Moving git host yet again - personal-pages/mirror-world - A parallel world, population: one soul

diff options

author	Aoi Koizumi <novaburst@kalli.st>	2022-04-29 10:06:13 -0300
committer	Aoi Koizumi <novaburst@kalli.st>	2022-04-29 10:06:13 -0300
commit	09bb2854478e552523c15de6f80c9f92d71376a3 (patch)
tree	687a026d177e28e25f47ba7c200164b7b0b67ef6 /gmi/a373033e.txt
parent	d13dfdbdd6f8f10552d744cdb137390e7a8dabf6 (diff)
download	mirror-world-09bb2854478e552523c15de6f80c9f92d71376a3.tar.gz

Moving git host yet again galaxy

Diffstat (limited to 'gmi/a373033e.txt')

-rw-r--r--

gmi/a373033e.txt

1 files changed, 34 insertions, 0 deletions

<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.01//EN" "http://www.w3.org/TR/html4/strict.dtd"> <html> <head> <meta http-equiv="content-type" content="text/html; charset=UTF-8"> <title>Mu - linux/bootstrap/012elf.cc</title> <meta name="Generator" content="Vim/8.1"> <meta name="plugin-version" content="vim8.1_v1"> <meta name="syntax" content="cpp"> <meta name="settings" content="number_lines,use_css,pre_wrap,no_foldcolumn,expand_tabs,line_ids,prevent_copy="> <meta name="colorscheme" content="minimal-dark"> <style type="text/css">  </style> <script type='text/javascript'>  </script> </head> <body onload='JumpToLine();'> <a href='https://github.com/akkartik/mu/blob/main/linux/bootstrap/012elf.cc'>https://github.com/akkartik/mu/blob/main/linux/bootstrap/012elf.cc</a> <pre id='vimCodeElement'> 1 //: Loading SubX programs from ELF binaries. 2 //: This will allow us to run them natively on a Linux kernel. 3 //: Based on <a href="https://github.com/kragen/stoneknifeforth/blob/702d2ebe1b/386.c">https://github.com/kragen/stoneknifeforth/blob/702d2ebe1b/386.c</a> 4 5 :(before "End Main") 6 assert(argc > 1); 7 if (<a href='001help.cc.html#L98'>is_equal</a>(argv[1], "run")) { 8 // Outside of tests, traces must be explicitly requested. 9 if (<a href='003trace.cc.html#L446'>Trace_file</a>.is_open()) <a href='003trace.cc.html#L72'>Trace_stream</a> = new trace_stream; 10 <a href='003trace.cc.html#L96'>trace</a>(2, "run") << "=== Starting to run" << end(); 11 if (argc <= 2) { 12 <a href='003trace.cc.html#L226'>raise</a> << "Not enough arguments provided.\n" << <a href='003trace.cc.html#L173'>die</a>(); 13 } 14 <a href='000organization.cc.html#L150'>reset</a>(); 15 cerr << std::hex; 16 <a href='012elf.cc.html#L24'>load_elf</a>(argv[2], argc, argv); 17 while (<a href='010vm.cc.html#L26'>EIP</a> < <a href='010vm.cc.html#L166'>End_of_program</a>) // weak final-gasp termination check 18 <a href='010vm.cc.html#L277'>run_one_instruction</a>(); 19 <a href='003trace.cc.html#L226'>raise</a> << "executed past end of the world: " << <a href='010vm.cc.html#L26'>EIP</a> << " vs " << <a href='010vm.cc.html#L166'>End_of_program</a> << '\n' << end(); 20 return 1; 21 } 22 23 :(code) 24 void <a href='012elf.cc.html#L24'>load_elf</a>(const string& filename, int argc, char* argv[]) { 25 int fd = open(filename.c_str(), O_RDONLY); 26 if (fd < 0) <a href='003trace.cc.html#L226'>raise</a> << filename.c_str() << ": open" << <a href='012elf.cc.html#L179'>perr</a>() << '\n' << <a href='003trace.cc.html#L173'>die</a>(); 27 off_t size = lseek(fd, 0, SEEK_END); 28 lseek(fd, 0, SEEK_SET); 29 uint8_t* elf_contents = static_cast<uint8_t*>(malloc(size)); 30 if (elf_contents == NULL) <a href='003trace.cc.html#L226'>raise</a> << "malloc(" << size << ')' << <a href='012elf.cc.html#L179'>perr</a>() << '\n' << <a href='003trace.cc.html#L173'>die</a>(); 31 ssize_t read_size = read(fd, elf_contents, size); 32 if (size != read_size) <a href='003trace.cc.html#L226'>raise</a> << "read → " << size << " (!