ef0d6eb598
* Disabled shared-library building, as some API breakage is still likely. (I didn't realize it was turned on by default.) If you have an existing /usr/lib/libarchive.so.2, I recommend deleting it. * Pax interchange format now correctly stores and reads UTF8 for extended attributes. In particular, pax format can portably handle arbitrarily long pathnames containing arbitrary characters. * Library compiles cleanly at -O2, -O3, and WARNS=6 on all FreeBSD-CURRENT platforms. * Minor portability improvements inspired by Juergen Lock and Greg Lewis. (Less reliance on stdint.h, isolating of various portability-challenged constructs.) * archive_entry transparently converts multi-byte <-> wide character strings, allowing clients and format handlers to deal with either one, as appropriate. * Support for reading 'L' and 'K' entries in standard tar archives for star compatibility. * Recognize (but don't yet handle) ACL entries from Solaris tar. * Pushed format-specific data for format readers down into format-specific storage and out of library-global storage. This should make it easier to maintain individual formats without mucking with the core library management. * Documentation updates to track the above changes. * Updates to tar.5 to correct a few mistakes and add some additional information about GNU tar and Solaris tar formats. Notes: * The basic 'tar' reader is getting more general; there's not much point in keeping the 'gnutar' reader separate. Merging the two would lose a bunch of duplicate code. * The libc ACL support is looking increasingly inadequate for my needs here. I might need to assemble some fairly significant code for parsing and building ACLs. <sigh>
501 lines
13 KiB
C
501 lines
13 KiB
C
/*-
|
|
* Copyright (c) 2003-2004 Tim Kientzle
|
|
* All rights reserved.
|
|
*
|
|
* Redistribution and use in source and binary forms, with or without
|
|
* modification, are permitted provided that the following conditions
|
|
* are met:
|
|
* 1. Redistributions of source code must retain the above copyright
|
|
* notice, this list of conditions and the following disclaimer
|
|
* in this position and unchanged.
|
|
* 2. Redistributions in binary form must reproduce the above copyright
|
|
* notice, this list of conditions and the following disclaimer in the
|
|
* documentation and/or other materials provided with the distribution.
|
|
*
|
|
* THIS SOFTWARE IS PROVIDED BY THE AUTHOR(S) ``AS IS'' AND ANY EXPRESS OR
|
|
* IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES
|
|
* OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.
|
|
* IN NO EVENT SHALL THE AUTHOR(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
|
|
* INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT
|
|
* NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
|
|
* DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
|
|
* THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
|
|
* (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF
|
|
* THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
|
|
*/
|
|
|
|
#include "archive_platform.h"
|
|
__FBSDID("$FreeBSD$");
|
|
|
|
#ifdef HAVE_DMALLOC
|
|
#include <dmalloc.h>
|
|
#endif
|
|
#include <errno.h>
|
|
#include <stdlib.h>
|
|
#include <string.h>
|
|
#include <unistd.h>
|
|
#include <zlib.h>
|
|
|
|
#include <err.h> /* zlib.h is borked, so must precede err.h */
|
|
|
|
#include "archive.h"
|
|
#include "archive_private.h"
|
|
|
|
struct private_data {
|
|
z_stream stream;
|
|
unsigned char *uncompressed_buffer;
|
|
size_t uncompressed_buffer_size;
|
|
unsigned char *read_next;
|
|
int64_t total_out;
|
|
unsigned long crc;
|
|
char header_done;
|
|
};
|
|
|
|
static int bid(const void *, size_t);
|
|
static int finish(struct archive *);
|
|
static int init(struct archive *, const void *, size_t);
|
|
static ssize_t read_ahead(struct archive *, const void **, size_t);
|
|
static ssize_t read_consume(struct archive *, size_t);
|
|
static int drive_decompressor(struct archive *a, struct private_data *);
|
|
|
|
int
|
|
archive_read_support_compression_gzip(struct archive *a)
|
|
{
|
|
return (__archive_read_register_compression(a, bid, init));
|
|
}
|
|
|
|
/*
|
|
* Test whether we can handle this data.
