Andrew Walbran | 3d2c197 | 2020-04-07 12:24:26 +0100 | [diff] [blame] | 1 | //===- FunctionInfo.h -------------------------------------------*- C++ -*-===// |
| 2 | // |
Olivier Deprez | f4ef2d0 | 2021-04-20 13:36:24 +0200 | [diff] [blame] | 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 |
Andrew Walbran | 3d2c197 | 2020-04-07 12:24:26 +0100 | [diff] [blame] | 6 | // |
| 7 | //===----------------------------------------------------------------------===// |
| 8 | |
| 9 | #ifndef LLVM_DEBUGINFO_GSYM_FUNCTIONINFO_H |
| 10 | #define LLVM_DEBUGINFO_GSYM_FUNCTIONINFO_H |
| 11 | |
Olivier Deprez | f4ef2d0 | 2021-04-20 13:36:24 +0200 | [diff] [blame] | 12 | #include "llvm/ADT/Optional.h" |
Andrew Walbran | 3d2c197 | 2020-04-07 12:24:26 +0100 | [diff] [blame] | 13 | #include "llvm/DebugInfo/GSYM/InlineInfo.h" |
Olivier Deprez | f4ef2d0 | 2021-04-20 13:36:24 +0200 | [diff] [blame] | 14 | #include "llvm/DebugInfo/GSYM/LineTable.h" |
| 15 | #include "llvm/DebugInfo/GSYM/LookupResult.h" |
Andrew Walbran | 3d2c197 | 2020-04-07 12:24:26 +0100 | [diff] [blame] | 16 | #include "llvm/DebugInfo/GSYM/Range.h" |
| 17 | #include "llvm/DebugInfo/GSYM/StringTable.h" |
| 18 | #include <tuple> |
| 19 | #include <vector> |
| 20 | |
| 21 | namespace llvm { |
| 22 | class raw_ostream; |
| 23 | namespace gsym { |
| 24 | |
Olivier Deprez | f4ef2d0 | 2021-04-20 13:36:24 +0200 | [diff] [blame] | 25 | class GsymReader; |
| 26 | /// Function information in GSYM files encodes information for one contiguous |
| 27 | /// address range. If a function has discontiguous address ranges, they will |
| 28 | /// need to be encoded using multiple FunctionInfo objects. |
| 29 | /// |
| 30 | /// ENCODING |
| 31 | /// |
| 32 | /// The function information gets the function start address as an argument |
| 33 | /// to the FunctionInfo::decode(...) function. This information is calculated |
| 34 | /// from the GSYM header and an address offset from the GSYM address offsets |
| 35 | /// table. The encoded FunctionInfo information must be aligned to a 4 byte |
| 36 | /// boundary. |
| 37 | /// |
| 38 | /// The encoded data for a FunctionInfo starts with fixed data that all |
| 39 | /// function info objects have: |
| 40 | /// |
| 41 | /// ENCODING NAME DESCRIPTION |
| 42 | /// ========= =========== ==================================================== |
| 43 | /// uint32_t Size The size in bytes of this function. |
| 44 | /// uint32_t Name The string table offset of the function name. |
| 45 | /// |
| 46 | /// The optional data in a FunctionInfo object follows this fixed information |
| 47 | /// and consists of a stream of tuples that consist of: |
| 48 | /// |
| 49 | /// ENCODING NAME DESCRIPTION |
| 50 | /// ========= =========== ==================================================== |
| 51 | /// uint32_t InfoType An "InfoType" enumeration that describes the type |
| 52 | /// of optional data that is encoded. |
| 53 | /// uint32_t InfoLength The size in bytes of the encoded data that |
| 54 | /// immediately follows this length if this value is |
| 55 | /// greater than zero. |
| 56 | /// uint8_t[] InfoData Encoded bytes that represent the data for the |
| 57 | /// "InfoType". These bytes are only present if |
| 58 | /// "InfoLength" is greater than zero. |
| 59 | /// |
| 60 | /// The "InfoType" is an enumeration: |
| 61 | /// |
| 62 | /// enum InfoType { |
| 63 | /// EndOfList = 0u, |
| 64 | /// LineTableInfo = 1u, |
| 65 | /// InlineInfo = 2u |
| 66 | /// }; |
| 67 | /// |
| 68 | /// This stream of tuples is terminated by a "InfoType" whose value is |
| 69 | /// InfoType::EndOfList and a zero for "InfoLength". This signifies the end of |
| 70 | /// the optional information list. This format allows us to add new optional |
| 71 | /// information data to a FunctionInfo object over time and allows older |
| 72 | /// clients to still parse the format and skip over any data that they don't |
| 73 | /// understand or want to parse. |
