diff options
author | Mark Wielaard <mjw@redhat.com> | 2015-12-01 14:55:07 +0100 |
---|---|---|
committer | Mark Wielaard <mjw@redhat.com> | 2016-01-05 15:03:13 +0100 |
commit | c9ce0de100980a998c54d852bb64e52b874198af (patch) | |
tree | 62e25f6c4758f7535b6e46bbeab4a6b564a4b5ca /src/elfcompress.c | |
parent | 923c75f42c956374f845edbefd04fcd9807c842b (diff) | |
download | elfutils-mjw/compress4.tar.gz |
elfcompress: New utility.mjw/compress4
Usage: elfcompress [OPTION...] FILE...
Compress or decompress sections in an ELF file.
-f, --force Force compression of section even if it would
become larger
-n, --name=SECTION SECTION name to (de)compress, SECTION is an
extended wildcard pattern (defaults to
'.?(z)debug*')
-o, --output=FILE Place (de)compressed output into FILE
-p, --permissive Relax a few rules to handle slightly broken ELF
files
-q, --quiet Be silent when a section cannot be compressed
-t, --type=TYPE What type of compression to apply. TYPE can be
'none' (decompress), 'zlib' (ELF ZLIB compression,
the default, 'zlib-gabi' is an alias) or
'zlib-gnu' (.zdebug GNU style compression, 'gnu'
is an alias)
-v, --verbose Print a message for each section being
(de)compressed
-?, --help Give this help list
--usage Give a short usage message
-V, --version Print program version
Signed-off-by: Mark Wielaard <mjw@redhat.com>
Diffstat (limited to 'src/elfcompress.c')
-rw-r--r-- | src/elfcompress.c | 1321 |
1 files changed, 1321 insertions, 0 deletions
diff --git a/src/elfcompress.c b/src/elfcompress.c new file mode 100644 index 00000000..5be2375d --- /dev/null +++ b/src/elfcompress.c @@ -0,0 +1,1321 @@ +/* Compress or decompress an ELF file. + Copyright (C) 2015 Red Hat, Inc. + This file is part of elfutils. + + This file is free software; you can redistribute it and/or modify + it under the terms of the GNU General Public License as published by + the Free Software Foundation; either version 3 of the License, or + (at your option) any later version. + + elfutils is distributed in the hope that it will be useful, but + WITHOUT ANY WARRANTY; without even the implied warranty of + MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + GNU General Public License for more details. + + You should have received a copy of the GNU General Public License + along with this program. If not, see <http://www.gnu.org/licenses/>. */ + +#include <config.h> +#include <assert.h> +#include <argp.h> +#include <error.h> +#include <stdbool.h> +#include <stdlib.h> +#include <inttypes.h> +#include <stdio.h> +#include <string.h> +#include <locale.h> +#include <fcntl.h> +#include <fnmatch.h> +#include <sys/types.h> +#include <sys/stat.h> +#include <unistd.h> +#include ELFUTILS_HEADER(elf) +#include ELFUTILS_HEADER(ebl) +#include <gelf.h> +#include "system.h" + +/* Name and version of program. */ +static void print_version (FILE *stream, struct argp_state *state); +ARGP_PROGRAM_VERSION_HOOK_DEF = print_version; + +/* Bug report address. */ +ARGP_PROGRAM_BUG_ADDRESS_DEF = PACKAGE_BUGREPORT; + +static int verbose = 0; /* < 0, no warnings, > 0 extra verbosity. */ +static bool force = false; +static bool permissive = false; +static const char *foutput = NULL; + +#define T_UNSET 0 +#define T_DECOMPRESS 1 /* none */ +#define T_COMPRESS_ZLIB 2 /* zlib */ +#define T_COMPRESS_GNU 3 /* zlib-gnu */ +static int type = T_UNSET; + +static void +print_version (FILE *stream, struct argp_state *state __attribute__ ((unused))) +{ + fprintf (stream, "elfcompress (%s) %s\n", PACKAGE_NAME, PACKAGE_VERSION); +} + +struct section_pattern +{ + char *pattern; + struct section_pattern *next; +}; + +static struct section_pattern *patterns = NULL; + +static void +add_pattern (const char *pattern) +{ + struct section_pattern *p = xmalloc (sizeof *p); + p->pattern = xstrdup (pattern); + p->next = patterns; + patterns = p; +} + +static void +free_patterns (void) +{ + struct section_pattern *pattern = patterns; + while (pattern != NULL) + { + struct section_pattern *p = pattern; + pattern = p->next; + free (p->pattern); + free (p); + } +} + +static error_t +parse_opt (int key, char *arg __attribute__ ((unused)), + struct argp_state *state __attribute__ ((unused))) +{ + switch (key) + { + case 'v': + verbose++; + break; + + case 'q': + verbose--; + + case 'f': + force = true; + break; + + case 'p': + permissive = true; + break; + + case 'n': + add_pattern (arg); + break; + + case 'o': + if (foutput != NULL) + argp_error (state, N_("-o option specified twice")); + else + foutput = arg; + break; + + case 't': + if (type != T_UNSET) + argp_error (state, N_("-t option specified twice")); + + if (strcmp ("none", arg) == 0) + type = T_DECOMPRESS; + else if (strcmp ("zlib", arg) == 0 || strcmp ("zlib-gabi", arg) == 0) + type = T_COMPRESS_ZLIB; + else if (strcmp ("zlib-gnu", arg) == 0 || strcmp ("gnu", arg) == 0) + type = T_COMPRESS_GNU; + else + argp_error (state, N_("unknown compression type '%s'"), arg); + break; + + case ARGP_KEY_SUCCESS: + if (type == T_UNSET) + type = T_COMPRESS_ZLIB; + if (patterns == NULL) + add_pattern (".?