|
|
*
|
|
* This logic returns zero if any part of the signature fails. It
|
|
* also tries to Do The Right Thing if a very short buffer prevents us
|
|
* from verifying as much as we would like.
|
|
*/
|
|
static int
|
|
bid(const void *buff, size_t len)
|
|
{
|
|
const unsigned char *buffer;
|
|
int bits_checked;
|
|
|
|
if (len < 1)
|
|
return (0);
|
|
|
|
buffer = buff;
|
|
bits_checked = 0;
|
|
if (buffer[0] != 037) /* Verify first ID byte. */
|
|
return (0);
|
|
bits_checked += 8;
|
|
if (len < 2)
|
|
return (bits_checked);
|
|
|
|
if (buffer[1] != 0213) /* Verify second ID byte. */
|
|
return (0);
|
|
bits_checked += 8;
|
|
if (len < 3)
|
|
return (bits_checked);
|
|
|
|
if (buffer[2] != 8) /* Compression must be 'deflate'. */
|
|
return (0);
|
|
bits_checked += 8;
|
|
if (len < 4)
|
|
return (bits_checked);
|
|
|
|
if ((buffer[3] & 0xE0)!= 0) /* No reserved flags set. */
|
|
return (0);
|
|
bits_checked += 3;
|
|
if (len < 5)
|
|
return (bits_checked);
|
|
|
|
/*
|
|
* TODO: Verify more; in particular, gzip has an optional
|
|
* header CRC, which would give us 16 more verified bits. We
|
|
* may also be able to verify certain constraints on other
|
|
* fields.
|
|
*/
|
|
|
|
return (bits_checked);
|
|
}
|
|
|
|
/*
|
|
* Setup the callbacks.
|
|
*/
|
|
static int
|
|
init(struct archive *a, const void *buff, size_t n)
|
|
{
|
|
struct private_data *state;
|
|
int ret;
|
|
|
|
a->compression_code = ARCHIVE_COMPRESSION_GZIP;
|
|
a->compression_name = "gzip";
|
|
|
|
state = malloc(sizeof(*state));
|
|
if (state == NULL) {
|
|
archive_set_error(a, ENOMEM,
|
|
"Can't allocate data for %s decompression",
|
|
a->compression_name);
|
|
return (ARCHIVE_FATAL);
|
|
}
|
|
memset(state, 0, sizeof(*state));
|
|
|
|
state->crc = crc32(0L, NULL, 0);
|
|
state->header_done = 0; /* We've not yet begun to parse header... */
|
|
|
|
state->uncompressed_buffer_size = 64 * 1024;
|
|
state->uncompressed_buffer = malloc(state->uncompressed_buffer_size);
|
|
state->stream.next_out = state->uncompressed_buffer;
|
|
state->read_next = state->uncompressed_buffer;
|
|
state->stream.avail_out = state->uncompressed_buffer_size;
|
|
|
|
if (state->uncompressed_buffer == NULL) {
|
|
archive_set_error(a, ENOMEM,
|
|
"Can't allocate %s decompression buffers",
|
|
a->compression_name);
|
|
free(state);
|
|
return (ARCHIVE_FATAL);
|
|
}
|
|
|
|
/*
|
|
* A bug in zlib.h: stream.next_in should be marked 'const'
|
|
* but isn't (the library never alters data through the
|
|
* next_in pointer, only reads it). The result: this ugly
|
|
* cast to remove 'const'.
|
|
*/
|
|
state->stream.next_in = (void *)(uintptr_t)(const void *)buff;
|
|
state->stream.avail_in = n;
|
|
|
|
a->compression_read_ahead = read_ahead;
|
|
a->compression_read_consume = read_consume;
|
|
a->compression_finish = finish;
|
|
|
|
/*
|
|
* TODO: Do I need to parse the gzip header before calling
|
|
* inflateInit2()? In particular, one of the header bytes
|
|
* marks "best compression" or "fastest", which may be
|
|
* appropriate for setting the second parameter here.
