1 //===-- llvm/BinaryFormat/XCOFF.h - The XCOFF file format -------*- C++/-*-===//
2 //
3 // Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
4 // See https://llvm.org/LICENSE.txt for license information.
5 // SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
6 //
7 //===----------------------------------------------------------------------===//
8 //
9 // This file defines manifest constants for the XCOFF object file format.
10 //
11 //===----------------------------------------------------------------------===//
12 
13 #ifndef LLVM_BINARYFORMAT_XCOFF_H
14 #define LLVM_BINARYFORMAT_XCOFF_H
15 
16 #include <stddef.h>
17 #include <stdint.h>
18 
19 namespace llvm {
20 class StringRef;
21 template <unsigned> class SmallString;
22 template <typename T> class Expected;
23 
24 namespace XCOFF {
25 
26 // Constants used in the XCOFF definition.
27 
28 constexpr size_t FileNamePadSize = 6;
29 constexpr size_t NameSize = 8;
30 constexpr size_t FileHeaderSize32 = 20;
31 constexpr size_t FileHeaderSize64 = 24;
32 constexpr size_t AuxFileHeaderSize32 = 72;
33 constexpr size_t AuxFileHeaderSize64 = 110;
34 constexpr size_t AuxFileHeaderSizeShort = 28;
35 constexpr size_t SectionHeaderSize32 = 40;
36 constexpr size_t SectionHeaderSize64 = 72;
37 constexpr size_t SymbolTableEntrySize = 18;
38 constexpr size_t RelocationSerializationSize32 = 10;
39 constexpr size_t RelocationSerializationSize64 = 14;
40 constexpr size_t ExceptionSectionEntrySize32 = 6;
41 constexpr size_t ExceptionSectionEntrySize64 = 10;
42 constexpr uint16_t RelocOverflow = 65535;
43 constexpr uint8_t AllocRegNo = 31;
44 
45 enum ReservedSectionNum : int16_t { N_DEBUG = -2, N_ABS = -1, N_UNDEF = 0 };
46 
47 enum MagicNumber : uint16_t { XCOFF32 = 0x01DF, XCOFF64 = 0x01F7 };
48 
49 // Masks for packing/unpacking the r_rsize field of relocations.
50 
51 // The msb is used to indicate if the bits being relocated are signed or
52 // unsigned.
53 static constexpr uint8_t XR_SIGN_INDICATOR_MASK = 0x80;
54 // The 2nd msb is used to indicate that the binder has replaced/modified the
55 // original instruction.
56 static constexpr uint8_t XR_FIXUP_INDICATOR_MASK = 0x40;
57 // The remaining bits specify the bit length of the relocatable reference
58 // minus one.
59 static constexpr uint8_t XR_BIASED_LENGTH_MASK = 0x3f;
60 
61 // This field only exists in the XCOFF64 definition.
62 enum AuxHeaderFlags64 : uint16_t {
63   SHR_SYMTAB = 0x8000,  ///< At exec time, create shared symbol table for program
64                         ///< (main program only).
65   FORK_POLICY = 0x4000, ///< Forktree policy specified (main program only).
66   FORK_COR = 0x2000     ///< If _AOUT_FORK_POLICY is set, specify copy-on-reference
67                         ///< if this bit is set. Specify copy-on- write otherwise.
68                         ///< If _AOUT_FORK_POLICY is 0, this bit is reserved for
69                         ///< future use and should be set to 0.
70 };
71 
72 enum XCOFFInterpret : uint16_t {
73   OLD_XCOFF_INTERPRET = 1,
74   NEW_XCOFF_INTERPRET = 2
75 };
76 
77 enum FileFlag : uint16_t {
78   F_RELFLG = 0x0001,    ///< relocation info stripped from file
79   F_EXEC = 0x0002,      ///< file is executable (i.e., it
80                         ///< has a loader section)
81   F_LNNO = 0x0004,      ///< line numbers stripped from file
82   F_LSYMS = 0x0008,     ///< local symbols stripped from file
83   F_FDPR_PROF = 0x0010, ///< file was profiled with FDPR
84   F_FDPR_OPTI = 0x0020, ///< file was reordered with FDPR
85   F_DSA = 0x0040,       ///< file uses Dynamic Segment Allocation (32-bit
86                         ///< only)
87   F_DEP_1 = 0x0080,     ///< Data Execution Protection bit 1
88   F_VARPG = 0x0100,     ///< executable requests using variable size pages
89   F_LPTEXT = 0x0400,    ///< executable requires large pages for text
90   F_LPDATA = 0x0800,    ///< executable requires large pages for data
91   F_DYNLOAD = 0x1000,   ///< file is dynamically loadable and
92                         ///< executable (equivalent to F_EXEC on AIX)
93   F_SHROBJ = 0x2000,    ///< file is a shared object
94   F_LOADONLY =
95       0x4000,      ///< file can be loaded by the system loader, but it is
96                    ///< ignored by the linker if it is a member of an archive.
