[PATCH v3 bpf-next 0/3] enable BPF_PROG_TEST_RUN for raw_tp

[PATCH v3 bpf-next 0/3] enable BPF_PROG_TEST_RUN for raw_tp

[Song Liu]

This set enables BPF_PROG_TEST_RUN for raw_tracepoint type programs. This
set also enables running the raw_tp program on a specific CPU. This feature
can be used by user space to trigger programs that access percpu resources,
e.g. perf_event, percpu variables.

Changes v2 => v3:
1. Fix memory leak in the selftest. (Andrii)
2. Use __u64 instead of unsigned long long. (Andrii)

Changes v1 => v2:
1. More checks for retval in the selftest. (John)
2. Remove unnecessary goto in bpf_prog_test_run_raw_tp. (John)

Song Liu (3):
  bpf: enable BPF_PROG_TEST_RUN for raw_tracepoint
  libbpf: introduce bpf_prog_test_run_xattr_opts
  selftests/bpf: add raw_tp_test_run

 include/linux/bpf.h                           |  3 +
 include/uapi/linux/bpf.h                      |  5 ++
 kernel/bpf/syscall.c                          |  2 +-
 kernel/trace/bpf_trace.c                      |  1 +
 net/bpf/test_run.c                            | 88 +++++++++++++++++++
 tools/include/uapi/linux/bpf.h                |  5 ++
 tools/lib/bpf/bpf.c                           | 13 ++-
 tools/lib/bpf/bpf.h                           | 11 +++
 tools/lib/bpf/libbpf.map                      |  1 +
 .../bpf/prog_tests/raw_tp_test_run.c          | 75 ++++++++++++++++
 .../bpf/progs/test_raw_tp_test_run.c          | 25 ++++++
 11 files changed, 227 insertions(+), 2 deletions(-)
 create mode 100644 tools/testing/selftests/bpf/prog_tests/raw_tp_test_run.c
 create mode 100644 tools/testing/selftests/bpf/progs/test_raw_tp_test_run.c

--
2.24.1

[PATCH v3 bpf-next 1/3] bpf: enable BPF_PROG_TEST_RUN for raw_tracepoint

[Song Liu]

Add .test_run for raw_tracepoint. Also, introduce a new feature that runs
the target program on a specific CPU. This is achieved by a new flag in
bpf_attr.test, cpu_plus. For compatibility, cpu_plus == 0 means run the
program on current cpu, cpu_plus > 0 means run the program on cpu with id
(cpu_plus - 1). This feature is needed for BPF programs that handle
perf_event and other percpu resources, as the program can access these
resource locally.

Acked-by: John Fastabend <john.fastabend@gmail.com>
Signed-off-by: Song Liu <songliubraving@fb.com>
---
 include/linux/bpf.h            |  3 ++
 include/uapi/linux/bpf.h       |  5 ++
 kernel/bpf/syscall.c           |  2 +-
 kernel/trace/bpf_trace.c       |  1 +
 net/bpf/test_run.c             | 88 ++++++++++++++++++++++++++++++++++
 tools/include/uapi/linux/bpf.h |  5 ++
 6 files changed, 103 insertions(+), 1 deletion(-)

diff --git a/include/linux/bpf.h b/include/linux/bpf.h
index d7c5a6ed87e30..23758c282eb4b 100644
--- a/include/linux/bpf.h
+++ b/include/linux/bpf.h
@@ -1376,6 +1376,9 @@ int bpf_prog_test_run_tracing(struct bpf_prog *prog,
 int bpf_prog_test_run_flow_dissector(struct bpf_prog *prog,
 				     const union bpf_attr *kattr,
 				     union bpf_attr __user *uattr);
+int bpf_prog_test_run_raw_tp(struct bpf_prog *prog,
+			     const union bpf_attr *kattr,
+			     union bpf_attr __user *uattr);
 bool btf_ctx_access(int off, int size, enum bpf_access_type type,
 		    const struct bpf_prog *prog,
 		    struct bpf_insn_access_aux *info);
diff --git a/include/uapi/linux/bpf.h b/include/uapi/linux/bpf.h
index a22812561064a..89acf41913e70 100644
--- a/include/uapi/linux/bpf.h
+++ b/include/uapi/linux/bpf.h
@@ -566,6 +566,11 @@ union bpf_attr {
 						 */
 		__aligned_u64	ctx_in;
 		__aligned_u64	ctx_out;
+		__u32		cpu_plus;	/* run this program on cpu
+						 * (cpu_plus - 1).
+						 * If cpu_plus == 0, run on
+						 * current cpu.
+						 */
 	} test;
 
 	struct { /* anonymous struct used by BPF_*_GET_*_ID */
diff --git a/kernel/bpf/syscall.c b/kernel/bpf/syscall.c
index ec68d3a23a2b7..4664531ff92ea 100644
--- a/kernel/bpf/syscall.c
+++ b/kernel/bpf/syscall.c
@@ -2975,7 +2975,7 @@ static int bpf_prog_query(const union bpf_attr *attr,
 	}
 }
 