|
|
* However, I think the only penalty for not setting it
|
|
* correctly is wasted memory. If this is necessary, it
|
|
* should probably go into drive_decompressor() below.
|
|
*/
|
|
|
|
/* Initialize compression library. */
|
|
ret = inflateInit2(&(state->stream),
|
|
-15 /* Don't check for zlib header */);
|
|
if (ret == Z_OK) {
|
|
a->compression_data = state;
|
|
return (ARCHIVE_OK);
|
|
}
|
|
|
|
/* Library setup failed: Clean up. */
|
|
archive_set_error(a, ARCHIVE_ERRNO_MISC,
|
|
"Internal error initializing %s library", a->compression_name);
|
|
free(state->uncompressed_buffer);
|
|
free(state);
|
|
|
|
/* Override the error message if we know what really went wrong. */
|
|
switch (ret) {
|
|
case Z_STREAM_ERROR:
|
|
archive_set_error(a, ARCHIVE_ERRNO_MISC,
|
|
"Internal error initializing compression library: "
|
|
"invalid setup parameter");
|
|
break;
|
|
case Z_MEM_ERROR:
|
|
archive_set_error(a, ENOMEM,
|
|
"Internal error initializing compression library: "
|
|
"out of memory");
|
|
break;
|
|
case Z_VERSION_ERROR:
|
|
archive_set_error(a, ARCHIVE_ERRNO_MISC,
|
|
"Internal error initializing compression library: "
|
|
"invalid library version");
|
|
break;
|
|
}
|
|
|
|
return (ARCHIVE_FATAL);
|
|
}
|
|
|
|
/*
|
|
* Return a block of data from the decompression buffer. Decompress more
|
|
* as necessary.
|
|
*/
|
|
static ssize_t
|
|
read_ahead(struct archive *a, const void **p, size_t min)
|
|
{
|
|
struct private_data *state;
|
|
int read_avail, was_avail, ret;
|
|
|
|
state = a->compression_data;
|
|
was_avail = -1;
|
|
if (!a->client_reader) {
|
|
archive_set_error(a, ARCHIVE_ERRNO_PROGRAMMER,
|
|
"No read callback is registered? "
|
|
"This is probably an internal programming error.");
|
|
return (ARCHIVE_FATAL);
|
|
}
|
|
|
|
read_avail = state->stream.next_out - state->read_next;
|
|
|
|
if (read_avail + state->stream.avail_out < min) {
|
|
memmove(state->uncompressed_buffer, state->read_next,
|
|
read_avail);
|
|
state->read_next = state->uncompressed_buffer;
|
|
state->stream.next_out = state->read_next + read_avail;
|
|
state->stream.avail_out
|
|
= state->uncompressed_buffer_size - read_avail;
|
|
}
|
|
|
|
while (was_avail < read_avail && /* Made some progress. */
|
|
read_avail < (int)min && /* Haven't satisfied min. */
|
|
read_avail < (int)state->uncompressed_buffer_size) { /* !full */
|
|
if ((ret = drive_decompressor(a, state)) != ARCHIVE_OK)
|
|
return (ret);
|
|
was_avail = read_avail;
|
|
read_avail = state->stream.next_out - state->read_next;
|
|
}
|
|
|
|
*p = state->read_next;
|
|
return (read_avail);
|
|
}
|
|
|
|
/*
|
|
* Mark a previously-returned block of data as read.
