diff options
author | Elliott Hughes <enh@google.com> | 2020-01-10 15:56:23 -0800 |
---|---|---|
committer | Elliott Hughes <enh@google.com> | 2020-01-10 15:57:25 -0800 |
commit | 1dab2b948c97f26fe85e822bd7742d15e0cf3c63 (patch) | |
tree | a760a344d22ac6498d80b4bc55772169ec7d98cf /toys | |
parent | 50d7116796a1c7f323d5f81f65f0fe5ecbc86403 (diff) | |
parent | d3493991bf5b65c376f40129cfacd566eefe4e2e (diff) | |
download | toybox-1dab2b948c97f26fe85e822bd7742d15e0cf3c63.tar.gz |
Merge remote-tracking branch 'toybox/master' into HEAD
Change-Id: I1d41c672abacac0b92df7f7d1cd8fae8cdd1524d
Diffstat (limited to 'toys')
-rw-r--r-- | toys/pending/readelf.c | 594 | ||||
-rw-r--r-- | toys/pending/sh.c | 192 | ||||
-rw-r--r-- | toys/pending/vi.c | 93 | ||||
-rw-r--r-- | toys/posix/file.c | 22 | ||||
-rw-r--r-- | toys/posix/ls.c | 28 |
5 files changed, 810 insertions, 119 deletions
diff --git a/toys/pending/readelf.c b/toys/pending/readelf.c new file mode 100644 index 00000000..58e31432 --- /dev/null +++ b/toys/pending/readelf.c @@ -0,0 +1,594 @@ +/* readelf.c - display information about ELF files. + * + * Copyright 2019 The Android Open Source Project + * + * See http://pubs.opengroup.org/onlinepubs/9699919799/utilities/nm.html + +USE_READELF(NEWTOY(readelf, "<1(dyn-syms)adhlnp:SsWx:", TOYFLAG_USR|TOYFLAG_BIN)) + +config READELF + bool "readelf" + default y + help + usage: readelf [-adhlnSsW] [-p SECTION] [-x SECTION] [file...] + + Displays information about ELF files. + + -a Equivalent to -dhlnSs + -d Show dynamic section + -h Show ELF header + -l Show program headers + -n Show notes + -p S Dump strings found in named/numbered section + -S Show section headers + -s Show symbol tables (.dynsym and .symtab) + -W Don't truncate fields (default in toybox) + -x S Hex dump of named/numbered section + + --dyn-syms Show just .dynsym symbol table +*/ + +#define FOR_readelf +#include "toys.h" + +GLOBALS( + char *x, *p; + + char *elf, *shstrtab, *f; + off_t shoff, phoff, size; + int bits, shnum, shentsize, phentsize; + int64_t (*elf_int)(void *ptr, unsigned size); +) + +// Section header. +struct sh { + int type, link, info; + long flags, addr, offset, size, addralign, entsize; + char *name; +}; + +// Program header. +struct ph { + int type, flags; + long offset, vaddr, paddr, filesz, memsz, align; +}; + +static void get_sh(int i, struct sh *s) +{ + char *shdr = TT.elf+TT.shoff+i*TT.shentsize; + + if (i >= TT.shnum || shdr > TT.elf+TT.size-TT.shentsize) { + error_exit("%s: bad shdr %d",TT.f,i); + } + + s->type = TT.elf_int(shdr+4, 4); + s->flags = TT.elf_int(shdr+8, 4*(TT.bits+1)); + s->addr = TT.elf_int(shdr+8+4*(TT.bits+1), 4*(TT.bits+1)); + s->offset = TT.elf_int(shdr+8+8*(TT.bits+1), 4*(TT.bits+1)); + s->size = TT.elf_int(shdr+8+12*(TT.bits+1), 4*(TT.bits+1)); + s->link = TT.elf_int(shdr+8+16*(TT.bits+1), 4); + s->info = TT.elf_int(shdr+12+16*(TT.bits+1), 4); + s->addralign = TT.elf_int(shdr+16+16*(TT.bits+1), 4*(TT.bits+1)); + s->entsize = TT.elf_int(shdr+16+20*(TT.bits+1), 4*(TT.bits+1)); + + if (!TT.shstrtab) s->name = "?"; + else { + s->name = TT.shstrtab + TT.elf_int(shdr, 4); + if (s->name >= TT.elf+TT.size) error_exit("%s: bad shdr name %d",TT.f,i); + if (s->offset >= TT.size-s->size && s->type != 8 /*SHT_NOBITS*/) + error_exit("%s: bad section %d",TT.f,i); + } +} + +static int find_section(char *spec, struct sh *s) +{ + char *end; + int i; + + // Valid section number? + errno = 0; + i = strtoul(spec, &end, 0); + if (!errno && !*end && i < TT.shnum) { + get_sh(i, s); + return 1; + } + + // Search the section names. + for (i=0; i<TT.shnum; i++) { + get_sh(i, s); + if (!strcmp(s->name, spec)) return 1; + } + + error_msg("%s: no section '%s", TT.f, spec); + return 0; +} + +static void get_ph(int i, struct ph *ph) +{ + char *phdr = TT.elf+TT.phoff+i*TT.phentsize; + + if (phdr > TT.elf+TT.size-TT.phentsize) error_exit("%s: bad phdr %d",TT.f,i); + + // Elf64_Phdr reordered fields. + ph->type = TT.elf_int(phdr, 4); + if (TT.bits) { + ph->flags = TT.elf_int(phdr+=4, 4); + ph->offset = TT.elf_int(phdr+=4, 8); + ph->vaddr = TT.elf_int(phdr+=8, 8); + ph->paddr = TT.elf_int(phdr+=8, 8); + ph->filesz = TT.elf_int(phdr+=8, 8); + ph->memsz = TT.elf_int(phdr+=8, 