-#define BPF_PROG_TEST_RUN_LAST_FIELD test.ctx_out
+#define BPF_PROG_TEST_RUN_LAST_FIELD test.cpu_plus
 
 static int bpf_prog_test_run(const union bpf_attr *attr,
 			     union bpf_attr __user *uattr)
diff --git a/kernel/trace/bpf_trace.c b/kernel/trace/bpf_trace.c
index b2a5380eb1871..4553aebf53862 100644
--- a/kernel/trace/bpf_trace.c
+++ b/kernel/trace/bpf_trace.c
@@ -1675,6 +1675,7 @@ const struct bpf_verifier_ops raw_tracepoint_verifier_ops = {
 };
 
 const struct bpf_prog_ops raw_tracepoint_prog_ops = {
+	.test_run = bpf_prog_test_run_raw_tp,
 };
 
 const struct bpf_verifier_ops tracing_verifier_ops = {
diff --git a/net/bpf/test_run.c b/net/bpf/test_run.c
index 99eb8c6c0fbcc..b32e22f0ee16f 100644
--- a/net/bpf/test_run.c
+++ b/net/bpf/test_run.c
@@ -11,6 +11,7 @@
 #include <net/sock.h>
 #include <net/tcp.h>
 #include <linux/error-injection.h>
+#include <linux/smp.h>
 
 #define CREATE_TRACE_POINTS
 #include <trace/events/bpf_test_run.h>
@@ -204,6 +205,9 @@ int bpf_prog_test_run_tracing(struct bpf_prog *prog,
 	int b = 2, err = -EFAULT;
 	u32 retval = 0;
 
+	if (kattr->test.cpu_plus)
+		return -EINVAL;
+
 	switch (prog->expected_attach_type) {
 	case BPF_TRACE_FENTRY:
 	case BPF_TRACE_FEXIT:
@@ -236,6 +240,84 @@ int bpf_prog_test_run_tracing(struct bpf_prog *prog,
 	return err;
 }
 
+struct bpf_raw_tp_test_run_info {
+	struct bpf_prog *prog;
+	void *ctx;
+	u32 retval;
+};
+
+static void
+__bpf_prog_test_run_raw_tp(void *data)
+{
+	struct bpf_raw_tp_test_run_info *info = data;
+
+	rcu_read_lock();
+	migrate_disable();
+	info->retval = BPF_PROG_RUN(info->prog, info->ctx);
+	migrate_enable();
+	rcu_read_unlock();
+}
+
+int bpf_prog_test_run_raw_tp(struct bpf_prog *prog,
+			     const union bpf_attr *kattr,
+			     union bpf_attr __user *uattr)
+{
+	void __user *ctx_in = u64_to_user_ptr(kattr->test.ctx_in);
+	__u32 ctx_size_in = kattr->test.ctx_size_in;
+	struct bpf_raw_tp_test_run_info info;
+	int cpu, err = 0;
+
+	/* doesn't support data_in/out, ctx_out, duration, or repeat */
+	if (kattr->test.data_in || kattr->test.data_out ||
+	    kattr->test.ctx_out || kattr->test.duration ||
+	    kattr->test.repeat)
+		return -EINVAL;
+
+	if (ctx_size_in < prog->aux->max_ctx_offset)
+		return -EINVAL;
+
+	if (ctx_size_in) {
+		info.ctx = kzalloc(ctx_size_in, GFP_USER);
+		if (!info.ctx)
+			return -ENOMEM;
+		if (copy_from_user(info.ctx, ctx_in, ctx_size_in)) {
+			err = -EFAULT;
+			goto out;
+		}
+	} else {
+		info.ctx = NULL;
+	}
+
+	info.prog = prog;
+	cpu = kattr->test.cpu_plus - 1;
+
+	if (!kattr->test.cpu_plus || cpu == smp_processor_id()) {
+		__bpf_prog_test_run_raw_tp(&info);
+	} else {
+		/* smp_call_function_single() also checks cpu_online()
+		 * after csd_lock(). However, since cpu_plus is from user
+		 * space, let's do an extra quick check to filter out
+		 * invalid value before smp_call_function_single().
+		 */
+		if (!cpu_online(cpu)) {
+			err = -ENXIO;
+			goto out;
+		}
+
+		err = smp_call_function_single(cpu, __bpf_prog_test_run_raw_tp,
+					       &info, 1);
+		if (err)
+			goto out;
+	}
+
+	if (copy_to_user(&uattr->test.retval, &info.retval, sizeof(u32)))
+		err = -EFAULT;
+
+out:
+	kfree(info.ctx);
+	return err;
+}
+
 static void *bpf_ctx_init(const union bpf_attr *kattr, u32 max_size)
 {
 	void __user *data_in = u64_to_user_ptr(kattr->test.ctx_in);
@@ -410,6 +492,9 @@ int bpf_prog_test_run_skb(struct bpf_prog *prog, const union bpf_attr *kattr,
 	void *data;
 	int ret;
 