| 74 | /// |
| 75 | /// So the function information encoding essientially looks like: |
| 76 | /// |
| 77 | /// struct { |
| 78 | /// uint32_t Size; |
| 79 | /// uint32_t Name; |
| 80 | /// struct { |
| 81 | /// uint32_t InfoType; |
| 82 | /// uint32_t InfoLength; |
| 83 | /// uint8_t InfoData[InfoLength]; |
| 84 | /// }[N]; |
| 85 | /// } |
| 86 | /// |
| 87 | /// Where "N" is the number of tuples. |
Andrew Walbran | 3d2c197 | 2020-04-07 12:24:26 +0100 | [diff] [blame] | 88 | struct FunctionInfo { |
| 89 | AddressRange Range; |
| 90 | uint32_t Name; ///< String table offset in the string table. |
Olivier Deprez | f4ef2d0 | 2021-04-20 13:36:24 +0200 | [diff] [blame] | 91 | llvm::Optional<LineTable> OptLineTable; |
| 92 | llvm::Optional<InlineInfo> Inline; |
Andrew Walbran | 3d2c197 | 2020-04-07 12:24:26 +0100 | [diff] [blame] | 93 | |
| 94 | FunctionInfo(uint64_t Addr = 0, uint64_t Size = 0, uint32_t N = 0) |
| 95 | : Range(Addr, Addr + Size), Name(N) {} |
| 96 | |
Olivier Deprez | f4ef2d0 | 2021-04-20 13:36:24 +0200 | [diff] [blame] | 97 | /// Query if a FunctionInfo has rich debug info. |
| 98 | /// |
| 99 | /// \returns A bool that indicates if this object has something else than |
| 100 | /// range and name. When converting information from a symbol table and from |
| 101 | /// debug info, we might end up with multiple FunctionInfo objects for the |
| 102 | /// same range and we need to be able to tell which one is the better object |
| 103 | /// to use. |
Andrew Walbran | 3d2c197 | 2020-04-07 12:24:26 +0100 | [diff] [blame] | 104 | bool hasRichInfo() const { |
Olivier Deprez | f4ef2d0 | 2021-04-20 13:36:24 +0200 | [diff] [blame] | 105 | return OptLineTable.hasValue() || Inline.hasValue(); |
Andrew Walbran | 3d2c197 | 2020-04-07 12:24:26 +0100 | [diff] [blame] | 106 | } |
| 107 | |
Olivier Deprez | f4ef2d0 | 2021-04-20 13:36:24 +0200 | [diff] [blame] | 108 | /// Query if a FunctionInfo object is valid. |
| 109 | /// |
| 110 | /// Address and size can be zero and there can be no line entries for a |
| 111 | /// symbol so the only indication this entry is valid is if the name is |
| 112 | /// not zero. This can happen when extracting information from symbol |
| 113 | /// tables that do not encode symbol sizes. In that case only the |
| 114 | /// address and name will be filled in. |
| 115 | /// |
| 116 | /// \returns A boolean indicating if this FunctionInfo is valid. |
Andrew Walbran | 3d2c197 | 2020-04-07 12:24:26 +0100 | [diff] [blame] | 117 | bool isValid() const { |
Andrew Walbran | 3d2c197 | 2020-04-07 12:24:26 +0100 | [diff] [blame] | 118 | return Name != 0; |
| 119 | } |
| 120 | |
Olivier Deprez | f4ef2d0 | 2021-04-20 13:36:24 +0200 | [diff] [blame] | 121 | /// Decode an object from a binary data stream. |
| 122 | /// |
| 123 | /// \param Data The binary stream to read the data from. This object must |
| 124 | /// have the data for the object starting at offset zero. The data |
| 125 | /// can contain more data than needed. |
| 126 | /// |
| 127 | /// \param BaseAddr The FunctionInfo's start address and will be used as the |
| 128 | /// base address when decoding any contained information like the line table |
| 129 | /// and the inline info. |
| 130 | /// |
| 131 | /// \returns An FunctionInfo or an error describing the issue that was |
| 132 | /// encountered during decoding. |
| 133 | static llvm::Expected<FunctionInfo> decode(DataExtractor &Data, |
| 134 | uint64_t BaseAddr); |
| 135 | |
| 136 | /// Encode this object into FileWriter stream. |
| 137 | /// |
| 138 | /// \param O The binary stream to write the data to at the current file |
| 139 | /// position. |
| 140 | /// |
| 141 | /// \returns An error object that indicates failure or the offset of the |
| 142 | /// function info that was successfully written into the stream. |
| 143 | llvm::Expected<uint64_t> encode(FileWriter &O) const; |
| 144 | |
| 145 | |
| 146 | /// Lookup an address within a FunctionInfo object's data stream. |