8); + ph->align = TT.elf_int(phdr+=8, 8); + } else { + ph->offset = TT.elf_int(phdr+=4, 4); + ph->vaddr = TT.elf_int(phdr+=4, 4); + ph->paddr = TT.elf_int(phdr+=4, 4); + ph->filesz = TT.elf_int(phdr+=4, 4); + ph->memsz = TT.elf_int(phdr+=4, 4); + ph->flags = TT.elf_int(phdr+=4, 4); + ph->align = TT.elf_int(phdr+=4, 4); + } +} + +#define MAP(...) __VA_ARGS__ +#define DECODER(name, values) \ + static char *name(int type) { \ + static char unknown[20]; \ + struct {int v; char *s;} a[] = values; \ + int i; \ + \ + for (i=0; i<ARRAY_LEN(a); i++) if (type==a[i].v) return a[i].s; \ + sprintf(unknown, "0x%x", type); \ + return unknown; \ + } + +DECODER(dt_type, MAP({{0,"x(NULL)"},{1,"N(NEEDED)"},{2,"b(PLTRELSZ)"}, + {3,"x(PLTGOT)"},{4,"x(HASH)"},{5,"x(STRTAB)"},{6,"x(SYMTAB)"},{7,"x(RELA)"}, + {8,"b(RELASZ)"},{9,"b(RELAENT)"},{10,"b(STRSZ)"},{11,"b(SYMENT)"}, + {12,"x(INIT)"},{13,"x(FINI)"},{14,"S(SONAME)"},{15,"R(RPATH)"}, + {16,"x(SYMBOLIC)"},{17,"x(REL)"},{18,"b(RELSZ)"},{19,"b(RELENT)"}, + {20,"P(PLTREL)"},{21,"x(DEBUG)"},{22,"x(TEXTREL)"},{23,"x(JMPREL)"}, + {24,"d(BIND_NOW)"},{25,"x(INIT_ARRAY)"},{26,"x(FINI_ARRAY)"}, + {27,"b(INIT_ARRAYSZ)"},{28,"b(FINI_ARRAYSZ)"},{29,"R(RUNPATH)"}, + {30,"f(FLAGS)"},{32,"x(PREINIT_ARRAY)"},{33,"x(PREINIT_ARRAYSZ)"}, + {35,"b(RELRSZ)"},{36,"x(RELR)"},{37,"b(RELRENT)"}, + {0x6000000f,"x(ANDROID_REL)"},{0x60000010,"b(ANDROID_RELSZ)"}, + {0x60000011,"x(ANDROID_RELA)"},{0x60000012,"b(ANDROID_RELASZ)"}, + {0x6fffe000,"x(ANDROID_RELR)"},{0x6fffe001,"b(ANDROID_RELRSZ)"}, + {0x6fffe003,"x(ANDROID_RELRENT)"},{0x6ffffef5,"x(GNU_HASH)"}, + {0x6ffffef6,"x(TLSDESC_PLT)"},{0x6ffffef7,"x(TLSDESC_GOT)"}, + {0x6ffffff0,"x(VERSYM)"},{0x6ffffff9,"d(RELACOUNT)"}, + {0x6ffffffa,"d(RELCOUNT)"},{0x6ffffffb,"F(FLAGS_1)"}, + {0x6ffffffc," (VERDEF)"},{0x6ffffffd,"d(VERDEFNUM)"}, + {0x6ffffffe,"x(VERNEED)"},{0x6fffffff,"d(VERNEEDNUM)"}})) + +DECODER(et_type, MAP({{0,"NONE (None)"},{1,"REL (Relocatable file)"}, + {2,"EXEC (Executable file)"},{3,"DYN (Shared object file)"}, + {4,"CORE (Core file)"}})) + +DECODER(nt_type_core, MAP({{1,"NT_PRSTATUS"},{2,"NT_FPREGSET"}, + {3,"NT_PRPSINFO"},{5,"NT_PLATFORM"},{6,"NT_AUXV"}, + {0x46494c45,"NT_FILE"},{0x53494749,"NT_SIGINFO"}})) + +DECODER(nt_type_linux, MAP({{0x200,"NT_386_TLS"},{0x202, "NT_X86_XSTATE"}, + {0x400,"NT_ARM_VFP"},{0x401,"NT_ARM_TLS"},{0x405,"NT_ARM_SVE"}})) + +DECODER(os_abi, MAP({{0,"UNIX - System V"}})) + +DECODER(ph_type, MAP({{0,"NULL"},{1,"LOAD"},{2,"DYNAMIC"},{3,"INTERP"}, + {4,"NOTE"},{5,"SHLIB"},{6,"PHDR"},{7,"TLS"},{0x6474e550,"GNU_EH_FRAME"}, + {0x6474e551,"GNU_STACK"},{0x6474e552,"GNU_RELRO"},{0x70000001,"EXIDX"}})) + +DECODER(sh_type, MAP({{0,"NULL"},{1,"PROGBITS"},{2,"SYMTAB"},{3,"STRTAB"}, + {4,"RELA"},{5,"HASH"},{6,"DYNAMIC"},{7,"NOTE"},{8,"NOBITS"},{9,"REL"}, + {10,"SHLIB"},{11,"DYNSYM"},{14,"INIT_ARRAY"},{15,"FINI_ARRAY"}, + {16,"PREINIT_ARRAY"},{17,"GROUP"},{18,"SYMTAB_SHNDX"},{19,"RELR"}, + {0x60000001,"ANDROID_REL"},{0x60000002,"ANDROID_RELA"}, + {0x6fffff00,"ANDROID_RELR"},{0x6ffffff6,"GNU_HASH"}, + {0x6ffffffd,"VERDEF"},{0x6ffffffe,"VERNEED"}, + {0x6fffffff,"VERSYM"},{0x70000001,"ARM_EXIDX"}, + {0x70000003,"ARM_ATTRIBUTES"}})) + +DECODER(stb_type, MAP({{0,"LOCAL"},{1,"GLOBAL"},{2,"WEAK"}})) + +DECODER(stt_type, MAP({{0,"NOTYPE"},{1,"OBJECT"},{2,"FUNC"},{3,"SECTION"}, + {4,"FILE"},{5,"COMMON"},{6,"TLS"},{10,"GNU_IFUNC"}})) + +DECODER(stv_type, MAP({{0,"DEFAULT"},{1,"INTERNAL"},{2,"HIDDEN"}, + {3,"PROTECTED"}})) + +static void show_symbols(struct sh *table, struct sh *strtab) +{ + char *symtab = TT.elf+table->offset, *ndx; + int sym_size = (TT.bits ? 24 : 16), numsym = table->size/sym_size, i; + + if (numsym == 0) return; + + xputc('\n'); + printf("Symbol table '%s' contains %d entries:\n" + " Num: %*s Size Type Bind Vis Ndx Name\n", + table->name, numsym, 5+8*TT.bits, "Value"); + for (i=0; i<numsym; i++) { + int st_name = TT.elf_int(symtab, 4), st_value, st_shndx; + unsigned char st_info, st_other; + long st_size; + char *name; + + // The various fields were moved around for 64-bit. + if (TT.bits) { + st_info = symtab[4]; + st_other = symtab[5]; + st_shndx = TT.elf_int(symtab+6, 2); + st_value = TT.elf_int(symtab+8, 8); + st_size = TT.elf_int(symtab+16, 8); + } else { + st_value = TT.elf_int(symtab+4, 4); + st_size = TT.elf_int(symtab+8, 4); + st_info = symtab[12]; + st_other = symtab[13]; + st_shndx = TT.elf_int(symtab+14, 2); + } + + name = TT.elf + strtab->offset + st_name; + if (name >= TT.elf+TT.size) error_exit("%s: bad symbol name", TT.f); + + if (!st_shndx) ndx = "UND"; + else if (st_shndx==0xfff1) ndx = "ABS"; + else sprintf(ndx = toybuf, "%d", st_shndx); + + // TODO: look up and show any symbol versions with @ or @@. + + printf("%6d: %0*x %5ld %-7s %-6s %-9s%3s %s\n", i, 8*(TT.bits+1), + st_value, st_size, stt_type(st_info & 0xf), stb_type(st_info >> 4), + stv_type(st_other & 3), ndx, name); + symtab += sym_size; + } +} + +static void show_notes(long offset, long size) +{ + char *note = TT.elf + offset; + + printf(" %-20s %10s\tDescription\n", "Owner", "Data size"); + while (note < TT.elf+offset+size) { + int namesz = TT.elf_int(note, 4), descsz = TT.elf_int(note+4, 4), + type = TT.elf_int(note+8, 4), j = 0; + char *name = note+12; + + printf(" %-20.*s 0x%08x\t", namesz, name, descsz); + if (!memcmp(name, "GNU", 4)) { + if (type == 1) { + printf("NT_GNU_ABI_TAG\tOS: %s, ABI: %ld.%ld.%ld", + !TT.elf_int(note+16, 4)?"Linux":"?", + TT.elf_int(note+20, 4), TT.elf_int(note+24, 4), + TT.elf_int(note+28, 4)), j=1; + } else if (type == 3) { + printf("NT_GNU_BUILD_ID\t"); + for (;j<descsz;j++) printf("%02x",note[16+j]); + } else if (type == 4) { + printf("NT_GNU_GOLD_VERSION\t%.*s", descsz, note+16), j=1; + } + } else if (!memcmp(name, "Android", 8)) { + if (type == 1) { + printf("NT_VERSION\tAPI level %ld", TT.elf_int(note+20, 4)), j=1; + if (descsz>=132) printf(", NDK %.64s (%.64s)",note+24,note+24+64); + } + } else if (!memcmp(name, "CORE", 5) || !memcmp(name, "LINUX", 6)) { + char *desc = *name=='C' ? nt_type_core(type) : nt_type_linux(type); + + if (*desc != '0') printf("%s", desc), j=1; + } + + // If we didn't do custom output above, show a hex dump. + if (!j) { + printf("0x%x\t", type); + for (;j<descsz;j++) printf("%c%02x",!j?'\t':' ',note[16+j]); + } + xputc('\n'); + + note += 3*4 + ((namesz+3)&~3) + ((descsz+3)&~3); + } +} + +static void scan_elf() +{ + struct sh dynamic = {}, dynstr = {}, dynsym = {}, shstr = {}, strtab = {}, + symtab = {}, s; + struct ph ph; + int endian, version, elf_type, flags, entry, ehsize, phnum, shstrndx, i,j,w; + + if (TT.size < 45 || memcmp(TT.elf, "\177ELF", 4)) { + error_msg("%s: not ELF", TT.f); + return; + } + + TT.bits = TT.elf[4] - 1; + endian = TT.elf[5]; + version = TT.elf[6]; + TT.elf_int = (endian==2) ? peek_be : peek_le; + if (TT.bits < 0 || TT.bits > 1 || endian < 1 || endian > 2 || version != 1) { + error_msg("%s: bad ELF", TT.f); + return; + } + + elf_type = TT.elf_int(TT.elf+16, 2); + entry = TT.elf_int(TT.elf+24, 4+4*TT.bits); + TT.phoff = TT.elf_int(TT.elf+28+4*TT.bits, 4+4*TT.bits); + TT.shoff = TT.elf_int(TT.elf+32+8*TT.bits, 4+4*TT.bits); + flags = TT.elf_int(TT.elf+36+12*TT.bits, 4); + ehsize = TT.elf_int(TT.elf+40+12*TT.bits, 2); + TT.phentsize = TT.elf_int(TT.elf+42+12*TT.bits, 2); + phnum = TT.elf_int(TT.elf+44+12*TT.bits, 2); + TT.shentsize = TT.elf_int(TT.elf+46+12*TT.bits, 2); + TT.shnum = TT.elf_int(TT.elf+48+12*TT.bits, 2); + shstrndx = TT.elf_int(TT.elf+50+12*TT.bits, 2); + + // Set up the section header string table so we can use section header names. + // Core files have shstrndx == 0. + TT.shstrtab = 0; + if (shstrndx != 0) { + get_sh(shstrndx, &shstr); + if (shstr.type != 3 /*SHT_STRTAB*/) { + error_msg("%s: bad shstrndx", TT.f); + return; + } + TT.shstrtab = TT.elf+shstr.offset; + } + + if (toys.optc > 1) printf("\nFile: %s\n", TT.f); + + if (FLAG(h)) { + printf("ELF Header:\n"); + printf(" Magic: "); + for (i=0; i<16; i++) printf("%02x%c", TT.elf[i], i==15?'\n':' '); + printf(" Class: ELF%d\n", TT.bits?64:32); + printf(" Data: 2's complement, %s endian\n", + (endian==2)?"big":"little"); + printf(" Version: 1 (current)\n"); + printf(" OS/ABI: %s\n", os_abi(TT.elf[7])); + printf(" ABI Version: %d\n", TT.elf[8]); + printf(" Type: %s\n", et_type(elf_type)); + printf(" Machine: %s\n", + elf_arch_name(TT.elf_int(TT.elf+18, 2))); + printf(" Version: 0x%x\n", + (int) TT.elf_int(TT.elf+20, 4)); + printf(" Entry point address: 0x%x\n", entry); + printf(" Start of program headers: %ld (bytes into file)\n", + TT.phoff); + printf(" Start of section headers: %ld (bytes into file)\n", + TT.shoff); + printf(" Flags: 0x%x\n", flags); + printf(" Size of this header: %d (bytes)\n", ehsize); + printf(" Size of program headers: %d (bytes)\n", TT.phentsize); + printf(" Number of program headers: %d\n", phnum); + printf(" Size of section headers: %d (bytes)\n", TT.shentsize); + printf(" Number of section headers: %d\n", TT.shnum); + printf(" Section header string table index: %d\n", shstrndx); + } + + w = 8*(TT.bits+1); + if (FLAG(S)) { + if (!TT.shnum) printf("\nThere are no sections in this file.\n"); + else { + if (!FLAG(h)) { + printf("There are %d section headers, starting at offset %#lx:\n", + TT.shnum, TT.shoff); + } + printf("\n" + "Section Headers:\n" + " [Nr] %-20s %-14s %-*s %-6s %-6s ES Flg Lk Inf Al\n", + "Name", "Type", w, "Address", "Off", "Size"); + } + } + // We need to iterate through the section headers even if we're not + // dumping them, to find specific sections. + for (i=0; i<TT.shnum; i++) { + get_sh(i, &s); + if (s.type == 2 /*SHT_SYMTAB*/) symtab = s; + else if (s.type == 6 /*SHT_DYNAMIC*/) dynamic = s; + else if (s.type == 11 /*SHT_DYNSYM*/) dynsym = s; + else if (s.type == 3 /*SHT_STRTAB*/) { + if (!strcmp(s.name, ".strtab")) strtab = s; + else if (!strcmp(s.name, ".dynstr")) dynstr = s; + } + + if (FLAG(S)) { + char sh_flags[12] = {}, *p = sh_flags; + + for (j=0; j<12; j++) if (s.flags&(1<<j)) *p++="WAXxMSILOTC"[j]; + printf(" [%2d] %-20s %-14s %0*lx %06lx %06lx %02lx %3s %2d %2d %2ld\n", + i, s.name, sh_type(s.type), w, s.addr, s.offset, s.size, + s.entsize, sh_flags, s.link, s.info, s.addralign); + } + } + if (FLAG(S) && TT.shnum) { + printf("Key:\n" + " (W)rite, (A)lloc, e(X)ecute, (M)erge, (S)trings, (I)nfo\n" + " (L)ink order, (O)S, (G)roup, (T)LS, (C)ompressed, x=unknown\n"); + } + + if (FLAG(l)) { + xputc('\n'); + if (!phnum) printf("There are no program headers in this file.\n"); + else { + if (!FLAG(h)) { + printf("Elf file type is %s\n" + "Entry point %#x\n" + "There are %d program headers, starting at offset %ld\n" + "\n", + et_type(elf_type), entry, phnum, TT.phoff); + } + printf("Program Headers:\n" + " %-14s %-8s %-*s %-*s %-7s %-7s Flg Align\n", "Type", + "Offset", w, "VirtAddr", w, "PhysAddr", "FileSiz", "MemSiz"); + for (i=0; i<phnum; i++) { + get_ph(i, &ph); + printf(" %-14s 0x%06lx 0x%0*lx 0x%0*lx 0x%05lx 0x%05lx %c%c%c %#lx\n", + ph_type(ph.type), ph.offset, w, ph.vaddr, w, ph.paddr, + ph.filesz, ph.memsz, ph.flags&4?'R':' ', ph.flags&2?'W':' ', + ph.flags&1?'E':' ', ph.align); + if (ph.type == 3 /*PH_INTERP*/) { + printf(" [Requesting program interpreter: %*s]\n", + (int) ph.filesz-1, TT.elf+ph.offset); + } + } + + printf("\n" + " Section to Segment mapping:\n" + " Segment Sections...\n"); + for (i=0; i<phnum; i++) { + get_ph(i, &ph); + printf(" %02d ", i); + for (j=0; j<TT.shnum; j++) { + get_sh(j, &s); + if (s.offset >= ph.offset && s.offset+s.size <= ph.offset+ph.filesz) + printf(" %s", s.name); + } + xputc('\n'); + } + } + } + + // binutils ld emits a bunch of extra DT_NULL entries, so binutils readelf + // uses two passes here! We just tell the truth, which matches -h. + if (FLAG(d)) { + char *dyn = TT.elf+dynamic.offset, *end = dyn+dynamic.size; + + xputc('\n'); + if (!dynamic.size) printf("There is no dynamic section in this file.\n"); + else printf("Dynamic section at offset 0x%lx contains %ld entries:\n" + " %-*s %-20s %s\n", + dynamic.offset, dynamic.size/dynamic.entsize, + w+2, "Tag", "Type", "Name/Value"); + for (; dyn < end; dyn += dynamic.entsize) { + int es = 4*(TT.bits+1); + long tag = TT.elf_int(dyn, es), val = TT.elf_int(dyn+es, es); + char *type = dt_type(tag); + + printf(" 0x%0*lx %-20s ", w, tag, *type=='0' ? type : type+1); + if (*type == 'd') printf("%ld\n", val); + else if (*type == 'b') printf("%ld (bytes)\n", val); + else if (*type == 's') printf("%s\n", TT.elf+dynstr.offset+val); + else if (*type == 'f' || *type == 'F') { + struct bitname { int bit; char *s; } + df_names[] = {{0, "ORIGIN"},{1,"SYMBOLIC"},{2,"TEXTREL"}, + {3,"BIND_NOW"},{4,"STATIC_TLS"},{}}, + df_1_names[]={{0,"NOW"},{1,"GLOBAL"},{2,"GROUP"},{3,"NODELETE"}, + {5,"INITFIRST"},{27,"PIE"},{}}, + *names = *type == 'f' ? df_names : df_1_names; + int mask; + + if (*type == 'F') printf("Flags: "); + for (j=0; names[j].s; j++) { + if (val & (mask=(1<<names[j].bit))) { + printf("%s%s", names[j].s, (val &= ~mask) ? " " : ""); + } + } + if (val) printf("0x%lx", val); + xputc('\n'); + } else if (*type == 'N' || *type == 'R' || *type == 'S') { + printf("%s: [%s]\n", *type=='N' ? "Shared library" : + (*type=='R' ? "Library runpath" : "Library soname"), + TT.elf+dynstr.offset+val); + } else if (*type == 'P') { + type = dt_type(val); + j = strlen(type); + if (*type != '0') type += 2, j -= 3; + printf("%*.*s\n", j, j, type); + } else printf("0x%lx\n", val); + } + } + + if (FLAG(dyn_syms)) show_symbols(&dynsym, &dynstr); + if (FLAG(s)) show_symbols(&symtab, &strtab); + + if (FLAG(n)) { + int found = 0; + + for (i=0; i<TT.shnum; i++) { + get_sh(i, &s); + if (s.type == 7 /*SHT_NOTE*/) { + printf("\nDisplaying notes found in: %s\n", s.name); + show_notes(s.offset, s.size); + found = 1; + } + } + for (i=0; !found && i<phnum; i++) { + get_ph(i, &ph); + if (ph.type == 4 /*PT_NOTE*/) { + printf("\n" + "Displaying notes found at file offset 0x%lx with length 0x%lx:\n", + ph.offset, ph.filesz); + show_notes(ph.offset, ph.filesz); + } + } + } + + if (FLAG(x)) { + if (find_section(TT.x, &s)) { + char *p = TT.elf+s.offset; + long offset = 0; + + printf("\nHex dump of section '%s':\n", s.name); + while (offset < s.size) { + int space = 2*16 + 16/4; + + printf(" 0x%08lx ", offset); + for (i=0; i<16 && offset < s.size; offset++) { + space -= printf("%02x%s", *p++, ++i%4 ? "" : " "); + } + printf("%*s", space, ""); + for (p-=i; i; i--, p++) putchar(*p>=' ' && *p<='~' ? *p : '.'); + xputc('\n'); + } + printf("\n"); + } + } + + if (FLAG(p)) { + if (find_section(TT.p, &s)) { + char *begin = TT.elf+s.offset, *end = begin + s.size, *p = begin; + int any = 0; + + printf("\nString dump of section '%s':\n", s.name); + for (; p < end; p++) { + if (isprint(*p)) { + printf(" [%6tx] ", p-begin); + while (p < end && isprint(*p)) putchar(*p++); + xputc('\n'); + any=1; + } + } + if (!any) printf(" No strings found in this section.\n"); + printf("\n"); + } + } +} + +void readelf_main(void) +{ + char **arg; + int all = FLAG_d|FLAG_h|FLAG_l|FLAG_n|FLAG_S|FLAG_s|FLAG_dyn_syms; + + if (FLAG(a)) toys.optflags |= all; + if (FLAG(s)) toys.optflags |= FLAG_dyn_syms; + if (!(toys.optflags & (all|FLAG_p|FLAG_x))) help_exit("needs a flag"); + + for (arg = toys.optargs; *arg; arg++) { + int fd = open(TT.f = *arg, O_RDONLY); + struct stat sb; + + if (fd == -1) perror_msg("%s", TT.f); + else { + if (fstat(fd, &sb)) perror_msg("%s", TT.f); + else if (!sb.st_size) error_msg("%s: empty", TT.f); + else if (!S_ISREG(sb.st_mode)) error_msg("%s: not a regular file",TT.f); + else { + TT.elf = xmmap(NULL, TT.size=sb.st_size, PROT_READ, MAP_SHARED, fd, 0); + scan_elf(); + munmap(TT.elf, TT.size); + } + close(fd); + } + } +} diff --git a/toys/pending/sh.c b/toys/pending/sh.c index dc4b774e..482639da 100644 --- a/toys/pending/sh.c +++ b/toys/pending/sh.c @@ -217,6 +217,32 @@ static char *getvar(char *s) return getvarlen(s, strlen(s)); } + + +// returns pointer to next unquoted (or double quoted if dquot) char. +// handle \ '' "" `` $() +int skip_quote(char *s, int dquot, int *depth) +{ + int i, q = dquot ? *depth : 0; + + // quotes were checked for balance and overflow by parse_word() + for (i = 0; s[i]; i++) { + char c = s[i], qq = q ? toybuf[q-1] : 0; + + if (c == '\\') i++; + else if (qq!='\'' && c=='$' && s[1]=='(') { + toybuf[q++] = ')'; + i++; + } else if (q && qq==c) q--; + else if ((!q || qq==')') && (c=='"' || c=='\'' || c=='`')) toybuf[q++] = c; + else if (!q || (dquot && q==1 && qq=='"')) break; + } + + if (dquot) *depth = q; + + return i; +} + // quote removal, brace, tilde, parameter/variable, $(command), // $((arithmetic)), split, path #define NO_PATH (1<<0) @@ -225,6 +251,7 @@ static char *getvar(char *s) #define NO_TILDE (1<<3) #define NO_QUOTE (1<<4) #define FORCE_COPY (1<<31) +#define FORCE_KEEP (1<<30) // TODO: ${name:?error} causes an error/abort here (syntax_err longjmp?) // TODO: $1 $@ $* need args marshalled down here: function+structure? // arg = append to this @@ -232,18 +259,19 @@ static char *getvar(char *s) // flags = type of expansions (not) to do // delete = append new allocations to this so they can be freed later // TODO: at_args: $1 $2 $3 $* $@ -static void expand_arg(struct sh_arg *arg, char *old, unsigned flags, +static void expand_arg_nobrace(struct sh_arg *arg, char *old, unsigned flags, struct arg_list **delete) { - char *new = old, *s, *ss, quote = 0; + char *new = old; -// TODO ls -l /proc/$$/fd + if (flags&FORCE_KEEP) old = 0; -// ${ $(( $( $[ $' ` +// TODO ls -l /proc/$$/fd // Tilde expansion if (!