llvm-mc: Fix .s output to quote section & symbol names when necessary.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@77749 91177308-0d34-0410-b5e6-96231b3b80d8

3 files changed, 59 insertions, 20 deletions

diff --git a/lib/MC/MCAsmStreamer.cpp b/lib/MC/MCAsmStreamer.cpp
index b67250383a..a71174a77b 100644
--- a/lib/MC/MCAsmStreamer.cpp
+++ b/lib/MC/MCAsmStreamer.cpp
@@ -75,12 +75,51 @@ namespace {
 
 }
 
+/// NeedsQuoting - Return true if the string \arg Str needs quoting, i.e., it
+/// does not match [a-zA-Z_.][a-zA-Z0-9_.]*.
+//
+// FIXME: This could be more permissive, do we care?
+static inline bool NeedsQuoting(const StringRef &Str) {
+  if (Str.empty())
+    return true;
+
+  // Check that first character is in [a-zA-Z_.].
+  if (!((Str[0] >= 'a' && Str[0] <= 'z') ||
+        (Str[0] >= 'A' && Str[0] <= 'Z') ||
+        (Str[0] == '_' || Str[0] == '.')))
+    return true;
+
+  // Check subsequent characters are in [a-zA-Z0-9_.].
+  for (unsigned i = 1, e = Str.size(); i != e; ++i)
+    if (!((Str[i] >= 'a' && Str[i] <= 'z') ||
+          (Str[i] >= 'A' && Str[i] <= 'Z') ||
+          (Str[i] >= '0' && Str[i] <= '9') ||
+          (Str[i] == '_' || Str[i] == '.')))
+      return true;
+
+  return false;
+}
+
+/// Allow printing sections directly to a raw_ostream with proper quoting.
+static inline raw_ostream &operator<<(raw_ostream &os, const MCSection *S) {
+  if (NeedsQuoting(S->getName()))
+    return os << '"' << S->getName() << '"';
+  return os << S->getName();
+}
+
+/// Allow printing symbols directly to a raw_ostream with proper quoting.
+static inline raw_ostream &operator<<(raw_ostream &os, const MCSymbol *S) {
+  if (NeedsQuoting(S->getName()))
+    return os << '"' << S->getName() << '"';
+  return os << S->getName();
+}
+
 /// Allow printing values directly to a raw_ostream.
 static inline raw_ostream &operator<<(raw_ostream &os, const MCValue &Value) {
   if (Value.getSymA()) {
-    os << Value.getSymA()->getName();
+    os << Value.getSymA();
     if (Value.getSymB())
-      os << " - " << Value.getSymB()->getName();
+      os << " - " << Value.getSymB();
     if (Value.getConstant())
       os << " + " << Value.getConstant();
   } else {
@@ -108,7 +147,7 @@ void MCAsmStreamer::SwitchSection(MCSection *Section) {
     // FIXME: Really we would like the segment, flags, etc. to be separate
     // values instead of embedded in the name. Not all assemblers understand all
     // this stuff though.
-    OS << ".section " << Section->getName() << "\n";
+    OS << ".section " << Section << "\n";
   }
 }
 
@@ -118,7 +157,7 @@ void MCAsmStreamer::EmitLabel(MCSymbol *Symbol) {
   assert(!getContext().GetSymbolValue(Symbol) && 
          "Cannot emit symbol which was directly assigned to!");
 
-  OS << Symbol->getName() << ":\n";
+  OS << Symbol << ":\n";
   Symbol->setSection(CurSection);
   Symbol->setExternal(false);
 }
@@ -135,9 +174,9 @@ void MCAsmStreamer::EmitAssignment(MCSymbol *Symbol, const MCValue &Value,
   assert(!Symbol->getSection() && "Cannot assign to a label!");
 
   if (MakeAbsolute) {
-    OS << ".set " << Symbol->getName() << ", " << Value << '\n';
+    OS << ".set " << Symbol << ", " << Value << '\n';
   } else {
-    OS << Symbol->getName() << " = " << Value << '\n';
+    OS << Symbol << " = " << Value << '\n';
   }
 
   getContext().SetSymbolValue(Symbol, Value);
@@ -160,15 +199,15 @@ void MCAsmStreamer::EmitSymbolAttribute(MCSymbol *Symbol,
   case WeakReference: OS << ".weak_reference"; break;
   }
 