+	if (kattr->test.cpu_plus)
+		return -EINVAL;
+
 	data = bpf_test_init(kattr, size, NET_SKB_PAD + NET_IP_ALIGN,
 			     SKB_DATA_ALIGN(sizeof(struct skb_shared_info)));
 	if (IS_ERR(data))
@@ -607,6 +692,9 @@ int bpf_prog_test_run_flow_dissector(struct bpf_prog *prog,
 	if (prog->type != BPF_PROG_TYPE_FLOW_DISSECTOR)
 		return -EINVAL;
 
+	if (kattr->test.cpu_plus)
+		return -EINVAL;
+
 	if (size < ETH_HLEN)
 		return -EINVAL;
 
diff --git a/tools/include/uapi/linux/bpf.h b/tools/include/uapi/linux/bpf.h
index a22812561064a..89acf41913e70 100644
--- a/tools/include/uapi/linux/bpf.h
+++ b/tools/include/uapi/linux/bpf.h
@@ -566,6 +566,11 @@ union bpf_attr {
 						 */
 		__aligned_u64	ctx_in;
 		__aligned_u64	ctx_out;
+		__u32		cpu_plus;	/* run this program on cpu
+						 * (cpu_plus - 1).
+						 * If cpu_plus == 0, run on
+						 * current cpu.
+						 */
 	} test;
 
 	struct { /* anonymous struct used by BPF_*_GET_*_ID */
-- 
2.24.1

[PATCH v3 bpf-next 2/3] libbpf: introduce bpf_prog_test_run_xattr_opts

[Song Liu]

This API supports new field cpu_plus in bpf_attr.test.

Acked-by: John Fastabend <john.fastabend@gmail.com>
Signed-off-by: Song Liu <songliubraving@fb.com>
---
 tools/lib/bpf/bpf.c      | 13 ++++++++++++-
 tools/lib/bpf/bpf.h      | 11 +++++++++++
 tools/lib/bpf/libbpf.map |  1 +
 3 files changed, 24 insertions(+), 1 deletion(-)

diff --git a/tools/lib/bpf/bpf.c b/tools/lib/bpf/bpf.c
index 2baa1308737c8..3228dd60fa32f 100644
--- a/tools/lib/bpf/bpf.c
+++ b/tools/lib/bpf/bpf.c
@@ -684,7 +684,8 @@ int bpf_prog_test_run(int prog_fd, int repeat, void *data, __u32 size,
 	return ret;
 }
 
-int bpf_prog_test_run_xattr(struct bpf_prog_test_run_attr *test_attr)
+int bpf_prog_test_run_xattr_opts(struct bpf_prog_test_run_attr *test_attr,
+				 const struct bpf_prog_test_run_opts *opts)
 {
 	union bpf_attr attr;
 	int ret;
@@ -693,6 +694,11 @@ int bpf_prog_test_run_xattr(struct bpf_prog_test_run_attr *test_attr)
 		return -EINVAL;
 
 	memset(&attr, 0, sizeof(attr));
+	if (opts) {
+		if (!OPTS_VALID(opts, bpf_prog_test_run_opts))
+			return -EINVAL;
+		attr.test.cpu_plus = opts->cpu_plus;
+	}
 	attr.test.prog_fd = test_attr->prog_fd;
 	attr.test.data_in = ptr_to_u64(test_attr->data_in);
 	attr.test.data_out = ptr_to_u64(test_attr->data_out);
@@ -712,6 +718,11 @@ int bpf_prog_test_run_xattr(struct bpf_prog_test_run_attr *test_attr)
 	return ret;
 }
 
