preproc: generate .size in script_data sections (#9233)

Makefile

@@ -529,7 +529,7 @@ ifneq ($(NODEP),1)
 endif
 
 $(DATA_ASM_BUILDDIR)/%.o: $(DATA_ASM_SUBDIR)/%.s
-	$(PREPROC) $< charmap.txt | $(CPP) $(CPPFLAGS) $(INCLUDE_SCANINC_ARGS) - | $(PREPROC) -ie $< charmap.txt | $(AS) $(ASFLAGS) -o $@
+	$(PREPROC) -s $< charmap.txt | $(CPP) $(CPPFLAGS) $(INCLUDE_SCANINC_ARGS) - | $(PREPROC) -ie $< charmap.txt | $(AS) $(ASFLAGS) -o $@
 
 $(DATA_ASM_BUILDDIR)/%.d: $(DATA_ASM_SUBDIR)/%.s
 	$(SCANINC) -M $@ $(INCLUDE_SCANINC_ARGS) -I "" $<

tools/preproc/asm_file.cpp

@@ -160,9 +160,9 @@ Directive AsmFile::GetDirective()
         return Directive::Unknown;
 }
 
-// Checks if we're at label that ends with '::'.
+// Checks if we're at label and if so, returns its symbol and scope.
-// Returns the name if so and an empty string if not.
+// Returns 'label::none' if not.
-std::string AsmFile::GetGlobalLabel()
+Label AsmFile::GetLabel()
 {
     long start = m_pos;
     long pos = m_pos;
@@ -175,14 +175,57 @@ std::string AsmFile::GetGlobalLabel()
             pos++;
     }
 
-    if (m_buffer[pos] == ':' && m_buffer[pos + 1] == ':')
+    if (m_buffer[pos] == ':')
     {
-        m_pos = pos + 2;
+        std::string symbol(&m_buffer[start], pos - start);
-        ExpectEmptyRestOfLine();
+        if (m_buffer[pos + 1] == ':')
-        return std::string(&m_buffer[start], pos - start);
+        {
+            m_pos = pos + 2;
+            ExpectEmptyRestOfLine();
+            return Label(symbol, Label::global);
+        }
+        else
+        {
+            m_pos = pos + 1;
+            return Label(symbol, Label::local);
+        }
     }
 
-    return std::string();
+    return Label("", Label::none);
+}
+
+std::string AsmFile::PeekSection()
+{
+    long oldPos = m_pos;
+    std::string section;
+
+    SkipWhitespace();
+
+    // TODO: Support 'pushsection', 'popsection', '.previous'.
+    if (CheckForDirective(".bss"))
+    {
+        section = ".bss";
+    }
+    else if (CheckForDirective(".data"))
+    {
+        section = ".data";
+    }
+    else if (CheckForDirective(".rodata"))
+    {
+        section = ".rodata";
+    }
+    else if (CheckForDirective(".text"))
+    {
+        section = ".text";
+    }
+    else if (CheckForDirective(".section"))
+    {
+        SkipWhitespace();
+        section = ReadIdentifier();
+    }
+
+    m_pos = oldPos;
+    return section;
 }
 
 // Skips tabs and spaces.

tools/preproc/asm_file.h

@@ -35,6 +35,17 @@ enum class Directive
     Unknown
 };
 
+struct Label
+{
+    std::string symbol;
+    enum Type { none, global, local } type;
+
+    Label() : symbol(""), type(none) {}
+    Label(const Label&) = default;
+    Label(std::string symbol_, Type type_) : symbol(symbol_), type(type_) {}
+    explicit operator bool() { return !symbol.empty() && type != none; }
+};
+
 class AsmFile
 {
 public:
@@ -43,7 +54,8 @@ public:
     AsmFile(const AsmFile&) = delete;
     ~AsmFile();
     Directive GetDirective();
-    std::string GetGlobalLabel();
+    Label GetLabel();
+    std::string PeekSection();
     std::string ReadPath();
     int ReadString(unsigned char* s);
     int ReadBraille(unsigned char* s);
@@ -60,6 +72,7 @@ private:
     long m_lineNum;
     long m_lineStart;
     std::string m_filename;
+    bool m_scriptData;
 
