X-Git-Url: https://git.librecmc.org/?a=blobdiff_plain;f=archival%2Ftar.c;h=375e838d279a54a4a3ae7fbd748c058cff7eb313;hb=0d0260fd1e55c39525660370094d090bc5412fe5;hp=6a4c4e7e8b0dffdc6007904cfd51ce91469fc706;hpb=c61852a02bd2d61682235ba3185173f527313827;p=oweals%2Fbusybox.git diff --git a/archival/tar.c b/archival/tar.c index 6a4c4e7e8..375e838d2 100644 --- a/archival/tar.c +++ b/archival/tar.c @@ -3,7 +3,7 @@ * Mini tar implementation for busybox * * Modified to use common extraction code used by ar, cpio, dpkg-deb, dpkg - * Glenn McGrath + * by Glenn McGrath * * Note, that as of BusyBox-0.43, tar has been completely rewritten from the * ground up. It still has remnants of the old code lying about, but it is @@ -15,80 +15,88 @@ * Copyright (c) 1999 by David I. Bell * Permission is granted to use, distribute, or modify this source, * provided that this copyright notice remains intact. - * Permission to distribute sash derived code under the GPL has been granted. + * Permission to distribute sash derived code under GPL has been granted. * * Based in part on the tar implementation from busybox-0.28 * Copyright (C) 1995 Bruce Perens * - * Licensed under GPLv2 or later, see file LICENSE in this tarball for details. + * Licensed under GPLv2 or later, see file LICENSE in this source tree. + */ + +/* TODO: security with -C DESTDIR option can be enhanced. + * Consider tar file created via: + * $ tar cvf bug.tar anything.txt + * $ ln -s /tmp symlink + * $ tar --append -f bug.tar symlink + * $ rm symlink + * $ mkdir symlink + * $ tar --append -f bug.tar symlink/evil.py + * + * This will result in an archive which contains: + * $ tar --list -f bug.tar + * anything.txt + * symlink + * symlink/evil.py + * + * Untarring it puts evil.py in '/tmp' even if the -C DESTDIR is given. + * This doesn't feel right, and IIRC GNU tar doesn't do that. */ -#include "busybox.h" -#include "unarchive.h" #include -#include +#include "libbb.h" +#include "archive.h" +/* FIXME: Stop using this non-standard feature */ +#ifndef FNM_LEADING_DIR +# define FNM_LEADING_DIR 0 +#endif -#if ENABLE_FEATURE_TAR_CREATE -/* Tar file constants */ - -#define TAR_BLOCK_SIZE 512 - -/* POSIX tar Header Block, from POSIX 1003.1-1990 */ -#define NAME_SIZE 100 -#define NAME_SIZE_STR "100" -struct TarHeader { /* byte offset */ - char name[NAME_SIZE]; /* 0-99 */ - char mode[8]; /* 100-107 */ - char uid[8]; /* 108-115 */ - char gid[8]; /* 116-123 */ - char size[12]; /* 124-135 */ - char mtime[12]; /* 136-147 */ - char chksum[8]; /* 148-155 */ - char typeflag; /* 156-156 */ - char linkname[NAME_SIZE]; /* 157-256 */ - char magic[6]; /* 257-262 */ - char version[2]; /* 263-264 */ - char uname[32]; /* 265-296 */ - char gname[32]; /* 297-328 */ - char devmajor[8]; /* 329-336 */ - char devminor[8]; /* 337-344 */ - char prefix[155]; /* 345-499 */ - char padding[12]; /* 500-512 (pad to exactly the TAR_BLOCK_SIZE) */ -}; -typedef struct TarHeader TarHeader; +//#define DBG(fmt, ...) bb_error_msg("%s: " fmt, __func__, ## __VA_ARGS__) +#define DBG(...) ((void)0) + + +#define block_buf bb_common_bufsiz1 + + +#if !ENABLE_FEATURE_SEAMLESS_GZ && !ENABLE_FEATURE_SEAMLESS_BZ2 +/* Do not pass gzip flag to writeTarFile() */ +#define writeTarFile(tar_fd, verboseFlag, dereferenceFlag, include, exclude, gzip) \ + writeTarFile(tar_fd, verboseFlag, dereferenceFlag, include, exclude) +#endif + + +#if ENABLE_FEATURE_TAR_CREATE /* ** writeTarFile(), writeFileToTarball(), and writeTarHeader() are ** the only functions that deal with the HardLinkInfo structure. ** Even these functions use the xxxHardLinkInfo() functions. */ -typedef struct HardLinkInfo HardLinkInfo; -struct HardLinkInfo { - HardLinkInfo *next; /* Next entry in list */ - dev_t dev; /* Device number */ - ino_t ino; /* Inode number */ - short linkCount; /* (Hard) Link Count */ - char name[1]; /* Start of filename (must be last) */ -}; +typedef struct HardLinkInfo { + struct HardLinkInfo *next; /* Next entry in list */ + dev_t dev; /* Device number */ + ino_t ino; /* Inode number */ +// short linkCount; /* (Hard) Link Count */ + char name[1]; /* Start of filename (must be last) */ +} HardLinkInfo; /* Some info to be carried along when creating a new tarball */ -struct TarBallInfo { - int tarFd; /* Open-for-write file descriptor - for the tarball */ - struct stat statBuf; /* Stat info for the tarball, letting - us know the inode and device that the - tarball lives, so we can avoid trying - to include the tarball into itself */ - int verboseFlag; /* Whether to print extra stuff or not */ - const llist_t *excludeList; /* List of files to not include */ - HardLinkInfo *hlInfoHead; /* Hard Link Tracking Information */ - HardLinkInfo *hlInfo; /* Hard Link Info for the current file */ -}; -typedef struct TarBallInfo TarBallInfo; +typedef struct TarBallInfo { + int tarFd; /* Open-for-write file descriptor + * for the tarball */ + int verboseFlag; /* Whether to print extra stuff or not */ + const llist_t *excludeList; /* List of files to not include */ + HardLinkInfo *hlInfoHead; /* Hard Link Tracking Information */ + HardLinkInfo *hlInfo; /* Hard Link Info for the current file */ +//TODO: save only st_dev + st_ino + struct stat tarFileStatBuf; /* Stat info for the tarball, letting + * us know the inode and device that the + * tarball lives, so we can avoid trying + * to include the tarball into itself */ +} TarBallInfo; /* A nice enum with all the possible tar file content types */ -enum TarFileType { +enum { REGTYPE = '0', /* regular file */ REGTYPE0 = '\0', /* regular file (ancient bug compat) */ LNKTYPE = '1', /* hard link */ @@ -101,10 +109,9 @@ enum TarFileType { GNULONGLINK = 'K', /* GNU long (>100 chars) link name */ GNULONGNAME = 'L', /* GNU long (>100 chars) file name */ }; -typedef enum TarFileType TarFileType; /* Might be faster (and bigger) if the dev/ino were stored in numeric order;) */ -static void addHardLinkInfo(HardLinkInfo ** hlInfoHeadPtr, +static void addHardLinkInfo(HardLinkInfo **hlInfoHeadPtr, struct stat *statbuf, const char *fileName) { @@ -116,11 +123,11 @@ static void addHardLinkInfo(HardLinkInfo ** hlInfoHeadPtr, *hlInfoHeadPtr = hlInfo; hlInfo->dev = statbuf->st_dev; hlInfo->ino = statbuf->st_ino; - hlInfo->linkCount = statbuf->st_nlink; +// hlInfo->linkCount = statbuf->st_nlink; strcpy(hlInfo->name, fileName); } -static void freeHardLinkInfo(HardLinkInfo ** hlInfoHeadPtr) +static void freeHardLinkInfo(HardLinkInfo **hlInfoHeadPtr) { HardLinkInfo *hlInfo; HardLinkInfo *hlInfoNext; @@ -134,15 +141,18 @@ static void freeHardLinkInfo(HardLinkInfo ** hlInfoHeadPtr) } *hlInfoHeadPtr = NULL; } - return; } -/* Might be faster (and bigger) if the dev/ino were stored in numeric order;) */ -static HardLinkInfo *findHardLinkInfo(HardLinkInfo * hlInfo, struct stat *statbuf) +/* Might be faster (and bigger) if the dev/ino were stored in numeric order ;) */ +static HardLinkInfo *findHardLinkInfo(HardLinkInfo *hlInfo, struct stat *statbuf) { while (hlInfo) { - if ((statbuf->st_ino == hlInfo->ino) && (statbuf->st_dev == hlInfo->dev)) + if (statbuf->st_ino == hlInfo->ino + && statbuf->st_dev == hlInfo->dev + ) { + DBG("found hardlink:'%s'", hlInfo->name); break; + } hlInfo = hlInfo->next; } return hlInfo; @@ -153,7 +163,7 @@ static HardLinkInfo *findHardLinkInfo(HardLinkInfo * hlInfo, struct stat *statbu * Stores low-order bits only if whole value does not fit. */ static void putOctal(char *cp, int len, off_t value) { - char tempBuffer[sizeof(off_t)*3+1]; + char tempBuffer[sizeof(off_t)*3 + 1]; char *tempString = tempBuffer; int width; @@ -171,8 +181,11 @@ static void putOctal(char *cp, int len, off_t value) } #define PUT_OCTAL(a, b) putOctal((a), sizeof(a), (b)) -static void chksum_and_xwrite(int fd, struct TarHeader* hp) +static void chksum_and_xwrite(int fd, struct tar_header_t* hp) { + /* POSIX says that checksum is done on unsigned bytes + * (Sun and HP-UX gets it wrong... more details in + * GNU tar source) */ const unsigned char *cp; int chksum, size; @@ -210,14 +223,14 @@ static void writeLongname(int fd, int type, const char *name, int dir) "00000000000", "00000000000", }; - struct TarHeader header; + struct tar_header_t header; int size; dir = !!dir; /* normalize: 0/1 */ size = strlen(name) + 1 + dir; /* GNU tar uses strlen+1 */ /* + dir: account for possible '/' */ - bzero(&header, sizeof(header)); + memset(&header, 0, sizeof(header)); strcpy(header.name, "././@LongLink"); memcpy(header.mode, prefilled.mode, sizeof(prefilled)); PUT_OCTAL(header.size, size); @@ -231,22 +244,18 @@ static void writeLongname(int fd, int type, const char *name, int dir) xwrite(fd, name, size - dir); xwrite(fd, "/", dir); size = (-size) & (TAR_BLOCK_SIZE-1); - bzero(&header, size); + memset(&header, 0, size); xwrite(fd, &header, size); } #endif /* Write out a tar header for the specified file/directory/whatever */ -void BUG_tar_header_size(void); static int writeTarHeader(struct TarBallInfo *tbInfo, const char *header_name, const char *fileName, struct stat *statbuf) { - struct TarHeader header; + struct tar_header_t header; - if (sizeof(header) != 512) - BUG_tar_header_size(); - - bzero(&header, sizeof(struct TarHeader)); + memset(&header, 0, sizeof(header)); strncpy(header.name, header_name, sizeof(header.name)); @@ -255,7 +264,8 @@ static int writeTarHeader(struct TarBallInfo *tbInfo, PUT_OCTAL(header.uid, statbuf->st_uid); PUT_OCTAL(header.gid, statbuf->st_gid); memset(header.size, '0', sizeof(header.size)-1); /* Regular file size is handled later */ - PUT_OCTAL(header.mtime, statbuf->st_mtime); + /* users report that files with negative st_mtime cause trouble, so: */ + PUT_OCTAL(header.mtime, statbuf->st_mtime >= 0 ? statbuf->st_mtime : 0); /* Enter the user and group