1 //===- lib/MC/MCAsmStreamer.cpp - Text Assembly Output --------------------===//
2 //
3 //                     The LLVM Compiler Infrastructure
4 //
5 // This file is distributed under the University of Illinois Open Source
6 // License. See LICENSE.TXT for details.
7 //
8 //===----------------------------------------------------------------------===//
9 
10 #include "llvm/MC/MCStreamer.h"
11 
12 #include "llvm/MC/MCContext.h"
13 #include "llvm/MC/MCInst.h"
14 #include "llvm/MC/MCSection.h"
15 #include "llvm/MC/MCSymbol.h"
16 #include "llvm/MC/MCValue.h"
17 #include "llvm/Support/ErrorHandling.h"
18 #include "llvm/Support/raw_ostream.h"
19 
20 using namespace llvm;
21 
22 namespace {
23 
24   class MCAsmStreamer : public MCStreamer {
25     raw_ostream &OS;
26 
27     MCSection *CurSection;
28 
29   public:
30     MCAsmStreamer(MCContext &Context, raw_ostream &_OS)
31       : MCStreamer(Context), OS(_OS), CurSection(0) {}
32     ~MCAsmStreamer() {}
33 
34     /// @name MCStreamer Interface
35     /// @{
36 
37     virtual void SwitchSection(MCSection *Section);
38 
39     virtual void EmitLabel(MCSymbol *Symbol);
40 
41     virtual void EmitAssemblerFlag(AssemblerFlag Flag);
42 
43     virtual void EmitAssignment(MCSymbol *Symbol, const MCValue &Value,
44                                 bool MakeAbsolute = false);
45 
46     virtual void EmitSymbolAttribute(MCSymbol *Symbol, SymbolAttr Attribute);
47 
48     virtual void EmitSymbolDesc(MCSymbol *Symbol, unsigned DescValue);
49 
50     virtual void EmitLocalSymbol(MCSymbol *Symbol, const MCValue &Value);
51 
52     virtual void EmitCommonSymbol(MCSymbol *Symbol, unsigned Size,
53                                   unsigned Pow2Alignment, bool IsLocal);
54 
55     virtual void EmitZerofill(MCSection *Section, MCSymbol *Symbol = NULL,
56                               unsigned Size = 0, unsigned Pow2Alignment = 0);
57 
58     virtual void EmitBytes(const StringRef &Data);
59 
60     virtual void EmitValue(const MCValue &Value, unsigned Size);
61 
62     virtual void EmitValueToAlignment(unsigned ByteAlignment, int64_t Value = 0,
63                                       unsigned ValueSize = 1,
64                                       unsigned MaxBytesToEmit = 0);
65 
66     virtual void EmitValueToOffset(const MCValue &Offset,
67                                    unsigned char Value = 0);
68 
69     virtual void EmitInstruction(const MCInst &Inst);
70 
71     virtual void Finish();
72 
73     /// @}
74   };
75 
76 }
77 
78 /// NeedsQuoting - Return true if the string \arg Str needs quoting, i.e., it
79 /// does not match [a-zA-Z_.][a-zA-Z0-9_.]*.
80 //
81 // FIXME: This could be more permissive, do we care?
82 static inline bool NeedsQuoting(const StringRef &Str) {
83   if (Str.empty())
84     return true;
85 
86   // Check that first character is in [a-zA-Z_.].
87   if (!((Str[0] >= 'a' && Str[0] <= 'z') ||
88         (Str[0] >= 'A' && Str[0] <= 'Z') ||
89         (Str[0] == '_' || Str[0] == '.')))
90     return true;
91 
92   // Check subsequent characters are in [a-zA-Z0-9_.].
93   for (unsigned i = 1, e = Str.size(); i != e; ++i)
94     if (!((Str[i] >= 'a' && Str[i] <= 'z') ||
95           (Str[i] >= 'A' && Str[i] <= 'Z') ||
96           (Str[i] >= '0' && Str[i] <= '9') ||
97           (Str[i] == '_' || Str[i] == '.')))
98       return true;
99 
100   return false;
101 }
102 
103 /// Allow printing sections directly to a raw_ostream with proper quoting.
104 static inline raw_ostream &operator<<(raw_ostream &os, const MCSection *S) {
105   if (NeedsQuoting(S->getName()))
106     return os << '"' << S->getName() << '"';
107   return os << S->getName();
108 }
109 
110 /// Allow printing symbols directly to a raw_ostream with proper quoting.
