bpf: btf: support 128 bit integer type

Currently, btf only supports up to 64-bit integer.
On the other hand, 128bit support for gcc and clang
has existed for a long time. For example, both gcc 4.8
and llvm 3.7 supports types "__int128" and
"unsigned __int128" for virtually all 64bit architectures
including bpf.

The requirement for __int128 support comes from two areas:
  . bpf program may use __int128. For example, some bcc tools
    (https://github.com/iovisor/bcc/tree/master/tools),
    mostly tcp v6 related, tcpstates.py, tcpaccept.py, etc.,
    are using __int128 to represent the ipv6 addresses.
  . linux itself is using __int128 types. Hence supporting
    __int128 type in BTF is required for vmlinux BTF,
    which will be used by "compile once and run everywhere"
    and other projects.

For 128bit integer, instead of base-10, hex numbers are pretty
printed out as large decimal number is hard to decipher, e.g.,
for ipv6 addresses.

Acked-by: Martin KaFai Lau <kafai@fb.com>
Signed-off-by: Yonghong Song <yhs@fb.com>
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>

diff --git a/kernel/bpf/btf.c b/kernel/bpf/btf.c
index a2f53642592b4d7fe3e8bffd12c25f91376c7caf..022ef9ca1296b97636a5b39976d716cc71809d9e 100644 (file)
--- a/kernel/bpf/btf.c
+++ b/kernel/bpf/btf.c
@@ -157,7 +157,7 @@
  *
  */
 
-#define BITS_PER_U64 (sizeof(u64) * BITS_PER_BYTE)
+#define BITS_PER_U128 (sizeof(u64) * BITS_PER_BYTE * 2)
 #define BITS_PER_BYTE_MASK (BITS_PER_BYTE - 1)
 #define BITS_PER_BYTE_MASKED(bits) ((bits) & BITS_PER_BYTE_MASK)
 #define BITS_ROUNDDOWN_BYTES(bits) ((bits) >> 3)
@@ -525,7 +525,7 @@ const struct btf_type *btf_type_by_id(const struct btf *btf, u32 type_id)
 
 /*
  * Regular int is not a bit field and it must be either
- * u8/u16/u32/u64.
+ * u8/u16/u32/u64 or __int128.
  */
 static bool btf_type_int_is_regular(const struct btf_type *t)
 {
@@ -538,7 +538,8 @@ static bool btf_type_int_is_regular(const struct btf_type *t)
        if (BITS_PER_BYTE_MASKED(nr_bits) ||
            BTF_INT_OFFSET(int_data) ||
            (nr_bytes != sizeof(u8) && nr_bytes != sizeof(u16) &&
-            nr_bytes != sizeof(u32) && nr_bytes != sizeof(u64))) {
+            nr_bytes != sizeof(u32) && nr_bytes != sizeof(u64) &&
+            nr_bytes != (2 * sizeof(u64)))) {
                return false;
        }
 
@@ -1063,9 +1064,9 @@ static int btf_int_check_member(struct btf_verifier_env *env,
        nr_copy_bits = BTF_INT_BITS(int_data) +
                BITS_PER_BYTE_MASKED(struct_bits_off);
 
-       if (nr_copy_bits > BITS_PER_U64) {
+       if (nr_copy_bits > BITS_PER_U128) {
                btf_verifier_log_member(env, struct_type, member,
-                                       "nr_copy_bits exceeds 64");
+                                       "nr_copy_bits exceeds 128");
                return -EINVAL;
        }
 
@@ -1119,9 +1120,9 @@ static int btf_int_check_kflag_member(struct btf_verifier_env *env,
 
        bytes_offset = BITS_ROUNDDOWN_BYTES(struct_bits_off);
        nr_copy_bits = nr_bits + BITS_PER_BYTE_MASKED(struct_bits_off);
-       if (nr_copy_bits > BITS_PER_U64) {
+       if (nr_copy_bits > BITS_PER_U128) {
                btf_verifier_log_member(env, struct_type, member,
-                                       "nr_copy_bits exceeds 64");
+                                       "nr_copy_bits exceeds 128");
                return -EINVAL;
        }
 
@@ -1168,9 +1169,9 @@ static s32 btf_int_check_meta(struct btf_verifier_env *env,
 
        nr_bits = BTF_INT_BITS(int_data) + BTF_INT_OFFSET(int_data);
 
-       if (nr_bits > BITS_PER_U64) {
+       if (nr_bits > BITS_PER_U128) {
                btf_verifier_log_type(env, t, "nr_bits exceeds %zu",
-                                     BITS_PER_U64);
+                                     BITS_PER_U128);
                return -EINVAL;
        }
 
@@ -1211,31 +1212,93 @@ static void btf_int_log(struct btf_verifier_env *env,
                         btf_int_encoding_str(BTF_INT_ENCODING(int_data)));
 }
 
+static void btf_int128_print(struct seq_file *m, void *data)
+{
+       /* 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 (upper_num == 0)
+               seq_printf(m, "0x%llx", lower_num);
+       else
+               seq_printf(m, "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_bitfield_seq_show(void *data, u8 bits_offset,
                                  u8 nr_bits, struct seq_file *m)
 {
        u16 left_shift_bits, right_shift_bits;
        u8 nr_copy_bytes;
        u8 nr_copy_bits;
-       u64 print_num;
+       u64 print_num[2] = {};
 
        nr_copy_bits = nr_bits + bits_offset;
        nr_copy_bytes = BITS_ROUNDUP_BYTES(nr_copy_bits);
 
-       print_num = 0;
-       memcpy(&print_num, data, nr_copy_bytes);
+       memcpy(print_num, data, nr_copy_bytes);
 
 #ifdef __BIG_ENDIAN_BITFIELD
        left_shift_bits = bits_offset;
 #else
-       left_shift_bits = BITS_PER_U64 - nr_copy_bits;
+       left_shift_bits = BITS_PER_U128 - nr_copy_bits;
 #endif
-       right_shift_bits = BITS_PER_U64 - nr_bits;
-
-       print_num <<= left_shift_bits;
-       print_num >>= right_shift_bits;
+       right_shift_bits = BITS_PER_U128 - nr_bits;
 
-       seq_printf(m, "0x%llx", print_num);
+       btf_int128_shift(print_num, left_shift_bits, right_shift_bits);
+       btf_int128_print(m, print_num);
 }
 
 
@@ -1250,7 +1313,7 @@ static void btf_int_bits_seq_show(const struct btf *btf,
 
        /*
         * bits_offset is at most 7.
-        * BTF_INT_OFFSET() cannot exceed 64 bits.
+        * BTF_INT_OFFSET() cannot exceed 128 bits.
         */
        total_bits_offset = bits_offset + BTF_INT_OFFSET(int_data);
        data += BITS_ROUNDDOWN_BYTES(total_bits_offset);
@@ -1274,6 +1337,9 @@ static void btf_int_seq_show(const struct btf *btf, const struct btf_type *t,
        }
 
        switch (nr_bits) {
+       case 128:
+               btf_int128_print(m, data);
+               break;
        case 64:
                if (sign)
                        seq_printf(m, "%lld", *(s64 *)data);