Fix unused label issues

Treat labels which are used only by directives in other labels as used
if the other label itself is used by an opcode: that is, propagate the
transitive usage of labels.

Fixes #35

lib/asm.js

@@ -23,6 +23,7 @@
 // POSSIBILITY OF SUCH DAMAGE.
 
 (function () {
+    var _ = require('underscore-node');
     var tabsRe = /\t/g;
 
     function expandTabs(line) {
@@ -35,35 +36,91 @@
         });
     }
 
-    function processAsm(asm, filters) {
-        if (filters.binary) return processBinaryAsm(asm, filters);
+    var labelFind = /[.a-zA-Z_][a-zA-Z0-9$_.]*/g;
+    var dataDefn = /\.(string|asciz|ascii|[1248]?byte|short|word|long|quad|value|zero)/;
+    var fileFind = /^\s*\.file\s+(\d+)\s+"([^"]+)".*/;
+    var hasOpcode = /^\s*([a-zA-Z$_][a-zA-Z0-9$_.]*:\s*)?[a-zA-Z].*/;
+    var labelDef = /^([.a-zA-Z_][a-zA-Z0-9$_.]+):/;
+    var hasOpcode = /^\s*([a-zA-Z$_][a-zA-Z0-9$_.]*:\s*)?[a-zA-Z].*/;
 
-        var result = [];
-        var asmLines = asm.split("\n");
+    function findUsedLabels(asmLines, filterDirectives) {
         var labelsUsed = {};
-        var labelFind = /[.a-zA-Z0-9_][a-zA-Z0-9$_.]*/g;
-        var files = {};
-        var prevLabel = "";
-        var dataDefn = /\.(string|asciz|ascii|[1248]?byte|short|word|long|quad|value|zero)/;
-        var fileFind = /^\s*\.file\s+(\d+)\s+"([^"]+)".*/;
-        var hasOpcode = /^\s*([a-zA-Z0-9$_][a-zA-Z0-9$_.]*:\s*)?[a-zA-Z].*/;
+        var weakUsages = {};
+        var currentLabel = "";
+
+        // Scan through looking for definite label usages (ones used by opcodes),
+        // and ones that are weakly used: that is, their use is conditional on another label.
+        // For example:
+        // .foo: .string "moo"
+        // .baz: .quad .foo
+        //       mov eax, .baz
+        // In this case, the '.baz' is used by an opcode, and so is strongly used.
+        // The '.foo' is weakly used by .baz.
         asmLines.forEach(function (line) {
-            if (line === "" || line[0] === ".") return;
-            var match = line.match(labelFind);
-            if (match && (!filters.directives || line.match(hasOpcode))) {
+            var match = line.match(labelDef);
+            if (match)
+                currentLabel = match[1];
+            if (!line || line[0] === '.') return;
+
+            match = line.match(labelFind);
+            if (!match) return;
+
+            if (!filterDirectives || line.match(hasOpcode)) {
                 // Only count a label as used if it's used by an opcode, or else we're not filtering directives.
                 match.forEach(function (label) {
                     labelsUsed[label] = true;
                 });
+            } else if (currentLabel) {
+                // Note any "weak" usages by this label; that is, usages that are only
+                // interesting if the currentLabel is used by an opcode.
+                if (!weakUsages[currentLabel]) weakUsages[currentLabel] = [];
+                match.forEach(function (label) {
+                    weakUsages[currentLabel].push(label);
+                });
             }
-            match = line.match(fileFind);
+        });
+
+        // Now follow the chains of used labels, marking any weak references they refer
+        // to as also used. We iteratively do this until either no new labels are found,
+        // or we hit a limit (only here to prevent a pathological case from hanging).
+        var MaxLabelIterations = 10;
+        for (var iter = 0; iter < MaxLabelIterations; ++iter) {
+            var toAdd = [];
+            _.each(labelsUsed, function (t, label) {
+                _.each(weakUsages[label], function (nowused) {
+                    if (labelsUsed[nowused]) return;
+                    toAdd.push(nowused);
+                });
+            });
+            if (!toAdd) break;
+            _.each(toAdd, function (label) {
+                labelsUsed[label] = true;
+            });
+        }
+        return labelsUsed;
+    }
+
+    function parseFiles(asmLines) {
+        var files = {};
+        asmLines.forEach(function (line) {
+            var match = line.match(fileFind);
             if (match) {
                 files[parseInt(match[1])] = match[2];
             }
         });
+        return files;
+    }
+
+    function processAsm(asm, filters) {
+        if (filters.binary) return processBinaryAsm(asm, filters);
+
+        var result = [];
+        var asmLines = asm.split("\n");
+        var labelsUsed = findUsedLabels(asmLines, filters.directives);
+        var files = parseFiles(asmLines);
+        var prevLabel = "";
 