names */ safe_strncpy(header.uname, get_cached_username(statbuf->st_uid), sizeof(header.uname)); @@ -273,8 +283,8 @@ static int writeTarHeader(struct TarBallInfo *tbInfo, tbInfo->hlInfo->name, 0); #endif } else if (S_ISLNK(statbuf->st_mode)) { - char *lpath = xreadlink(fileName); - if (!lpath) /* Already printed err msg inside xreadlink() */ + char *lpath = xmalloc_readlink_or_warn(fileName); + if (!lpath) return FALSE; header.typeflag = SYMTYPE; strncpy(header.linkname, lpath, sizeof(header.linkname)); @@ -307,15 +317,42 @@ static int writeTarHeader(struct TarBallInfo *tbInfo, } else if (S_ISFIFO(statbuf->st_mode)) { header.typeflag = FIFOTYPE; } else if (S_ISREG(statbuf->st_mode)) { - if (sizeof(statbuf->st_size) > 4 - && statbuf->st_size > (off_t)0777777777777LL + /* header.size field is 12 bytes long */ + /* Does octal-encoded size fit? */ + uoff_t filesize = statbuf->st_size; + if (sizeof(filesize) <= 4 + || filesize <= (uoff_t)0777777777777LL ) { - bb_error_msg_and_die("cannot store file '%s' " - "of size %"OFF_FMT"d, aborting", + PUT_OCTAL(header.size, filesize); + } + /* Does base256-encoded size fit? + * It always does unless off_t is wider than 64 bits. + */ + else if (ENABLE_FEATURE_TAR_GNU_EXTENSIONS +#if ULLONG_MAX > 0xffffffffffffffffLL /* 2^64-1 */ + && (filesize <= 0x3fffffffffffffffffffffffLL) +#endif + ) { + /* GNU tar uses "base-256 encoding" for very large numbers. + * Encoding is binary, with highest bit always set as a marker + * and sign in next-highest bit: + * 80 00 .. 00 - zero + * bf ff .. ff - largest positive number + * ff ff .. ff - minus 1 + * c0 00 .. 00 - smallest negative number + */ + char *p8 = header.size + sizeof(header.size); + do { + *--p8 = (uint8_t)filesize; + filesize >>= 8; + } while (p8 != header.size); + *p8 |= 0x80; + } else { + bb_error_msg_and_die("can't store file '%s' " + "of size %"OFF_FMT"u, aborting", fileName, statbuf->st_size); } header.typeflag = REGTYPE; - PUT_OCTAL(header.size, statbuf->st_size); } else { bb_error_msg("%s: unknown file type", fileName); return FALSE; @@ -336,7 +373,8 @@ static int writeTarHeader(struct TarBallInfo *tbInfo, if (tbInfo->verboseFlag) { FILE *vbFd = stdout; - if (tbInfo->tarFd == STDOUT_FILENO) /* If the archive goes to stdout, verbose to stderr */ + /* If archive goes to stdout, verbose goes to stderr */ + if (tbInfo->tarFd == STDOUT_FILENO) vbFd = stderr; /* GNU "tar cvvf" prints "extended" listing a-la "ls -l" */ /* We don't have such excesses here: for us "v" == "vv" */ @@ -348,22 +386,25 @@ static int writeTarHeader(struct TarBallInfo *tbInfo, return TRUE; } -# if ENABLE_FEATURE_TAR_FROM +#if ENABLE_FEATURE_TAR_FROM static int exclude_file(const llist_t *excluded_files, const char *file) { while (excluded_files) { if (excluded_files->data[0] == '/') { if (fnmatch(excluded_files->data, file, - FNM_PATHNAME | FNM_LEADING_DIR) == 0) + FNM_PATHNAME | FNM_LEADING_DIR) == 0) return 1; } else { const char *p; for (p = file; p[0] != '\0'; p++) { - if ((p == file || p[-1] == '/') && p[0] != '/' && - fnmatch(excluded_files->data, p, - FNM_PATHNAME | FNM_LEADING_DIR) == 0) + if ((p == file || p[-1] == '/') + && p[0] != '/' + && fnmatch(excluded_files->data, p, + FNM_PATHNAME | FNM_LEADING_DIR) == 0 + ) { return 1; + } } } excluded_files = excluded_files->link; @@ -371,17 +412,31 @@ static int exclude_file(const llist_t *excluded_files, const char *file) return 0; } -# else -#define exclude_file(excluded_files, file) 0 -# endif +#else +# define exclude_file(excluded_files, file) 0 +#endif -static int writeFileToTarball(const char *fileName, struct stat *statbuf, - void *userData, int depth) +static int FAST_FUNC writeFileToTarball(const char *fileName, struct stat *statbuf, + void *userData, int depth UNUSED_PARAM) { struct TarBallInfo *tbInfo = (struct TarBallInfo *) userData; const char *header_name; int inputFileFd = -1; + DBG("writeFileToTarball('%s')", fileName); + + /* Strip leading '/' and such (must be before memorizing hardlink's name) */ + header_name = strip_unsafe_prefix(fileName); + + if (header_name[0] == '\0') + return TRUE; + + /* It is against the rules to archive a socket */ + if (S_ISSOCK(statbuf->st_mode)) { + bb_error_msg("%s: socket ignored", fileName); + return TRUE; + } + /* * Check to see if we are dealing with a hard link. * If so - @@ -390,57 +445,40 @@ static int writeFileToTarball(const char *fileName, struct stat *statbuf, * by adding the file information to the HardLinkInfo linked list. */ tbInfo->hlInfo = NULL; - if (statbuf->st_nlink > 1) { + if (!S_ISDIR(statbuf->st_mode) && statbuf->st_nlink > 1) { + DBG("'%s': st_nlink > 1", header_name); tbInfo->hlInfo = findHardLinkInfo(tbInfo->hlInfoHead, statbuf); - if (tbInfo->hlInfo == NULL) - addHardLinkInfo(&tbInfo->hlInfoHead, statbuf, fileName); - } - - /* It is against the rules to archive a socket */ - if (S_ISSOCK(statbuf->st_mode)) { - bb_error_msg("%s: socket ignored", fileName); - return TRUE; + if (tbInfo->hlInfo == NULL) { + DBG("'%s': addHardLinkInfo", header_name); + addHardLinkInfo(&tbInfo->hlInfoHead, statbuf, header_name); + } } /* It is a bad idea to store the archive we are in the process of creating, * so check the device and inode to be sure that this particular file isn't * the new tarball */ - if (tbInfo->statBuf.st_dev == statbuf->st_dev && - tbInfo->statBuf.st_ino == statbuf->st_ino) { + if (tbInfo->tarFileStatBuf.st_dev == statbuf->st_dev + && tbInfo->tarFileStatBuf.st_ino == statbuf->st_ino + ) { bb_error_msg("%s: file is the archive; skipping", fileName); return TRUE; } - header_name = fileName; - while (header_name[0] == '/') { - static int alreadyWarned = FALSE; - - if (alreadyWarned == FALSE) { - bb_error_msg("removing leading '/' from member names"); - alreadyWarned = TRUE; - } - header_name++; - } + if (exclude_file(tbInfo->excludeList, header_name)) + return SKIP; #if !ENABLE_FEATURE_TAR_GNU_EXTENSIONS - if (strlen(fileName) >= NAME_SIZE) { + if (strlen(header_name) >= NAME_SIZE) { bb_error_msg("names longer than "NAME_SIZE_STR" chars not supported"); return TRUE; } #endif - if (header_name[0] == '\0') - return TRUE; - - if (exclude_file(tbInfo->excludeList, header_name)) - return SKIP; - /* Is this a regular file? */ if (tbInfo->hlInfo == NULL && S_ISREG(statbuf->st_mode)) { /* open the file we want to archive, and make sure all is well */ - inputFileFd = open(fileName, O_RDONLY); + inputFileFd = open_or_warn(fileName, O_RDONLY); if (inputFileFd < 0) { - bb_perror_msg("%s: cannot open", fileName); return FALSE; } } @@ -452,123 +490,148 @@ static int writeFileToTarball(const char *fileName, struct stat *statbuf, /* If it was a regular file, write out the body */ if (inputFileFd >= 0) { - off_t readSize = 0; - - /* write the file to the archive */ - readSize = bb_copyfd_size(inputFileFd, tbInfo->tarFd, statbuf->st_size); - /* readSize < 0 means that error was already reported */ - if (readSize != statbuf->st_size && readSize >= 0) { - /* Deadly. We record size into header first, */ - /* and then write out file. If file shrinks in between, */ - /* tar will be corrupted. So bail out. */ - /* NB: GNU tar 1.16 warns and pads with zeroes */ - /* or even seeks back and updates header */ - bb_error_msg_and_die("short read from %s, aborting", fileName); - } + size_t readSize; + /* Write the file to the archive. */ + /* We record size into header first, */ + /* and then write out file. If file shrinks in between, */ + /* tar will be corrupted. So we don't allow for that. */ + /* NB: GNU tar 1.16 warns and pads with zeroes */ + /* or even seeks back and updates header */ + bb_copyfd_exact_size(inputFileFd, tbInfo->tarFd, statbuf->st_size); + ////off_t readSize; + ////readSize = bb_copyfd_size(inputFileFd, tbInfo->tarFd, statbuf->st_size); + ////if (readSize != statbuf->st_size && readSize >= 0) { + //// bb_error_msg_and_die("short read from %s, aborting", fileName); + ////} + /* Check that file did not grow in between? */ - /* if (safe_read(inputFileFd,1) == 1) warn but continue? */ + /* if (safe_read(inputFileFd, 1) == 1) warn but continue? */ + close(inputFileFd); /* Pad the file up to the tar block size */ /* (a few tricks here in the name of code size) */ - readSize = (-(int)readSize) & (TAR_BLOCK_SIZE-1); - bzero(bb_common_bufsiz1, readSize); - xwrite(tbInfo->tarFd, bb_common_bufsiz1, readSize); + readSize = (-(int)statbuf->st_size) & (TAR_BLOCK_SIZE-1); + memset(block_buf, 0, readSize); + xwrite(tbInfo->tarFd, block_buf, readSize); } return TRUE; } -static int writeTarFile(const int tar_fd, const int verboseFlag, - const unsigned long dereferenceFlag, const llist_t *include, - const llist_t *exclude, const int gzip) +#if ENABLE_FEATURE_SEAMLESS_GZ || ENABLE_FEATURE_SEAMLESS_BZ2 +# if !(ENABLE_FEATURE_SEAMLESS_GZ && ENABLE_FEATURE_SEAMLESS_BZ2) +# define vfork_compressor(tar_fd, gzip) vfork_compressor(tar_fd) +# endif +/* Don't inline: vfork scares gcc and pessimizes code */ +static void NOINLINE vfork_compressor(int tar_fd, int gzip) { - pid_t gzipPid = 0; - int errorFlag = FALSE; - struct TarBallInfo tbInfo; - - tbInfo.hlInfoHead = NULL; - - fchmod(tar_fd, 0644); - tbInfo.tarFd = tar_fd; - tbInfo.verboseFlag = verboseFlag; - - /* Store the stat info for the tarball's file, so - * can avoid including the tarball into itself.... */ - if (fstat(tbInfo.tarFd, &tbInfo.statBuf) < 0) - bb_perror_msg_and_die("cannot stat tar file"); - - if ((ENABLE_FEATURE_TAR_GZIP || ENABLE_FEATURE_TAR_BZIP2) && gzip) { - int gzipDataPipe[2] = { -1, -1 }; - int gzipStatusPipe[2] = { -1, -1 }; - volatile int vfork_exec_errno = 0; - char *zip_exec = (gzip == 1) ? "gzip" : "bzip2"; - - - if (pipe(gzipDataPipe) < 0 || pipe(gzipStatusPipe) < 0) - bb_perror_msg_and_die("pipe"); - - signal(SIGPIPE, SIG_IGN); /* we