= " << read_size << ')' << <a href='012elf.cc.html#L179'>perr</a>() << '\n' << <a href='003trace.cc.html#L173'>die</a>(); 33 <a href='012elf.cc.html#L37'>load_elf_contents</a>(elf_contents, size, argc, argv); 34 free(elf_contents); 35 } 36 37 void <a href='012elf.cc.html#L37'>load_elf_contents</a>(uint8_t* elf_contents, size_t size, int argc, char* argv[]) { 38 uint8_t magic[5] = {0}; 39 memcpy(magic, elf_contents, 4); 40 if (memcmp(magic, "\177ELF", 4) != 0) 41 <a href='003trace.cc.html#L226'>raise</a> << "Invalid ELF file; starts with \"" << magic << '"' << <a href='003trace.cc.html#L173'>die</a>(); 42 if (elf_contents[4] != 1) 43 <a href='003trace.cc.html#L226'>raise</a> << "Only 32-bit ELF files (4-byte words; virtual addresses up to 4GB) supported.\n" << <a href='003trace.cc.html#L173'>die</a>(); 44 if (elf_contents[5] != 1) 45 <a href='003trace.cc.html#L226'>raise</a> << "Only little-endian ELF files supported.\n" << <a href='003trace.cc.html#L173'>die</a>(); 46 // unused: remaining 10 bytes of e_ident 47 uint32_t e_machine_type = <a href='012elf.cc.html#L170'>u32_in</a>(&elf_contents[16]); 48 if (e_machine_type != 0x00030002) 49 <a href='003trace.cc.html#L226'>raise</a> << "ELF type/machine 0x" << <a href='010vm.cc.html#L408'>HEXWORD</a> << e_machine_type << " isn't i386 executable\n" << die(); 50 // unused: e_version. We only support version 1, and later versions will be backwards compatible. 51 uint32_t e_entry = <a href='012elf.cc.html#L170'>u32_in</a>(&elf_contents[24]); 52 uint32_t e_phoff = <a href='012elf.cc.html#L170'>u32_in</a>(&elf_contents[28]); 53 // unused: e_shoff 54 // unused: e_flags 55 uint32_t e_ehsize = <a href='012elf.cc.html#L174'>u16_in</a>(&elf_contents[40]); 56 if (e_ehsize < 52) <a href='003trace.cc.html#L226'>raise</a> << "Invalid binary; ELF header too small\n" << <a href='003trace.cc.html#L173'>die</a>(); 57 uint32_t e_phentsize = <a href='012elf.cc.html#L174'>u16_in</a>(&elf_contents[42]); 58 uint32_t e_phnum = <a href='012elf.cc.html#L174'>u16_in</a>(&elf_contents[44]); 59 <a href='003trace.cc.html#L96'>trace</a>(90, "load") << e_phnum << " entries in the program header, each " << e_phentsize << " bytes long" << end(); 60 // unused: e_shentsize 61 // unused: e_shnum 62 // unused: e_shstrndx 63 64 set<uint32_t> overlap; // to detect overlapping segments 65 for (size_t i = 0; i < e_phnum; ++i) 66 <a href='012elf.cc.html#L107'>load_segment_from_program_header</a>(elf_contents, i, size, e_phoff + i*e_phentsize, e_ehsize, overlap); 67 68 // initialize code and stack 69 assert(overlap.find(<a href='012elf.cc.html#L152'>STACK_SEGMENT</a>) == overlap.end()); 70 <a href='010vm.cc.html#L163'>Mem</a>.push_back(vma(<a href='012elf.cc.html#L152'>STACK_SEGMENT</a>)); 71 assert(overlap.find(<a href='012elf.cc.html#L153'>AFTER_STACK</a>) == overlap.end()); 72 // The stack grows downward. 