97   F_DEP_2 = 0x8000 ///< Data Execution Protection bit 2
98 };
99 
100 // x_smclas field of x_csect from system header: /usr/include/syms.h
101 /// Storage Mapping Class definitions.
102 enum StorageMappingClass : uint8_t {
103   //     READ ONLY CLASSES
104   XMC_PR = 0,      ///< Program Code
105   XMC_RO = 1,      ///< Read Only Constant
106   XMC_DB = 2,      ///< Debug Dictionary Table
107   XMC_GL = 6,      ///< Global Linkage (Interfile Interface Code)
108   XMC_XO = 7,      ///< Extended Operation (Pseudo Machine Instruction)
109   XMC_SV = 8,      ///< Supervisor Call (32-bit process only)
110   XMC_SV64 = 17,   ///< Supervisor Call for 64-bit process
111   XMC_SV3264 = 18, ///< Supervisor Call for both 32- and 64-bit processes
112   XMC_TI = 12,     ///< Traceback Index csect
113   XMC_TB = 13,     ///< Traceback Table csect
114 
115   //       READ WRITE CLASSES
116   XMC_RW = 5,   ///< Read Write Data
117   XMC_TC0 = 15, ///< TOC Anchor for TOC Addressability
118   XMC_TC = 3,   ///< General TOC item
119   XMC_TD = 16,  ///< Scalar data item in the TOC
120   XMC_DS = 10,  ///< Descriptor csect
121   XMC_UA = 4,   ///< Unclassified - Treated as Read Write
122   XMC_BS = 9,   ///< BSS class (uninitialized static internal)
123   XMC_UC = 11,  ///< Un-named Fortran Common
124 
125   XMC_TL = 20, ///< Initialized thread-local variable
126   XMC_UL = 21, ///< Uninitialized thread-local variable
127   XMC_TE = 22  ///< Symbol mapped at the end of TOC
128 };
129 
130 // Flags for defining the section type. Masks for use with the (signed, 32-bit)
131 // s_flags field of the section header structure, selecting for values in the
132 // lower 16 bits. Defined in the system header `scnhdr.h`.
133 enum SectionTypeFlags : int32_t {
134   STYP_PAD = 0x0008,
135   STYP_DWARF = 0x0010,
136   STYP_TEXT = 0x0020,
137   STYP_DATA = 0x0040,
138   STYP_BSS = 0x0080,
139   STYP_EXCEPT = 0x0100,
140   STYP_INFO = 0x0200,
141   STYP_TDATA = 0x0400,
142   STYP_TBSS = 0x0800,
143   STYP_LOADER = 0x1000,
144   STYP_DEBUG = 0x2000,
145   STYP_TYPCHK = 0x4000,
146   STYP_OVRFLO = 0x8000
147 };
148 
149 /// Values for defining the section subtype of sections of type STYP_DWARF as
150 /// they would appear in the (signed, 32-bit) s_flags field of the section
151 /// header structure, contributing to the 16 most significant bits. Defined in
152 /// the system header `scnhdr.h`.
