bpf, libbpf: simplify and cleanup perf ring buffer walk

Simplify bpf_perf_event_read_simple() a bit and fix up some minor
things along the way: the return code in the header is not of type
int but enum bpf_perf_event_ret instead. Once callback indicated
to break the loop walking event data, it also needs to be consumed
in data_tail since it has been processed already.

Moreover, bpf_perf_event_print_t callback should avoid void * as
we actually get a pointer to struct perf_event_header and thus
applications can make use of container_of() to have type checks.
The walk also doesn't have to use modulo op since the ring size is
required to be power of two.

Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
Signed-off-by: Alexei Starovoitov <ast@kernel.org>

tools/bpf/bpftool/map_perf_ring.c

@@ -50,15 +50,17 @@ static void int_exit(int signo)
 	stop = true;
 }
 
-static enum bpf_perf_event_ret print_bpf_output(void *event, void *priv)
+static enum bpf_perf_event_ret
+print_bpf_output(struct perf_event_header *event, void *private_data)
 {
-	struct event_ring_info *ring = priv;
-	struct perf_event_sample *e = event;
+	struct perf_event_sample *e = container_of(event, struct perf_event_sample,
+						   header);
+	struct event_ring_info *ring = private_data;
 	struct {
 		struct perf_event_header header;
 		__u64 id;
 		__u64 lost;
-	} *lost = event;
+	} *lost = (typeof(lost))event;
 
 	if (json_output) {
 		jsonw_start_object(json_wtr);

tools/lib/bpf/libbpf.c

@@ -2415,56 +2415,47 @@ int bpf_prog_load_xattr(const struct bpf_prog_load_attr *attr,
 }
 
 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)
+bpf_perf_event_read_simple(void *mmap_mem, size_t mmap_size, size_t page_size,
+			   void **copy_mem, size_t *copy_size,
+			   bpf_perf_event_print_t fn, void *private_data)
 {
-	struct perf_event_mmap_page *header = mem;
+	struct perf_event_mmap_page *header = mmap_mem;
 	__u64 data_head = ring_buffer_read_head(header);
 	__u64 data_tail = header->data_tail;
-	int ret = LIBBPF_PERF_EVENT_ERROR;
-	void *base, *begin, *end;
+	void *base = ((__u8 *)header) + page_size;
+	int ret = LIBBPF_PERF_EVENT_CONT;
+	struct perf_event_header *ehdr;
+	size_t ehdr_size;
 
-	if (data_head == data_tail)
-		return LIBBPF_PERF_EVENT_CONT;
+	while (data_head != data_tail) {
+		ehdr = base + (data_tail & (mmap_size - 1));
+		ehdr_size = ehdr->size;
 
-	base = ((char *)header) + page_size;
+		if (((void *)ehdr) + ehdr_size > base + mmap_size) {
+			void *copy_start = ehdr;
+			size_t len_first = base + mmap_size - copy_start;
+			size_t len_secnd = ehdr_size - len_first;
 
-	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) {
+			if (*copy_size < ehdr_size) {
+				free(*copy_mem);
+				*copy_mem = malloc(ehdr_size);
+				if (!*copy_mem) {
+					*copy_size = 0;
 					ret = LIBBPF_PERF_EVENT_ERROR;
 					break;
 				}
-				*buf_len = ehdr->size;
+				*copy_size = 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;
+			memcpy(*copy_mem, copy_start, len_first);
+			memcpy(*copy_mem + len_first, base, len_secnd);
+			ehdr = *copy_mem;
 		}
 
-		ret = fn(ehdr, priv);
+		ret = fn(ehdr, private_data);
+		data_tail += ehdr_size;
 		if (ret != LIBBPF_PERF_EVENT_CONT)
 			break;
-
-		data_tail += ehdr->size;
 	}
 
 	ring_buffer_write_tail(header, data_tail);

tools/lib/bpf/libbpf.h

@@ -297,13 +297,14 @@ enum bpf_perf_event_ret {
 	LIBBPF_PERF_EVENT_CONT	= -2,
 };
 
-typedef enum bpf_perf_event_ret (*bpf_perf_event_print_t)(void *event,
-							  void *priv);
-LIBBPF_API 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);
+struct perf_event_header;
+typedef enum bpf_perf_event_ret
+	(*bpf_perf_event_print_t)(struct perf_event_header *hdr,
+				  void *private_data);
+LIBBPF_API enum bpf_perf_event_ret
+bpf_perf_event_read_simple(void *mmap_mem, size_t mmap_size, size_t page_size,
+			   void **copy_mem, size_t *copy_size,
+			   bpf_perf_event_print_t fn, void *private_data);
 
 struct nlattr;
 typedef int (*libbpf_dump_nlmsg_t)(void *cookie, void *msg, struct nlattr **tb);

tools/testing/selftests/bpf/trace_helpers.c

@@ -125,10 +125,11 @@ struct perf_event_sample {
 	char data[];
 };
 
-static enum bpf_perf_event_ret bpf_perf_event_print(void *event, void *priv)
+static enum bpf_perf_event_ret
+bpf_perf_event_print(struct perf_event_header *hdr, void *private_data)
 {
-	struct perf_event_sample *e = event;
-	perf_event_print_fn fn = priv;
+	struct perf_event_sample *e = (struct perf_event_sample *)hdr;
+	perf_event_print_fn fn = private_data;
 	int ret;
 
 	if (e->header.type == PERF_RECORD_SAMPLE) {