     bool ConsumeComma();
     int ReadPadLength();

tools/preproc/preproc.cpp

@@ -51,9 +51,11 @@ void PrintAsmBytes(unsigned char *s, int length)
     }
 }
 
-void PreprocAsmFile(std::string filename, bool isStdin, bool doEnum)
+void PreprocAsmFile(std::string filename, bool isStdin, bool doEnum, bool doSize)
 {
     std::stack<AsmFile> stack;
+    Label prevLabel;
+    bool inScriptData = false;
 
     stack.push(AsmFile(filename, isStdin, doEnum));
     std::printf("# 1 \"%s\"\n", filename.c_str());
@@ -62,6 +64,11 @@ void PreprocAsmFile(std::string filename, bool isStdin, bool doEnum)
     {
         while (stack.top().IsAtEnd())
         {
+            const char *ps = prevLabel.symbol.c_str();
+            if (doSize && inScriptData && prevLabel)
+                std::printf(".ifdef %s ; .size %s, . - %s ; .endif\n", ps, ps, ps);
+            prevLabel = Label();
+
             stack.pop();
 
             if (stack.empty())
@@ -100,15 +107,28 @@ void PreprocAsmFile(std::string filename, bool isStdin, bool doEnum)
         }
         case Directive::Unknown:
         {
-            std::string globalLabel = stack.top().GetGlobalLabel();
+            Label label = stack.top().GetLabel();
 
-            if (globalLabel.length() != 0)
+            if (label)
             {
-                const char *s = globalLabel.c_str();
+                const char *s = label.symbol.c_str();
-                std::printf("%s: ; .global %s\n", s, s);
+                const char *ps = prevLabel.symbol.c_str();
+
+                if (doSize && inScriptData && prevLabel)
+                    std::printf(".ifdef %s ; .size %s, . - %s ; .endif ; ", ps, ps, ps);
+
+                if (label.type == Label::global)
+                    std::printf(".global %s\n%s:\n", s, s);
+
+                prevLabel = label;
             }
             else
             {
+                std::string section = stack.top().PeekSection();
+                if (section == "script_data")
+                    inScriptData = true;
+                else if (section != "")
+                    inScriptData = false;
                 stack.top().OutputLine();
             }
 
@@ -147,7 +167,7 @@ const char* GetFileExtension(const char* filename)
 
 static void UsageAndExit(const char *program)
 {
-    std::fprintf(stderr, "Usage: %s [-i] [-e] SRC_FILE CHARMAP_FILE\nwhere -i denotes if input is from stdin\n      -e enables enum handling\n", program);
+    std::fprintf(stderr, "Usage: %s [-i] [-e] [-s] SRC_FILE CHARMAP_FILE\nwhere -i denotes if input is from stdin\n      -e enables enum handling\n        -s enables '.size' handling\n", program);
     std::exit(EXIT_FAILURE);
 }
 
@@ -158,9 +178,10 @@ int main(int argc, char **argv)
     const char *charmap = NULL;
     bool isStdin = false;
     bool doEnum = false;
+    bool doSize = false;
 
-    /* preproc [-i] [-e] SRC_FILE CHARMAP_FILE */
+    /* preproc [-i] [-e] [-s] SRC_FILE CHARMAP_FILE */
-    while ((opt = getopt(argc, argv, "ie")) != -1)
+    while ((opt = getopt(argc, argv, "ies")) != -1)
     {
         switch (opt)
         {
@@ -170,6 +191,9 @@ int main(int argc, char **argv)
         case 'e':
             doEnum = true;
             break;
+        case 's':
+            doSize = true;
+            break;
         default:
             UsageAndExit(argv[0]);
             break;
@@ -196,7 +220,7 @@ int main(int argc, char **argv)
 
     if ((extension[0] == 's') && extension[1] == 0)
     {
-        PreprocAsmFile(source, isStdin, doEnum);
+        PreprocAsmFile(source, isStdin, doEnum, doSize);
     }
     else if ((extension[0] == 'c' || extension[0] == 'i') && extension[1] == 0)
     {