1*2a6b7db3Sskrll // binary.cc -- binary input files for gold
2*2a6b7db3Sskrll 
3*2a6b7db3Sskrll // Copyright 2008 Free Software Foundation, Inc.
4*2a6b7db3Sskrll // Written by Ian Lance Taylor <iant@google.com>.
5*2a6b7db3Sskrll 
6*2a6b7db3Sskrll // This file is part of gold.
7*2a6b7db3Sskrll 
8*2a6b7db3Sskrll // This program is free software; you can redistribute it and/or modify
9*2a6b7db3Sskrll // it under the terms of the GNU General Public License as published by
10*2a6b7db3Sskrll // the Free Software Foundation; either version 3 of the License, or
11*2a6b7db3Sskrll // (at your option) any later version.
12*2a6b7db3Sskrll 
13*2a6b7db3Sskrll // This program is distributed in the hope that it will be useful,
14*2a6b7db3Sskrll // but WITHOUT ANY WARRANTY; without even the implied warranty of
15*2a6b7db3Sskrll // MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
16*2a6b7db3Sskrll // GNU General Public License for more details.
17*2a6b7db3Sskrll 
18*2a6b7db3Sskrll // You should have received a copy of the GNU General Public License
19*2a6b7db3Sskrll // along with this program; if not, write to the Free Software
20*2a6b7db3Sskrll // Foundation, Inc., 51 Franklin Street - Fifth Floor, Boston,
21*2a6b7db3Sskrll // MA 02110-1301, USA.
22*2a6b7db3Sskrll 
23*2a6b7db3Sskrll #include "gold.h"
24*2a6b7db3Sskrll 
25*2a6b7db3Sskrll #include <cerrno>
26*2a6b7db3Sskrll #include <cstring>
27*2a6b7db3Sskrll #include "safe-ctype.h"
28*2a6b7db3Sskrll 
29*2a6b7db3Sskrll #include "elfcpp.h"
30*2a6b7db3Sskrll #include "stringpool.h"
31*2a6b7db3Sskrll #include "fileread.h"
32*2a6b7db3Sskrll #include "output.h"
33*2a6b7db3Sskrll #include "binary.h"
34*2a6b7db3Sskrll 
35*2a6b7db3Sskrll // Support for reading binary files as input.  These become blobs in
36*2a6b7db3Sskrll // the final output.  These files are treated as though they have a
37*2a6b7db3Sskrll // single .data section and define three symbols:
38*2a6b7db3Sskrll // _binary_FILENAME_start, _binary_FILENAME_end, _binary_FILENAME_end.
39*2a6b7db3Sskrll // The FILENAME is the name of the input file, with any
40*2a6b7db3Sskrll // non-alphanumeric character changed to an underscore.
41*2a6b7db3Sskrll 
42*2a6b7db3Sskrll // We implement this by creating an ELF file in memory.
43*2a6b7db3Sskrll 
44*2a6b7db3Sskrll namespace gold
45*2a6b7db3Sskrll {
46*2a6b7db3Sskrll 
47*2a6b7db3Sskrll // class Binary_to_elf.
48*2a6b7db3Sskrll 
49*2a6b7db3Sskrll Binary_to_elf::Binary_to_elf(elfcpp::EM machine, int size, bool big_endian,
50*2a6b7db3Sskrll 			     const std::string& filename)
51*2a6b7db3Sskrll   : elf_machine_(machine), size_(size), big_endian_(big_endian),
52*2a6b7db3Sskrll     filename_(filename), data_(NULL), filesize_(0)
53*2a6b7db3Sskrll {
54*2a6b7db3Sskrll }
55*2a6b7db3Sskrll 
56*2a6b7db3Sskrll Binary_to_elf::~Binary_to_elf()
57*2a6b7db3Sskrll {
58*2a6b7db3Sskrll   if (this->data_ != NULL)
59*2a6b7db3Sskrll     delete[] this->data_;
60*2a6b7db3Sskrll }
61*2a6b7db3Sskrll 
62*2a6b7db3Sskrll // Given FILENAME, create a buffer which looks like an ELF file with
63*2a6b7db3Sskrll // the contents of FILENAME as the contents of the only section.  The
64*2a6b7db3Sskrll // TASK parameters is mainly for debugging, and records who holds
65*2a6b7db3Sskrll // locks.