111 static inline raw_ostream &operator<<(raw_ostream &os, const MCSymbol *S) {
112   if (NeedsQuoting(S->getName()))
113     return os << '"' << S->getName() << '"';
114   return os << S->getName();
115 }
116 
117 /// Allow printing values directly to a raw_ostream.
118 static inline raw_ostream &operator<<(raw_ostream &os, const MCValue &Value) {
119   if (Value.getSymA()) {
120     os << Value.getSymA();
121     if (Value.getSymB())
122       os << " - " << Value.getSymB();
123     if (Value.getConstant())
124       os << " + " << Value.getConstant();
125   } else {
126     assert(!Value.getSymB() && "Invalid machine code value!");
127     os << Value.getConstant();
128   }
129 
130   return os;
131 }
132 
133 static inline int64_t truncateToSize(int64_t Value, unsigned Bytes) {
134   assert(Bytes && "Invalid size!");
135   return Value & ((uint64_t) (int64_t) -1 >> (64 - Bytes * 8));
136 }
137 
138 static inline MCValue truncateToSize(const MCValue &Value, unsigned Bytes) {
139   return MCValue::get(Value.getSymA(), Value.getSymB(),
140                       truncateToSize(Value.getConstant(), Bytes));
141 }
142 
143 void MCAsmStreamer::SwitchSection(MCSection *Section) {
144   if (Section != CurSection) {
145     CurSection = Section;
146 
147     // FIXME: Really we would like the segment, flags, etc. to be separate
148     // values instead of embedded in the name. Not all assemblers understand all
149     // this stuff though.
150     OS << ".section " << Section << "\n";
151   }
152 }
153 
154 void MCAsmStreamer::EmitLabel(MCSymbol *Symbol) {
155   assert(Symbol->getSection() == 0 && "Cannot emit a symbol twice!");
156   assert(CurSection && "Cannot emit before setting section!");
157   assert(!getContext().GetSymbolValue(Symbol) &&
158          "Cannot emit symbol which was directly assigned to!");
159 
160   OS << Symbol << ":\n";
161   Symbol->setSection(CurSection);
162   Symbol->setExternal(false);
163 }
164 
165 void MCAsmStreamer::EmitAssemblerFlag(AssemblerFlag Flag) {
166   switch (Flag) {
167   case SubsectionsViaSymbols: OS << ".subsections_via_symbols"; break;
168   }
169   OS << '\n';
170 }
171 
172 void MCAsmStreamer::EmitAssignment(MCSymbol *Symbol, const MCValue &Value,
173                                    bool MakeAbsolute) {
174   assert(!Symbol->getSection() && "Cannot assign to a label!");
175 
176   if (MakeAbsolute) {
177     OS << ".set " << Symbol << ", " << Value << '\n';
178   } else {
179     OS << Symbol << " = " << Value << '\n';
180   }
181 
182   getContext().SetSymbolValue(Symbol, Value);
183 }
184 
185 void MCAsmStreamer::EmitSymbolAttribute(MCSymbol *Symbol,
186                                         SymbolAttr Attribute) {
187   switch (Attribute) {
188   case Global: OS << ".globl"; break;
189   case Hidden: OS << ".hidden"; break;
190   case IndirectSymbol: OS << ".indirect_symbol"; break;
191   case Internal: OS << ".internal"; break;
192   case LazyReference: OS << ".lazy_reference"; break;
193   case NoDeadStrip: OS << ".no_dead_strip"; break;
194   case PrivateExtern: OS << ".private_extern"; break;
195   case Protected: OS << ".protected"; break;
196   case Reference: OS << ".reference"; break;
197   case Weak: OS << ".weak"; break;
198   case WeakDefinition: OS << ".weak_definition"; break;
199   case WeakReference: OS << ".weak_reference"; break;
200   }
201 
202   OS << ' ' << Symbol << '\n';
203 }
204 
205 void MCAsmStreamer::EmitSymbolDesc(MCSymbol *Symbol, unsigned DescValue) {
206   OS << ".desc" << ' ' << Symbol << ',' << DescValue << '\n';
207 }
208 
209 void MCAsmStreamer::EmitLocalSymbol(MCSymbol *Symbol, const MCValue &Value) {
210   OS << ".lsym" << ' ' << Symbol << ',' << Value << '\n';
211 }
212 
213 void MCAsmStreamer::EmitCommonSymbol(MCSymbol *Symbol, unsigned Size,
214                                      unsigned Pow2Alignment, bool IsLocal) {
215   if (IsLocal)
216     OS << ".lcomm";
217   else
218     OS << ".comm";
219   OS << ' ' << Symbol << ',' << Size;
220   if (Pow2Alignment != 0)
221     OS << ',' << Pow2Alignment;
222   OS << '\n';
223 }
224 
225 void MCAsmStreamer::EmitZerofill(MCSection *Section, MCSymbol *Symbol,
226                                  unsigned Size, unsigned Pow2Alignment) {
227   // Note: a .zerofill directive does not switch sections
228   // FIXME: Really we would like the segment and section names as well as the
229   // section type to be separate values instead of embedded in the name. Not
230   // all assemblers understand all this stuff though.
231   OS << ".zerofill " << Section;
232   if (Symbol != NULL) {
233     OS << ',' << Symbol << ',' << Size;
234     if (Pow2Alignment != 0)
235       OS << ',' << Pow2Alignment;
236   }
237   OS << '\n';
238 }
239 
240 void MCAsmStreamer::EmitBytes(const StringRef &Data) {
241   assert(CurSection && "Cannot emit contents before setting section!");
242   for (unsigned i = 0, e = Data.size(); i != e; ++i)
243     OS << ".byte " << (unsigned) Data[i] << '\n';
244 }
245 
246 void MCAsmStreamer::EmitValue(const MCValue &Value, unsigned Size) {
247   assert(CurSection && "Cannot emit contents before setting section!");
248   // Need target hooks to know how to print this.
249   switch (Size) {
250   default:
251     llvm_unreachable("Invalid size for machine code value!");
252   case 1: OS << ".byte"; break;
253   case 2: OS << ".short"; break;
254   case 4: OS << ".long"; break;
255   case 8: OS << ".quad"; break;
256   }
257 
258   OS << ' ' << truncateToSize(Value, Size) << '\n';
259 }
260 
261 void MCAsmStreamer::EmitValueToAlignment(unsigned ByteAlignment, int64_t Value,
262                                          unsigned ValueSize,
263                                          unsigned MaxBytesToEmit) {
264   // Some assemblers don't support .balign, so we always emit as .p2align if
265   // this is a power of two. Otherwise we assume the client knows the target
266   // supports .balign and use that.
267   unsigned Pow2 = Log2_32(ByteAlignment);
268   bool IsPow2 = (1U << Pow2) == ByteAlignment;
269 
270   switch (ValueSize) {
271   default:
272     llvm_unreachable("Invalid size for machine code value!");
273   case 8:
274     llvm_unreachable("Unsupported alignment size!");
275   case 1: OS << (IsPow2 ? ".p2align" : ".balign"); break;
276   case 2: OS << (IsPow2 ? ".p2alignw" : ".balignw"); break;
277   case 4: OS << (IsPow2 ? ".p2alignl" : ".balignl"); break;
278   }
279 
280   OS << ' ' << (IsPow2 ? Pow2 : ByteAlignment);
281 
282   OS << ", " << truncateToSize(Value, ValueSize);
283   if (MaxBytesToEmit)
284     OS << ", " << MaxBytesToEmit;
285   OS << '\n';
286 }
287 
288 void MCAsmStreamer::EmitValueToOffset(const MCValue &Offset,
289                                       unsigned char Value) {
290   // FIXME: Verify that Offset is associated with the current section.
291   OS << ".org " << Offset << ", " << (unsigned) Value << '\n';
292 }
293 
294 static raw_ostream &operator<<(raw_ostream &OS, const MCOperand &Op) {
295   if (Op.isReg())
296     return OS << "reg:" << Op.getReg();
297   if (Op.isImm())
298     return OS << "imm:" << Op.getImm();
299   if (Op.isMBBLabel())
300     return OS << "mbblabel:("
301               << Op.getMBBLabelFunction() << ", " << Op.getMBBLabelBlock();
302   assert(Op.isMCValue() && "Invalid operand!");
303   return OS << "val:" << Op.getMCValue();
304 }
305 
306 void MCAsmStreamer::EmitInstruction(const MCInst &Inst) {
307   assert(CurSection && "Cannot emit contents before setting section!");
308   // FIXME: Implement proper printing.
309   OS << "MCInst("
310      << "opcode=" << Inst.getOpcode() << ", "
311      << "operands=[";
312   for (unsigned i = 0, e = Inst.getNumOperands(); i != e; ++i) {
313     if (i)
314       OS << ", ";
315     OS << Inst.getOperand(i);
316   }
317   OS << "])\n";
318 }
319 
320 void MCAsmStreamer::Finish() {
321   OS.flush();
322 }
323 
324 MCStreamer *llvm::createAsmStreamer(MCContext &Context, raw_ostream &OS) {
325   return new MCAsmStreamer(Context, OS);
326 }
327