offsetof: Support sub-field

Since bpftrace PR [1] support offsetof() function. As previously discussed [2],
the current offsetof() does not support sub-fields, and this PR implements
offsetof(struct foo, a.b.c).

Example:

    struct Foo {
        struct e {
            int a;
            int b;
            struct c {
                int a, b, c;
            } c;
        } e;
    }

    struct offsetof {
        struct {
            int comm;
            int ustack;
        } kstack;
    }

    BEGIN {
        printf("struct Foo e.b offset %ld\n", offsetof(struct Foo, e.b));
        printf("struct Foo e.c offset %ld\n", offsetof(struct Foo, e.c));
        printf("struct Foo e.c.b offset %ld\n", offsetof(struct Foo, e.c.b));

        printf("struct offsetof kstack.ustack offset %ld\n", offsetof(struct offsetof, kstack.ustack));

        exit();
    }

Link: bpftrace#2579 [1] merged
Link: bpftrace#2216 [2]
Signed-off-by: Rong Tao <rongtao@cestc.cn>

of.bt

@@ -0,0 +1,54 @@
+#!./src/bpftrace
+
+struct Foo {
+	int a;
+	long b;
+	char c;
+	struct {
+		int d;
+	};
+	struct e {
+		int a;
+		int b;
+		struct c {
+			int a, b, c;
+		} c;
+	} e;
+}
+struct offsetof {
+	int bswap;
+	int comm;
+	int kstack;
+	int ustack;
+	int pid;
+	int ctx;
+	int arg;
+}
+
+union Bar {
+	int a;
+};
+
+BEGIN {
+	printf("%ld\n", offsetof(struct Foo, c));
+
+	printf("%ld\n", offsetof(struct Foo, d));
+//	printf("%ld\n", offsetof(struct Foo_1, d));
+
+	printf("e.a %ld\n", offsetof(struct Foo, e.a));
+	printf("e.b %ld\n", offsetof(struct Foo, e.b));
+	printf("e.c %ld\n", offsetof(struct Foo, e.c));
+	printf("e.c.a %ld\n", offsetof(struct Foo, e.c.a));
+	printf("e.c.b %ld\n", offsetof(struct Foo, e.c.b));
+
+	printf("%ld union\n", offsetof(union Bar, a));
+
+	printf("%ld\n", offsetof(struct offsetof, bswap));
+
+	printf("%ld\n", offsetof(struct task_struct, comm));
+	printf("%ld\n", offsetof(*curtask, comm));
+
+	printf("%ld\n", offsetof(*(struct task_struct *)0, comm));
+
+	exit();
+}

src/ast/passes/codegen_llvm.cpp

@@ -1522,8 +1522,21 @@ void CodegenLLVM::visit(Sizeof &szof)
 
 void CodegenLLVM::visit(Offsetof &ofof)
 {
-  auto &field = ofof.record.GetField(ofof.field);
-  expr_ = b_.getInt64(field.offset);
+  ssize_t offset = 0;
+  std::string token;
+  std::stringstream ss(ofof.field);
+  std::vector<std::string> tokens;
+
+  while (std::getline(ss, token, '.')) {
+    tokens.push_back(token);
+  }
+
+  const SizedType *record = &ofof.record;
+  for (const auto& token : tokens) {
+    offset += record->GetField(token).offset;
+    record = &record->GetField(token).type;
+  }
+  expr_ = b_.getInt64(offset);
 }
 
 void CodegenLLVM::visit(Map &map)

src/ast/passes/semantic_analyser.cpp

@@ -1615,17 +1615,33 @@ void SemanticAnalyser::visit(Offsetof &ofof)
     Visit(ofof.expr);
     ofof.record = ofof.expr->type;
   }
+
   resolve_struct_type(ofof.record, ofof.loc);
 
-  if (!ofof.record.IsRecordTy()) {
-    LOG(ERROR, ofof.loc, err_)
-        << "offsetof() 1st argument is not of a record type.";
-  } else if (!bpftrace_.structs.Has(ofof.record.GetName())) {
-    LOG(ERROR, ofof.loc, err_) << "'" << ofof.record << "' does not exist.";
-  } else if (!ofof.record.HasField(ofof.field)) {
-    LOG(ERROR, ofof.loc, err_) << "'" << ofof.record << "' "
-                               << "has no field named "
-                               << "'" << ofof.field << "'";
+  std::string token;
+  std::stringstream ss(ofof.field);
+  std::vector<std::string> tokens;
+  while (std::getline(ss, token, '.')) {
+    tokens.push_back(token);
+  }
+
+  // Check if all sub-fields are present.
+  SizedType record = ofof.record;
+  for (const auto& token : tokens) {
+    if (!record.IsRecordTy()) {
+      LOG(ERROR, ofof.loc, err_) << "'" << record << "' "
+                                 << "is not of a record type.";
+    } else if (!bpftrace_.structs.Has(record.GetName())) {
+      LOG(ERROR, ofof.loc, err_) << "'" << record.GetName()
+                                 << "' does not exist.";
+    } else if (!record.HasField(token)) {
+      LOG(ERROR, ofof.loc, err_) << "'" << record.GetName() << "' "
+                                 << "has no field named "
+                                 << "'" << token << "'";
+    } else {
+      // Get next sub-field
+      record = record.GetField(token).type;
+    }
   }
 }
 

src/parser.yy

@@ -131,7 +131,7 @@ void yyerror(bpftrace::Driver &driver, const char *s);
 
 
 %type <ast::Operator> unary_op compound_op
-%type <std::string> attach_point_def c_definitions ident keyword external_name
+%type <std::string> attach_point_def c_definitions ident keyword external_name struct_field
 
 %type <ast::AttachPoint *> attach_point
 %type <ast::AttachPointList> attach_points
@@ -626,10 +626,15 @@ sizeof_expr:
         |       SIZEOF "(" expr ")"                         { $$ = driver.ctx.make_node<ast::Sizeof>($3, @$); }
                 ;
 
+struct_field:
+                external_name                       { $$ = $1; }
+        |       struct_field DOT external_name      { $$ = $1 + "." + $3; }
+        ;
+
 offsetof_expr:
-                OFFSETOF "(" struct_type "," external_name ")"      { $$ = driver.ctx.make_node<ast::Offsetof>($3, $5, @$); }
-/* For example: offsetof(*curtask, comm) */
-        |       OFFSETOF "(" expr "," external_name ")"             { $$ = driver.ctx.make_node<ast::Offsetof>($3, $5, @$); }
+                OFFSETOF "(" struct_type "," struct_field ")"      { $$ = driver.ctx.make_node<ast::Offsetof>($3, $5, @$); }
+                /* For example: offsetof(*curtask, comm) */
+        |       OFFSETOF "(" expr "," struct_field ")"             { $$ = driver.ctx.make_node<ast::Offsetof>($3, $5, @$); }
                 ;
 
 int: