Merge pull request #126 from iovisor/tracepoint-adjust-integer-types

Tracepoint adjust integer types

scripts/tracepoint_variable_sized_types.py

+# This script lists all the types in the kernel's tracepoint format files
+# which appear with more than one size. This script's output should be
+# compared to the code in TracepointFormatParser::adjust_integer_types()
+
+import glob
+
+field_types = {}
+
+for format_file in glob.iglob("/sys/kernel/debug/tracing/events/*/*/format"):
+    for line in open(format_file):
+        if not line.startswith("\tfield:"):
+            continue
+
+        size_section = line.split(";")[2].split(":")
+        if size_section[0] != "\tsize":
+            continue
+        size_val = size_section[1]
+
+        field_section = line.split(";")[0].split(":")
+        if field_section[0] != "\tfield":
+            continue
+        field_val = field_section[1]
+        if "[" in field_val or "*" in field_val:
+            continue
+
+        field_type = " ".join(field_val.split()[:-1])
+
+        if field_type not in field_types:
+            field_types[field_type] = set()
+        field_types[field_type].add(size_val)
+
+for t in sorted(field_types):
+    sizes = field_types[t]
+    if len(sizes) > 1:
+        sizes_str = ",".join(sorted(sizes))
+        print(f"{t}: {sizes_str}")

src/tracepoint_format_parser.cpp

@@ -98,29 +98,13 @@ std::string TracepointFormatParser::adjust_integer_types(const std::string &fiel
 {
   std::string new_type = field_type;
   // Adjust integer fields to correctly sized types
-  if (size == 2)
+  if (size == 8)
   {
-    if (new_type == "char" || new_type == "int8_t")
-      new_type = "s16";
-    if (new_type == "unsigned char" || new_type == "uint8_t")
-      new_type = "u16";
-  } else if (size == 4)
-  {
-    if (new_type == "char" || new_type == "short" ||
-        new_type == "int8_t" || new_type == "int16_t")
-      new_type = "s32";
-    if (new_type == "unsigned char" || new_type == "unsigned short" ||
-        new_type == "uint8_t" || new_type == "uint16_t")
-      new_type = "u32";
-  } else if (size == 8)
-  {
-    if (new_type == "char" || new_type == "short" || new_type == "int" ||
-        new_type == "int8_t" || new_type == "int16_t" ||
-        new_type == "int32_t")
+    if (field_type == "int")
       new_type = "s64";
-    if (new_type == "unsigned char" || new_type == "unsigned short" ||
-        new_type == "unsigned int" || new_type == "uint8_t" ||
-        new_type == "uint16_t" || new_type == "uint32_t")
+    if (field_type == "unsigned int" || field_type == "unsigned" ||
+        field_type == "u32" || field_type == "pid_t" ||
+        field_type == "uid_t" || field_type == "gid_t")
       new_type = "u64";
   }
 

tests/tracepoint_format_parser.cpp

@@ -89,6 +89,66 @@ TEST(tracepoint_format_parser, data_loc)
   EXPECT_EQ(expected, result);
 }
 
+TEST(tracepoint_format_parser, adjust_integer_types)
+{
+  std::string input =
+    "	field:int arr[8];	offset:0;	size:32;	signed:1;\n"
+
+    "	field:int int_a;	offset:0;	size:4;	signed:1;\n"
+    "	field:int int_b;	offset:0;	size:8;	signed:1;\n"
+
+    "	field:u32 u32_a;	offset:0;	size:4;	signed:0;\n"
+    "	field:u32 u32_b;	offset:0;	size:8;	signed:0;\n"
+
+    "	field:unsigned int uint_a;	offset:0;	size:4;	signed:0;\n"
+    "	field:unsigned int uint_b;	offset:0;	size:8;	signed:0;\n"
+
+    "	field:unsigned unsigned_a;	offset:0;	size:4;	signed:0;\n"
+    "	field:unsigned unsigned_b;	offset:0;	size:8;	signed:0;\n"
+
+    "	field:uid_t uid_a;	offset:0;	size:4;	signed:0;\n"
+    "	field:uid_t uid_b;	offset:0;	size:8;	signed:0;\n"
+
+    "	field:gid_t gid_a;	offset:0;	size:4;	signed:0;\n"
+    "	field:gid_t gid_b;	offset:0;	size:8;	signed:0;\n"
+
+    "	field:pid_t pid_a;	offset:0;	size:4;	signed:1;\n"
+    "	field:pid_t pid_b;	offset:0;	size:8;	signed:0;\n";
+
+  std::string expected =
+    "struct _tracepoint_syscalls_sys_enter_read\n"
+    "{\n"
+    "  int arr[8];\n"
+
+    "  int int_a;\n"
+    "  s64 int_b;\n"
+
+    "  u32 u32_a;\n"
+    "  u64 u32_b;\n"
+
+    "  unsigned int uint_a;\n"
+    "  u64 uint_b;\n"
+
+    "  unsigned unsigned_a;\n"
+    "  u64 unsigned_b;\n"
+
+    "  uid_t uid_a;\n"
+    "  u64 uid_b;\n"
+
+    "  gid_t gid_a;\n"
+    "  u64 gid_b;\n"
+
+    "  pid_t pid_a;\n"
+    "  u64 pid_b;\n"
+    "};\n";
+
+  std::istringstream format_file(input);
+
+  std::string result = MockTracepointFormatParser::get_tracepoint_struct_public(format_file, "syscalls", "sys_enter_read");
+
+  EXPECT_EQ(expected, result);
+}
+
 } // namespace tracepoint_format_parser
 } // namespace test
 } // namespace bpftrace