66*2a6b7db3Sskrll 
67*2a6b7db3Sskrll bool
68*2a6b7db3Sskrll Binary_to_elf::convert(const Task* task)
69*2a6b7db3Sskrll {
70*2a6b7db3Sskrll   if (this->size_ == 32)
71*2a6b7db3Sskrll     {
72*2a6b7db3Sskrll       if (!this->big_endian_)
73*2a6b7db3Sskrll 	{
74*2a6b7db3Sskrll #ifdef HAVE_TARGET_32_LITTLE
75*2a6b7db3Sskrll 	  return this->sized_convert<32, false>(task);
76*2a6b7db3Sskrll #else
77*2a6b7db3Sskrll 	  gold_unreachable();
78*2a6b7db3Sskrll #endif
79*2a6b7db3Sskrll 	}
80*2a6b7db3Sskrll       else
81*2a6b7db3Sskrll 	{
82*2a6b7db3Sskrll #ifdef HAVE_TARGET_32_BIG
83*2a6b7db3Sskrll 	  return this->sized_convert<32, true>(task);
84*2a6b7db3Sskrll #else
85*2a6b7db3Sskrll 	  gold_unreachable();
86*2a6b7db3Sskrll #endif
87*2a6b7db3Sskrll 	}
88*2a6b7db3Sskrll     }
89*2a6b7db3Sskrll   else if (this->size_ == 64)
90*2a6b7db3Sskrll     {
91*2a6b7db3Sskrll       if (!this->big_endian_)
92*2a6b7db3Sskrll 	{
93*2a6b7db3Sskrll #ifdef HAVE_TARGET_64_LITTLE
94*2a6b7db3Sskrll 	  return this->sized_convert<64, false>(task);
95*2a6b7db3Sskrll #else
96*2a6b7db3Sskrll 	  gold_unreachable();
97*2a6b7db3Sskrll #endif
98*2a6b7db3Sskrll 	}
99*2a6b7db3Sskrll       else
100*2a6b7db3Sskrll 	{
101*2a6b7db3Sskrll #ifdef HAVE_TARGET_64_BIG
102*2a6b7db3Sskrll 	  return this->sized_convert<64, true>(task);
103*2a6b7db3Sskrll #else
104*2a6b7db3Sskrll 	  gold_unreachable();
105*2a6b7db3Sskrll #endif
106*2a6b7db3Sskrll 	}
107*2a6b7db3Sskrll     }
108*2a6b7db3Sskrll   else
109*2a6b7db3Sskrll     gold_unreachable();
110*2a6b7db3Sskrll }
111*2a6b7db3Sskrll 
112*2a6b7db3Sskrll // We are going to create:
113*2a6b7db3Sskrll // * The ELF file header.
114*2a6b7db3Sskrll // * Five sections: null section, .data, .symtab, .strtab, .shstrtab
115*2a6b7db3Sskrll // * The contents of the file.
116*2a6b7db3Sskrll // * Four symbols: null, begin, end, size.
117*2a6b7db3Sskrll // * Three symbol names.
118*2a6b7db3Sskrll // * Four section names.
119*2a6b7db3Sskrll 
120*2a6b7db3Sskrll template<int size, bool big_endian>
121*2a6b7db3Sskrll bool
122*2a6b7db3Sskrll Binary_to_elf::sized_convert(const Task* task)
123*2a6b7db3Sskrll {
124*2a6b7db3Sskrll   // Read the input file.
125*2a6b7db3Sskrll 
126*2a6b7db3Sskrll   File_read f;
127*2a6b7db3Sskrll   if (!f.open(task, this->filename_))
128*2a6b7db3Sskrll     {
129*2a6b7db3Sskrll       gold_error(_("cannot open %s: %s:"), this->filename_.c_str(),
130*2a6b7db3Sskrll 		 strerror(errno));
131*2a6b7db3Sskrll       return false;
132*2a6b7db3Sskrll     }
133*2a6b7db3Sskrll 
134*2a6b7db3Sskrll   section_size_type filesize = convert_to_section_size_type(f.filesize());
135*2a6b7db3Sskrll   const unsigned char* fileview = f.get_view(0, 0, filesize, false, false);
136*2a6b7db3Sskrll 
137*2a6b7db3Sskrll   unsigned int align;
138*2a6b7db3Sskrll   if (size == 32)
139*2a6b7db3Sskrll     align = 4;
140*2a6b7db3Sskrll   else if (size == 64)
141*2a6b7db3Sskrll     align = 8;
142*2a6b7db3Sskrll   else
143*2a6b7db3Sskrll     gold_unreachable();
144*2a6b7db3Sskrll   section_size_type aligned_filesize = align_address(filesize, align);
145*2a6b7db3Sskrll 
146*2a6b7db3Sskrll   // Build the stringpool for the symbol table.
147*2a6b7db3Sskrll 
148*2a6b7db3Sskrll   std::string mangled_name = this->filename_;
149*2a6b7db3Sskrll   for (std::string::iterator p = mangled_name.begin();
150*2a6b7db3Sskrll        p != mangled_name.end();
151*2a6b7db3Sskrll        ++p)
152*2a6b7db3Sskrll     if (!ISALNUM(*p))
153*2a6b7db3Sskrll       *p = '_';
154*2a6b7db3Sskrll   mangled_name = "_binary_" + mangled_name;
155*2a6b7db3Sskrll   std::string start_symbol_name = mangled_name + "_start";
156*2a6b7db3Sskrll   std::string end_symbol_name = mangled_name + "_end";
157*2a6b7db3Sskrll   std::string size_symbol_name = mangled_name + "_size";
158*2a6b7db3Sskrll 
159*2a6b7db3Sskrll   Stringpool strtab;
160*2a6b7db3Sskrll   strtab.add(start_symbol_name.c_str(), false, NULL);
161*2a6b7db3Sskrll   strtab.add(end_symbol_name.c_str(), false, NULL);
162*2a6b7db3Sskrll   strtab.add(size_symbol_name.c_str(), false, NULL);
163*2a6b7db3Sskrll   strtab.set_string_offsets();
164*2a6b7db3Sskrll 
165*2a6b7db3Sskrll   // Build the stringpool for the section name table.
166*2a6b7db3Sskrll 
167*2a6b7db3Sskrll   Stringpool shstrtab;
168*2a6b7db3Sskrll   shstrtab.add(".data", false, NULL);
169*2a6b7db3Sskrll   shstrtab.add(".symtab", false, NULL);
170*2a6b7db3Sskrll   shstrtab.add(".strtab", false, NULL);
171*2a6b7db3Sskrll   shstrtab.add(".shstrtab", false, NULL);
172*2a6b7db3Sskrll   shstrtab.set_string_offsets();
173*2a6b7db3Sskrll 
174*2a6b7db3Sskrll   // Work out the size of the generated file, and the offsets of the
175*2a6b7db3Sskrll   // various sections, and allocate a buffer.
176*2a6b7db3Sskrll 
177*2a6b7db3Sskrll   const int sym_size = elfcpp::Elf_sizes<size>::sym_size;
178*2a6b7db3Sskrll 
179*2a6b7db3Sskrll   size_t output_size = (elfcpp::Elf_sizes<size>::ehdr_size
180*2a6b7db3Sskrll 			+ 5 * elfcpp::Elf_sizes<size>::shdr_size);
181*2a6b7db3Sskrll   size_t data_offset = output_size;
182*2a6b7db3Sskrll   output_size += aligned_filesize;
183*2a6b7db3Sskrll   size_t symtab_offset = output_size;
184*2a6b7db3Sskrll   output_size += 4 * sym_size;
185*2a6b7db3Sskrll   size_t strtab_offset = output_size;
186*2a6b7db3Sskrll   output_size += strtab.get_strtab_size();
187*2a6b7db3Sskrll   size_t shstrtab_offset = output_size;
188*2a6b7db3Sskrll   output_size += shstrtab.get_strtab_size();
189*2a6b7db3Sskrll 
190*2a6b7db3Sskrll   unsigned char* buffer = new unsigned char[output_size];
191*2a6b7db3Sskrll 
192*2a6b7db3Sskrll   // Write out the data.
193*2a6b7db3Sskrll 
194*2a6b7db3Sskrll   unsigned char* pout = buffer;
195*2a6b7db3Sskrll 
196*2a6b7db3Sskrll   this->write_file_header<size, big_endian>(&pout);
197*2a6b7db3Sskrll 
198*2a6b7db3Sskrll   this->write_section_header<size, big_endian>("", &shstrtab, elfcpp::SHT_NULL,
199*2a6b7db3Sskrll 					       0, 0, 0, 0, 0,
200*2a6b7db3Sskrll 					       0, 0, &pout);
201*2a6b7db3Sskrll   // Having the section be named ".data", having it be writable, and
202*2a6b7db3Sskrll   // giving it an alignment of 1 is because the GNU linker does it
203*2a6b7db3Sskrll   // that way, and existing linker script expect it.
204*2a6b7db3Sskrll   this->write_section_header<size, big_endian>(".data", &shstrtab,
205*2a6b7db3Sskrll 					       elfcpp::SHT_PROGBITS,
206*2a6b7db3Sskrll 					       (elfcpp::SHF_ALLOC
207*2a6b7db3Sskrll 						| elfcpp::SHF_WRITE),
208*2a6b7db3Sskrll 					       data_offset,
209*2a6b7db3Sskrll 					       filesize, 0, 0,
210*2a6b7db3Sskrll 					       1, 0, &pout);
211*2a6b7db3Sskrll   this->write_section_header<size, big_endian>(".symtab", &shstrtab,
212*2a6b7db3Sskrll 					       elfcpp::SHT_SYMTAB,
213*2a6b7db3Sskrll 					       0, symtab_offset, 4 * sym_size,
214*2a6b7db3Sskrll 					       3, 1, align, sym_size, &pout);
215*2a6b7db3Sskrll   this->write_section_header<size, big_endian>(".strtab", &shstrtab,
216*2a6b7db3Sskrll 					       elfcpp::SHT_STRTAB,
217*2a6b7db3Sskrll 					       0, strtab_offset,
218*2a6b7db3Sskrll 					       strtab.get_strtab_size(),
219*2a6b7db3Sskrll 					       0, 0, 1, 0, &pout);
220*2a6b7db3Sskrll   this->write_section_header<size, big_endian>(".shstrtab", &shstrtab,
221*2a6b7db3Sskrll 					       elfcpp::SHT_STRTAB,
222*2a6b7db3Sskrll 					       0, shstrtab_offset,
223*2a6b7db3Sskrll 					       shstrtab.get_strtab_size(),
224*2a6b7db3Sskrll 					       0, 0, 1, 0, &pout);
225*2a6b7db3Sskrll 
226*2a6b7db3Sskrll   memcpy(pout, fileview, filesize);
227*2a6b7db3Sskrll   pout += filesize;
228*2a6b7db3Sskrll   memset(pout, 0, aligned_filesize - filesize);
229*2a6b7db3Sskrll   pout += aligned_filesize - filesize;
230*2a6b7db3Sskrll 
231*2a6b7db3Sskrll   this->write_symbol<size, big_endian>("", &strtab, 0, 0, &pout);
232*2a6b7db3Sskrll   this->write_symbol<size, big_endian>(start_symbol_name, &strtab, 0, 1,
233*2a6b7db3Sskrll 				       &pout);
234*2a6b7db3Sskrll   this->write_symbol<size, big_endian>(end_symbol_name, &strtab, filesize, 1,
235*2a6b7db3Sskrll 				       &pout);
236*2a6b7db3Sskrll   this->write_symbol<size, big_endian>(size_symbol_name, &strtab, filesize,
237*2a6b7db3Sskrll 				       elfcpp::SHN_ABS, &pout);
238*2a6b7db3Sskrll 
239*2a6b7db3Sskrll   strtab.write_to_buffer(pout, strtab.get_strtab_size());
240*2a6b7db3Sskrll   pout += strtab.get_strtab_size();
241*2a6b7db3Sskrll 
242*2a6b7db3Sskrll   shstrtab.write_to_buffer(pout, shstrtab.get_strtab_size());
243*2a6b7db3Sskrll   pout += shstrtab.get_strtab_size();
244*2a6b7db3Sskrll 
245*2a6b7db3Sskrll   gold_assert(static_cast<size_t>(pout - buffer) == output_size);
246*2a6b7db3Sskrll 
247*2a6b7db3Sskrll   this->data_ = buffer;
248*2a6b7db3Sskrll   this->filesize_ = output_size;
249*2a6b7db3Sskrll 
250*2a6b7db3Sskrll   f.unlock(task);
251*2a6b7db3Sskrll 
252*2a6b7db3Sskrll   return true;
253*2a6b7db3Sskrll }
254*2a6b7db3Sskrll 
255*2a6b7db3Sskrll // Write out the file header.