         var directive = /^\s*\..*$/;
-        var labelDefinition = /^([a-zA-Z0-9$_.]+):/;
         var commentOnly = /^\s*(#|@|\/\/).*/;
         var sourceTag = /^\s*\.loc\s+(\d+)\s+(\d+).*/;
         var stdInLooking = /.*<stdin>|-/;
@@ -89,7 +146,7 @@
 
             if (filters.commentOnly && line.match(commentOnly)) return;
 
-            match = line.match(labelDefinition);
+            match = line.match(labelDef);
             if (match) {
                 // It's a label definition.
                 if (labelsUsed[match[1]] === undefined) {

package.json

@@ -28,7 +28,8 @@
     "nopt": "3.0.x",
     "serve-favicon": "2.3.x",
     "serve-static": "1.10.x",
-    "temp": "0.8.x"
+    "temp": "0.8.x",
+    "underscore-node": "*"
   },
   "devDependencies": {
     "supervisor": "0.3.1",

test/cases/string-constant.asm

@@ -0,0 +1,46 @@
+        .file   "example.cpp"
+        .intel_syntax noprefix
+        .text
+.Ltext0:
+        .globl  s_sA
+        .section        .rodata
+.LC0:
+        .string "hello world!"
+        .data
+        .align 8
+        .type   s_sA, @object
+        .size   s_sA, 8
+s_sA:
+        .quad   .LC0
+        .section        .rodata
+        .align 8
+        .type   s_sB, @object
+        .size   s_sB, 10
+s_sB:
+        .string "hey there"
+        .text
+        .globl  main
+        .type   main, @function
+main:
+.LFB0:
+        .file 1 "/tmp/gcc-explorer-compiler116820-58-ewfj5u/example.cpp"
+        .loc 1 6 0
+        .cfi_startproc
+        push    rbp
+        .cfi_def_cfa_offset 16
+        .cfi_offset 6, -16
+        mov     rbp, rsp
+        .cfi_def_cfa_register 6
+        .loc 1 7 0
+        mov     rax, QWORD PTR s_sA[rip]
+        mov     rdi, rax
+        call    puts
+        .loc 1 8 0
+        mov     edi, OFFSET FLAT:s_sB
+        call    puts
+        .loc 1 9 0
+        mov     eax, 0
+        pop     rbp
+        .cfi_def_cfa 7, 8
+        ret
+        .cfi_endproc

test/cases/string-constant.asm.directives.labels.comments

@@ -0,0 +1,17 @@
+.LC0:
+        .string "hello world!"
+s_sA:
+        .quad   .LC0
+s_sB:
+        .string "hey there"
+main:
+        push    rbp
+        mov     rbp, rsp
+        mov     rax, QWORD PTR s_sA[rip]
+        mov     rdi, rax
+        call    puts
+        mov     edi, OFFSET FLAT:s_sB
+        call    puts
+        mov     eax, 0
+        pop     rbp
+        ret

test/test.js

@@ -68,7 +68,6 @@ function testFilter(filename, suffix, filters, withSource) {
         assertEq(file[i], lineExpected, expected + ":" + (i + 1));
     }
 }
-
 cases.forEach(function (x) {
     testFilter(x, "", {})
 });
@@ -91,4 +90,5 @@ cases.forEach(function (x) {
 
 if (failures) {
     console.log(failures + " failures");
+    process.exit(1);
 }