153 enum DwarfSectionSubtypeFlags : int32_t {
154   SSUBTYP_DWINFO = 0x1'0000,  ///< DWARF info section
155   SSUBTYP_DWLINE = 0x2'0000,  ///< DWARF line section
156   SSUBTYP_DWPBNMS = 0x3'0000, ///< DWARF pubnames section
157   SSUBTYP_DWPBTYP = 0x4'0000, ///< DWARF pubtypes section
158   SSUBTYP_DWARNGE = 0x5'0000, ///< DWARF aranges section
159   SSUBTYP_DWABREV = 0x6'0000, ///< DWARF abbrev section
160   SSUBTYP_DWSTR = 0x7'0000,   ///< DWARF str section
161   SSUBTYP_DWRNGES = 0x8'0000, ///< DWARF ranges section
162   SSUBTYP_DWLOC = 0x9'0000,   ///< DWARF loc section
163   SSUBTYP_DWFRAME = 0xA'0000, ///< DWARF frame section
164   SSUBTYP_DWMAC = 0xB'0000    ///< DWARF macinfo section
165 };
166 
167 // STORAGE CLASSES, n_sclass field of syment.
168 // The values come from `storclass.h` and `dbxstclass.h`.
169 enum StorageClass : uint8_t {
170   // Storage classes used for symbolic debugging symbols.
171   C_FILE = 103,  // File name
172   C_BINCL = 108, // Beginning of include file
173   C_EINCL = 109, // Ending of include file
174   C_GSYM = 128,  // Global variable
175   C_STSYM = 133, // Statically allocated symbol
176   C_BCOMM = 135, // Beginning of common block
177   C_ECOMM = 137, // End of common block
178   C_ENTRY = 141, // Alternate entry
179   C_BSTAT = 143, // Beginning of static block
180   C_ESTAT = 144, // End of static block
181   C_GTLS = 145,  // Global thread-local variable
182   C_STTLS = 146, // Static thread-local variable
183 
184   // Storage classes used for DWARF symbols.
185   C_DWARF = 112, // DWARF section symbol
186 
187   // Storage classes used for absolute symbols.
188   C_LSYM = 129,  // Automatic variable allocated on stack
189   C_PSYM = 130,  // Argument to subroutine allocated on stack
190   C_RSYM = 131,  // Register variable
191   C_RPSYM = 132, // Argument to function or procedure stored in register
192   C_ECOML = 136, // Local member of common block
193   C_FUN = 142,   // Function or procedure
194 
195   // Storage classes used for undefined external symbols or
196   // symbols of general sections.
197   C_EXT = 2,       // External symbol
198   C_WEAKEXT = 111, // Weak external symbol
199 
200   // Storage classes used for symbols of general sections.
201   C_NULL = 0,
202   C_STAT = 3,     // Static
203   C_BLOCK = 100,  // ".bb" or ".eb"
204   C_FCN = 101,    // ".bf" or ".ef"
205   C_HIDEXT = 107, // Un-named external symbol
206   C_INFO = 110,   // Comment string in .info section
207   C_DECL = 140,   // Declaration of object (type)
208 
209   // Storage classes - Obsolete/Undocumented.
210   C_AUTO = 1,     // Automatic variable
211   C_REG = 4,      // Register variable
212   C_EXTDEF = 5,   // External definition
213   C_LABEL = 6,    // Label
214   C_ULABEL = 7,   // Undefined label
215   C_MOS = 8,      // Member of structure
216   C_ARG = 9,      // Function argument
217   C_STRTAG = 10,  // Structure tag
218   C_MOU = 11,     // Member of union
219   C_UNTAG = 12,   // Union tag
220   C_TPDEF = 13,   // Type definition
221   C_USTATIC = 14, // Undefined static
222   C_ENTAG = 15,   // Enumeration tag
223   C_MOE = 16,     // Member of enumeration
224   C_REGPARM = 17, // Register parameter
225   C_FIELD = 18,   // Bit field
226   C_EOS = 102,    // End of structure
227   C_LINE = 104,
228   C_ALIAS = 105,  // Duplicate tag
229   C_HIDDEN = 106, // Special storage class for external
230   C_EFCN = 255,   // Physical end of function
231 
232   // Storage classes - reserved
233   C_TCSYM = 134 // Reserved
234 };
235 
236 // Flags for defining the symbol type. Values to be encoded into the lower 3
237 // bits of the (unsigned, 8-bit) x_smtyp field of csect auxiliary symbol table
238 // entries. Defined in the system header `syms.h`.
239 enum SymbolType : uint8_t {
240   XTY_ER = 0, ///< External reference.
241   XTY_SD = 1, ///< Csect definition for initialized storage.
242   XTY_LD = 2, ///< Label definition.
243               ///< Defines an entry point to an initialized csect.