73 Reg[ESP].u = <a href='012elf.cc.html#L153'>AFTER_STACK</a>; 74 Reg[EBP].u = 0; 75 <a href='010vm.cc.html#L26'>EIP</a> = e_entry; 76 77 // initialize args on stack 78 // no envp for now 79 // we wastefully use a separate page of memory for argv 80 <a href='010vm.cc.html#L163'>Mem</a>.push_back(vma(<a href='012elf.cc.html#L154'>ARGV_DATA_SEGMENT</a>)); 81 uint32_t argv_data = <a href='012elf.cc.html#L154'>ARGV_DATA_SEGMENT</a>; 82 for (int i = argc-1; i >= /*skip 'subx_bin' and 'run'*/2; --i) { 83 <a href='012elf.cc.html#L94'>push</a>(argv_data); 84 for (size_t j = 0; j <= strlen(argv[i]); ++j) { 85 assert(overlap.find(argv_data) == overlap.end()); // don't bother comparing ARGV and STACK 86 <a href='010vm.cc.html#L244'>write_mem_u8</a>(argv_data, argv[i][j]); 87 argv_data += sizeof(char); 88 assert(argv_data < <a href='012elf.cc.html#L154'>ARGV_DATA_SEGMENT</a> + <a href='010vm.cc.html#L98'>SEGMENT_ALIGNMENT</a>); 89 } 90 } 91 <a href='012elf.cc.html#L94'>push</a>(argc-/*skip 'subx_bin' and 'run'*/2); 92 } 93 94 void <a href='012elf.cc.html#L94'>push</a>(uint32_t val) { 95 Reg[ESP].u -= 4; 96 if (Reg[ESP].u < <a href='012elf.cc.html#L152'>STACK_SEGMENT</a>) { 97 <a href='003trace.cc.html#L226'>raise</a> << "The stack overflowed its segment. " 98 << "Maybe SPACE_FOR_SEGMENT should be larger? " 99 << "Or you need to carve out an exception for the stack segment " 100 << "to be larger.\n" << <a href='003trace.cc.html#L173'>die</a>(); 101 } 102 <a href='003trace.cc.html#L96'>trace</a>(Callstack_depth+1, "run") << "decrementing <a href='010vm.cc.html#L14'>ESP</a> to 0x" << <a href='010vm.cc.html#L408'>HEXWORD</a> << Reg[ESP].u << end(); 103 <a href='003trace.cc.html#L96'>trace</a>(Callstack_depth+1, "run") << "pushing value 0x" << <a href='010vm.cc.html#L408'>HEXWORD</a> << val << end(); 104 <a href='010vm.cc.html#L252'>write_mem_u32</a>(Reg[ESP].u, val); 105 } 106 107 void <a href='012elf.cc.html#L107'>load_segment_from_program_header</a>(uint8_t* elf_contents, int segment_index, size_t size, uint32_t offset, uint32_t e_ehsize, set<uint32_t>& overlap) { 108 uint32_t p_type = <a href='012elf.cc.html#L170'>u32_in</a>(&elf_contents[offset]); 109 <a href='003trace.cc.html#L96'>trace</a>(90, "load") << "program header at offset " << offset << ": type " << p_type << end(); 110 if (p_type != 1) { 111 <a href='003trace.cc.html#L96'>trace</a>(90, "load") << "ignoring segment at offset " << offset << " of non PT_LOAD type " << p_type << " (see <a href="http://refspecs.linuxbase.org/elf/elf.pdf)">http://refspecs.linuxbase.org/elf/elf.pdf)</a>" << end(); 112 return; 113 } 114 uint32_t p_offset = <a href='012elf.cc.html#L170'>u32_in</a>(&elf_contents[offset + 4]); 115 uint32_t p_vaddr = <a href='012elf.cc.html#L170'>u32_in</a>(&elf_contents[offset + 8]); 116 if (e_ehsize > p_vaddr) <a href='003trace.cc.html#L226'>raise</a> << "Invalid binary; program header overlaps ELF header\n" << <a href='003trace.cc.html#L173'>die</a>(); 117 // unused: p_paddr 118 uint32_t p_filesz = <a href='012elf.cc.html#L170'>u32_in</a>(&elf_contents[offset + 16]); 119 uint32_t p_memsz = <a href='012elf.cc.html#L170'>u32_in</a>(&elf_contents[offset + 20]); 120 if (p_filesz != p_memsz) 121 <a href='003trace.cc.html#L226'>raise</a> << "Can't yet handle segments where p_filesz != p_memsz (see <a href="http://refspecs.linuxbase.org/elf/elf.pdf)">http://refspecs.linuxbase.org/elf/elf.pdf)</a>\n" << die(); 122 123 if (p_offset + p_filesz > size) 124 <a href='003trace.cc.html#L226'>raise</a> << "Invalid binary; segment at offset " << offset << " is too large: wants to end at " << p_offset+p_filesz << " but the file ends at " << size << '\n' << <a href='003trace.cc.html#L173'>die</a>(); 125 if (p_memsz >= <a href='010vm.cc.html#L98'>SEGMENT_ALIGNMENT</a>) { 126 <a href='003trace.cc.html#L226'>raise</a> << "Code segment too small for SubX; for now please manually increase SEGMENT_ALIGNMENT.