+int bpf_prog_test_run_xattr(struct bpf_prog_test_run_attr *test_attr)
+{
+	return bpf_prog_test_run_xattr_opts(test_attr, NULL);
+}
+
 static int bpf_obj_get_next_id(__u32 start_id, __u32 *next_id, int cmd)
 {
 	union bpf_attr attr;
diff --git a/tools/lib/bpf/bpf.h b/tools/lib/bpf/bpf.h
index 8c1ac4b42f908..61318f47c8e1b 100644
--- a/tools/lib/bpf/bpf.h
+++ b/tools/lib/bpf/bpf.h
@@ -251,6 +251,17 @@ struct bpf_prog_bind_opts {
 
 LIBBPF_API int bpf_prog_bind_map(int prog_fd, int map_fd,
 				 const struct bpf_prog_bind_opts *opts);
+
+struct bpf_prog_test_run_opts {
+	size_t sz; /* size of this struct for forward/backward compatibility */
+	__u32 cpu_plus;
+};
+#define bpf_prog_test_run_opts__last_field cpu_plus
+
+LIBBPF_API
+int bpf_prog_test_run_xattr_opts(struct bpf_prog_test_run_attr *test_attr,
+				 const struct bpf_prog_test_run_opts *opts);
+
 #ifdef __cplusplus
 } /* extern "C" */
 #endif
diff --git a/tools/lib/bpf/libbpf.map b/tools/lib/bpf/libbpf.map
index 5f054dadf0829..c84a8bec57634 100644
--- a/tools/lib/bpf/libbpf.map
+++ b/tools/lib/bpf/libbpf.map
@@ -303,6 +303,7 @@ LIBBPF_0.1.0 {
 LIBBPF_0.2.0 {
 	global:
 		bpf_prog_bind_map;
+		bpf_prog_test_run_xattr_opts;
 		bpf_program__section_name;
 		perf_buffer__buffer_cnt;
 		perf_buffer__buffer_fd;
-- 
2.24.1

[PATCH v3 bpf-next 3/3] selftests/bpf: add raw_tp_test_run

[Song Liu]

This test runs test_run for raw_tracepoint program. The test covers ctx
input, retval output, and proper handling of cpu_plus field.

Signed-off-by: Song Liu <songliubraving@fb.com>
---
 .../bpf/prog_tests/raw_tp_test_run.c          | 75 +++++++++++++++++++
 .../bpf/progs/test_raw_tp_test_run.c          | 25 +++++++
 2 files changed, 100 insertions(+)
 create mode 100644 tools/testing/selftests/bpf/prog_tests/raw_tp_test_run.c
 create mode 100644 tools/testing/selftests/bpf/progs/test_raw_tp_test_run.c

diff --git a/tools/testing/selftests/bpf/prog_tests/raw_tp_test_run.c b/tools/testing/selftests/bpf/prog_tests/raw_tp_test_run.c
new file mode 100644
index 0000000000000..492cd8a8c0e00
--- /dev/null
+++ b/tools/testing/selftests/bpf/prog_tests/raw_tp_test_run.c
@@ -0,0 +1,75 @@
+// SPDX-License-Identifier: GPL-2.0-only
+/* Copyright (c) 2019 Facebook */
+#include <test_progs.h>
+#include "bpf/libbpf_internal.h"
+#include "test_raw_tp_test_run.skel.h"
+
+static int duration;
+
+void test_raw_tp_test_run(void)
+{
+	struct bpf_prog_test_run_attr test_attr = {};
+	__u64 args[2] = {0x1234ULL, 0x5678ULL};
+	int comm_fd = -1, err, nr_online, i;
+	int expected_retval = 0x1234 + 0x5678;
+	struct test_raw_tp_test_run *skel;
+	char buf[] = "new_name";
+	bool *online = NULL;
+
+	err = parse_cpu_mask_file("/sys/devices/system/cpu/online", &online,
+				  &nr_online);
+	if (CHECK(err, "parse_cpu_mask_file", "err %d\n", err))
+		return;
+
+	skel = test_raw_tp_test_run__open_and_load();
+	if (CHECK(!skel, "skel_open", "failed to open skeleton\n")) {
+		free(online);
+		return;
+	}
+	err = test_raw_tp_test_run__attach(skel);
+	if (CHECK(err, "skel_attach", "skeleton attach failed: %d\n", err))
+		goto cleanup;
+
+	comm_fd = open("/proc/self/comm", O_WRONLY|O_TRUNC);
+	if (CHECK(comm_fd < 0, "open /proc/self/comm", "err %d\n", errno))
+		goto cleanup;
+
+	err = write(comm_fd, buf, sizeof(buf));
+	CHECK(err < 0, "task rename", "err %d", errno);
+
+	CHECK(skel->bss->count == 0, "check_count", "didn't increase\n");
+	CHECK(skel->data->on_cpu != 0xffffffff, "check_on_cpu", "got wrong value\n");
+
+	test_attr.prog_fd = bpf_program__fd(skel->progs.rename);
+	test_attr.ctx_in = args;
+	test_attr.ctx_size_in = sizeof(__u64);
+
+	err = bpf_prog_test_run_xattr(&test_attr);
+	CHECK(err == 0, "test_run", "should fail for too small ctx\n");
+
+	test_attr.ctx_size_in = sizeof(args);
+	err = bpf_prog_test_run_xattr(&test_attr);
+	CHECK(err < 0, "test_run", "err %d\n", errno);
+	CHECK(test_attr.retval != expected_retval, "check_retval",
+	      "expect 0x%x, got 0x%x\n", expected_retval, test_attr.retval);
+
+	for (i = 0; i < nr_online; i++)
+		if (online[i]) {
+			DECLARE_LIBBPF_OPTS(bpf_prog_test_run_opts, opts,
+				.cpu_plus = i + 1,
+			);
+
+			test_attr.retval = 0;
+			err = bpf_prog_test_run_xattr_opts(&test_attr, &opts);
+			CHECK(err < 0, "test_run_with_opts", "err %d\n", errno);
+			CHECK(skel->data->on_cpu != i, "check_on_cpu",
+			      "got wrong value\n");
+			CHECK(test_attr.retval != expected_retval,
+			      "check_retval", "expect 0x%x, got 0x%x\n",
+			      expected_retval, test_attr.retval);
+		}
+cleanup:
+	close(comm_fd);
+	test_raw_tp_test_run__destroy(skel);
+	free(online);
+}
diff --git a/tools/testing/selftests/bpf/progs/test_raw_tp_test_run.c b/tools/testing/selftests/bpf/progs/test_raw_tp_test_run.c
new file mode 100644
index 0000000000000..6b356e003d16c
--- /dev/null
+++ b/tools/testing/selftests/bpf/progs/test_raw_tp_test_run.c
@@ -0,0 +1,25 @@
+// SPDX-License-Identifier: GPL-2.0
+/* Copyright (c) 2020 Facebook */
+
+#include "vmlinux.h"
+#include <bpf/bpf_helpers.h>
+#include <bpf/bpf_endian.h>
+#include <bpf/bpf_tracing.h>
+
+__u32 count = 0;
+__u32 on_cpu = 0xffffffff;
+
+SEC("raw_tp/task_rename")
+int BPF_PROG(rename, struct task_struct *task, char *comm)
+{
+
+	count++;
+	if ((__u64) task == 0x1234ULL && (__u64) comm == 0x5678ULL) {
+		on_cpu = bpf_get_smp_processor_id();
+		return (int)task + (int)comm;
+	}
+
+	return 0;
+}
+
+char _license[] SEC("license") = "GPL";
-- 
2.24.1

Re: [PATCH v3 bpf-next 0/3] enable BPF_PROG_TEST_RUN for raw_tp

[Song Liu]

> On Sep 23, 2020, at 2:33 PM, Song Liu <songliubraving@fb.com> wrote:
> 
> This set enables BPF_PROG_TEST_RUN for raw_tracepoint type programs. This
> set also enables running the raw_tp program on a specific CPU. This feature
> can be used by user space to trigger programs that access percpu resources,
> e.g. perf_event, percpu variables.
> 
> Changes v2 => v3:
> 1. Fix memory leak in the selftest. (Andrii)
> 2. Use __u64 instead of unsigned long long. (Andrii)
> 
> Changes v1 => v2:
> 1. More checks for retval in the selftest. (John)
> 2. Remove unnecessary goto in bpf_prog_test_run_raw_tp. (John)

I somehow missed Andrii's other comments and sent v3 too fast. 

Please ignore this version.