(flags&NO_TILDE) && *new == '~') { struct passwd *pw = 0; + char *s, *ss; // first expansion so don't need to free previous new ss = 0; @@ -257,9 +285,13 @@ static void expand_arg(struct sh_arg *arg, char *old, unsigned flags, } if (pw && pw->pw_dir) ss = pw->pw_dir; if (!ss || !*ss) ss = "/"; - new = xmprintf("%s%s", ss, s); + s = xmprintf("%s%s", ss, s); + if (old != new) free(new); + new = s; } +// ${ $(( $( $[ $' ` " ' + /* while (*s) { if (!quote && !(flags&NO_BRACE) && *s == '{') { @@ -289,7 +321,21 @@ TODO this recurses } */ - // We have a result. Append it. + // quote removal + if (!(flags&NO_QUOTE)) { + int to = 0, from = 0; + + for (;;) { + char c = new[from++]; + + if (c == '"' || c=='\'') continue; + if (c == '\\' && new[from]) c = new[from++]; + if (from != to && old == new) new = xstrdup(new); + if (!(new[to++] = c)) break; + } + } + + // Record result. if (old==new && (flags&FORCE_COPY)) new = xstrdup(new); if (old!=new && delete) { struct arg_list *al = xmalloc(sizeof(struct arg_list)); @@ -301,6 +347,135 @@ TODO this recurses array_add(&arg->v, arg->c++, new); } +// expand braces (ala {a,b,c}) and call expand_arg_nobrace() each permutation +static void expand_arg(struct sh_arg *arg, char *old, unsigned flags, + struct arg_list **delete) +{ + struct brace { + struct brace *next, *prev, *stack; + int active, cnt, idx, commas[]; + } *bb = 0, *blist = 0, *bstk, *bnext; + int i, j; + char *s, *ss; + + // collect brace spans + if (!(flags&NO_BRACE)) for (i = 0; ; i++) { + i += skip_quote(old+i, 0, 0); + if (!bb && !old[i]) break; + if (bb && (!old[i] || old[i] == '}')) { + bb->active = bb->commas[bb->cnt+1] = i; + for (bnext = bb; bb && bb->active; bb = (bb==blist)?0:bb->prev); + if (!old[i] || !bnext->cnt) // discard commaless brace from start/middle + free(dlist_pop((blist == bnext) ? &blist : &bnext)); + } else if (old[i] == '{') { + dlist_add_nomalloc((void *)&blist, + (void *)(bb = xzalloc(sizeof(struct brace)+34*4))); + bb->commas[0] = i; + } else if (!bb) continue; + else if (bb && old[i] == ',') { + if (bb->cnt && !(bb->cnt&31)) { + dlist_lpop(&blist); + dlist_add_nomalloc((void *)&blist, + (void *)(bb = xrealloc(bb, sizeof(struct brace)+(bb->cnt+34)*4))); + } + bb->commas[++bb->cnt] = i; + } + } + +// TODO NOSPLIT with braces? (Collate with spaces?) + // If none, pass on verbatim + if (!blist) return expand_arg_nobrace(arg, old, flags, delete); + + // enclose entire range in top level brace. + (bstk = xzalloc(sizeof(struct brace)+8))->commas[1] = strlen(old)+1; + bstk->commas[0] = -1; + + // loop through each combination + for (;;) { + + // Brace expansion can't be longer than original string. Keep start to { + s = ss = xmalloc(bstk->commas[1]); + + // Append output from active braces (in "saved" list) + for (bb = blist; bb;) { + + // keep prefix and push self onto stack + if (bstk == bb) bstk = bstk->stack; // pop self + i = bstk->commas[bstk->idx]+1; + if (bstk->commas[bstk->cnt+1]>bb->commas[0]) + s = stpncpy(s, old+i, bb->commas[0]-i); + + // pop sibling + if (bstk->commas[bstk->cnt+1]<bb->commas[0]) bstk = bstk->stack; + + bb->stack = bstk; // push + bb->active = 1; + bstk = bnext = bb; + + // skip inactive spans from earlier or later commas + while ((bnext = (bnext->next==blist) ? 0 : bnext->next)) { + i = bnext->commas[0]; + + // past end of this brace + if (i>bb->commas[bb->cnt+1]) break; + + // in this brace but not this selection + if (i<bb->commas[bb->idx] || i>bb->commas[bb->idx+1]) { + bnext->active = 0; + bnext->stack = 0; + + // in this selection + } else break; + } + + // is next span past this range? + if (!bnext || bnext->commas[0]>bb->commas[bb->idx+1]) { + + // output uninterrupted span + i = bb->commas[bstk->idx]+1; + s = stpncpy(s, old+i, bb->commas[bb->idx+1]-i); + + // While not sibling, output tail and pop + while (!bnext || bnext->commas[0] > bstk->commas[bstk->cnt+1]) { + if (!(bb = bstk->stack)) break; + i = bstk->commas[bstk->cnt+1]+1; // start of span + j = bb->commas[bb->idx+1]; // enclosing comma span + + while (bnext) { + if (bnext->commas[0]<j) { + j = bnext->commas[0];// sibling + break; + } else if (bb->commas[bb->cnt+1]>bnext->commas[0]) + bnext = (bnext->next == blist) ? 0 : bnext->next; + else break; + } + s = stpncpy(s, old+i, j-i); + + // if next is sibling but parent _not_ a sibling, don't pop + if (bnext && bnext->commas[0]<bstk->stack->commas[bstk->stack->cnt+1]) + break; + bstk = bstk->stack; + } + } + bb = (bnext == blist) ? 0 : bnext; + } + + // Save result + expand_arg_nobrace(arg, ss, flags|FORCE_KEEP, delete); + + // increment + for (bb = blist->prev; bb; bb = (bb == blist) ? 0 : bb->prev) { + if (!bb->stack) continue; + else if (++bb->idx > bb->cnt) bb->idx = 0; + else break; + } + + // if increment went off left edge, done expanding + if (!bb) return llist_traverse(blist, free); + } +} + + // Expand exactly one arg, returning NULL if it split. // If return != new you need to free it. static char *expand_one_arg(char *new, unsigned flags, struct arg_list **del) @@ -1646,6 +1821,11 @@ void sh_main(void) TT.hfd = 10; signal(SIGPIPE, SIG_IGN); + // Ensure environ copied and toys.envc set + xunsetenv(""); + + // TODO: traverse and unset illegal environment variables named "$" and such + // TODO euid stuff? // TODO read profile, read rc diff --git a/toys/pending/vi.c b/toys/pending/vi.c index 1cf8bc7b..9906ca93 100644 --- a/toys/pending/vi.c +++ b/toys/pending/vi.c @@ -64,7 +64,6 @@ static void draw_page(); //utf8 support static int utf8_lnw(int* width, char* str, int bytes); static int utf8_dec(char key, char *utf8_scratch, int *sta_p); -static int utf8_len(char *str); static int utf8_width(char *str, int bytes); static char* utf8_last(char* str, int size); @@ -1114,6 +1113,7 @@ static void draw_page() struct linelist *scr_buf = 0; unsigned y = 0; int x = 0; + wchar_t wc; char *line = 0, *end = 0; int utf_l = 0, bytes = 0; @@ -1246,33 +1246,21 @@ static void draw_page() tty_jump(0, TT.screen_height); tty_esc("2K"); - switch (TT.vi_mode) { - case 0: - tty_esc("30;44m"); - printf("COMMAND|"); - break; - case 1: - tty_esc("30;42m"); - printf("NORMAL|"); - break; - case 2: - tty_esc("30;41m"); - printf("INSERT|"); - break; + if (TT.vi_mode == 2) printf("\x1b[1m-- INSERT --\x1b[m"); - } //DEBUG - tty_esc("m"); - utf_l = utf8_len(&c_r->line->data[TT.cur_col]); - if (utf_l) { + utf_l=utf8towc(&wc, &c_r->line->data[TT.cur_col], c_r->line->len-TT.cur_col); + if (utf_l > 1) { char t[5] = {0, 0, 0, 0, 0}; strncpy(t, &c_r->line->data[TT.cur_col], utf_l); - printf("utf: %d %s", utf_l, t); + printf(" (utf: %d %s)", utf_l, t); } - printf("| %d, %d\n", cx_scr, cy_scr); //screen coord + //DEBUG tty_jump(TT.screen_width-12, TT.screen_height); - printf("| %d, %d\n", TT.cur_row, TT.cur_col); + printf("%d,%d", TT.cur_row+1, TT.cur_col+1); + if (TT.cur_col != cx_scr) printf("-%d", cx_scr+1); + putchar('\n'); tty_esc("m"); tty_jump(0, TT.screen_height+1); @@ -1325,71 +1313,34 @@ static void adjust_screen_buffer() } -//return 0 if not ASCII nor UTF-8 -//this is not fully tested -//naive implementation with branches -//there is better branchless lookup table versions out there -//1 0xxxxxxx -//2 110xxxxx 10xxxxxx -//3 1110xxxx 10xxxxxx 10xxxxxx -//4 11110xxx 10xxxxxx 10xxxxxx 10xxxxxx -static int utf8_len(char *str) -{ - int len = 0; - int i = 0; - uint8_t *c = (uint8_t*)str; - if (!c || !