\n" << end(); 127 return; 128 } 129 <a href='003trace.cc.html#L96'>trace</a>(90, "load") << "blitting file offsets (" << p_offset << ", " << (p_offset+p_filesz) << ") to addresses (" << p_vaddr << ", " << (p_vaddr+p_memsz) << ')' << end(); 130 if (size > p_memsz) size = p_memsz; 131 <a href='010vm.cc.html#L163'>Mem</a>.push_back(vma(p_vaddr)); 132 for (size_t i = 0; i < p_filesz; ++i) { 133 assert(overlap.find(p_vaddr+i) == overlap.end()); 134 <a href='010vm.cc.html#L244'>write_mem_u8</a>(p_vaddr+i, elf_contents[p_offset+i]); 135 overlap.insert(p_vaddr+i); 136 } 137 if (segment_index == 0 && <a href='010vm.cc.html#L166'>End_of_program</a> < p_vaddr+p_memsz) 138 <a href='010vm.cc.html#L166'>End_of_program</a> = p_vaddr+p_memsz; 139 } 140 141 :(before "End Includes") 142 // Very primitive/fixed/insecure ELF segments for now. 143 // --- inaccessible: 0x00000000 -> 0x08047fff 144 // code: 0x09000000 -> 0x09ffffff (specified in ELF binary) 145 // data: 0x0a000000 -> 0x0affffff (specified in ELF binary) 146 // --- heap gets mmap'd somewhere here --- 147 // stack: 0xbdffffff -> 0xbd000000 (downward; not in ELF binary) 148 // argv hack: 0xbf000000 -> 0xbfffffff (not in ELF binary) 149 // --- reserved for kernel: 0xc0000000 -> ... 150 const uint32_t <a href='012elf.cc.html#L150'>START_HEAP</a> = 0x0b000000; 151 const uint32_t <a href='012elf.cc.html#L151'>END_HEAP</a> = 0xbd000000; 152 const uint32_t <a href='012elf.cc.html#L152'>STACK_SEGMENT</a> = 0xbd000000; 153 const uint32_t <a href='012elf.cc.html#L153'>AFTER_STACK</a> = 0xbe000000; 154 const uint32_t <a href='012elf.cc.html#L154'>ARGV_DATA_SEGMENT</a> = 0xbf000000; 155 // When updating the above memory map, don't forget to update `mmap`'s 156 // implementation in the 'syscalls' layer. 157 :(before "End Dump Info for Instruction") 158 //? dump_stack(); // slow 159 :(code) 160 void dump_stack() { 161 ostringstream out; 162 <a href='003trace.cc.html#L96'>trace</a>(Callstack_depth+1, "run") << "stack:" << end(); 163 for (uint32_t a = AFTER_STACK-4; a > Reg[ESP].u; a -= 4) 164 <a href='003trace.cc.html#L96'>trace</a>(Callstack_depth+2, "run") << " 0x" << <a href='010vm.cc.html#L408'>HEXWORD</a> << a << " => 0x" << <a href='010vm.cc.html#L408'>HEXWORD</a> << <a href='010vm.cc.html#L181'>read_mem_u32</a>(a) << end(); 165 <a href='003trace.cc.html#L96'>trace</a>(Callstack_depth+2, "run") << " 0x" << <a href='010vm.cc.html#L408'>HEXWORD</a> << Reg[ESP].u << " => 0x" << <a href='010vm.cc.html#L408'>HEXWORD</a> << <a href='010vm.cc.html#L181'>read_mem_u32</a>(Reg[ESP].u) << " <=== ESP" << end(); 166 for (uint32_t a = Reg[ESP].u-4; a > Reg[ESP].u-40; a -= 4) 167 <a href='003trace.cc.html#L96'>trace</a>(Callstack_depth+2, "run") << " 0x" << <a href='010vm.cc.html#L408'>HEXWORD</a> << a << " => 0x" << <a href='010vm.cc.html#L408'>HEXWORD</a> << <a href='010vm.cc.html#L181'>read_mem_u32</a>(a) << end(); 168 } 169 170 inline uint32_t <a href='012elf.cc.html#L170'>u32_in</a>(uint8_t* p) { 171 return p[0] | p[1] << 8 | p[2] << 16 | p[3] << 24; 172 } 173 174 inline uint16_t <a href='012elf.cc.html#L174'>u16_in</a>(uint8_t* p) { 175 return p[0] | p[1] << 8; 176 } 177 178 :(before "End Types") 179 struct <a href='012elf.cc.html#L179'>perr</a> {}; 180 :(code) 181 ostream& operator<<(ostream& os, <a href='012elf.cc.html#L179'>perr</a> /*unused*/) { 182 if (errno) 183 os << ": " << strerror(errno); 184 return os; 185 } 186 187 :(before "End Includes") 188 #include <sys/types.h> 189 #include <sys/stat.h> 190 #include <fcntl.h> 191 #include <stdarg.h> 192 #include <errno.h> 193 #include <unistd.h> </pre> </body> </html>


context:
space:
mode: