Merge branch 'bpf-int128-btf'

Yonghong Song says:

====================
Previous maximum supported integer bit width is 64. But
the __int128 type has been supported by most (if not all)
64bit architectures including bpf for both gcc and clang.

The kernel itself uses __int128 for x64 and arm64. Some bcc
tools are using __int128 in bpf programs to describe ipv6
addresses. Without 128bit int support, the vmlinux BTF won't
work and those bpf programs using __int128 cannot utilize BTF.

This patch set therefore implements BTF __int128 support.
====================

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

2 files changed, 333 insertions, 49 deletions

diff --git a/tools/bpf/bpftool/btf_dumper.c b/tools/bpf/bpftool/btf_dumper.c
index 6ba5f567a9d8..e63bce0755eb 100644
--- a/tools/bpf/bpftool/btf_dumper.c
+++ b/tools/bpf/bpftool/btf_dumper.c
@@ -73,35 +73,104 @@ static int btf_dumper_array(const struct btf_dumper *d, __u32 type_id,
 	return ret;
 }
 
+static void btf_int128_print(json_writer_t *jw, const void *data,
+			     bool is_plain_text)
+{
+	/* data points to a __int128 number.
+	 * Suppose
+	 *     int128_num = *(__int128 *)data;
+	 * The below formulas shows what upper_num and lower_num represents:
+	 *     upper_num = int128_num >> 64;
+	 *     lower_num = int128_num & 0xffffffffFFFFFFFFULL;
+	 */
+	__u64 upper_num, lower_num;
+
+#ifdef __BIG_ENDIAN_BITFIELD
+	upper_num = *(__u64 *)data;
+	lower_num = *(__u64 *)(data + 8);
+#else
+	upper_num = *(__u64 *)(data + 8);
+	lower_num = *(__u64 *)data;
+#endif
+
+	if (is_plain_text) {
+		if (upper_num == 0)
+			jsonw_printf(jw, "0x%llx", lower_num);
+		else
+			jsonw_printf(jw, "0x%llx%016llx", upper_num, lower_num);
+	} else {
+		if (upper_num == 0)
+			jsonw_printf(jw, "\"0x%llx\"", lower_num);
+		else
+			jsonw_printf(jw, "\"0x%llx%016llx\"", upper_num, lower_num);
+	}
+}
+
+static void btf_int128_shift(__u64 *print_num, u16 left_shift_bits,
+			     u16 right_shift_bits)
+{
+	__u64 upper_num, lower_num;
+
+#ifdef __BIG_ENDIAN_BITFIELD
+	upper_num = print_num[0];
+	lower_num = print_num[1];
+#else
+	upper_num = print_num[1];
+	lower_num = print_num[0];
+#endif
+
+	/* shake out un-needed bits by shift/or operations */
+	if (left_shift_bits >= 64) {
+		upper_num = lower_num << (left_shift_bits - 64);
+		lower_num = 0;
+	} else {
+		upper_num = (upper_num << left_shift_bits) |
+			    (lower_num >> (64 - left_shift_bits));
+		lower_num = lower_num << left_shift_bits;
+	}
+
+	if (right_shift_bits >= 64) {
+		lower_num = upper_num >> (right_shift_bits - 64);
+		upper_num = 0;
+	} else {
+		lower_num = (lower_num >> right_shift_bits) |
+			    (upper_num << (64 - right_shift_bits));
+		upper_num = upper_num >> right_shift_bits;
+	}
+
+#ifdef __BIG_ENDIAN_BITFIELD
+	print_num[0] = upper_num;
+	print_num[1] = lower_num;
+#else
+	print_num[0] = lower_num;
+	print_num[1] = upper_num;
+#endif
+}
+
 static void btf_dumper_bitfield(__u32 nr_bits, __u8 bit_offset,
 				const void *data, json_writer_t *jw,
 				bool is_plain_text)
 {
 	int left_shift_bits, right_shift_bits;
+	__u64 print_num[2] = {};
 	int bytes_to_copy;
 	int bits_to_copy;
-	__u64 print_num;
 
 	bits_to_copy = bit_offset + nr_bits;
 	bytes_to_copy = BITS_ROUNDUP_BYTES(bits_to_copy);
 