|
|
*/
|
|
static ssize_t
|
|
read_consume(struct archive *a, size_t n)
|
|
{
|
|
struct private_data *state;
|
|
|
|
state = a->compression_data;
|
|
a->file_position += n;
|
|
state->read_next += n;
|
|
if (state->read_next > state->stream.next_out)
|
|
errx(1, "Internal error: Request to consume too many "
|
|
"bytes from %s decompressor.\n",
|
|
a->compression_name);
|
|
return (n);
|
|
}
|
|
|
|
/*
|
|
* Clean up the decompressor.
|
|
*/
|
|
static int
|
|
finish(struct archive *a)
|
|
{
|
|
struct private_data *state;
|
|
int ret;
|
|
|
|
state = a->compression_data;
|
|
ret = ARCHIVE_OK;
|
|
switch (inflateEnd(&(state->stream))) {
|
|
case Z_OK:
|
|
break;
|
|
default:
|
|
archive_set_error(a, ARCHIVE_ERRNO_MISC,
|
|
"Failed to clean up %s compressor", a->compression_name);
|
|
ret = ARCHIVE_FATAL;
|
|
}
|
|
|
|
free(state->uncompressed_buffer);
|
|
free(state);
|
|
|
|
a->compression_data = NULL;
|
|
if (a->client_closer != NULL)
|
|
(a->client_closer)(a, a->client_data);
|
|
|
|
return (ret);
|
|
}
|
|
|
|
/*
|
|
* Utility function to pull data through decompressor, reading input
|
|
* blocks as necessary.
|
|
*/
|
|
static int
|
|
drive_decompressor(struct archive *a, struct private_data *state)
|
|
{
|
|
ssize_t ret;
|
|
int decompressed, total_decompressed;
|
|
int count, flags, header_state;
|
|
unsigned char *output;
|
|
unsigned char b;
|
|
|
|
flags = 0;
|
|
count = 0;
|
|
header_state = 0;
|
|
total_decompressed = 0;
|
|
for (;;) {
|
|
if (state->stream.avail_in == 0) {
|
|
ret = (a->client_reader)(a, a->client_data,
|
|
(const void **)&state->stream.next_in);
|
|
if (ret < 0) {
|
|
/*
|
|
* TODO: Find a better way to handle
|
|
* this read failure.
|
|
*/
|
|
goto fatal;
|
|
}
|
|
if (ret == 0 && total_decompressed == 0) {
|
|
archive_set_error(a, EIO,
|
|
"Premature end of %s compressed data",
|
|
a->compression_name);
|
|
return (ARCHIVE_FATAL);
|
|
}
|
|
state->stream.avail_in = ret;
|
|
}
|
|
|
|
if (!state->header_done) {
|
|
/*
|
|
* If still parsing the header, interpret the
|
|
* next byte.
|
|
*/
|
|
b = *(state->stream.next_in++);
|
|
state->stream.avail_in--;
|
|
|
|
/*
|
|
* Yes, this is somewhat crude, but it works,
|
|
* GZip format isn't likely to change anytime
|
|
* in the near future, and header parsing is
|
|
* certainly not a performance issue, so
|
|
* there's little point in making this more
|
|
* elegant. Of course, if you see an easy way
|
|
* to make this more elegant, please let me
|
|
* know.. ;-)
|
|
*/
|
|
switch (header_state) {
|
|
case 0: /* First byte of signature. */
|
|
if (b != 037)
|
|
goto fatal;
|
|
header_state = 1;
|
|
break;
|
|
case 1: /* Second byte of signature. */
|
|
if (b != 0213)
|
|
goto fatal;
|
|
header_state = 2;
|
|
break;
|
|
case 2: /* Compression type must be 8. */
|
|
if (b != 8)
|
|
goto fatal;
|
|
header_state = 3;
|
|
break;
|
|
case 3: /* GZip flags. */
|
|
flags = b;
|
|
header_state = 4;
|
|
break;
|
|
case 4: case 5: case 6: case 7: /* Mod time. */
|
|
header_state++;
|
|
break;
|
|
case 8: /* Deflate flags. */
|
|
header_state = 9;
|
|
break;
|
|
case 9: /* OS. */
|
|
header_state = 10;
|
|
break;
|
|
case 10: /* Optional Extra: First byte of Length. */
|
|
if ((flags & 4)) {
|
|
count = 255 & (int)b;
|
|
header_state = 11;
|
|
break;
|
|
}
|
|
/*
|
|
* Fall through if there is no
|
|
* Optional Extra field.