(*c)) return 0; - if (*c < 0x7F) return 1; - if ((*c & 0xE0) == 0xc0) len = 2; - else if ((*c & 0xF0) == 0xE0 ) len = 3; - else if ((*c & 0xF8) == 0xF0 ) len = 4; - else return 0; - c++; - for (i = len-1; i > 0; i--) { - if ((*c++ & 0xc0) != 0x80) return 0; - } - return len; -} - //get utf8 length and width at same time -static int utf8_lnw(int* width, char* str, int bytes) +static int utf8_lnw(int* width, char* s, int bytes) { wchar_t wc; - int length = 1; - *width = 1; - if (*str == 0x09) { + int length; + + *width = 0; + if (*s == '\t') { *width = TT.tabstop; return 1; } - length = mbtowc(&wc, str, bytes); - switch (length) { - case -1: - mbtowc(0,0,4); - case 0: - *width = 0; - length = 0; - break; - default: + length = utf8towc(&wc, s, bytes); + if (length < 1) return 0; *width = wcwidth(wc); - } return length; } //try to estimate width of next "glyph" in terminal buffer //combining chars 0x300-0x36F shall be zero width -static int utf8_width(char *str, int bytes) +static int utf8_width(char *s, int bytes) { wchar_t wc; - if (*str == 0x09) return TT.tabstop; - switch (mbtowc(&wc, str, bytes)) { - case -1: - mbtowc(0,0,4); - case 0: - return -1; - default: + int length; + + if (*s == '\t') return TT.tabstop; + length = utf8towc(&wc, s, bytes); + if (length < 1) return -1; return wcwidth(wc); - } - return 0; } static int utf8_dec(char key, char *utf8_scratch, int *sta_p) diff --git a/toys/posix/file.c b/toys/posix/file.c index 5d6eef12..f2cd3af8 100644 --- a/toys/posix/file.c +++ b/toys/posix/file.c @@ -36,21 +36,6 @@ static void do_elf_file(int fd) int endian = toybuf[5], bits = toybuf[4], i, j, dynamic = 0, stripped = 1, phentsize, phnum, shsize, shnum; int64_t (*elf_int)(void *ptr, unsigned size); - // Values from include/linux/elf-em.h (plus arch/*/include/asm/elf.h) - // Names are linux/arch/ directory (sometimes before 32/64 bit merges) - struct {int val; char *name;} type[] = {{0x9026, "alpha"}, {93, "arc"}, - {195, "arcv2"}, {40, "arm"}, {183, "arm64"}, {0x18ad, "avr32"}, - {247, "bpf"}, {106, "blackfin"}, {140, "c6x"}, {23, "cell"}, {76, "cris"}, - {252, "csky"}, {0x5441, "frv"}, {46, "h8300"}, {164, "hexagon"}, - {50, "ia64"}, {88, "m32r"}, {0x9041, "m32r"}, {4, "m68k"}, {174, "metag"}, - {189, "microblaze"}, {0xbaab, "microblaze-old"}, {8, "mips"}, - {10, "mips-old"}, {89, "mn10300"}, {0xbeef, "mn10300-old"}, {113, "nios2"}, - {92, "openrisc"}, {0x8472, "openrisc-old"}, {15, "parisc"}, {20, "ppc"}, - {21, "ppc64"}, {243, "riscv"}, {22, "s390"}, {0xa390, "s390-old"}, - {135, "score"}, {42, "sh"}, {2, "sparc"}, {18, "sparc8+"}, {43, "sparc9"}, - {188, "tile"}, {191, "tilegx"}, {3, "386"}, {6, "486"}, {62, "x86-64"}, - {94, "xtensa"}, {0xabc7, "xtensa-old"} - }; char *map = 0; off_t phoff, shoff; @@ -83,11 +68,8 @@ static void do_elf_file(int fd) endian = 0; } - // e_machine, ala "x86", from big table above - j = elf_int(toybuf+18, 2); - for (i = 0; i<ARRAY_LEN(type); i++) if (j==type[i].val) break; - if (i<ARRAY_LEN(type)) printf("%s", type[i].name); - else printf("(unknown arch %d)", j); + // "x86". + printf("%s", elf_arch_name(elf_int(toybuf+18, 2))); bits--; // If what we've seen so far doesn't seem consistent, bail. diff --git a/toys/posix/ls.c b/toys/posix/ls.c index c2286f6f..47620590 100644 --- a/toys/posix/ls.c +++ b/toys/posix/ls.c @@ -187,29 +187,13 @@ static int filter(struct dirtree *new) if (FLAG(Z)) { if (!CFG_TOYBOX_LSM_NONE) { + // Linux doesn't support fgetxattr(2) on O_PATH file descriptors (though + // bionic works around that), and there are no *xattrat(2) calls, so we + // just use lgetxattr(2). + char *path = dirtree_path(new, 0); - // (Wouldn't it be nice if the lsm functions worked like openat(), - // fchmodat(), mknodat(), readlinkat() so we could do this without - // even O_PATH? But no, this is 1990's tech.) - int fd = openat(dirtree_parentfd(new), new->name, - O_PATH|(O_NOFOLLOW*!FLAG(L))); - - if (fd != -1) { - if (-1 == lsm_fget_context(fd, (char **)&new->extra) && errno == EBADF) - { - char hack[32]; - - // Work around kernel bug that won't let us read this "metadata" from - // the filehandle unless we have permission to read the data. (We can - // query the same data in by path, but can't do it through an O_PATH - // filehandle, because reasons. But for some reason, THIS is ok? If - // they ever fix the kernel, this should stop triggering.) - - sprintf(hack, "/proc/self/fd/%d", fd); - lsm_lget_context(hack, (char **)&new->extra); - } - close(fd); - } + (FLAG(L) ? lsm_get_context : lsm_lget_context)(path,(char **)&new->extra); + free(path); } if (CFG_TOYBOX_LSM_NONE || !new->extra) new->extra = (long)xstrdup("?"); } |