPF_SYSCALL
+static int bpf_stats_handler(struct ctl_table *table, int write,
+			     void __user *buffer, size_t *lenp,
+			     loff_t *ppos)
+{
+	struct static_key *key = (struct static_key *)table->data;
+	static int saved_val;
+	int val, ret;
+	struct ctl_table tmp = {
+		.data   = &val,
+		.maxlen = sizeof(val),
+		.mode   = table->mode,
+		.extra1 = SYSCTL_ZERO,
+		.extra2 = SYSCTL_ONE,
+	};
+
+	if (write && !capable(CAP_SYS_ADMIN))
+		return -EPERM;
+
+	mutex_lock(&bpf_stats_enabled_mutex);
+	val = saved_val;
+	ret = proc_dointvec_minmax(&tmp, write, buffer, lenp, ppos);
+	if (write && !ret && val != saved_val) {
+		if (val)
+			static_key_slow_inc(key);
+		else
+			static_key_slow_dec(key);
+		saved_val = val;
+	}
+	mutex_unlock(&bpf_stats_enabled_mutex);
+	return ret;
+}
+#endif
+
 static struct ctl_table kern_table[] = {
 	{
 		.procname	= "sched_child_runs_first",
@@ -1244,7 +1278,7 @@ static struct ctl_table kern_table[] = {
 		.data		= &bpf_stats_enabled_key.key,
 		.maxlen		= sizeof(bpf_stats_enabled_key),
 		.mode		= 0644,
-		.proc_handler	= proc_do_static_key,
+		.proc_handler	= bpf_stats_handler,
 	},
 #endif
 #if defined(CONFIG_TREE_RCU)
diff --git a/tools/include/uapi/linux/bpf.h b/tools/include/uapi/linux/bpf.h
index 2e29a671d67e..e16dba8d466a 100644
--- a/tools/include/uapi/linux/bpf.h
+++ b/tools/include/uapi/linux/bpf.h
@@ -113,6 +113,7 @@ enum bpf_cmd {
 	BPF_MAP_DELETE_BATCH,
 	BPF_LINK_CREATE,
 	BPF_LINK_UPDATE,
+	BPF_ENABLE_STATS,
 };
 
 enum bpf_map_type {
@@ -379,6 +380,12 @@ enum {
  */
 #define BPF_F_QUERY_EFFECTIVE	(1U << 0)
 
+/* type for BPF_ENABLE_STATS */
+enum bpf_stats_type {
+	/* enabled run_time_ns and run_cnt */
+	BPF_STATS_RUNTIME_CNT = 0,
+};
+
 enum bpf_stack_build_id_status {
 	/* user space need an empty entry to identify end of a trace */
 	BPF_STACK_BUILD_ID_EMPTY = 0,
@@ -589,6 +596,10 @@ union bpf_attr {
 		__u32		old_prog_fd;
 	} link_update;
 
+	struct { /* struct used by BPF_ENABLE_STATS command */
+		__u32		type;
+	} enable_stats;
+
 } __attribute__((aligned(8)));
 
 /* The description below is an attempt at providing documentation to eBPF
-- 
2.24.1

[PATCH v4 bpf-next 2/3] libbpf: add support for command BPF_ENABLE_STATS

[Song Liu]

bpf_enable_stats() is added to enable given stats.

Signed-off-by: Song Liu <songliubraving@fb.com>
---
 tools/lib/bpf/bpf.c      | 9 +++++++++
 tools/lib/bpf/bpf.h      | 1 +
 tools/lib/bpf/libbpf.map | 5 +++++
 3 files changed, 15 insertions(+)

diff --git a/tools/lib/bpf/bpf.c b/tools/lib/bpf/bpf.c
index 5cc1b0785d18..c06c25293ab7 100644
--- a/tools/lib/bpf/bpf.c
+++ b/tools/lib/bpf/bpf.c
@@ -826,3 +826,12 @@ int bpf_task_fd_query(int pid, int fd, __u32 flags, char *buf, __u32 *buf_len,
 
 	return err;
 }
+
+int bpf_enable_stats(enum bpf_stats_type type)
+{
+	union bpf_attr attr = {};
+
+	attr.enable_stats.type = type;
+
+	return sys_bpf(BPF_ENABLE_STATS, &attr, sizeof(attr));
+}
diff --git a/tools/lib/bpf/bpf.h b/tools/lib/bpf/bpf.h
index 46d47afdd887..5996e64d324c 100644
--- a/tools/lib/bpf/bpf.h
+++ b/tools/lib/bpf/bpf.h
@@ -229,6 +229,7 @@ LIBBPF_API int bpf_load_btf(void *btf, __u32 btf_size, char *log_buf,
 LIBBPF_API int bpf_task_fd_query(int pid, int fd, __u32 flags, char *buf,
 				 __u32 *buf_len, __u32 *prog_id, __u32 *fd_type,
 				 __u64 *probe_offset, __u64 *probe_addr);
+LIBBPF_API int bpf_enable_stats(enum bpf_stats_type type);
 
 #ifdef __cplusplus
 } /* extern "C" */
diff --git a/tools/lib/bpf/libbpf.map b/tools/lib/bpf/libbpf.map
index bb8831605b25..ebd946faada5 100644
--- a/tools/lib/bpf/libbpf.map
+++ b/tools/lib/bpf/libbpf.map
@@ -254,3 +254,8 @@ LIBBPF_0.0.8 {
 		bpf_program__set_lsm;
 		bpf_set_link_xdp_fd_opts;
 } LIBBPF_0.0.7;
+
+LIBBPF_0.0.9 {
+	global:
+		bpf_enable_stats;
+} LIBBPF_0.0.8;
\ No newline at end of file
-- 
2.24.1

Re: [PATCH v4 bpf-next 2/3] libbpf: add support for command BPF_ENABLE_STATS

[Andrii Nakryiko]

On Thu, Apr 23, 2020 at 11:19 PM Song Liu <songliubraving@fb.com> wrote:
>
> bpf_enable_stats() is added to enable given stats.
>
> Signed-off-by: Song Liu <songliubraving@fb.com>
> ---
>  tools/lib/bpf/bpf.c      | 9 +++++++++
>  tools/lib/bpf/bpf.h      | 1 +
>  tools/lib/bpf/libbpf.map | 5 +++++
>  3 files changed, 15 insertions(+)
>
> diff --git a/tools/lib/bpf/bpf.c b/tools/lib/bpf/bpf.c
> index 5cc1b0785d18..c06c25293ab7 100644
> --- a/tools/lib/bpf/bpf.c
> +++ b/tools/lib/bpf/bpf.c
> @@ -826,3 +826,12 @@ int bpf_task_fd_query(int pid, int fd, __u32 flags, char *buf, __u32 *buf_len,
>
>         return err;
>  }
> +
> +int bpf_enable_stats(enum bpf_stats_type type)
> +{
> +       union bpf_attr attr = {};

this has to be memset(0) explicitly due to possibility of compiler not
zero-initializing padding

> +
> +       attr.enable_stats.type = type;
> +
> +       return sys_bpf(BPF_ENABLE_STATS, &attr, sizeof(attr));
> +}
> diff --git a/tools/lib/bpf/bpf.h b/tools/lib/bpf/bpf.h
> index 46d47afdd887..5996e64d324c 100644
> --- a/tools/lib/bpf/bpf.h
> +++ b/tools/lib/bpf/bpf.h
> @@ -229,6 +229,7 @@ LIBBPF_API int bpf_load_btf(void *btf, __u32 btf_size, char *log_buf,
>  LIBBPF_API int bpf_task_fd_query(int pid, int fd, __u32 flags, char *buf,
>                                  __u32 *buf_len, __u32 *prog_id, __u32 *fd_type,
>                                  __u64 *probe_offset, __u64 *probe_addr);
> +LIBBPF_API int bpf_enable_stats(enum bpf_stats_type type);
>
>  #ifdef __cplusplus
>  } /* extern "C" */
> diff --git a/tools/lib/bpf/libbpf.map b/tools/lib/bpf/libbpf.map
> index bb8831605b25..ebd946faada5 100644
> --- a/tools/lib/bpf/libbpf.map
> +++ b/tools/lib/bpf/libbpf.map
> @@ -254,3 +254,8 @@ LIBBPF_0.0.8 {
>                 bpf_program__set_lsm;
>                 bpf_set_link_xdp_fd_opts;
>  } LIBBPF_0.0.7;
> +
> +LIBBPF_0.0.9 {
> +       global:
> +               bpf_enable_stats;
> +} LIBBPF_0.0.8;
> \ No newline at end of file
> --
> 2.24.1
>

[PATCH v4 bpf-next 3/3] bpf: add selftest for BPF_ENABLE_STATS

[Song Liu]

Add test for  BPF_ENABLE_STATS, which should enable run_time_ns stats.

~/selftests/bpf# ./test_progs -t enable_stats  -v
test_enable_stats:PASS:skel_open_and_load 0 nsec
test_enable_stats:PASS:get_stats_fd 0 nsec
test_enable_stats:PASS:attach_raw_tp 0 nsec
test_enable_stats:PASS:get_prog_info 0 nsec
test_enable_stats:PASS:check_stats_enabled 0 nsec
Summary: 1/0 PASSED, 0 SKIPPED, 0 FAILED

Signed-off-by: Song Liu <songliubraving@fb.com>
---
 .../selftests/bpf/prog_tests/enable_stats.c   | 45 +++++++++++++++++++
 .../selftests/bpf/progs/test_enable_stats.c   | 28 ++++++++++++
 2 files changed, 73 insertions(+)
 create mode 100644 tools/testing/selftests/bpf/prog_tests/enable_stats.c
 create mode 100644 tools/testing/selftests/bpf/progs/test_enable_stats.c

diff --git a/tools/testing/selftests/bpf/prog_tests/enable_stats.c b/tools/testing/selftests/bpf/prog_tests/enable_stats.c
new file mode 100644
index 000000000000..987fc743ab75
--- /dev/null
+++ b/tools/testing/selftests/bpf/prog_tests/enable_stats.c
@@ -0,0 +1,45 @@
+// SPDX-License-Identifier: GPL-2.0
+#include <test_progs.h>
+#include <sys/mman.h>
+#include "test_enable_stats.skel.h"
+
+void test_enable_stats(void)
+{
+	struct test_enable_stats *skel;
+	struct bpf_prog_info info = {};
+	__u32 info_len = sizeof(info);
+	int stats_fd, err, prog_fd;
+	int duration = 0;
+
+	skel = test_enable_stats__open_and_load();
+
+	if (CHECK(!skel, "skel_open_and_load", "skeleton open/load failed\n"))
+		return;
+
+	stats_fd = bpf_enable_stats(BPF_STATS_RUNTIME_CNT);
+
+	if (CHECK(stats_fd < 0, "get_stats_fd", "failed %d\n", errno))
+		goto cleanup;
+
+	err = test_enable_stats__attach(skel);
+
+	if (CHECK(err, "attach_raw_tp", "err %d\n", err))
+		goto cleanup;
+
+	usleep(1000);
+
+	prog_fd = bpf_program__fd(skel->progs.test_enable_stats);
+	err = bpf_obj_get_info_by_fd(prog_fd, &info, &info_len);
+
+	if (CHECK(err, "get_prog_info",
+		  "failed to get bpf_prog_info for fd %d\n", prog_fd))
+		goto cleanup;
+
+	CHECK(info.run_time_ns == 0, "check_stats_enabled",
+	      "failed to enable run_time_ns stats\n");
+
+cleanup:
+	test_enable_stats__destroy(skel);
+	if (stats_fd >= 0)
+		close(stats_fd);
+}
diff --git a/tools/testing/selftests/bpf/progs/test_enable_stats.c b/tools/testing/selftests/bpf/progs/test_enable_stats.c
new file mode 100644
index 000000000000..f95ac0c94639
--- /dev/null
+++ b/tools/testing/selftests/bpf/progs/test_enable_stats.c
@@ -0,0 +1,28 @@
+// SPDX-License-Identifier: GPL-2.0
+// Copyright (c) 2020 Facebook
+
+#include <linux/bpf.h>
+#include <stdint.h>
+#include <bpf/bpf_helpers.h>
+
+char _license[] SEC("license") = "GPL";
+
+struct {
+	__uint(type, BPF_MAP_TYPE_ARRAY);
+	__uint(max_entries, 1);
+	__type(key, __u32);
+	__type(value, __u64);
+} count SEC(".maps");
+
+SEC("raw_tracepoint/sys_enter")
+int test_enable_stats(void *ctx)
+{
+	__u32 key = 0;
+	__u64 *val;
+
+	val = bpf_map_lookup_elem(&count, &key);
+	if (val)
+		*val += 1;
+
+	return 0;
+}
-- 
2.24.1