256*2a6b7db3Sskrll 
257*2a6b7db3Sskrll template<int size, bool big_endian>
258*2a6b7db3Sskrll void
259*2a6b7db3Sskrll Binary_to_elf::write_file_header(unsigned char** ppout)
260*2a6b7db3Sskrll {
261*2a6b7db3Sskrll   elfcpp::Ehdr_write<size, big_endian> oehdr(*ppout);
262*2a6b7db3Sskrll 
263*2a6b7db3Sskrll   unsigned char e_ident[elfcpp::EI_NIDENT];
264*2a6b7db3Sskrll   memset(e_ident, 0, elfcpp::EI_NIDENT);
265*2a6b7db3Sskrll   e_ident[elfcpp::EI_MAG0] = elfcpp::ELFMAG0;
266*2a6b7db3Sskrll   e_ident[elfcpp::EI_MAG1] = elfcpp::ELFMAG1;
267*2a6b7db3Sskrll   e_ident[elfcpp::EI_MAG2] = elfcpp::ELFMAG2;
268*2a6b7db3Sskrll   e_ident[elfcpp::EI_MAG3] = elfcpp::ELFMAG3;
269*2a6b7db3Sskrll   if (size == 32)
270*2a6b7db3Sskrll     e_ident[elfcpp::EI_CLASS] = elfcpp::ELFCLASS32;
271*2a6b7db3Sskrll   else if (size == 64)
272*2a6b7db3Sskrll     e_ident[elfcpp::EI_CLASS] = elfcpp::ELFCLASS64;
273*2a6b7db3Sskrll   else
274*2a6b7db3Sskrll     gold_unreachable();
275*2a6b7db3Sskrll   e_ident[elfcpp::EI_DATA] = (big_endian
276*2a6b7db3Sskrll 			      ? elfcpp::ELFDATA2MSB
277*2a6b7db3Sskrll 			      : elfcpp::ELFDATA2LSB);
278*2a6b7db3Sskrll   e_ident[elfcpp::EI_VERSION] = elfcpp::EV_CURRENT;
279*2a6b7db3Sskrll   oehdr.put_e_ident(e_ident);
280*2a6b7db3Sskrll 
281*2a6b7db3Sskrll   oehdr.put_e_type(elfcpp::ET_REL);
282*2a6b7db3Sskrll   oehdr.put_e_machine(this->elf_machine_);
283*2a6b7db3Sskrll   oehdr.put_e_version(elfcpp::EV_CURRENT);
284*2a6b7db3Sskrll   oehdr.put_e_entry(0);
285*2a6b7db3Sskrll   oehdr.put_e_phoff(0);
286*2a6b7db3Sskrll   oehdr.put_e_shoff(elfcpp::Elf_sizes<size>::ehdr_size);
287*2a6b7db3Sskrll   oehdr.put_e_flags(0);
288*2a6b7db3Sskrll   oehdr.put_e_ehsize(elfcpp::Elf_sizes<size>::ehdr_size);
289*2a6b7db3Sskrll   oehdr.put_e_phentsize(0);
290*2a6b7db3Sskrll   oehdr.put_e_phnum(0);
291*2a6b7db3Sskrll   oehdr.put_e_shentsize(elfcpp::Elf_sizes<size>::shdr_size);
292*2a6b7db3Sskrll   oehdr.put_e_shnum(5);
293*2a6b7db3Sskrll   oehdr.put_e_shstrndx(4);
294*2a6b7db3Sskrll 
295*2a6b7db3Sskrll   *ppout += elfcpp::Elf_sizes<size>::ehdr_size;
296*2a6b7db3Sskrll }
297*2a6b7db3Sskrll 
298*2a6b7db3Sskrll // Write out a section header.