only want EPIPE on errors */ - -# if __GNUC__ - /* Avoid vfork clobbering */ - (void) &include; - (void) &errorFlag; - (void) &zip_exec; + pid_t gzipPid; +# if ENABLE_FEATURE_SEAMLESS_GZ && ENABLE_FEATURE_SEAMLESS_BZ2 + const char *zip_exec = (gzip == 1) ? "gzip" : "bzip2"; +# elif ENABLE_FEATURE_SEAMLESS_GZ + const char *zip_exec = "gzip"; +# else /* only ENABLE_FEATURE_SEAMLESS_BZ2 */ + const char *zip_exec = "bzip2"; # endif + // On Linux, vfork never unpauses parent early, although standard + // allows for that. Do we want to waste bytes checking for it? +# define WAIT_FOR_CHILD 0 + volatile int vfork_exec_errno = 0; + struct fd_pair gzipDataPipe; +# if WAIT_FOR_CHILD + struct fd_pair gzipStatusPipe; + xpiped_pair(gzipStatusPipe); +# endif + xpiped_pair(gzipDataPipe); - gzipPid = vfork(); - - if (gzipPid == 0) { - dup2(gzipDataPipe[0], 0); - close(gzipDataPipe[1]); + signal(SIGPIPE, SIG_IGN); /* we only want EPIPE on errors */ - dup2(tbInfo.tarFd, 1); +# if defined(__GNUC__) && __GNUC__ + /* Avoid vfork clobbering */ + (void) &zip_exec; +# endif - close(gzipStatusPipe[0]); - fcntl(gzipStatusPipe[1], F_SETFD, FD_CLOEXEC); /* close on exec shows success */ + gzipPid = xvfork(); + + if (gzipPid == 0) { + /* child */ + /* NB: close _first_, then move fds! */ + close(gzipDataPipe.wr); +# if WAIT_FOR_CHILD + close(gzipStatusPipe.rd); + /* gzipStatusPipe.wr will close only on exec - + * parent waits for this close to happen */ + fcntl(gzipStatusPipe.wr, F_SETFD, FD_CLOEXEC); +# endif + xmove_fd(gzipDataPipe.rd, 0); + xmove_fd(tar_fd, 1); + /* exec gzip/bzip2 program/applet */ + BB_EXECLP(zip_exec, zip_exec, "-f", NULL); + vfork_exec_errno = errno; + _exit(EXIT_FAILURE); + } - execlp(zip_exec, zip_exec, "-f", NULL); - vfork_exec_errno = errno; + /* parent */ + xmove_fd(gzipDataPipe.wr, tar_fd); + close(gzipDataPipe.rd); +# if WAIT_FOR_CHILD + close(gzipStatusPipe.wr); + while (1) { + char buf; + int n; + + /* Wait until child execs (or fails to) */ + n = full_read(gzipStatusPipe.rd, &buf, 1); + if (n < 0 /* && errno == EAGAIN */) + continue; /* try it again */ + } + close(gzipStatusPipe.rd); +# endif + if (vfork_exec_errno) { + errno = vfork_exec_errno; + bb_perror_msg_and_die("can't execute '%s'", zip_exec); + } +} +#endif /* ENABLE_FEATURE_SEAMLESS_GZ || ENABLE_FEATURE_SEAMLESS_BZ2 */ - close(gzipStatusPipe[1]); - exit(-1); - } else if (gzipPid > 0) { - close(gzipDataPipe[0]); - close(gzipStatusPipe[1]); - while (1) { - char buf; +/* gcc 4.2.1 inlines it, making code bigger */ +static NOINLINE int writeTarFile(int tar_fd, int verboseFlag, + int dereferenceFlag, const llist_t *include, + const llist_t *exclude, int gzip) +{ + int errorFlag = FALSE; + struct TarBallInfo tbInfo; - int n = full_read(gzipStatusPipe[0], &buf, 1); + tbInfo.hlInfoHead = NULL; + tbInfo.tarFd = tar_fd; + tbInfo.verboseFlag = verboseFlag; - if (n == 0 && vfork_exec_errno != 0) { - errno = vfork_exec_errno; - bb_perror_msg_and_die("cannot exec %s", zip_exec); - } else if ((n < 0) && (errno == EAGAIN || errno == EINTR)) - continue; /* try it again */ - break; - } - close(gzipStatusPipe[0]); + /* Store the stat info for the tarball's file, so + * can avoid including the tarball into itself.... */ + xfstat(tbInfo.tarFd, &tbInfo.tarFileStatBuf, "can't stat tar file"); - tbInfo.tarFd = gzipDataPipe[1]; - } else bb_perror_msg_and_die("vfork gzip"); - } +#if ENABLE_FEATURE_SEAMLESS_GZ || ENABLE_FEATURE_SEAMLESS_BZ2 + if (gzip) + vfork_compressor(tbInfo.tarFd, gzip); +#endif tbInfo.excludeList = exclude; /* Read the directory/files and iterate over them one at a time */ while (include) { - if (!recursive_action(include->data, TRUE, dereferenceFlag, - FALSE, writeFileToTarball, writeFileToTarball, &tbInfo, 0)) - { + if (!recursive_action(include->data, ACTION_RECURSE | + (dereferenceFlag ? ACTION_FOLLOWLINKS : 0), + writeFileToTarball, writeFileToTarball, &tbInfo, 0) + ) { errorFlag = TRUE; } include = include->link; } /* Write two empty blocks to the end of the archive */ - bzero(bb_common_bufsiz1, 2*TAR_BLOCK_SIZE); - xwrite(tbInfo.tarFd, bb_common_bufsiz1, 2*TAR_BLOCK_SIZE); + memset(block_buf, 0, 2*TAR_BLOCK_SIZE); + xwrite(tbInfo.tarFd, block_buf, 2*TAR_BLOCK_SIZE); /* To be pedantically correct, we would check if the tarball * is smaller than 20 tar blocks, and pad it if it was smaller, @@ -585,32 +648,34 @@ static int writeTarFile(const int tar_fd, const int verboseFlag, if (errorFlag) bb_error_msg("error exit delayed from previous errors"); - if (gzipPid && waitpid(gzipPid, NULL, 0) == -1) - bb_error_msg("waitpid failed"); - - return !errorFlag; +#if ENABLE_FEATURE_SEAMLESS_GZ || ENABLE_FEATURE_SEAMLESS_BZ2 + if (gzip) { + int status; + if (safe_waitpid(-1, &status, 0) == -1) + bb_perror_msg("waitpid"); + else if (!WIFEXITED(status) || WEXITSTATUS(status)) + /* gzip was killed or has exited with nonzero! */ + errorFlag = TRUE; + } +#endif + return errorFlag; } #else -int writeTarFile(const int tar_fd, const int verboseFlag, - const unsigned long dereferenceFlag, const llist_t *include, - const llist_t *exclude, const int gzip); -#endif /* tar_create */ +int writeTarFile(int tar_fd, int verboseFlag, + int dereferenceFlag, const llist_t *include, + const llist_t *exclude, int gzip); +#endif /* FEATURE_TAR_CREATE */ #if ENABLE_FEATURE_TAR_FROM static llist_t *append_file_list_to_list(llist_t *list) { FILE *src_stream; - llist_t *cur = list; - llist_t *tmp; char *line; llist_t *newlist = NULL; - while (cur) { - src_stream = xfopen(cur->data, "r"); - tmp = cur; - cur = cur->link; - free(tmp); - while ((line = xmalloc_getline(src_stream)) != NULL) { + while (list) { + src_stream = xfopen_stdin(llist_pop(&list)); + while ((line = xmalloc_fgetline(src_stream)) != NULL) { /* kill trailing '/' unless the string is just "/" */ char *cp = last_char_is(line, '/'); if (cp > line) @@ -622,11 +687,11 @@ static llist_t *append_file_list_to_list(llist_t *list) return newlist; } #else -#define append_file_list_to_list(x) 0 +# define append_file_list_to_list(x) 0 #endif -#if ENABLE_FEATURE_TAR_COMPRESS -static char get_header_tar_Z(archive_handle_t *archive_handle) +#if ENABLE_FEATURE_SEAMLESS_Z +static char FAST_FUNC get_header_tar_Z(archive_handle_t *archive_handle) { /* Can't lseek over pipes */ archive_handle->seek = seek_by_read; @@ -638,136 +703,286 @@ static char get_header_tar_Z(archive_handle_t *archive_handle) bb_error_msg_and_die("invalid magic"); } - archive_handle->src_fd = open_transformer(archive_handle->src_fd, uncompress); + open_transformer(archive_handle->src_fd, unpack_Z_stream, "uncompress"); archive_handle->offset = 0; while (get_header_tar(archive_handle) == EXIT_SUCCESS) - /* nothing */; + continue; /* Can only do one file at a time */ return EXIT_FAILURE; } #else -#define get_header_tar_Z 0 +# define get_header_tar_Z NULL +#endif + +#ifdef CHECK_FOR_CHILD_EXITCODE +/* Looks like it isn't needed - tar detects malformed (truncated) + * archive if e.g. bunzip2 fails */ +static int child_error; + +static void handle_SIGCHLD(int status) +{ + /* Actually, 'status' is a signo. We reuse it for other needs */ + + /* Wait for any child without blocking */ + if (wait_any_nohang(&status) < 0) + /* wait failed?! I'm confused... */ + return; + + if (WIFEXITED(status) && WEXITSTATUS(status) == 0) + /* child exited with 0 */ + return; + /* Cannot happen? + if (!WIFSIGNALED(status) && !WIFEXITED(status)) return; */ + child_error = 1; +} #endif +//usage:#define tar_trivial_usage +//usage: "-[" IF_FEATURE_TAR_CREATE("c") "xt" +//usage: IF_FEATURE_SEAMLESS_Z("Z") +//usage: IF_FEATURE_SEAMLESS_GZ("z") +//usage: IF_FEATURE_SEAMLESS_BZ2("j") +//usage: IF_FEATURE_SEAMLESS_LZMA("a") +//usage: IF_FEATURE_TAR_CREATE("h") +//usage: IF_FEATURE_TAR_NOPRESERVE_TIME("m") +//usage: "vO] " +//usage: IF_FEATURE_TAR_FROM("[-X FILE] [-T FILE] ") +//usage: "[-f TARFILE] [-C DIR] [FILE]..." +//usage:#define tar_full_usage "\n\n" +//usage: IF_FEATURE_TAR_CREATE("Create, extract, ") +//usage: IF_NOT_FEATURE_TAR_CREATE("Extract ") +//usage: "or list files from a tar file\n" +//usage: "\nOperation:" +//usage: IF_FEATURE_TAR_CREATE( +//usage: "\n c Create" +//usage: ) +//usage: "\n x Extract" +//usage: "\n t List" +//usage: "\n f Name of TARFILE ('-' for stdin/out)" +//usage: "\n C Change to DIR before operation" +//usage: "\n v Verbose" +//usage: IF_FEATURE_SEAMLESS_Z( +//usage: "\n Z (De)compress using compress" +//usage: ) +//usage: IF_FEATURE_SEAMLESS_GZ( +//usage: "\n z (De)compress using gzip" +//usage: ) +//usage: IF_FEATURE_SEAMLESS_BZ2( +//usage: "\n j (De)compress using bzip2" +//usage: ) +//usage: IF_FEATURE_SEAMLESS_LZMA( +//usage: "\n a (De)compress using lzma" +//usage: ) +//usage: "\n O Extract to stdout" +//usage: IF_FEATURE_TAR_CREATE( +//usage: "\n h Follow symlinks" +//usage: ) +//usage: IF_FEATURE_TAR_NOPRESERVE_TIME( +//usage: "\n m Don't restore mtime" +//usage: ) +//usage: IF_FEATURE_TAR_FROM( +//usage: IF_FEATURE_TAR_LONG_OPTIONS( +//usage: "\n exclude File to exclude" +//usage: ) +//usage: "\n X File with names to exclude" +//usage: "\n T File with names to include" +//usage: ) +//usage: +//usage:#define tar_example_usage +//usage: "$ zcat /tmp/tarball.tar.gz | tar -xf -\n" +//usage: "$ tar -cf /tmp/tarball.tar /usr/local\n" + +// Supported but aren't in --help: +// o no-same-owner +// p same-permissions +// k keep-old +// numeric-owner +// no-same-permissions +// overwrite +//IF_FEATURE_TAR_TO_COMMAND( +// to-command +//) + enum { - OPTBIT_KEEP_OLD = 7, - USE_FEATURE_TAR_CREATE( OPTBIT_CREATE ,) - USE_FEATURE_TAR_CREATE( OPTBIT_DEREFERENCE ,) - USE_FEATURE_TAR_BZIP2( OPTBIT_BZIP2 ,) - USE_FEATURE_TAR_LZMA( OPTBIT_LZMA ,) - USE_FEATURE_TAR_FROM( OPTBIT_INCLUDE_FROM,) - USE_FEATURE_TAR_FROM( OPTBIT_EXCLUDE_FROM,) - USE_FEATURE_TAR_GZIP( OPTBIT_GZIP ,) - USE_FEATURE_TAR_COMPRESS(OPTBIT_COMPRESS ,) - OPTBIT_NOPRESERVE_OWN, + OPTBIT_KEEP_OLD = 8, + IF_FEATURE_TAR_CREATE( OPTBIT_CREATE ,) + IF_FEATURE_TAR_CREATE( OPTBIT_DEREFERENCE ,) + IF_FEATURE_SEAMLESS_BZ2( OPTBIT_BZIP2 ,) + IF_FEATURE_SEAMLESS_LZMA(OPTBIT_LZMA ,) + IF_FEATURE_TAR_FROM( OPTBIT_INCLUDE_FROM,) + IF_FEATURE_TAR_FROM( OPTBIT_EXCLUDE_FROM,) + IF_FEATURE_SEAMLESS_GZ( OPTBIT_GZIP ,) + IF_FEATURE_SEAMLESS_Z( OPTBIT_COMPRESS ,) // 16th bit + IF_FEATURE_TAR_NOPRESERVE_TIME(OPTBIT_NOPRESERVE_TIME,) +#if ENABLE_FEATURE_TAR_LONG_OPTIONS + IF_FEATURE_TAR_TO_COMMAND(OPTBIT_2COMMAND ,) + OPTBIT_NUMERIC_OWNER, OPTBIT_NOPRESERVE_PERM, + OPTBIT_OVERWRITE, +#endif OPT_TEST = 1 << 0, // t OPT_EXTRACT = 1 << 1, // x OPT_BASEDIR = 1 << 2, // C OPT_TARNAME = 1 << 3, // f OPT_2STDOUT = 1 << 4, // O - OPT_P = 1 << 5, // p - OPT_VERBOSE = 1 << 6, // v - OPT_KEEP_OLD = 1 << 7, // k - OPT_CREATE = USE_FEATURE_TAR_CREATE( (1<flags = ARCHIVE_CREATE_LEADING_DIRS - | ARCHIVE_PRESERVE_DATE - | ARCHIVE_EXTRACT_UNCONDITIONAL; + tar_handle->ah_flags = ARCHIVE_CREATE_LEADING_DIRS + | ARCHIVE_RESTORE_DATE + | ARCHIVE_UNLINK_OLD; + + /* Apparently only root's tar preserves perms (see bug 3844) */ + if (getuid() != 0) + tar_handle->ah_flags |= ARCHIVE_DONT_RESTORE_PERM; /* Prepend '-' to the first argument if required */ opt_complementary = "--:" // first arg is options "tt:vv:" // count -t,-v - "?:" // bail out with usage instead of error return - "X::T::" // cumulative lists + IF_FEATURE_TAR_FROM("X::T::") // cumulative lists +#if ENABLE_FEATURE_TAR_LONG_OPTIONS && ENABLE_FEATURE_TAR_FROM "\xff::" // cumulative lists for --exclude - USE_FEATURE_TAR_CREATE("c:") "t:x:" // at least one of these is reqd - USE_FEATURE_TAR_CREATE("c--tx:t--cx:x--ct") // mutually exclusive - SKIP_FEATURE_TAR_CREATE("t--x:x--t"); // mutually exclusive +#endif + IF_FEATURE_TAR_CREATE("c:") "t:x:" // at least one of these is reqd + IF_FEATURE_TAR_CREATE("c--tx:t--cx:x--ct") // mutually exclusive + IF_NOT_FEATURE_TAR_CREATE("t--x:x--t"); // mutually exclusive #if ENABLE_FEATURE_TAR_LONG_OPTIONS - applet_long_options = tar_long_options; + applet_long_options = tar_longopts; +#endif +#if ENABLE_DESKTOP + if (argv[1] && argv[1][0] != '-') { + /* Compat: + * 1st argument without dash handles options with parameters + * differently from dashed one: it takes *next argv[i]* + * as paramenter even if there are more chars in 1st argument: + * "tar fx TARFILE" - "x" is not taken as f's param + * but is interpreted as -x option + * "tar -xf TARFILE" - dashed equivalent of the above + * "tar -fx ..." - "x" is taken as f's param + * getopt32 wouldn't handle 1st command correctly. + * Unfortunately, people do use such commands. + * We massage argv[1] to work around it by moving 'f' + * to the end of the string. + * More contrived "tar fCx TARFILE DIR" still fails, + * but such commands are much less likely to be used. + */ + char *f = strchr(argv[1], 'f'); + if (f) { + while (f[1] != '\0') { + *f = f[1]; + f++; + } + *f = 'f'; + } + } +#endif + opt = getopt32(argv, + "txC:f:Oopvk" + IF_FEATURE_TAR_CREATE( "ch" ) + IF_FEATURE_SEAMLESS_BZ2( "j" ) + IF_FEATURE_SEAMLESS_LZMA("a" ) + IF_FEATURE_TAR_FROM( "T:X:") + IF_FEATURE_SEAMLESS_GZ( "z" ) + IF_FEATURE_SEAMLESS_Z( "Z" ) + IF_FEATURE_TAR_NOPRESERVE_TIME("m") + , &base_dir // -C dir + , &tar_filename // -f filename + IF_FEATURE_TAR_FROM(, &(tar_handle->accept)) // T + IF_FEATURE_TAR_FROM(, &(tar_handle->reject)) // X + IF_FEATURE_TAR_TO_COMMAND(, &(tar_handle->tar__to_command)) // --to-command +#if ENABLE_FEATURE_TAR_LONG_OPTIONS && ENABLE_FEATURE_TAR_FROM + , &excludes // --exclude #endif - opt = getopt32(argc, argv, - "txC:f:Opvk" - USE_FEATURE_TAR_CREATE( "ch" ) - USE_FEATURE_TAR_BZIP2( "j" ) - USE_FEATURE_TAR_LZMA( "a" ) - USE_FEATURE_TAR_FROM( "T:X:") - USE_FEATURE_TAR_GZIP( "z" ) - USE_FEATURE_TAR_COMPRESS("Z" ) - , - &base_dir, // -C dir - &tar_filename, // -f filename - USE_FEATURE_TAR_FROM(&(tar_handle->accept),) // T - USE_FEATURE_TAR_FROM(&(tar_handle->reject),) // X - USE_FEATURE_TAR_FROM(&excludes ,) // --exclude - &verboseFlag, // combined count for -t and -v - &verboseFlag // combined count for -t and -v + , &verboseFlag // combined count for -t and -v + , &verboseFlag // combined count for -t and -v ); + //bb_error_msg("opt:%08x", opt); + argv += optind; if (verboseFlag) tar_handle->action_header = header_verbose_list; if (verboseFlag == 1) tar_handle->action_header = header_list; @@ -778,14 +993,29 @@ int tar_main(int argc, char **argv) if (opt & OPT_2STDOUT) tar_handle->action_data = data_extract_to_stdout; + if (opt & OPT_2COMMAND) { + putenv((char*)"TAR_FILETYPE=f"); + signal(SIGPIPE, SIG_IGN); + tar_handle->action_data = data_extract_to_command; + IF_FEATURE_TAR_TO_COMMAND(tar_handle->tar__to_command_shell = xstrdup(get_shell_name());) + } + if (opt & OPT_KEEP_OLD) - tar_handle->flags &= ~ARCHIVE_EXTRACT_UNCONDITIONAL; + tar_handle->ah_flags &= ~ARCHIVE_UNLINK_OLD; + + if (opt & OPT_NUMERIC_OWNER) + tar_handle->ah_flags |= ARCHIVE_NUMERIC_OWNER; - if (opt & OPT_NOPRESERVE_OWN) - tar_handle->flags |= ARCHIVE_NOPRESERVE_OWN; + if (opt & OPT_NOPRESERVE_OWNER) + tar_handle->ah_flags |= ARCHIVE_DONT_RESTORE_OWNER; if (opt & OPT_NOPRESERVE_PERM) - tar_handle->flags |= ARCHIVE_NOPRESERVE_PERM; + tar_handle->ah_flags |= ARCHIVE_DONT_RESTORE_PERM; + + if (opt & OPT_OVERWRITE) { + tar_handle->ah_flags &= ~ARCHIVE_UNLINK_OLD; + tar_handle->ah_flags |= ARCHIVE_O_TRUNC; + } if (opt & OPT_GZIP) get_header_ptr = get_header_tar_gz; @@ -799,85 +1029,94 @@ int tar_main(int argc, char **argv) if (opt & OPT_COMPRESS) get_header_ptr = get_header_tar_Z; - if (ENABLE_FEATURE_TAR_FROM) { - tar_handle->reject = append_file_list_to_list(tar_handle->reject); - /* Append excludes to reject */ - while (excludes) { - llist_t *temp = excludes->link; - excludes->link = tar_handle->reject; - tar_handle->reject = excludes; - excludes = temp; - } - tar_handle->accept = append_file_list_to_list(tar_handle->accept); - } + if (opt & OPT_NOPRESERVE_TIME) + tar_handle->ah_flags &= ~ARCHIVE_RESTORE_DATE; - /* Check if we are reading from stdin */ - if (argv[optind] && *argv[optind] == '-') { - /* Default is to read from stdin, so just skip to next arg */ - optind++; +#if ENABLE_FEATURE_TAR_FROM + tar_handle->reject = append_file_list_to_list(tar_handle->reject); +# if ENABLE_FEATURE_TAR_LONG_OPTIONS + /* Append excludes to reject */ + while (excludes) { + llist_t *next = excludes->link; + excludes->link = tar_handle->reject; + tar_handle->reject = excludes; + excludes = next; } +# endif + tar_handle->accept = append_file_list_to_list(tar_handle->accept); +#endif /* Setup an array of filenames to work with */ - /* TODO: This is the same as in ar, separate function ? */ - while (optind < argc) { + /* TODO: This is the same as in ar, make a separate function? */ + while (*argv) { /* kill trailing '/' unless the string is just "/" */ - char *cp = last_char_is(argv[optind], '/'); - if (cp > argv[optind]) + char *cp = last_char_is(*argv, '/'); + if (cp > *argv) *cp = '\0'; - llist_add_to(&tar_handle->accept, argv[optind]); - optind++; + llist_add_to_end(&tar_handle->accept, *argv); + argv++; } - tar_handle->accept = rev_llist(tar_handle->accept); if (tar_handle->accept || tar_handle->reject) tar_handle->filter = filter_accept_reject_list; /* Open the tar file */ { - FILE *tar_stream; - int flags; + int tar_fd = STDIN_FILENO; + int flags = O_RDONLY; if (opt & OPT_CREATE) { - /* Make sure there is at least one file to tar up. */ + /* Make sure there is at least one file to tar up */ if (tar_handle->accept == NULL) bb_error_msg_and_die("empty archive"); - tar_stream = stdout; + tar_fd = STDOUT_FILENO; /* Mimicking GNU tar 1.15.1: */ - flags = O_WRONLY|O_CREAT|O_TRUNC; - /* was doing unlink; open(O_WRONLY|O_CREAT|O_EXCL); why? */ - } else { - tar_stream = stdin; - flags = O_RDONLY; + flags = O_WRONLY | O_CREAT | O_TRUNC; } - if (tar_filename[0] == '-' && !tar_filename[1]) { - tar_handle->src_fd = fileno(tar_stream); + if (LONE_DASH(tar_filename)) { + tar_handle->src_fd = tar_fd; tar_handle->seek = seek_by_read; } else { - tar_handle->src_fd = xopen(tar_filename, flags); + if (ENABLE_FEATURE_TAR_AUTODETECT + && flags == O_RDONLY + && get_header_ptr == get_header_tar + ) { + tar_handle->src_fd = open_zipped(tar_filename); + if (tar_handle->src_fd < 0) + bb_perror_msg_and_die("can't open '%s'", tar_filename); + } else { + tar_handle->src_fd = xopen(tar_filename, flags); + } } } if (base_dir) xchdir(base_dir); - /* create an archive */ +#ifdef CHECK_FOR_CHILD_EXITCODE + /* We need to know whether child (gzip/bzip/etc) exits abnormally */ + signal(SIGCHLD, handle_SIGCHLD); +#endif + + /* Create an archive */ if (opt & OPT_CREATE) { +#if ENABLE_FEATURE_SEAMLESS_GZ || ENABLE_FEATURE_SEAMLESS_BZ2 int zipMode = 0; - if (ENABLE_FEATURE_TAR_GZIP && get_header_ptr == get_header_tar_gz) + if (ENABLE_FEATURE_SEAMLESS_GZ && (opt & OPT_GZIP)) zipMode = 1; - if (ENABLE_FEATURE_TAR_BZIP2 && get_header_ptr == get_header_tar_bz2) + if (ENABLE_FEATURE_SEAMLESS_BZ2 && (opt & OPT_BZIP2)) zipMode = 2; - writeTarFile(tar_handle->src_fd, verboseFlag, opt & OPT_DEREFERENCE, +#endif + /* NB: writeTarFile() closes tar_handle->src_fd */ + return writeTarFile(tar_handle->src_fd, verboseFlag, opt & OPT_DEREFERENCE, tar_handle->accept, tar_handle->reject, zipMode); - /* NB: writeTarFile() closes tar_handle->src_fd */ - return EXIT_SUCCESS; } while (get_header_ptr(tar_handle) == EXIT_SUCCESS) - /* nothing */; + continue; /* Check that every file that should have been extracted was */ while (tar_handle->accept) {