-	print_num = 0;
-	memcpy(&print_num, data, bytes_to_copy);
+	memcpy(print_num, data, bytes_to_copy);
 #if defined(__BIG_ENDIAN_BITFIELD)
 	left_shift_bits = bit_offset;
 #elif defined(__LITTLE_ENDIAN_BITFIELD)
-	left_shift_bits = 64 - bits_to_copy;
+	left_shift_bits = 128 - bits_to_copy;
 #else
 #error neither big nor little endian
 #endif
-	right_shift_bits = 64 - nr_bits;
+	right_shift_bits = 128 - nr_bits;
 
-	print_num <<= left_shift_bits;
-	print_num >>= right_shift_bits;
-	if (is_plain_text)
-		jsonw_printf(jw, "0x%llx", print_num);
-	else
-		jsonw_printf(jw, "%llu", print_num);
+	btf_int128_shift(print_num, left_shift_bits, right_shift_bits);
+	btf_int128_print(jw, print_num, is_plain_text);
 }
 
 
@@ -113,7 +182,7 @@ static void btf_dumper_int_bits(__u32 int_type, __u8 bit_offset,
 	int total_bits_offset;
 
 	/* bits_offset is at most 7.
-	 * BTF_INT_OFFSET() cannot exceed 64 bits.
+	 * BTF_INT_OFFSET() cannot exceed 128 bits.
 	 */
 	total_bits_offset = bit_offset + BTF_INT_OFFSET(int_type);
 	data += BITS_ROUNDDOWN_BYTES(total_bits_offset);
@@ -139,6 +208,11 @@ static int btf_dumper_int(const struct btf_type *t, __u8 bit_offset,
 		return 0;
 	}
 
+	if (nr_bits == 128) {
+		btf_int128_print(jw, data, is_plain_text);
+		return 0;
+	}
+
 	switch (BTF_INT_ENCODING(*int_type)) {
 	case 0:
 		if (BTF_INT_BITS(*int_type) == 64)
diff --git a/tools/testing/selftests/bpf/test_btf.c b/tools/testing/selftests/bpf/test_btf.c
index a0bd04befe87..bfd92182533b 100644
--- a/tools/testing/selftests/bpf/test_btf.c
+++ b/tools/testing/selftests/bpf/test_btf.c
@@ -18,6 +18,7 @@
 #include <unistd.h>
 #include <fcntl.h>
 #include <errno.h>
+#include <assert.h>
 #include <bpf/libbpf.h>
 #include <bpf/btf.h>
 
@@ -134,6 +135,12 @@ static struct btf_header hdr_tmpl = {
 	.hdr_len = sizeof(struct btf_header),
 };
 
+/* several different mapv kinds(types) supported by pprint */
+enum pprint_mapv_kind_t {
+	PPRINT_MAPV_KIND_BASIC = 0,
+	PPRINT_MAPV_KIND_INT128,
+};
+
 struct btf_raw_test {
 	const char *descr;
 	const char *str_sec;
@@ -156,6 +163,7 @@ struct btf_raw_test {
 	int type_off_delta;
 	int str_off_delta;
 	int str_len_delta;
+	enum pprint_mapv_kind_t mapv_kind;
 };
 
 #define BTF_STR_SEC(str) \
@@ -2707,6 +2715,99 @@ static struct btf_raw_test raw_tests[] = {
 	.err_str = "Invalid member offset",
 },
 
+{
+	.descr = "128-bit int",
+	.raw_types = {
+		BTF_TYPE_INT_ENC(0, BTF_INT_SIGNED, 0, 32, 4),			/* [1] */
+		BTF_TYPE_INT_ENC(0, BTF_INT_SIGNED, 0, 128, 16),		/* [2] */
+		BTF_END_RAW,
+	},
+	BTF_STR_SEC("\0A"),
+	.map_type = BPF_MAP_TYPE_ARRAY,
+	.map_name = "int_type_check_btf",
+	.key_size = sizeof(int),
+	.value_size = sizeof(int),
+	.key_type_id = 1,
+	.value_type_id = 1,
+	.max_entries = 4,
+},
+
+{
+	.descr = "struct, 128-bit int member",
+	.raw_types = {
+		BTF_TYPE_INT_ENC(0, BTF_INT_SIGNED, 0, 32, 4),			/* [1] */
+		BTF_TYPE_INT_ENC(0, BTF_INT_SIGNED, 0, 128, 16),		/* [2] */
+		BTF_TYPE_ENC(0, BTF_INFO_ENC(BTF_KIND_STRUCT, 0, 1), 16),	/* [3] */
+		BTF_MEMBER_ENC(NAME_TBD, 2, 0),
+		BTF_END_RAW,
+	},
+	BTF_STR_SEC("\0A"),
+	.map_type = BPF_MAP_TYPE_ARRAY,
+	.map_name = "struct_type_check_btf",
+	.key_size = sizeof(int),
+	.value_size = sizeof(int),
+	.key_type_id = 1,
+	.value_type_id = 1,
+	.max_entries = 4,
+},
+
+{
+	.descr = "struct, 120-bit int member bitfield",
+	.raw_types = {
+		BTF_TYPE_INT_ENC(0, BTF_INT_SIGNED, 0, 32, 4),			/* [1] */
+		BTF_TYPE_INT_ENC(0, BTF_INT_SIGNED, 0, 120, 16),		/* [2] */
+		BTF_TYPE_ENC(0, BTF_INFO_ENC(BTF_KIND_STRUCT, 0, 1), 16),	/* [3] */
+		BTF_MEMBER_ENC(NAME_TBD, 2, 0),
+		BTF_END_RAW,
+	},
+	BTF_STR_SEC("\0A"),
+	.map_type = BPF_MAP_TYPE_ARRAY,
+	.map_name = "struct_type_check_btf",
+	.key_size = sizeof(int),
+	.value_size = sizeof(int),
+	.key_type_id = 1,
+	.value_type_id = 1,
+	.max_entries = 4,
+},
+
+{
+	.descr = "struct, kind_flag, 128-bit int member",
+	.raw_types = {
+		BTF_TYPE_INT_ENC(0, BTF_INT_SIGNED, 0, 32, 4),			/* [1] */
+		BTF_TYPE_INT_ENC(0, BTF_INT_SIGNED, 0, 128, 16),		/* [2] */
+		BTF_TYPE_ENC(0, BTF_INFO_ENC(BTF_KIND_STRUCT, 1, 1), 16),	/* [3] */
+		BTF_MEMBER_ENC(NAME_TBD, 2, BTF_MEMBER_OFFSET(0, 0)),
+		BTF_END_RAW,
+	},
+	BTF_STR_SEC("\0A"),
+	.map_type = BPF_MAP_TYPE_ARRAY,
+	.map_name = "struct_type_check_btf",
+	.key_size = sizeof(int),
+	.value_size = sizeof(int),
+	.key_type_id = 1,
+	.value_type_id = 1,
+	.max_entries = 4,
+},
+
+{
+	.descr = "struct, kind_flag, 120-bit int member bitfield",
+	.raw_types = {
+		BTF_TYPE_INT_ENC(0, BTF_INT_SIGNED, 0, 32, 4),			/* [1] */
+		BTF_TYPE_INT_ENC(0, BTF_INT_SIGNED, 0, 128, 16),		/* [2] */
+		BTF_TYPE_ENC(0, BTF_INFO_ENC(BTF_KIND_STRUCT, 1, 1), 16),	/* [3] */
+		BTF_MEMBER_ENC(NAME_TBD, 2, BTF_MEMBER_OFFSET(120, 0)),
+		BTF_END_RAW,
+	},
+	BTF_STR_SEC("\0A"),
+	.map_type = BPF_MAP_TYPE_ARRAY,
+	.map_name = "struct_type_check_btf",
+	.key_size = sizeof(int),
+	.value_size = sizeof(int),
+	.key_type_id = 1,
+	.value_type_id = 1,
+	.max_entries = 4,
+},
+
 }; /* struct btf_raw_test raw_tests[] */
 
 static const char *get_next_str(const char *start, const char *end)
@@ -3530,6 +3631,16 @@ struct pprint_mapv {
 	uint32_t bits2c:2;
 };
 
+#ifdef __SIZEOF_INT128__
+struct pprint_mapv_int128 {
+	__int128 si128a;
+	__int128 si128b;
+	unsigned __int128 bits3:3;
+	unsigned __int128 bits80:80;
+	unsigned __int128 ui128;
+};
+#endif
+
 static struct btf_raw_test pprint_test_template[] = {
 {
 	.raw_types = {
@@ -3721,6 +3832,35 @@ static struct btf_raw_test pprint_test_template[] = {
 	.max_entries = 128 * 1024,
 },
 
+#ifdef __SIZEOF_INT128__
+{
+	/* test int128 */
+	.raw_types = {
+		/* unsigned int */				/* [1] */
+		BTF_TYPE_INT_ENC(NAME_TBD, 0, 0, 32, 4),
+		/* __int128 */					/* [2] */
+		BTF_TYPE_INT_ENC(NAME_TBD, BTF_INT_SIGNED, 0, 128, 16),
+		/* unsigned __int128 */				/* [3] */
+		BTF_TYPE_INT_ENC(NAME_TBD, 0, 0, 128, 16),
+		/* struct pprint_mapv_int128 */			/* [4] */
+		BTF_TYPE_ENC(NAME_TBD, BTF_INFO_ENC(BTF_KIND_STRUCT, 1, 5), 64),
+		BTF_MEMBER_ENC(NAME_TBD, 2, BTF_MEMBER_OFFSET(0, 0)),		/* si128a */
+		BTF_MEMBER_ENC(NAME_TBD, 2, BTF_MEMBER_OFFSET(0, 128)),		/* si128b */
+		BTF_MEMBER_ENC(NAME_TBD, 3, BTF_MEMBER_OFFSET(3, 256)),		/* bits3 */
+		BTF_MEMBER_ENC(NAME_TBD, 3, BTF_MEMBER_OFFSET(80, 259)),	/* bits80 */
+		BTF_MEMBER_ENC(NAME_TBD, 3, BTF_MEMBER_OFFSET(0, 384)),		/* ui128 */
+		BTF_END_RAW,
+	},
+	BTF_STR_SEC("\0unsigned int\0__int128\0unsigned __int128\0pprint_mapv_int128\0si128a\0si128b\0bits3\0bits80\0ui128"),
+	.key_size = sizeof(unsigned int),
+	.value_size = sizeof(struct pprint_mapv_int128),
+	.key_type_id = 1,
+	.value_type_id = 4,
+	.max_entries = 128 * 1024,
+	.mapv_kind = PPRINT_MAPV_KIND_INT128,
+},
+#endif
+
 };
 
 static struct btf_pprint_test_meta {
@@ -3787,24 +3927,108 @@ static struct btf_pprint_test_meta {
 
 };
 
+static size_t get_pprint_mapv_size(enum pprint_mapv_kind_t mapv_kind)
+{
+	if (mapv_kind == PPRINT_MAPV_KIND_BASIC)
+		return sizeof(struct pprint_mapv);
+
+#ifdef __SIZEOF_INT128__
+	if (mapv_kind == PPRINT_MAPV_KIND_INT128)
+		return sizeof(struct pprint_mapv_int128);
+#endif
+
+	assert(0);
+}
 
-static void set_pprint_mapv(struct pprint_mapv *v, uint32_t i,
+static void set_pprint_mapv(enum pprint_mapv_kind_t mapv_kind,
+			    void *mapv, uint32_t i,
 			    int num_cpus, int rounded_value_size)
 {
 	int cpu;
 