-  OS << ' ' << Symbol->getName() << '\n';
+  OS << ' ' << Symbol << '\n';
 }
 
 void MCAsmStreamer::EmitSymbolDesc(MCSymbol *Symbol, unsigned DescValue) {
-  OS << ".desc" << ' ' << Symbol->getName() << ',' << DescValue << '\n';
+  OS << ".desc" << ' ' << Symbol << ',' << DescValue << '\n';
 }
 
 void MCAsmStreamer::EmitLocalSymbol(MCSymbol *Symbol, const MCValue &Value) {
-  OS << ".lsym" << ' ' << Symbol->getName() << ',' << Value << '\n';
+  OS << ".lsym" << ' ' << Symbol << ',' << Value << '\n';
 }
 
 void MCAsmStreamer::EmitCommonSymbol(MCSymbol *Symbol, unsigned Size,
@@ -177,7 +216,7 @@ void MCAsmStreamer::EmitCommonSymbol(MCSymbol *Symbol, unsigned Size,
     OS << ".lcomm";
   else
     OS << ".comm";
-  OS << ' ' << Symbol->getName() << ',' << Size;
+  OS << ' ' << Symbol << ',' << Size;
   if (Pow2Alignment != 0)
     OS << ',' << Pow2Alignment;
   OS << '\n';
@@ -189,9 +228,9 @@ void MCAsmStreamer::EmitZerofill(MCSection *Section, MCSymbol *Symbol,
   // FIXME: Really we would like the segment and section names as well as the
   // section type to be separate values instead of embedded in the name. Not
   // all assemblers understand all this stuff though.
-  OS << ".zerofill " << Section->getName();
+  OS << ".zerofill " << Section;
   if (Symbol != NULL) {
-    OS << ',' << Symbol->getName() << ',' << Size;
+    OS << ',' << Symbol << ',' << Size;
     if (Pow2Alignment != 0)
       OS << ',' << Pow2Alignment;
   }
diff --git a/test/MC/AsmParser/directive_zerofill.s b/test/MC/AsmParser/directive_zerofill.s
index 4b26f9b68c..fd96812592 100644
--- a/test/MC/AsmParser/directive_zerofill.s
+++ b/test/MC/AsmParser/directive_zerofill.s
@@ -1,9 +1,9 @@
 # RUN: llvm-mc -triple i386-unknown-unknown %s | FileCheck %s
 
 # CHECK: TEST0:
-# CHECK: .zerofill __FOO,__bar,x,1
-# CHECK: .zerofill __FOO,__bar,y,8,2
-# CHECK: .zerofill __EMPTY,__NoSymbol
+# CHECK: .zerofill "__FOO,__bar",x,1
+# CHECK: .zerofill "__FOO,__bar",y,8,2
+# CHECK: .zerofill "__EMPTY,__NoSymbol"
 TEST0:  
 	.zerofill __FOO, __bar, x, 2-1
 	.zerofill __FOO,   __bar, y ,  8 , 1+1
diff --git a/test/MC/AsmParser/labels.s b/test/MC/AsmParser/labels.s
index f429b3bd25..1cdff8991d 100644
--- a/test/MC/AsmParser/labels.s
+++ b/test/MC/AsmParser/labels.s
@@ -7,20 +7,20 @@ a:
 // CHECK: b:
 "b":
         .long 0
-// FIXME(quoting): CHECK: a$b:
+// CHECK: "a$b":
 "a$b":
         .long 0
 
         .text:
 foo:    
-// FIXME(quoting): CHECK: val:a$b
+// CHECK: val:"a$b"
         addl $24, "a$b"(%eax)    
-// FIXME(quoting): CHECK: val:a$b + 10
+// CHECK: val:"a$b" + 10
         addl $24, ("a$b" + 10)(%eax)
         
-// FIXME(quoting): CHECK: b$c = 10
+// CHECK: "b$c" = 10
 "b$c" = 10
-// FIXME(quoting): CHECK: val:10
+// CHECK: val:10
         addl "b$c", %eax