299*2a6b7db3Sskrll 
300*2a6b7db3Sskrll template<int size, bool big_endian>
301*2a6b7db3Sskrll void
302*2a6b7db3Sskrll Binary_to_elf::write_section_header(
303*2a6b7db3Sskrll     const char* name,
304*2a6b7db3Sskrll     const Stringpool* shstrtab,
305*2a6b7db3Sskrll     elfcpp::SHT type,
306*2a6b7db3Sskrll     unsigned int flags,
307*2a6b7db3Sskrll     section_size_type offset,
308*2a6b7db3Sskrll     section_size_type section_size,
309*2a6b7db3Sskrll     unsigned int link,
310*2a6b7db3Sskrll     unsigned int info,
311*2a6b7db3Sskrll     unsigned int addralign,
312*2a6b7db3Sskrll     unsigned int entsize,
313*2a6b7db3Sskrll     unsigned char** ppout)
314*2a6b7db3Sskrll {
315*2a6b7db3Sskrll   elfcpp::Shdr_write<size, big_endian> oshdr(*ppout);
316*2a6b7db3Sskrll 
317*2a6b7db3Sskrll   oshdr.put_sh_name(*name == '\0' ? 0 : shstrtab->get_offset(name));
318*2a6b7db3Sskrll   oshdr.put_sh_type(type);
319*2a6b7db3Sskrll   oshdr.put_sh_flags(flags);
320*2a6b7db3Sskrll   oshdr.put_sh_addr(0);
321*2a6b7db3Sskrll   oshdr.put_sh_offset(offset);
322*2a6b7db3Sskrll   oshdr.put_sh_size(section_size);
323*2a6b7db3Sskrll   oshdr.put_sh_link(link);
324*2a6b7db3Sskrll   oshdr.put_sh_info(info);
325*2a6b7db3Sskrll   oshdr.put_sh_addralign(addralign);
326*2a6b7db3Sskrll   oshdr.put_sh_entsize(entsize);
327*2a6b7db3Sskrll 
328*2a6b7db3Sskrll   *ppout += elfcpp::Elf_sizes<size>::shdr_size;
329*2a6b7db3Sskrll }
330*2a6b7db3Sskrll 
331*2a6b7db3Sskrll // Write out a symbol.
332*2a6b7db3Sskrll 
333*2a6b7db3Sskrll template<int size, bool big_endian>
334*2a6b7db3Sskrll void
335*2a6b7db3Sskrll Binary_to_elf::write_symbol(
336*2a6b7db3Sskrll     const std::string& name,
337*2a6b7db3Sskrll     const Stringpool* strtab,
338*2a6b7db3Sskrll     section_size_type value,
339*2a6b7db3Sskrll     unsigned int shndx,
340*2a6b7db3Sskrll     unsigned char** ppout)
341*2a6b7db3Sskrll {
342*2a6b7db3Sskrll   unsigned char* pout = *ppout;
343*2a6b7db3Sskrll 
344*2a6b7db3Sskrll   elfcpp::Sym_write<size, big_endian> osym(pout);
345*2a6b7db3Sskrll   osym.put_st_name(name.empty() ? 0 : strtab->get_offset(name.c_str()));
346*2a6b7db3Sskrll   osym.put_st_value(value);
347*2a6b7db3Sskrll   osym.put_st_size(0);
348*2a6b7db3Sskrll   osym.put_st_info(name.empty() ? elfcpp::STB_LOCAL : elfcpp::STB_GLOBAL,
349*2a6b7db3Sskrll 		   elfcpp::STT_NOTYPE);
350*2a6b7db3Sskrll   osym.put_st_other(elfcpp::STV_DEFAULT, 0);
351*2a6b7db3Sskrll   osym.put_st_shndx(shndx);
352*2a6b7db3Sskrll 
353*2a6b7db3Sskrll   *ppout += elfcpp::Elf_sizes<size>::sym_size;
354*2a6b7db3Sskrll }
355*2a6b7db3Sskrll 
356*2a6b7db3Sskrll } // End namespace gold.
357