| 147 | /// |
| 148 | /// Instead of decoding an entire FunctionInfo object when doing lookups, |
| 149 | /// we can decode only the information we need from the FunctionInfo's data |
| 150 | /// for the specific address. The lookup result information is returned as |
| 151 | /// a LookupResult. |
| 152 | /// |
| 153 | /// \param Data The binary stream to read the data from. This object must |
| 154 | /// have the data for the object starting at offset zero. The data |
| 155 | /// can contain more data than needed. |
| 156 | /// |
| 157 | /// \param GR The GSYM reader that contains the string and file table that |
| 158 | /// will be used to fill in information in the returned result. |
| 159 | /// |
| 160 | /// \param FuncAddr The function start address decoded from the GsymReader. |
| 161 | /// |
| 162 | /// \param Addr The address to lookup. |
| 163 | /// |
| 164 | /// \returns An LookupResult or an error describing the issue that was |
| 165 | /// encountered during decoding. An error should only be returned if the |
| 166 | /// address is not contained in the FunctionInfo or if the data is corrupted. |
| 167 | static llvm::Expected<LookupResult> lookup(DataExtractor &Data, |
| 168 | const GsymReader &GR, |
| 169 | uint64_t FuncAddr, |
| 170 | uint64_t Addr); |
| 171 | |
Andrew Walbran | 3d2c197 | 2020-04-07 12:24:26 +0100 | [diff] [blame] | 172 | uint64_t startAddress() const { return Range.Start; } |
| 173 | uint64_t endAddress() const { return Range.End; } |
| 174 | uint64_t size() const { return Range.size(); } |
| 175 | void setStartAddress(uint64_t Addr) { Range.Start = Addr; } |
| 176 | void setEndAddress(uint64_t Addr) { Range.End = Addr; } |
| 177 | void setSize(uint64_t Size) { Range.End = Range.Start + Size; } |
| 178 | |
| 179 | void clear() { |
| 180 | Range = {0, 0}; |
| 181 | Name = 0; |
Olivier Deprez | f4ef2d0 | 2021-04-20 13:36:24 +0200 | [diff] [blame] | 182 | OptLineTable = None; |
| 183 | Inline = None; |
Andrew Walbran | 3d2c197 | 2020-04-07 12:24:26 +0100 | [diff] [blame] | 184 | } |
| 185 | }; |
| 186 | |
| 187 | inline bool operator==(const FunctionInfo &LHS, const FunctionInfo &RHS) { |
| 188 | return LHS.Range == RHS.Range && LHS.Name == RHS.Name && |
Olivier Deprez | f4ef2d0 | 2021-04-20 13:36:24 +0200 | [diff] [blame] | 189 | LHS.OptLineTable == RHS.OptLineTable && LHS.Inline == RHS.Inline; |
Andrew Walbran | 3d2c197 | 2020-04-07 12:24:26 +0100 | [diff] [blame] | 190 | } |
| 191 | inline bool operator!=(const FunctionInfo &LHS, const FunctionInfo &RHS) { |
| 192 | return !(LHS == RHS); |
| 193 | } |
| 194 | /// This sorting will order things consistently by address range first, but then |
| 195 | /// followed by inlining being valid and line tables. We might end up with a |
| 196 | /// FunctionInfo from debug info that will have the same range as one from the |
| 197 | /// symbol table, but we want to quickly be able to sort and use the best version |
| 198 | /// when creating the final GSYM file. |
| 199 | inline bool operator<(const FunctionInfo &LHS, const FunctionInfo &RHS) { |
| 200 | // First sort by address range |
| 201 | if (LHS.Range != RHS.Range) |
| 202 | return LHS.Range < RHS.Range; |
| 203 | |
| 204 | // Then sort by inline |
Olivier Deprez | f4ef2d0 | 2021-04-20 13:36:24 +0200 | [diff] [blame] | 205 | if (LHS.Inline.hasValue() != RHS.Inline.hasValue()) |
| 206 | return RHS.Inline.hasValue(); |
Andrew Walbran | 3d2c197 | 2020-04-07 12:24:26 +0100 | [diff] [blame] | 207 | |
Olivier Deprez | f4ef2d0 | 2021-04-20 13:36:24 +0200 | [diff] [blame] | 208 | return LHS.OptLineTable < RHS.OptLineTable; |
Andrew Walbran | 3d2c197 | 2020-04-07 12:24:26 +0100 | [diff] [blame] | 209 | } |
| 210 | |
| 211 | raw_ostream &operator<<(raw_ostream &OS, const FunctionInfo &R); |
| 212 | |
| 213 | } // namespace gsym |
| 214 | } // namespace llvm |
| 215 | |
| 216 | #endif // #ifndef LLVM_DEBUGINFO_GSYM_FUNCTIONINFO_H |