|
|
*/
|
|
case 11: /* Optional Extra: Second byte of Length. */
|
|
if ((flags & 4)) {
|
|
count = (count << 8) | (255 & (int)b);
|
|
header_state = 12;
|
|
break;
|
|
}
|
|
/*
|
|
* Fall through if there is no
|
|
* Optional Extra field.
|
|
*/
|
|
case 12: /* Optional Extra Field: counted length. */
|
|
if ((flags & 4)) {
|
|
--count;
|
|
if (count == 0) header_state = 13;
|
|
else header_state = 12;
|
|
break;
|
|
}
|
|
/*
|
|
* Fall through if there is no
|
|
* Optional Extra field.
|
|
*/
|
|
case 13: /* Optional Original Filename. */
|
|
if ((flags & 8)) {
|
|
if (b == 0) header_state = 14;
|
|
else header_state = 13;
|
|
break;
|
|
}
|
|
/*
|
|
* Fall through if no Optional
|
|
* Original Filename.
|
|
*/
|
|
case 14: /* Optional Comment. */
|
|
if ((flags & 16)) {
|
|
if (b == 0) header_state = 15;
|
|
else header_state = 14;
|
|
break;
|
|
}
|
|
/* Fall through if no Optional Comment. */
|
|
case 15: /* Optional Header CRC: First byte. */
|
|
if ((flags & 2)) {
|
|
header_state = 16;
|
|
break;
|
|
}
|
|
/* Fall through if no Optional Header CRC. */
|
|
case 16: /* Optional Header CRC: Second byte. */
|
|
if ((flags & 2)) {
|
|
header_state = 17;
|
|
break;
|
|
}
|
|
/* Fall through if no Optional Header CRC. */
|
|
case 17: /* First byte of compressed data. */
|
|
state->header_done = 1; /* done with header */
|
|
state->stream.avail_in++;
|
|
state->stream.next_in--;
|
|
}
|
|
|
|
/*
|
|
* TODO: Consider moving the inflateInit2 call
|
|
* here so it can include the compression type
|
|
* from the header?
|
|
*/
|
|
} else {
|
|
output = state->stream.next_out;
|
|
|
|
/* Decompress some data. */
|
|
ret = inflate(&(state->stream), 0);
|
|
decompressed = state->stream.next_out - output;
|
|
|
|
/* Accumulate the CRC of the uncompressed data. */
|
|
state->crc = crc32(state->crc, output, decompressed);
|
|
|
|
/* Accumulate the total bytes of output. */
|
|
state->total_out += decompressed;
|
|
total_decompressed += decompressed;
|
|
|
|
switch (ret) {
|
|
case Z_OK: /* Decompressor made some progress. */
|
|
if (decompressed > 0)
|
|
return (ARCHIVE_OK);
|
|
break;
|
|
case Z_STREAM_END: /* Found end of stream. */
|
|
/*
|
|
* TODO: Verify gzip trailer
|
|
* (uncompressed length and CRC).
|
|
*/
|
|
return (ARCHIVE_OK);
|
|
default:
|
|
/* Any other return value is an error. */
|
|
goto fatal;
|
|
}
|
|
}
|
|
}
|
|
return (ARCHIVE_OK);
|
|
|
|
/* Return a fatal error. */
|
|
fatal:
|
|
archive_set_error(a, ARCHIVE_ERRNO_MISC, "%s decompression failed",
|
|
a->compression_name);
|
|
return (ARCHIVE_FATAL);
|
|
}
|