-	for (cpu = 0; cpu < num_cpus; cpu++) {
-		v->ui32 = i + cpu;
-		v->si32 = -i;
-		v->unused_bits2a = 3;
-		v->bits28 = i;
-		v->unused_bits2b = 3;
-		v->ui64 = i;
-		v->aenum = i & 0x03;
-		v->ui32b = 4;
-		v->bits2c = 1;
-		v = (void *)v + rounded_value_size;
+	if (mapv_kind == PPRINT_MAPV_KIND_BASIC) {
+		struct pprint_mapv *v = mapv;
+
+		for (cpu = 0; cpu < num_cpus; cpu++) {
+			v->ui32 = i + cpu;
+			v->si32 = -i;
+			v->unused_bits2a = 3;
+			v->bits28 = i;
+			v->unused_bits2b = 3;
+			v->ui64 = i;
+			v->aenum = i & 0x03;
+			v->ui32b = 4;
+			v->bits2c = 1;
+			v = (void *)v + rounded_value_size;
+		}
 	}
+
+#ifdef __SIZEOF_INT128__
+	if (mapv_kind == PPRINT_MAPV_KIND_INT128) {
+		struct pprint_mapv_int128 *v = mapv;
+
+		for (cpu = 0; cpu < num_cpus; cpu++) {
+			v->si128a = i;
+			v->si128b = -i;
+			v->bits3 = i & 0x07;
+			v->bits80 = (((unsigned __int128)1) << 64) + i;
+			v->ui128 = (((unsigned __int128)2) << 64) + i;
+			v = (void *)v + rounded_value_size;
+		}
+	}
+#endif
+}
+
+ssize_t get_pprint_expected_line(enum pprint_mapv_kind_t mapv_kind,
+				 char *expected_line, ssize_t line_size,
+				 bool percpu_map, unsigned int next_key,
+				 int cpu, void *mapv)
+{
+	ssize_t nexpected_line = -1;
+
+	if (mapv_kind == PPRINT_MAPV_KIND_BASIC) {
+		struct pprint_mapv *v = mapv;
+
+		nexpected_line = snprintf(expected_line, line_size,
+					  "%s%u: {%u,0,%d,0x%x,0x%x,0x%x,"
+					  "{%lu|[%u,%u,%u,%u,%u,%u,%u,%u]},%s,"
+					  "%u,0x%x}\n",
+					  percpu_map ? "\tcpu" : "",
+					  percpu_map ? cpu : next_key,
+					  v->ui32, v->si32,
+					  v->unused_bits2a,
+					  v->bits28,
+					  v->unused_bits2b,
+					  v->ui64,
+					  v->ui8a[0], v->ui8a[1],
+					  v->ui8a[2], v->ui8a[3],
+					  v->ui8a[4], v->ui8a[5],
+					  v->ui8a[6], v->ui8a[7],
+					  pprint_enum_str[v->aenum],
+					  v->ui32b,
+					  v->bits2c);
+	}
+
+#ifdef __SIZEOF_INT128__
+	if (mapv_kind == PPRINT_MAPV_KIND_INT128) {
+		struct pprint_mapv_int128 *v = mapv;
+
+		nexpected_line = snprintf(expected_line, line_size,
+					  "%s%u: {0x%lx,0x%lx,0x%lx,"
+					  "0x%lx%016lx,0x%lx%016lx}\n",
+					  percpu_map ? "\tcpu" : "",
+					  percpu_map ? cpu : next_key,
+					  (uint64_t)v->si128a,
+					  (uint64_t)v->si128b,
+					  (uint64_t)v->bits3,
+					  (uint64_t)(v->bits80 >> 64),
+					  (uint64_t)v->bits80,
+					  (uint64_t)(v->ui128 >> 64),
+					  (uint64_t)v->ui128);
+	}
+#endif
+
+	return nexpected_line;
 }
 
