Merge branch 'bpf-perf-rb-libbpf'

Jakub Kicinski says:

====================
This series started out as a follow up to the bpftool perf event dumping
patches.

As suggested by Daniel patch 1 makes use of PERF_SAMPLE_TIME to simplify
code and improve accuracy of timestamps.

Remaining patches are trying to move perf event loop into libbpf as
suggested by Alexei.  One user for this new function is bpftool which
links with libbpf nicely, the other, unfortunately, is in samples/bpf.
Remaining patches make samples/bpf link against full libbpf.a (not just
a handful of objects).  Once we have full power of libbpf at our disposal
we can convert some of XDP samples to use libbpf loader instead of
bpf_load.c.  My understanding is that this is the desired direction,
at least for networking code.
====================

Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>

3 files changed, 138 insertions, 32 deletions

diff --git a/tools/lib/bpf/Makefile b/tools/lib/bpf/Makefile
index e6d5f8d1477f..f3fab4af4260 100644
--- a/tools/lib/bpf/Makefile
+++ b/tools/lib/bpf/Makefile
@@ -69,7 +69,7 @@ FEATURE_USER = .libbpf
 FEATURE_TESTS = libelf libelf-getphdrnum libelf-mmap bpf
 FEATURE_DISPLAY = libelf bpf
 
-INCLUDES = -I. -I$(srctree)/tools/include -I$(srctree)/tools/arch/$(ARCH)/include/uapi -I$(srctree)/tools/include/uapi
+INCLUDES = -I. -I$(srctree)/tools/include -I$(srctree)/tools/arch/$(ARCH)/include/uapi -I$(srctree)/tools/include/uapi -I$(srctree)/tools/perf
 FEATURE_CHECK_CFLAGS-bpf = $(INCLUDES)
 
 check_feat := 1
diff --git a/tools/lib/bpf/libbpf.c b/tools/lib/bpf/libbpf.c
index 7bcdca13083a..df54c4c9e48a 100644
--- a/tools/lib/bpf/libbpf.c
+++ b/tools/lib/bpf/libbpf.c
@@ -31,6 +31,7 @@
 #include <unistd.h>
 #include <fcntl.h>
 #include <errno.h>
+#include <perf-sys.h>
 #include <asm/unistd.h>
 #include <linux/err.h>
 #include <linux/kernel.h>
@@ -1437,9 +1438,37 @@ bpf_object__load_progs(struct bpf_object *obj)
 	return 0;
 }
 
-static int bpf_object__validate(struct bpf_object *obj)
+static bool bpf_prog_type__needs_kver(enum bpf_prog_type type)
+{
+	switch (type) {
+	case BPF_PROG_TYPE_SOCKET_FILTER:
+	case BPF_PROG_TYPE_SCHED_CLS:
+	case BPF_PROG_TYPE_SCHED_ACT:
+	case BPF_PROG_TYPE_XDP:
+	case BPF_PROG_TYPE_CGROUP_SKB:
+	case BPF_PROG_TYPE_CGROUP_SOCK:
+	case BPF_PROG_TYPE_LWT_IN:
+	case BPF_PROG_TYPE_LWT_OUT:
+	case BPF_PROG_TYPE_LWT_XMIT:
+	case BPF_PROG_TYPE_SOCK_OPS:
+	case BPF_PROG_TYPE_SK_SKB:
+	case BPF_PROG_TYPE_CGROUP_DEVICE:
+	case BPF_PROG_TYPE_SK_MSG:
+	case BPF_PROG_TYPE_CGROUP_SOCK_ADDR:
+		return false;
+	case BPF_PROG_TYPE_UNSPEC:
+	case BPF_PROG_TYPE_KPROBE:
+	case BPF_PROG_TYPE_TRACEPOINT:
+	case BPF_PROG_TYPE_PERF_EVENT:
+	case BPF_PROG_TYPE_RAW_TRACEPOINT:
+	default:
+		return true;
+	}
+}
+
+static int bpf_object__validate(struct bpf_object *obj, bool needs_kver)
 {
-	if (obj->kern_version == 0) {
+	if (needs_kver && obj->kern_version == 0) {
 		pr_warning("%s doesn't provide kernel version\n",
 			   obj->path);
 		return -LIBBPF_ERRNO__KVERSION;
@@ -1448,7 +1477,8 @@ static int bpf_object__validate(struct bpf_object *obj)
 }
 
 static struct bpf_object *
-__bpf_object__open(const char *path, void *obj_buf, size_t obj_buf_sz)
+__bpf_object__open(const char *path, void *obj_buf, size_t obj_buf_sz,
+		   bool needs_kver)
 {
 	struct bpf_object *obj;
 	int err;
@@ -1466,7 +1496,7 @@ __bpf_object__open(const char *path, void *obj_buf, size_t obj_buf_sz)
 	CHECK_ERR(bpf_object__check_endianness(obj), err, out);
 	CHECK_ERR(bpf_object__elf_collect(obj), err, out);
 	CHECK_ERR(bpf_object__collect_reloc(obj), err, out);
-	CHECK_ERR(bpf_object__validate(obj), err, out);
+	CHECK_ERR(bpf_object__validate(obj, needs_kver), err, out);
 
 	bpf_object__elf_finish(obj);
 	return obj;
@@ -1483,7 +1513,7 @@ struct bpf_object *bpf_object__open(const char *path)
 
 	pr_debug("loading %s\n", path);
 
-	return __bpf_object__open(path, NULL, 0);
+	return __bpf_object__open(path, NULL, 0, true);
 }
 
 struct bpf_object *bpf_object__open_buffer(void *obj_buf,
@@ -1506,7 +1536,7 @@ struct bpf_object *bpf_object__open_buffer(void *obj_buf,
 	pr_debug("loading object '%s' from buffer\n",
 		 name);
 
-	return __bpf_object__open(name, obj_buf, obj_buf_sz);
+	return __bpf_object__open(name, obj_buf, obj_buf_sz, true);
 }
 
 int bpf_object__unload(struct bpf_object *obj)
@@ -2163,8 +2193,11 @@ int bpf_prog_load_xattr(const struct bpf_prog_load_attr *attr,
 
 	if (!attr)
 		return -EINVAL;
+	if (!attr->file)
+		return -EINVAL;
 
-	obj = bpf_object__open(attr->file);
+	obj = __bpf_object__open(attr->file, NULL, 0,
+				 bpf_prog_type__needs_kver(attr->prog_type));
 	if (IS_ERR(obj))
 		return -ENOENT;
 
@@ -2210,3 +2243,63 @@ int bpf_prog_load_xattr(const struct bpf_prog_load_attr *attr,
 	*prog_fd = bpf_program__fd(first_prog);
 	return 0;
 }
+
+enum bpf_perf_event_ret
+bpf_perf_event_read_simple(void *mem, unsigned long size,
+			   unsigned long page_size, void **buf, size_t *buf_len,
+			   bpf_perf_event_print_t fn, void *priv)
+{
+	volatile struct perf_event_mmap_page *header = mem;
+	__u64 data_tail = header->data_tail;
+	__u64 data_head = header->data_head;
+	void *base, *begin, *end;
+	int ret;
+
+	asm volatile("" ::: "memory"); /* in real code it should be smp_rmb() */
+	if (data_head == data_tail)
+		return LIBBPF_PERF_EVENT_CONT;
+
+	base = ((char *)header) + page_size;
+
+	begin = base + data_tail % size;
+	end = base + data_head % size;
+
+	while (begin != end) {
+		struct perf_event_header *ehdr;
+
+		ehdr = begin;
+		if (begin + ehdr->size > base + size) {
+			long len = base + size - begin;
+
+			if (*buf_len < ehdr->size) {
+				free(*buf);
+				*buf = malloc(ehdr->size);
+				if (!*buf) {
+					ret = LIBBPF_PERF_EVENT_ERROR;
+					break;
+				}
+				*buf_len = ehdr->size;
+			}
+
+			memcpy(*buf, begin, len);
+			memcpy(*buf + len, base, ehdr->size - len);
+			ehdr = (void *)*buf;
+			begin = base + ehdr->size - len;
+		} else if (begin + ehdr->size == base + size) {
+			begin = base;
+		} else {
+			begin += ehdr->size;
+		}
+
+		ret = fn(ehdr, priv);
+		if (ret != LIBBPF_PERF_EVENT_CONT)
+			break;
+
+		data_tail += ehdr->size;
+	}
+
+	__sync_synchronize(); /* smp_mb() */
+	header->data_tail = data_tail;
+
+	return ret;
+}
diff --git a/tools/lib/bpf/libbpf.h b/tools/lib/bpf/libbpf.h
index 197f9ce2248c..4574b9563278 100644
--- a/tools/lib/bpf/libbpf.h
+++ b/tools/lib/bpf/libbpf.h
@@ -52,8 +52,8 @@ enum libbpf_errno {
 int libbpf_strerror(int err, char *buf, size_t size);
 
 /*
- * In include/linux/compiler-gcc.h, __printf is defined. However
- * it should be better if libbpf.h doesn't depend on Linux header file.
+ * __printf is defined in include/linux/compiler-gcc.h. However,
+ * it would be better if libbpf.h didn't depend on Linux header files.
  * So instead of __printf, here we use gcc attribute directly.
  */
 typedef int (*libbpf_print_fn_t)(const char *, ...)
@@ -92,7 +92,7 @@ int bpf_object__set_priv(struct bpf_object *obj, void *priv,
 			 bpf_object_clear_priv_t clear_priv);
 void *bpf_object__priv(struct bpf_object *prog);
 
-/* Accessors of bpf_program. */
+/* Accessors of bpf_program */
 struct bpf_program;
 struct bpf_program *bpf_program__next(struct bpf_program *prog,
 				      struct bpf_object *obj);
@@ -121,28 +121,28 @@ struct bpf_insn;
 
 /*
  * Libbpf allows callers to adjust BPF programs before being loaded
- * into kernel. One program in an object file can be transform into
- * multiple variants to be attached to different code.
+ * into kernel. One program in an object file can be transformed into
+ * multiple variants to be attached to different hooks.
  *
  * bpf_program_prep_t, bpf_program__set_prep and bpf_program__nth_fd
- * are APIs for this propose.
+ * form an API for this purpose.
  *
  * - bpf_program_prep_t:
- *   It defines 'preprocessor', which is a caller defined function
+ *   Defines a 'preprocessor', which is a caller defined function
  *   passed to libbpf through bpf_program__set_prep(), and will be
  *   called before program is loaded. The processor should adjust
- *   the program one time for each instances according to the number
+ *   the program one time for each instance according to the instance id
  *   passed to it.
  *
  * - bpf_program__set_prep:
- *   Attachs a preprocessor to a BPF program. The number of instances
- *   whould be created is also passed through this function.
+ *   Attaches a preprocessor to a BPF program. The number of instances
+ *   that should be created is also passed through this function.
  *
  * - bpf_program__nth_fd:
- *   After the program is loaded, get resuling fds from bpf program for
- *   each instances.
+ *   After the program is loaded, get resulting FD of a given instance
+ *   of the BPF program.
  *
- * If bpf_program__set_prep() is not used, the program whould be loaded
+ * If bpf_program__set_prep() is not used, the program would be loaded
  * without adjustment during bpf_object__load(). The program has only
  * one instance. In this case bpf_program__fd(prog) is equal to
  * bpf_program__nth_fd(prog, 0).
@@ -156,7 +156,7 @@ struct bpf_prog_prep_result {
 	struct bpf_insn *new_insn_ptr;
 	int new_insn_cnt;
 
-	/* If not NULL, result fd is set to it */
+	/* If not NULL, result FD is written to it. */
 	int *pfd;
 };
 
@@ -169,8 +169,8 @@ struct bpf_prog_prep_result {
  *  - res:	Output parameter, result of transformation.
  *
  * Return value:
- *  - Zero: pre-processing success.
- *  - Non-zero: pre-processing, stop loading.
+ *  - Zero:	pre-processing success.
+ *  - Non-zero:	pre-processing error, stop loading.
  */
 typedef int (*bpf_program_prep_t)(struct bpf_program *prog, int n,
 				  struct bpf_insn *insns, int insns_cnt,
@@ -182,7 +182,7 @@ int bpf_program__set_prep(struct bpf_program *prog, int nr_instance,
 int bpf_program__nth_fd(struct bpf_program *prog, int n);
 
 /*
- * Adjust type of bpf program. Default is kprobe.
+ * Adjust type of BPF program. Default is kprobe.
  */
 int bpf_program__set_socket_filter(struct bpf_program *prog);
 int bpf_program__set_tracepoint(struct bpf_program *prog);
@@ -206,10 +206,10 @@ bool bpf_program__is_xdp(struct bpf_program *prog);
 bool bpf_program__is_perf_event(struct bpf_program *prog);
 
 /*
- * We don't need __attribute__((packed)) now since it is
- * unnecessary for 'bpf_map_def' because they are all aligned.
- * In addition, using it will trigger -Wpacked warning message,
- * and will be treated as an error due to -Werror.
+ * No need for __attribute__((packed)), all members of 'bpf_map_def'
+ * are all aligned.  In addition, using __attribute__((packed))
+ * would trigger a -Wpacked warning message, and lead to an error
+ * if -Werror is set.
  */
 struct bpf_map_def {
 	unsigned int type;
@@ -220,8 +220,8 @@ struct bpf_map_def {
 };
 
 /*
- * There is another 'struct bpf_map' in include/linux/map.h. However,
- * it is not a uapi header so no need to consider name clash.
+ * The 'struct bpf_map' in include/linux/bpf.h is internal to the kernel,
+ * so no need to worry about a name clash.
  */
 struct bpf_map;
 struct bpf_map *
@@ -229,7 +229,7 @@ bpf_object__find_map_by_name(struct bpf_object *obj, const char *name);
 
 /*
  * Get bpf_map through the offset of corresponding struct bpf_map_def
- * in the bpf object file.
+ * in the BPF object file.
  */
 struct bpf_map *
 bpf_object__find_map_by_offset(struct bpf_object *obj, size_t offset);
@@ -267,4 +267,17 @@ int bpf_prog_load(const char *file, enum bpf_prog_type type,
 		  struct bpf_object **pobj, int *prog_fd);
 
 int bpf_set_link_xdp_fd(int ifindex, int fd, __u32 flags);
+
+enum bpf_perf_event_ret {
+	LIBBPF_PERF_EVENT_DONE	= 0,
+	LIBBPF_PERF_EVENT_ERROR	= -1,
+	LIBBPF_PERF_EVENT_CONT	= -2,
+};
+
+typedef enum bpf_perf_event_ret (*bpf_perf_event_print_t)(void *event,
+							  void *priv);
+int bpf_perf_event_read_simple(void *mem, unsigned long size,
+			       unsigned long page_size,
+			       void **buf, size_t *buf_len,
+			       bpf_perf_event_print_t fn, void *priv);
 #endif