(z)debug*"); + break; + + case ARGP_KEY_NO_ARGS: + /* We need at least one input file. */ + argp_error (state, N_("No input file given")); + break; + + case ARGP_KEY_ARGS: + if (foutput != NULL && state->argc - state->next > 1) + argp_error (state, + N_("Only one input file allowed together with '-o'")); + /* We only use this for checking the number of arguments, we don't + actually want to consume them, so fallthrough. */ + default: + return ARGP_ERR_UNKNOWN; + } + return 0; +} + +static bool +section_name_matches (const char *name) +{ + struct section_pattern *pattern = patterns; + while (pattern != NULL) + { + if (fnmatch (pattern->pattern, name, FNM_EXTMATCH) == 0) + return true; + pattern = pattern->next; + } + return false; +} + +static int +setshdrstrndx (Elf *elf, GElf_Ehdr *ehdr, size_t ndx) +{ + if (ndx < SHN_LORESERVE) + ehdr->e_shstrndx = ndx; + else + { + ehdr->e_shstrndx = SHN_XINDEX; + Elf_Scn *zscn = elf_getscn (elf, 0); + GElf_Shdr zshdr_mem; + GElf_Shdr *zshdr = gelf_getshdr (zscn, &zshdr_mem); + if (zshdr == NULL) + return -1; + zshdr->sh_link = ndx; + if (gelf_update_shdr (zscn, zshdr) == 0) + return -1; + } + + if (gelf_update_ehdr (elf, ehdr) == 0) + return -1; + + return 0; +} + +static int +compress_section (Elf_Scn *scn, size_t orig_size, const char *name, + const char *newname, size_t ndx, + bool gnu, bool compress, bool report_verbose) +{ + int res; + unsigned int flags = compress && force ? ELF_CHF_FORCE : 0; + if (gnu) + res = elf_compress_gnu (scn, compress ? 1 : 0, flags); + else + res = elf_compress (scn, compress ? ELFCOMPRESS_ZLIB : 0, flags); + + if (res < 0) + error (0, 0, "Couldn't decompress section [%zd] %s: %s", + ndx, name, elf_errmsg (-1)); + else + { + if (compress && res == 0) + { + if (verbose >= 0) + printf ("[%zd] %s NOT compressed, wouldn't be smaller\n", + ndx, name); + } + + if (report_verbose && res > 0) + { + printf ("[%zd] %s %s", ndx, name, + compress ? "compressed" : "decompressed"); + if (newname != NULL) + printf (" -> %s", newname); + + /* Reload shdr, it has changed. */ + GElf_Shdr shdr_mem; + GElf_Shdr *shdr = gelf_getshdr (scn, &shdr_mem); + if (shdr == NULL) + { + error (0, 0, "Couldn't get shdr for section [%zd]", ndx); + return -1; + } + float new = shdr->sh_size; + float orig = orig_size ?: 1; + printf (" (%" PRIu64 " => %" PRIu64 " %.2f%%)\n", + orig_size, shdr->sh_size, (new / orig) * 100); + } + } + + return res; +} + +static int +process_file (const char *fname) +{ + if (verbose > 0) + printf ("processing: %s\n", fname); + + /* The input ELF. */ + int fd = -1; + Elf *elf = NULL; + + /* The output ELF. */ + char *fnew = NULL; + int fdnew = -1; + Elf *elfnew = NULL; + + /* Buffer for (one) new section name if necessary. */ + char *snamebuf = NULL; + + /* String table (and symbol table), if section names need adjusting. */ + struct Ebl_Strtab *names = NULL; + struct Ebl_Strent **scnstrents = NULL; + struct Ebl_Strent **symstrents = NULL; + char **scnnames = NULL; + + /* Section data from names. */ + void *namesbuf = NULL; + + /* Which sections match and need to be (un)compressed. */ + unsigned int *sections = NULL; + + /* How many sections are we talking about? */ + size_t shnum = 0; + +#define WORD_BITS (8U * sizeof (unsigned int)) + void set_section (size_t ndx) + { + sections[ndx / WORD_BITS] |= (1U << (ndx % WORD_BITS)); + } + + bool get_section (size_t ndx) + { + return (sections[ndx / WORD_BITS] & (1U << (ndx % WORD_BITS))) != 0; + } + + int cleanup (int res) + { + elf_end (elf); + close (fd); + + elf_end (elfnew); + close (fdnew); + + if (fnew != NULL) + { + unlink (fnew); + free (fnew); + fnew = NULL; + } + + free (snamebuf); + if (names != NULL) + { + ebl_strtabfree (names); + free (scnstrents); + free (symstrents); + free (namesbuf); + if (scnnames != NULL) + { + for (size_t n = 0; n < shnum; n++) + free (scnnames[n]); + free (scnnames); + } + } + + free (sections); + + return res; + } + + fd = open (fname, O_RDONLY); + if (fd < 0) + { + error (0, errno, "Couldn't open %s\n", fname); + return cleanup (-1); + } + + elf = elf_begin (fd, ELF_C_READ, NULL); + if (elf == NULL) + { + error (0, 0, "Couldn't open ELF file %s for reading: %s", + fname, elf_errmsg (-1)); + return cleanup (-1); + } + + /* We dont' handle ar files (or anything else), we probably should. */ + Elf_Kind kind = elf_kind (elf); + if (kind != ELF_K_ELF) + { + if (kind == ELF_K_AR) + error (0, 0, "Cannot handle ar files: %s", fname); + else + error (0, 0, "Unknown file type: %s", fname); + return cleanup (-1); + } + + struct stat st; + if (fstat (fd, &st) != 0) + { + error (0, errno, "Couldn't fstat %s", fname); + return cleanup (-1); + } + + GElf_Ehdr ehdr; + if (gelf_getehdr (elf, &ehdr) == NULL) + { + error (0, 0, "Couldn't get ehdr for %s: %s", fname, elf_errmsg (-1)); + return cleanup (-1); + } + + /* Get the section header string table. */ + size_t shdrstrndx; + if (elf_getshdrstrndx (elf, &shdrstrndx) != 0) + { + error (0, 0, "Couldn't get section header string table index in %s: %s", + fname, elf_errmsg (-1)); + return cleanup (-1); + } + + /* How many sections are we talking about? */ + if (elf_getshdrnum (elf, &shnum) != 0) + { + error (0, 0, "Couldn't get number of sections in %s: %s", + fname, elf_errmsg (1)); + return cleanup (-1); + } + + if (shnum == 0) + { + error (0, 0, "ELF file %s has no sections", fname); + return cleanup (-1); + } + + sections = xcalloc (shnum / 8 + 1, sizeof (unsigned int)); + + size_t phnum; + if (elf_getphdrnum (elf, &phnum) != 0) + { + error (0, 0, "Couldn't get phdrnum: %s", elf_errmsg (-1)); + return cleanup (-1); + } + + /* Whether we need to adjust any section names (going to/from GNU + naming). If so we'll need to build a new section header string + table. */ + bool adjust_names = false; + + /* If there are phdrs we want to maintain the layout of the + allocated sections in the file. */ + bool layout = phnum != 0; + + /* While going through all sections keep track of last section data + offset if needed to keep the layout. We are responsible for + adding the section offsets and headers (e_shoff) in that case + (which we will place after the last section). */ + GElf_Off last_offset = 0; + if (layout) + last_offset = (ehdr.e_phoff + + gelf_fsize (elf, ELF_T_PHDR, phnum, EV_CURRENT)); + + /* Which section, if any, is a symbol table that shares a string + table with the section header string table? */ + size_t symtabndx = 0; + + /* We do three passes over all sections. + + First an inspection pass over the old Elf to see which section + data needs to be copied and/or transformed, which sections need a + names change and whether there is a symbol table that might need + to be adjusted be if the section header name table is changed. + + Second a collection pass that creates the Elf sections and copies + the data. This pass will compress/decompress section data when + needed. And it will collect all data needed if we'll need to + construct a new string table. Afterwards the new string table is + constructed. + + Third a fixup/adjustment pass over the new Elf that will adjust + any section references (names) and adjust the layout based on the + new sizes of the sections if necessary. This pass is optional if + we aren't responsible for the layout and the section header + string table hasn't been changed. */ + + /* Inspection pass. */ + size_t maxnamelen = 0; + Elf_Scn *scn = NULL; + while ((scn = elf_nextscn (elf, scn)) != NULL) + { + size_t ndx = elf_ndxscn (scn); + if (ndx > shnum) + { + error (0, 0, "Unexpected section number %zd, expected only %zd", + ndx, shnum); + cleanup (-1); + } + + GElf_Shdr shdr_mem; + GElf_Shdr *shdr = gelf_getshdr (scn, &shdr_mem); + if (shdr == NULL) + { + error (0, 0, "Couldn't get shdr for section %zd", ndx); + return cleanup (-1); + } + + const char *sname = elf_strptr (elf, shdrstrndx, shdr->sh_name); + if (sname == NULL) + { + error (0, 0, "Couldn't get name for section %zd", ndx); + return cleanup (-1); + } + + if (section_name_matches (sname)) + { + if (shdr->sh_type != SHT_NOBITS + && (shdr->sh_flags & SHF_ALLOC) == 0) + { + set_section (ndx); + /* Check if we might want to change this section name. */ + if (! adjust_names + && ((type != T_COMPRESS_GNU + && strncmp (sname, ".zdebug", + strlen (".zdebug")) == 0) + || (type == T_COMPRESS_GNU + && strncmp (sname, ".debug", + strlen (".debug")) == 0))) + adjust_names = true; + + /* We need a buffer this large if we change the names. */ + if (adjust_names) + { + size_t slen = strlen (sname); + if (slen > maxnamelen) + maxnamelen = slen; + } + } + else + if (verbose >= 0) + printf ("[%zd] %s ignoring %s section\n", ndx, sname, + (shdr->sh_type == SHT_NOBITS ? "no bits" : "allocated")); + } + + if (shdr->sh_type == SHT_SYMTAB) + { + /* Check if we might have to adjust the symbol name indexes. */ + if (shdr->sh_link == shdrstrndx) + { + if (symtabndx != 0) + { + error (0, 0, + "Multiple symbol tables (%zd, %zd) using the same string table unsupported", symtabndx, ndx); + return cleanup (-1); + } + symtabndx = ndx; + } + } + + /* Keep track of last allocated data offset. */ + if (layout) + if ((shdr->sh_flags & SHF_ALLOC) != 0) + { + GElf_Off off = shdr->sh_offset + (shdr->sh_type != SHT_NOBITS + ? shdr->sh_size : 0); + if (last_offset < off) + last_offset = off; + } + } + + if (adjust_names) + { + names = ebl_strtabinit (true); + if (names == NULL) + { + error (0, 0, "Not enough memory for new strtab"); + return cleanup (-1); + } + scnstrents = xmalloc (shnum + * sizeof (struct Ebl_Strent *)); + scnnames = xcalloc (shnum, sizeof (char *)); + } + + /* Create a new (temporary) ELF file for the result. */ + if (foutput == NULL) + { + size_t fname_len = strlen (fname); + fnew = xmalloc (fname_len + sizeof (".XXXXXX")); + strcpy (mempcpy (fnew, fname, fname_len), ".XXXXXX"); + fdnew = mkstemp (fnew); + } + else + { + fnew = xstrdup (foutput); + fdnew = open (fnew, O_WRONLY | O_CREAT, st.st_mode & ALLPERMS); + } + + if (fdnew < 0) + { + error (0, errno, "Couldn't create output file %s", fnew); + /* Since we didn't create it we don't want to try to unlink it. */ + free (fnew); + fnew = NULL; + return cleanup (-1); + } + + elfnew = elf_begin (fdnew, ELF_C_WRITE, NULL); + if (elfnew == NULL) + { + error (0, 0, "Couldn't open new ELF %s for writing: %s", + fnew, elf_errmsg (-1)); + return cleanup (-1); + } + + /* Create the new ELF header and copy over all the data. */ + if (gelf_newehdr (elfnew, gelf_getclass (elf)) == 0) + { + error (0, 0, "Couldn't create new ehdr: %s", elf_errmsg (-1)); + return cleanup (-1); + } + + GElf_Ehdr newehdr; + if (gelf_getehdr (elfnew, &newehdr) == NULL) + { + error (0, 0, "Couldn't get new ehdr: %s", elf_errmsg (-1)); + return cleanup (-1); + } + + newehdr.e_ident[EI_DATA] = ehdr.e_ident[EI_DATA]; + newehdr.e_type = ehdr.e_type; + newehdr.e_machine = ehdr.e_machine; + newehdr.e_version = ehdr.e_version; + newehdr.e_entry = ehdr.e_entry; + newehdr.e_flags = ehdr.e_flags; + + if (gelf_update_ehdr (elfnew, &newehdr) == 0) + { + error (0, 0, "Couldn't update ehdr: %s", elf_errmsg (-1)); + return cleanup (-1); + } + + /* Copy over the phdrs as is. */ + if (phnum != 0) + { + if (gelf_newphdr (elfnew, phnum) == 0) + { + error (0, 0, "Couldn't create phdrs: %s", elf_errmsg (-1)); + return cleanup (-1); + } + + for (size_t cnt = 0; cnt < phnum; ++cnt) + { + GElf_Phdr phdr_mem; + GElf_Phdr *phdr = gelf_getphdr (elf, cnt, &phdr_mem); + if (phdr == NULL) + { + error (0, 0, "Couldn't get phdr %zd: %s", cnt, elf_errmsg (-1)); + return cleanup (-1); + } + if (gelf_update_phdr (elfnew, cnt, phdr) == 0) + { + error (0, 0, "Couldn't create phdr %zd: %s", cnt, + elf_errmsg (-1)); + return cleanup (-1); + } + } + } + + /* Possibly add a 'z' and zero terminator. */ + if (maxnamelen > 0) + snamebuf = xmalloc (maxnamelen + 2); + + /* We might want to read/adjust the section header strings and + symbol tables. If so, and those sections are to be compressed + then we will have to decompress it during the collection pass and + compress it again in the fixup pass. Don't compress unnecessary + and keep track of whether or not to compress them (later in the + fixup pass). Also record the original size, so we can report the + difference later when we do compress. */ + int shstrtab_compressed = T_UNSET; + size_t shstrtab_size = 0; + char *shstrtab_name = NULL; + char *shstrtab_newname = NULL; + int symtab_compressed = T_UNSET; + size_t symtab_size = 0; + char *symtab_name = NULL; + char *symtab_newname = NULL; + + /* Collection pass. Copy over the sections, (de)compresses matching + sections, collect names of sections and symbol table if + necessary. */ + scn = NULL; + while ((scn = elf_nextscn (elf, scn)) != NULL) + { + size_t ndx = elf_ndxscn (scn); + assert (ndx < shnum); + + /* (de)compress if section matched. */ + char *sname = NULL; + char *newname = NULL; + if (get_section (ndx)) + { + GElf_Shdr shdr_mem; + GElf_Shdr *shdr = gelf_getshdr (scn, &shdr_mem); + if (shdr == NULL) + { + error (0, 0, "Couldn't get shdr for section %zd", ndx); + return cleanup (-1); + } + + uint64_t size = shdr->sh_size; + sname = elf_strptr (elf, shdrstrndx, shdr->sh_name); + if (sname == NULL) + { + error (0, 0, "Couldn't get name for section %zd", ndx); + return cleanup (-1); + } + + /* strdup sname, the shdrstrndx section itself might be + (de)compressed, invalidating the string pointers. */ + sname = xstrdup (sname); + + /* We might want to decompress (and rename), but not + compress during this pass since we might need the section + data in later passes. Skip those sections for now and + compress them in the fixup pass. */ + bool skip_compress_section = (adjust_names + && (ndx == shdrstrndx + || ndx == symtabndx)); + + switch (type) + { + case T_DECOMPRESS: + if ((shdr->sh_flags & SHF_COMPRESSED) != 0) + { + if (compress_section (scn, size, sname, NULL, ndx, + false, false, verbose > 0) < 0) + return cleanup (-1); + } + else if (strncmp (sname, ".zdebug", strlen (".zdebug")) == 0) + { + snamebuf[0] = '.'; + strcpy (&snamebuf[1], &sname[2]); + newname = snamebuf; + if (compress_section (scn, size, sname, newname, ndx, + true, false, verbose > 0) < 0) + return cleanup (-1); + } + else if (verbose > 0) + printf ("[%zd] %s already decompressed\n", ndx, sname); + break; + + case T_COMPRESS_GNU: + if (strncmp (sname, ".debug", strlen (".debug")) == 0) + { + if ((shdr->sh_flags & SHF_COMPRESSED) != 0) + { + /* First decompress to recompress GNU style. + Don't report even when verbose. */ + if (compress_section (scn, size, sname, NULL, ndx, + false, false, false) < 0) + return cleanup (-1); + } + + snamebuf[0] = '.'; + snamebuf[1] = 'z'; + strcpy (&snamebuf[2], &sname[1]); + newname = snamebuf; + + if (skip_compress_section) + { + if (ndx == shdrstrndx) + { + shstrtab_size = size; + shstrtab_compressed = T_COMPRESS_GNU; + shstrtab_name = xstrdup (sname); + shstrtab_newname = xstrdup (newname); + } + else + { + symtab_size = size; + symtab_compressed = T_COMPRESS_GNU; + symtab_name = xstrdup (sname); + symtab_newname = xstrdup (newname); + } + } + else + { + int res = compress_section (scn, size, sname, newname, + ndx, true, true, + verbose > 0); + if (res < 0) + return cleanup (-1); + + if (res == 0) + newname = NULL; + } + } + else if (verbose >= 0) + { + if (strncmp (sname, ".zdebug", strlen (".zdebug")) == 0) + printf ("[%zd] %s unchanged, already GNU compressed", + ndx, sname); + else + printf ("[%zd] %s cannot GNU compress section not starting with .debug\n", + ndx, sname); + } + break; + + case T_COMPRESS_ZLIB: + if ((shdr->sh_flags & SHF_COMPRESSED) == 0) + { + if (strncmp (sname, ".zdebug", strlen (".zdebug")) == 0) + { + /* First decompress to recompress zlib style. + Don't report even when verbose. */ + if (compress_section (scn, size, sname, NULL, ndx, + true, false, false) < 0) + return cleanup (-1); + + snamebuf[0] = '.'; + strcpy (&snamebuf[1], &sname[2]); + newname = snamebuf; + } + + if (skip_compress_section) + { + if (ndx == shdrstrndx) + { + shstrtab_size = size; + shstrtab_compressed = T_COMPRESS_ZLIB; + shstrtab_name = xstrdup (sname); + shstrtab_newname = (newname == NULL + ? NULL : xstrdup (newname)); + } + else + { + symtab_size = size; + symtab_compressed = T_COMPRESS_ZLIB; + symtab_name = xstrdup (sname); + symtab_newname = (newname == NULL + ? NULL : xstrdup (newname)); + } + } + else if (compress_section (scn, size, sname, newname, ndx, + false, true, verbose > 0) < 0) + return cleanup (-1); + } + else if (verbose > 0) + printf ("[%zd] %s already compressed\n", ndx, sname); + break; + } + + free (sname); + } + + Elf_Scn *newscn = elf_newscn (elfnew); + if (newscn == NULL) + { + error (0, 0, "Couldn't create new section %zd", ndx); + return cleanup (-1); + } + + GElf_Shdr shdr_mem; + GElf_Shdr *shdr = gelf_getshdr (scn, &shdr_mem); + if (shdr == NULL) + { + error (0, 0, "Couldn't get shdr for section %zd", ndx); + return cleanup (-1); + } + + if (gelf_update_shdr (newscn, shdr) == 0) + { + error (0, 0, "Couldn't update section header %zd", ndx); + return cleanup (-1); + } + + /* Except for the section header string table all data can be + copied as is. The section header string table will be + created later and the symbol table might be fixed up if + necessary. */ + if (! adjust_names || ndx != shdrstrndx) + { + Elf_Data *data = elf_getdata (scn, NULL); + if (data == NULL) + { + error (0, 0, "Couldn't get data from section %zd", ndx); + return cleanup (-1); + } + + Elf_Data *newdata = elf_newdata (newscn); + if (newdata == NULL) + { + error (0, 0, "Couldn't create new data for section %zd", ndx); + return cleanup (-1); + } + + *newdata = *data; + } + + /* Keep track of the (new) section names. */ + if (adjust_names) + { + char *name; + if (newname != NULL) + name = newname; + else + { + name = elf_strptr (elf, shdrstrndx, shdr->sh_name); + if (name == NULL) + { + error (0, 0, "Couldn't get name for section [%zd]", ndx); + return cleanup (-1); + } + } + + /* We need to keep a copy of the name till the strtab is done. */ + name = scnnames[ndx] = xstrdup (name); + if ((scnstrents[ndx] = ebl_strtabadd (names, name, 0)) == NULL) + { + error (0, 0, "No memory to add section name string table"); + return cleanup (-1); + } + + /* If the symtab shares strings then add those too. */ + if (ndx == symtabndx) + { + /* If the section is (still) compressed we'll need to + uncompress it first to adjust the data, then + recompress it in the fixup pass. */ + if (symtab_compressed == T_UNSET) + { + size_t size = shdr->sh_size; + if ((shdr->sh_flags == SHF_COMPRESSED) != 0) + { + /* Don't report the (internal) uncompression. */ + if (compress_section (newscn, size, sname, NULL, ndx, + false, false, false) < 0) + return cleanup (-1); + + symtab_size = size; + symtab_compressed = T_COMPRESS_ZLIB; + } + else if (strncmp (name, ".zdebug", strlen (".zdebug")) == 0) + { + /* Don't report the (internal) uncompression. */ + if (compress_section (newscn, size, sname, NULL, ndx, + true, false, false) < 0) + return cleanup (-1); + + symtab_size = size; + symtab_compressed = T_COMPRESS_GNU; + } + } + + Elf_Data *symd = elf_getdata (newscn, NULL); + if (symd == NULL) + { + error (0, 0, "Couldn't get symtab data for section [%zd] %s", + ndx, name); + return cleanup (-1); + } + size_t elsize = gelf_fsize (elfnew, ELF_T_SYM, 1, EV_CURRENT); + size_t syms = symd->d_size / elsize; + symstrents = xmalloc (syms * sizeof (struct Ebl_Strent *)); + for (size_t i = 0; i < syms; i++) + { + GElf_Sym sym_mem; + GElf_Sym *sym = gelf_getsym (symd, i, &sym_mem); + if (sym == NULL) + { + error (0, 0, "Couldn't get symbol %zd", i); + return cleanup (-1); + } + if (sym->st_name != 0) + { + /* Note we take the name from the original ELF, + since the new one will not have setup the + strtab yet. */ + const char *symname = elf_strptr (elf, shdrstrndx, + sym->st_name); + if (symname == NULL) + { + error (0, 0, "Couldn't get symbol %zd name", i); + return cleanup (-1); + } + symstrents[i] = ebl_strtabadd (names, symname, 0); + if (symstrents[i] == NULL) + { + error (0, 0, "No memory to add to symbol name"); + return cleanup (-1); + } + } + } + } + } + } + + if (adjust_names) + { + /* We got all needed strings, put the new data in the shstrtab. */ + if (verbose > 0) + printf ("[%zd] Updating section string table\n", shdrstrndx); + + scn = elf_getscn (elfnew, shdrstrndx); + if (scn == NULL) + { + error (0, 0, "Couldn't get new section header string table [%zd]", + shdrstrndx); + return cleanup (-1); + } + + Elf_Data *data = elf_newdata (scn); + if (data == NULL) + { + error (0, 0, "Couldn't create new section header string table data"); + return cleanup (-1); + } + ebl_strtabfinalize (names, data); + namesbuf = data->d_buf; + + GElf_Shdr shdr_mem; + GElf_Shdr *shdr = gelf_getshdr (scn, &shdr_mem); + if (shdr == NULL) + { + error (0, 0, "Couldn't get shdr for new section strings %zd", + shdrstrndx); + return cleanup (-1); + } + + /* Note that we also might have to compress and possibly set + sh_off below */ + shdr->sh_name = ebl_strtaboffset (scnstrents[shdrstrndx]); + shdr->sh_type = SHT_STRTAB; + shdr->sh_flags = 0; + shdr->sh_addr = 0; + shdr->sh_offset = 0; + shdr->sh_size = data->d_size; + shdr->sh_link = SHN_UNDEF; + shdr->sh_info = SHN_UNDEF; + shdr->sh_addralign = 1; + shdr->sh_entsize = 0; + + if (gelf_update_shdr (scn, shdr) == 0) + { + error (0, 0, "Couldn't update new section strings [%zd]", + shdrstrndx); + return cleanup (-1); + } + + /* We might have to compress the data if the user asked us to, + or if the section was already compressed (and the user didn't + ask for decompression). Note somewhat identical code for + symtab below. */ + if (shstrtab_compressed == T_UNSET) + { + /* The user didn't ask for compression, but maybe it was + compressed in the original ELF file. */ + Elf_Scn *oldscn = elf_getscn (elf, shdrstrndx); + if (oldscn == NULL) + { + error (0, 0, "Couldn't get section header string table [%zd]", + shdrstrndx); + return cleanup (-1); + } + + shdr = gelf_getshdr (oldscn, &shdr_mem); + if (shdr == NULL) + { + error (0, 0, "Couldn't get shdr for old section strings [%zd]", + shdrstrndx); + return cleanup (-1); + } + + shstrtab_name = elf_strptr (elf, shdrstrndx, shdr->sh_name); + if (shstrtab_name == NULL) + { + error (0, 0, "Couldn't get name for old section strings [%zd]", + shdrstrndx); + return cleanup (-1); + } + + shstrtab_size = shdr->sh_size; + if ((shdr->sh_flags & SHF_COMPRESSED) != 0) + shstrtab_compressed = T_COMPRESS_ZLIB; + else if (strncmp (shstrtab_name, ".zdebug", strlen (".zdebug")) == 0) + shstrtab_compressed = T_COMPRESS_GNU; + } + + /* Should we (re)compress? */ + if (shstrtab_compressed != T_UNSET) + { + if (compress_section (scn, shstrtab_size, shstrtab_name, + shstrtab_newname, shdrstrndx, + shstrtab_compressed == T_COMPRESS_GNU, + true, verbose > 0) < 0) + return cleanup (-1); + } + } + + /* Make sure to re-get the new ehdr. Adding phdrs and shdrs will + have changed it. */ + if (gelf_getehdr (elfnew, &newehdr) == NULL) + { + error (0, 0, "Couldn't re-get new ehdr: %s", elf_errmsg (-1)); + return cleanup (-1); + } + + /* Set this after the sections have been created, otherwise section + zero might not exist yet. */ + if (setshdrstrndx (elfnew, &newehdr, shdrstrndx) != 0) + { + error (0, 0, "Couldn't set new shdrstrndx: %s", elf_errmsg (-1)); + return cleanup (-1); + } + + /* Fixup pass. Adjust string table references, symbol table and + layout if necessary. */ + if (layout || adjust_names) + { + scn = NULL; + while ((scn = elf_nextscn (elfnew, scn)) != NULL) + { + size_t ndx = elf_ndxscn (scn); + + GElf_Shdr shdr_mem; + GElf_Shdr *shdr = gelf_getshdr (scn, &shdr_mem); + if (shdr == NULL) + { + error (0, 0, "Couldn't get shdr for section %zd", ndx); + return cleanup (-1); + } + + /* Keep the offset of allocated sections so they are at the + same place in the file. Add (possibly changed) + unallocated ones after the allocated ones. */ + if ((shdr->sh_flags & SHF_ALLOC) == 0) + { + /* Zero means one. No alignment constraints. */ + size_t addralign = shdr->sh_addralign ?: 1; + last_offset = (last_offset + addralign - 1) & ~(addralign - 1); + shdr->sh_offset = last_offset; + if (shdr->sh_type != SHT_NOBITS) + last_offset += shdr->sh_size; + } + + if (adjust_names) + shdr->sh_name = ebl_strtaboffset (scnstrents[ndx]); + + if (gelf_update_shdr (scn, shdr) == 0) + { + error (0, 0, "Couldn't update section header %zd", ndx); + return cleanup (-1); + } + + if (adjust_names && ndx == symtabndx) + { + if (verbose > 0) + printf ("[%zd] Updating symbol table\n", symtabndx); + + Elf_Data *symd = elf_getdata (scn, NULL); + if (symd == NULL) + { + error (0, 0, "Couldn't get new symtab data section [%zd]", + ndx); + return cleanup (-1); + } + size_t elsize = gelf_fsize (elfnew, ELF_T_SYM, 1, EV_CURRENT); + size_t syms = symd->d_size / elsize; + for (size_t i = 0; i < syms; i++) + { + GElf_Sym sym_mem; + GElf_Sym *sym = gelf_getsym (symd, i, &sym_mem); + if (sym == NULL) + { + error (0, 0, "2 Couldn't get symbol %zd", i); + return cleanup (-1); + } + + if (sym->st_name != 0) + { + sym->st_name = ebl_strtaboffset (symstrents[i]); + + if (gelf_update_sym (symd, i, sym) == 0) + { + error (0, 0, "Couldn't update symbol %zd", i); + return cleanup (-1); + } + } + } + + /* We might have to compress the data if the user asked + us to, or if the section was already compressed (and + the user didn't ask for decompression). Note + somewhat identical code for shstrtab above. */ + if (symtab_compressed == T_UNSET) + { + /* The user didn't ask for compression, but maybe it was + compressed in the original ELF file. */ + Elf_Scn *oldscn = elf_getscn (elf, symtabndx); + if (oldscn == NULL) + { + error (0, 0, "Couldn't get symbol table [%zd]", + symtabndx); + return cleanup (-1); + } + + shdr = gelf_getshdr (oldscn, &shdr_mem); + if (shdr == NULL) + { + error (0, 0, "Couldn't get old symbol table shdr [%zd]", + symtabndx); + return cleanup (-1); + } + + symtab_name = elf_strptr (elf, shdrstrndx, shdr->sh_name); + if (symtab_name == NULL) + { + error (0, 0, "Couldn't get old symbol table name [%zd]", + symtabndx); + return cleanup (-1); + } + + symtab_size = shdr->sh_size; + if ((shdr->sh_flags & SHF_COMPRESSED) != 0) + symtab_compressed = T_COMPRESS_ZLIB; + else if (strncmp (symtab_name, ".zdebug", + strlen (".zdebug")) == 0) + symtab_compressed = T_COMPRESS_GNU; + } + + /* Should we (re)compress? */ + if (symtab_compressed != T_UNSET) + { + if (compress_section (scn, symtab_size, symtab_name, + symtab_newname, symtabndx, + symtab_compressed == T_COMPRESS_GNU, + true, verbose > 0) < 0) + return cleanup (-1); + } + } + } + } + + /* If we have phdrs we want elf_update to layout the SHF_ALLOC + sections precisely as in the original file. In that case we are + also responsible for setting phoff and shoff */ + if (layout) + { + if (gelf_getehdr (elfnew, &newehdr) == NULL) + { + error (0, 0, "Couldn't get ehdr: %s", elf_errmsg (-1)); + return cleanup (-1); + } + + /* Position the shdrs after the last (unallocated) section. */ + const size_t offsize = gelf_fsize (elfnew, ELF_T_OFF, 1, EV_CURRENT); + newehdr.e_shoff = ((last_offset + offsize - 1) + & ~((GElf_Off) (offsize - 1))); + + /* The phdrs go in the same place as in the original file. + Normally right after the ELF header. */ + newehdr.e_phoff = ehdr.e_phoff; + + if (gelf_update_ehdr (elfnew, &newehdr) == 0) + { + error (0, 0, "Couldn't update ehdr: %s", elf_errmsg (-1)); + return cleanup (-1); + } + } + + elf_flagelf (elfnew, ELF_C_SET, ((layout ? ELF_F_LAYOUT : 0) + | (permissive ? ELF_F_PERMISSIVE : 0))); + + if (elf_update (elfnew, ELF_C_WRITE) < 0) + { + error (0, 0, "Couldn't write %s: %s", fnew, elf_errmsg (-1)); + return cleanup (-1); + } + + elf_end (elfnew); + elfnew = NULL; + + /* Try to match mode and owner.group of the original file. */ + if (fchmod (fdnew, st.st_mode & ALLPERMS) != 0) + if (verbose >= 0) + error (0, errno, "Couldn't fchmod %s", fnew); + if (fchown (fdnew, st.st_uid, st.st_gid) != 0) + if (verbose >= 0) + error (0, errno, "Couldn't fchown %s", fnew); + + /* Finally replace the old file with the new file. */ + if (foutput == NULL) + if (rename (fnew, fname) != 0) + { + error (0, errno, "Couldn't rename %s to %s", fnew, fname); + return cleanup (-1); + } + + /* We are finally done with the new file, don't unlink it now. */ + free (fnew); + fnew = NULL; + + return cleanup (0); +} + +int +main (int argc, char **argv) +{ + const struct argp_option options[] = + { + { "output", 'o', "FILE", 0, + N_("Place (de)compressed output into FILE"), + 0 }, + { "type", 't', "TYPE", 0, + N_("What type of compression to apply. TYPE can be 'none' (decompress), 'zlib' (ELF ZLIB compression, the default, 'zlib-gabi' is an alias) or 'zlib-gnu' (.zdebug GNU style compression, 'gnu' is an alias)"), + 0 }, + { "name", 'n', "SECTION", 0, + N_("SECTION name to (de)compress, SECTION is an extended wildcard pattern (defaults to '.?(z)debug*')"), + 0 }, + { "verbose", 'v', NULL, 0, + N_("Print a message for each section being (de)compressed"), + 0 }, + { "force", 'f', NULL, 0, + N_("Force compression of section even if it would become larger"), + 0 }, + { "permissive", 'p', NULL, 0, + N_("Relax a few rules to handle slightly broken ELF files"), + 0 }, + { "quiet", 'q', NULL, 0, + N_("Be silent when a section cannot be compressed"), + 0 }, + { NULL, 0, NULL, 0, NULL, 0 } + }; + + const struct argp argp = + { + .options = options, + .parser = parse_opt, + .args_doc = N_("FILE..."), + .doc = N_("Compress or decompress sections in an ELF file.") + }; + + int remaining; + if (argp_parse (&argp, argc, argv, 0, &remaining, NULL) != 0) + return EXIT_FAILURE; + + /* Should already be handled by ARGP_KEY_NO_ARGS case above, + just sanity check. */ + if (remaining >= argc) + error (EXIT_FAILURE, 0, N_("No input file given")); + + /* Likewise for the ARGP_KEY_ARGS case above, an extra sanity check. */ + if (foutput != NULL && remaining + 1 < argc) + error (EXIT_FAILURE, 0, + N_("Only one input file allowed together with '-o'")); + + elf_version (EV_CURRENT); + + /* Process all the remaining files. */ + int result = 0; + do + result |= process_file (argv[remaining]); + while (++remaining < argc); + + free_patterns (); + return result; +} |