 static int check_line(const char *expected_line, int nexpected_line,
@@ -3828,10 +4052,10 @@ static int check_line(const char *expected_line, int nexpected_line,
 static int do_test_pprint(int test_num)
 {
 	const struct btf_raw_test *test = &pprint_test_template[test_num];
+	enum pprint_mapv_kind_t mapv_kind = test->mapv_kind;
 	struct bpf_create_map_attr create_attr = {};
 	bool ordered_map, lossless_map, percpu_map;
 	int err, ret, num_cpus, rounded_value_size;
-	struct pprint_mapv *mapv = NULL;
 	unsigned int key, nr_read_elems;
 	int map_fd = -1, btf_fd = -1;
 	unsigned int raw_btf_size;
@@ -3840,6 +4064,7 @@ static int do_test_pprint(int test_num)
 	char pin_path[255];
 	size_t line_len = 0;
 	char *line = NULL;
+	void *mapv = NULL;
 	uint8_t *raw_btf;
 	ssize_t nread;
 
@@ -3892,7 +4117,7 @@ static int do_test_pprint(int test_num)
 
 	percpu_map = test->percpu_map;
 	num_cpus = percpu_map ? bpf_num_possible_cpus() : 1;
-	rounded_value_size = round_up(sizeof(struct pprint_mapv), 8);
+	rounded_value_size = round_up(get_pprint_mapv_size(mapv_kind), 8);
 	mapv = calloc(num_cpus, rounded_value_size);
 	if (CHECK(!mapv, "mapv allocation failure")) {
 		err = -1;
@@ -3900,7 +4125,7 @@ static int do_test_pprint(int test_num)
 	}
 
 	for (key = 0; key < test->max_entries; key++) {
-		set_pprint_mapv(mapv, key, num_cpus, rounded_value_size);
+		set_pprint_mapv(mapv_kind, mapv, key, num_cpus, rounded_value_size);
 		bpf_map_update_elem(map_fd, &key, mapv, 0);
 	}
 
@@ -3924,13 +4149,13 @@ static int do_test_pprint(int test_num)
 	ordered_map = test->ordered_map;
 	lossless_map = test->lossless_map;
 	do {
-		struct pprint_mapv *cmapv;
 		ssize_t nexpected_line;
 		unsigned int next_key;
+		void *cmapv;
 		int cpu;
 
 		next_key = ordered_map ? nr_read_elems : atoi(line);
-		set_pprint_mapv(mapv, next_key, num_cpus, rounded_value_size);
+		set_pprint_mapv(mapv_kind, mapv, next_key, num_cpus, rounded_value_size);
 		cmapv = mapv;
 
 		for (cpu = 0; cpu < num_cpus; cpu++) {
@@ -3963,31 +4188,16 @@ static int do_test_pprint(int test_num)
 					break;
 			}
 
-			nexpected_line = snprintf(expected_line, sizeof(expected_line),
-						  "%s%u: {%u,0,%d,0x%x,0x%x,0x%x,"
-						  "{%lu|[%u,%u,%u,%u,%u,%u,%u,%u]},%s,"
-						  "%u,0x%x}\n",
-						  percpu_map ? "\tcpu" : "",
-						  percpu_map ? cpu : next_key,
-						  cmapv->ui32, cmapv->si32,
-						  cmapv->unused_bits2a,
-						  cmapv->bits28,
-						  cmapv->unused_bits2b,
-						  cmapv->ui64,
-						  cmapv->ui8a[0], cmapv->ui8a[1],
-						  cmapv->ui8a[2], cmapv->ui8a[3],
-						  cmapv->ui8a[4], cmapv->ui8a[5],
-						  cmapv->ui8a[6], cmapv->ui8a[7],
-						  pprint_enum_str[cmapv->aenum],
-						  cmapv->ui32b,
-						  cmapv->bits2c);
-
+			nexpected_line = get_pprint_expected_line(mapv_kind, expected_line,
+								  sizeof(expected_line),
+								  percpu_map, next_key,
+								  cpu, cmapv);
 			err = check_line(expected_line, nexpected_line,
 					 sizeof(expected_line), line);
 			if (err == -1)
 				goto done;
 
-			cmapv = (void *)cmapv + rounded_value_size;
+			cmapv = cmapv + rounded_value_size;
 		}
 
 		if (percpu_map) {