244   XTY_CM = 3  ///< Common csect definition. For uninitialized storage.
245 };
246 
247 /// Values for visibility as they would appear when encoded in the high 4 bits
248 /// of the 16-bit unsigned n_type field of symbol table entries. Valid for
249 /// 32-bit XCOFF only when the vstamp in the auxiliary header is greater than 1.
250 enum VisibilityType : uint16_t {
251   SYM_V_UNSPECIFIED = 0x0000,
252   SYM_V_INTERNAL = 0x1000,
253   SYM_V_HIDDEN = 0x2000,
254   SYM_V_PROTECTED = 0x3000,
255   SYM_V_EXPORTED = 0x4000
256 };
257 
258 constexpr uint16_t VISIBILITY_MASK = 0x7000;
259 
260 // Relocation types, defined in `/usr/include/reloc.h`.
261 enum RelocationType : uint8_t {
262   R_POS = 0x00, ///< Positive relocation. Provides the address of the referenced
263                 ///< symbol.
264   R_RL = 0x0c,  ///< Positive indirect load relocation. Modifiable instruction.
265   R_RLA = 0x0d, ///< Positive load address relocation. Modifiable instruction.
266 
267   R_NEG = 0x01, ///< Negative relocation. Provides the negative of the address
268                 ///< of the referenced symbol.
269   R_REL = 0x02, ///< Relative to self relocation. Provides a displacement value
270                 ///< between the address of the referenced symbol and the
271                 ///< address being relocated.
272 
273   R_TOC = 0x03, ///< Relative to the TOC relocation. Provides a displacement
274                 ///< that is the difference between the address of the
275                 ///< referenced symbol and the TOC anchor csect.
276   R_TRL = 0x12, ///< TOC relative indirect load relocation. Similar to R_TOC,
277                 ///< but not modifiable instruction.
278 
279   R_TRLA =
280       0x13, ///< Relative to the TOC or to the thread-local storage base
281             ///< relocation. Compilers are not permitted to generate this
282             ///< relocation type. It is the result of a reversible
283             ///< transformation by the linker of an R_TOC relation that turned a
284             ///< load instruction into an add-immediate instruction.
285 
286   R_GL = 0x05, ///< Global linkage-external TOC address relocation. Provides the
287                ///< address of the external TOC associated with a defined
288                ///< external symbol.
289   R_TCL = 0x06, ///< Local object TOC address relocation. Provides the address
290                 ///< of the local TOC entry of a defined external symbol.
291 
292   R_REF = 0x0f, ///< A non-relocating relocation. Used to prevent the binder
293                 ///< from garbage collecting a csect (such as code used for
294                 ///< dynamic initialization of non-local statics) for which
295                 ///< another csect has an implicit dependency.
296 
297   R_BA = 0x08, ///< Branch absolute relocation. Provides the address of the
298                ///< referenced symbol. References a non-modifiable instruction.
299   R_BR = 0x0a, ///< Branch relative to self relocation. Provides the
300                ///< displacement that is the difference between the address of
301                ///< the referenced symbol and the address of the referenced
302                ///< branch instruction. References a non-modifiable instruction.
303   R_RBA = 0x18, ///< Branch absolute relocation. Similar to R_BA but
304                 ///< references a modifiable instruction.
305   R_RBR = 0x1a, ///< Branch relative to self relocation. Similar to the R_BR
306                 ///< relocation type, but references a modifiable instruction.
307 
308   R_TLS = 0x20,    ///< General-dynamic reference to TLS symbol.
309   R_TLS_IE = 0x21, ///< Initial-exec reference to TLS symbol.
310   R_TLS_LD = 0x22, ///< Local-dynamic reference to TLS symbol.
311   R_TLS_LE = 0x23, ///< Local-exec reference to TLS symbol.
312   R_TLSM = 0x24,  ///< Module reference to TLS. Provides a handle for the module
313                   ///< containing the referenced symbol.
314   R_TLSML = 0x25, ///< Module reference to the local TLS storage.
315 
316   R_TOCU = 0x30, ///< Relative to TOC upper. Specifies the high-order 16 bits of
317                  ///< a large code model TOC-relative relocation.
318   R_TOCL = 0x31 ///< Relative to TOC lower. Specifies the low-order 16 bits of a
319                 ///< large code model TOC-relative relocation.
320 };
321 
322 enum CFileStringType : uint8_t {
323   XFT_FN = 0,  ///< Specifies the source-file name.
324   XFT_CT = 1,  ///< Specifies the compiler time stamp.
325   XFT_CV = 2,  ///< Specifies the compiler version number.
326   XFT_CD = 128 ///< Specifies compiler-defined information.
327 };
328 
329 enum CFileLangId : uint8_t {
330   TB_C = 0,        ///< C language.
331   TB_CPLUSPLUS = 9 ///< C++ language.
332 };
333 
334 enum CFileCpuId : uint8_t {
335   TCPU_PPC64 = 2, ///< PowerPC common architecture 64-bit mode.
336   TCPU_COM = 3,   ///< POWER and PowerPC architecture common.
337   TCPU_970 = 19   ///< PPC970 - PowerPC 64-bit architecture.
338 };
339 
340 enum SymbolAuxType : uint8_t {
341   AUX_EXCEPT = 255, ///< Identifies an exception auxiliary entry.
342   AUX_FCN = 254,    ///< Identifies a function auxiliary entry.
343   AUX_SYM = 253,    ///< Identifies a symbol auxiliary entry.
344   AUX_FILE = 252,   ///< Identifies a file auxiliary entry.
345   AUX_CSECT = 251,  ///< Identifies a csect auxiliary entry.
346   AUX_SECT = 250    ///< Identifies a SECT auxiliary entry.
347 };                  // 64-bit XCOFF file only.
348 
349 StringRef getMappingClassString(XCOFF::StorageMappingClass SMC);
350 StringRef getRelocationTypeString(XCOFF::RelocationType Type);
351 Expected<SmallString<32>> parseParmsType(uint32_t Value, unsigned FixedParmsNum,
352                                          unsigned FloatingParmsNum);
353 Expected<SmallString<32>> parseParmsTypeWithVecInfo(uint32_t Value,
354                                                     unsigned FixedParmsNum,
355                                                     unsigned FloatingParmsNum,
356                                                     unsigned VectorParmsNum);
357 Expected<SmallString<32>> parseVectorParmsType(uint32_t Value,
358                                                unsigned ParmsNum);
359 
360 struct TracebackTable {
361   enum LanguageID : uint8_t {
362     C,
363     Fortran,
364     Pascal,
365     Ada,
366     PL1,
367     Basic,
368     Lisp,
369     Cobol,
370     Modula2,
371     CPlusPlus,
372     Rpg,
373     PL8,
374     PLIX = PL8,
375     Assembly,
376     Java,
377     ObjectiveC
378   };
379   // Byte 1
380   static constexpr uint32_t VersionMask = 0xFF00'0000;
381   static constexpr uint8_t VersionShift = 24;
382 
383   // Byte 2
384   static constexpr uint32_t LanguageIdMask = 0x00FF'0000;
385   static constexpr uint8_t LanguageIdShift = 16;
386 
387   // Byte 3
388   static constexpr uint32_t IsGlobaLinkageMask = 0x0000'8000;
389   static constexpr uint32_t IsOutOfLineEpilogOrPrologueMask = 0x0000'4000;
390   static constexpr uint32_t HasTraceBackTableOffsetMask = 0x0000'2000;
391   static constexpr uint32_t IsInternalProcedureMask = 0x0000'1000;
392   static constexpr uint32_t HasControlledStorageMask = 0x0000'0800;
393   static constexpr uint32_t IsTOClessMask = 0x0000'0400;
394   static constexpr uint32_t IsFloatingPointPresentMask = 0x0000'0200;
395   static constexpr uint32_t IsFloatingPointOperationLogOrAbortEnabledMask =
396       0x0000'0100;
397 
398   // Byte 4
399   static constexpr uint32_t IsInterruptHandlerMask = 0x0000'0080;
400   static constexpr uint32_t IsFunctionNamePresentMask = 0x0000'0040;
401   static constexpr uint32_t IsAllocaUsedMask = 0x0000'0020;
402   static constexpr uint32_t OnConditionDirectiveMask = 0x0000'001C;
403   static constexpr uint32_t IsCRSavedMask = 0x0000'0002;
404   static constexpr uint32_t IsLRSavedMask = 0x0000'0001;
405   static constexpr uint8_t OnConditionDirectiveShift = 2;
406 
407   // Byte 5
408   static constexpr uint32_t IsBackChainStoredMask = 0x8000'0000;
409   static constexpr uint32_t IsFixupMask = 0x4000'0000;
410   static constexpr uint32_t FPRSavedMask = 0x3F00'0000;
411   static constexpr uint32_t FPRSavedShift = 24;
412 
413   // Byte 6
414   static constexpr uint32_t HasExtensionTableMask = 0x0080'0000;
415   static constexpr uint32_t HasVectorInfoMask = 0x0040'0000;
416   static constexpr uint32_t GPRSavedMask = 0x003F'0000;
417   static constexpr uint32_t GPRSavedShift = 16;
418 
419   // Byte 7
420   static constexpr uint32_t NumberOfFixedParmsMask = 0x0000'FF00;
421   static constexpr uint8_t NumberOfFixedParmsShift = 8;
422 
423   // Byte 8
424   static constexpr uint32_t NumberOfFloatingPointParmsMask = 0x0000'00FE;
425   static constexpr uint32_t HasParmsOnStackMask = 0x0000'0001;
426   static constexpr uint8_t NumberOfFloatingPointParmsShift = 1;
427 
428   // Masks to select leftmost bits for decoding parameter type information.
429   // Bit to use when vector info is not presented.
430   static constexpr uint32_t ParmTypeIsFloatingBit = 0x8000'0000;
431   static constexpr uint32_t ParmTypeFloatingIsDoubleBit = 0x4000'0000;
432   // Bits to use when vector info is presented.
433   static constexpr uint32_t ParmTypeIsFixedBits = 0x0000'0000;
434   static constexpr uint32_t ParmTypeIsVectorBits = 0x4000'0000;
435   static constexpr uint32_t ParmTypeIsFloatingBits = 0x8000'0000;
436   static constexpr uint32_t ParmTypeIsDoubleBits = 0xC000'0000;
437   static constexpr uint32_t ParmTypeMask = 0xC000'0000;
438 
439   // Vector extension
440   static constexpr uint16_t NumberOfVRSavedMask = 0xFC00;
441   static constexpr uint16_t IsVRSavedOnStackMask = 0x0200;
442   static constexpr uint16_t HasVarArgsMask = 0x0100;
443   static constexpr uint8_t NumberOfVRSavedShift = 10;
444 
445   static constexpr uint16_t NumberOfVectorParmsMask = 0x00FE;
446   static constexpr uint16_t HasVMXInstructionMask = 0x0001;
447   static constexpr uint8_t NumberOfVectorParmsShift = 1;
448 
449   static constexpr uint32_t ParmTypeIsVectorCharBit = 0x0000'0000;
450   static constexpr uint32_t ParmTypeIsVectorShortBit = 0x4000'0000;
451   static constexpr uint32_t ParmTypeIsVectorIntBit = 0x8000'0000;
452   static constexpr uint32_t ParmTypeIsVectorFloatBit = 0xC000'0000;
453 
454   static constexpr uint8_t WidthOfParamType = 2;
455 };
456 
457 // Extended Traceback table flags.
458 enum ExtendedTBTableFlag : uint8_t {
459   TB_OS1 = 0x80,         ///< Reserved for OS use.
460   TB_RESERVED = 0x40,    ///< Reserved for compiler.
461   TB_SSP_CANARY = 0x20,  ///< stack smasher canary present on stack.
462   TB_OS2 = 0x10,         ///< Reserved for OS use.
463   TB_EH_INFO = 0x08,     ///< Exception handling info present.
464   TB_LONGTBTABLE2 = 0x01 ///< Additional tbtable extension exists.
465 };
466 
467 StringRef getNameForTracebackTableLanguageId(TracebackTable::LanguageID LangId);
468 SmallString<32> getExtendedTBTableFlagString(uint8_t Flag);
469 
470 struct CsectProperties {
CsectPropertiesCsectProperties471   CsectProperties(StorageMappingClass SMC, SymbolType ST)
472       : MappingClass(SMC), Type(ST) {}
473   StorageMappingClass MappingClass;
474   SymbolType Type;
475 };
476 
477 } // end namespace XCOFF
478 } // end namespace llvm
479 
480 #endif
481