diff options
Diffstat (limited to 'modules/libmar/src/mar_read.c')
-rw-r--r-- | modules/libmar/src/mar_read.c | 787 |
1 files changed, 787 insertions, 0 deletions
diff --git a/modules/libmar/src/mar_read.c b/modules/libmar/src/mar_read.c new file mode 100644 index 0000000000..679dc258d6 --- /dev/null +++ b/modules/libmar/src/mar_read.c @@ -0,0 +1,787 @@ +/* -*- Mode: C++; tab-width: 2; indent-tabs-mode: nil; c-basic-offset: 2 -*- */ +/* vim:set ts=2 sw=2 sts=2 et cindent: */ +/* This Source Code Form is subject to the terms of the Mozilla Public + * License, v. 2.0. If a copy of the MPL was not distributed with this + * file, You can obtain one at http://mozilla.org/MPL/2.0/. */ + +#include <sys/types.h> +#include <fcntl.h> +#include <stdlib.h> +#include <string.h> +#include "city.h" +#include "mar_private.h" +#include "mar.h" +#ifdef XP_WIN +# define strdup _strdup +#endif + +/* This block must be at most 104 bytes. + MAR channel name < 64 bytes, and product version < 32 bytes + 3 NULL + terminator bytes. We only check for 96 though because we remove 8 + bytes above from the additionalBlockSize: We subtract + sizeof(additionalBlockSize) and sizeof(additionalBlockID) */ +#define MAXADDITIONALBLOCKSIZE 96 + +static uint32_t mar_hash_name(const char* name) { + return CityHash64(name, strlen(name)) % TABLESIZE; +} + +static int mar_insert_item(MarFile* mar, const char* name, uint32_t namelen, + uint32_t offset, uint32_t length, uint32_t flags) { + MarItem *item, *root; + uint32_t hash; + + item = (MarItem*)malloc(sizeof(MarItem) + namelen); + if (!item) { + return -1; + } + item->next = NULL; + item->offset = offset; + item->length = length; + item->flags = flags; + memcpy(item->name, name, namelen + 1); + + hash = mar_hash_name(name); + + root = mar->item_table[hash]; + if (!root) { + mar->item_table[hash] = item; + } else { + /* append item */ + while (root->next) root = root->next; + root->next = item; + } + return 0; +} + +static int mar_consume_index(MarFile* mar, char** buf, const char* buf_end) { + /* + * Each item has the following structure: + * uint32_t offset (network byte order) + * uint32_t length (network byte order) + * uint32_t flags (network byte order) + * char name[N] (where N >= 1) + * char null_byte; + */ + uint32_t offset; + uint32_t length; + uint32_t flags; + const char* name; + int namelen; + + if ((buf_end - *buf) < (int)(3 * sizeof(uint32_t) + 2)) { + return -1; + } + + memcpy(&offset, *buf, sizeof(offset)); + *buf += sizeof(offset); + + memcpy(&length, *buf, sizeof(length)); + *buf += sizeof(length); + + memcpy(&flags, *buf, sizeof(flags)); + *buf += sizeof(flags); + + offset = ntohl(offset); + length = ntohl(length); + flags = ntohl(flags); + + name = *buf; + /* find namelen; must take care not to read beyond buf_end */ + while (**buf) { + /* buf_end points one byte past the end of buf's allocation */ + if (*buf == (buf_end - 1)) { + return -1; + } + ++(*buf); + } + namelen = (*buf - name); + /* must ensure that namelen is valid */ + if (namelen < 0) { + return -1; + } + /* consume null byte */ + if (*buf == buf_end) { + return -1; + } + ++(*buf); + + return mar_insert_item(mar, name, namelen, offset, length, flags); +} + +static int mar_read_index(MarFile* mar) { + char id[MAR_ID_SIZE], *buf, *bufptr, *bufend; + uint32_t offset_to_index, size_of_index; + size_t mar_position = 0; + + /* verify MAR ID */ + if (mar_read_buffer(mar, id, &mar_position, MAR_ID_SIZE) != 0) { + return -1; + } + if (memcmp(id, MAR_ID, MAR_ID_SIZE) != 0) { + return -1; + } + + if (mar_read_buffer(mar, &offset_to_index, &mar_position, sizeof(uint32_t)) != + 0) { + return -1; + } + offset_to_index = ntohl(offset_to_index); + + mar_position = 0; + if (mar_buffer_seek(mar, &mar_position, offset_to_index) != 0) { + return -1; + } + if (mar_read_buffer(mar, &size_of_index, &mar_position, sizeof(uint32_t)) != + 0) { + return -1; + } + size_of_index = ntohl(size_of_index); + + buf = (char*)malloc(size_of_index); + if (!buf) { + return -1; + } + if (mar_read_buffer(mar, buf, &mar_position, size_of_index) != 0) { + free(buf); + return -1; + } + + bufptr = buf; + bufend = buf + size_of_index; + while (bufptr < bufend && mar_consume_index(mar, &bufptr, bufend) == 0) + ; + + free(buf); + return (bufptr == bufend) ? 0 : -1; +} + +/** + * Adds an offset and length to the MarFile's index_list + * @param mar The MarFile that owns this offset length pair + * @param offset The byte offset in the archive to be marked as processed + * @param length The length corresponding to this byte offset + * @return int 1 on success, 0 if offset has been previously processed + * -1 if unable to allocate space for the SeenIndexes + */ +static int mar_insert_offset(MarFile* mar, uint32_t offset, uint32_t length) { + /* Ignore files with no length */ + if (length == 0) { + return 1; + } + + SeenIndex* index = (SeenIndex*)malloc(sizeof(SeenIndex)); + if (!index) { + return -1; + } + index->next = NULL; + index->offset = offset; + index->length = length; + uint32_t index_end = index->offset + index->length - 1; + + /* If this is our first index store it at the front */ + if (mar->index_list == NULL) { + mar->index_list = index; + return 1; + } + + /* Search for matching indexes in the list of those previously visited */ + SeenIndex* previous; + SeenIndex* current = mar->index_list; + while (current != NULL) { + uint32_t current_end = current->offset + current->length - 1; + + /* If index has collided with the front or end of current or if current has + collided with the front or end of index return false */ + if ((index->offset >= current->offset && index->offset <= current_end) || + (index_end >= current->offset && index_end <= current_end) || + (current->offset >= index->offset && current->offset <= index_end) || + (current_end >= index->offset && current_end <= index_end)) { + free(index); + return 0; + } + + /* else move to the next in the list */ + previous = current; + current = current->next; + } + + /* These indexes are valid, track them */ + previous->next = index; + return 1; +} + +/** + * Internal shared code for mar_open and mar_wopen. + * Reads the entire MAR into memory. Fails if it is bigger than + * MAX_SIZE_OF_MAR_FILE bytes. + */ +static MarReadResult mar_fpopen(FILE* fp, MarFile** out_mar) { + *out_mar = NULL; + MarFile* mar; + + mar = (MarFile*)malloc(sizeof(*mar)); + if (!mar) { + return MAR_MEM_ERROR; + } + + off_t buffer_size = -1; + if (fseeko(fp, 0, SEEK_END) == 0) { + buffer_size = ftello(fp); + } + rewind(fp); + if (buffer_size < 0) { + fprintf(stderr, "Warning: MAR size could not be determined\n"); + buffer_size = MAX_SIZE_OF_MAR_FILE; + } + if (buffer_size > MAX_SIZE_OF_MAR_FILE) { + fprintf(stderr, "ERROR: MAR exceeds maximum size (%lli)\n", + (long long int)buffer_size); + free(mar); + return MAR_FILE_TOO_BIG_ERROR; + } + + mar->buffer = malloc(buffer_size); + if (!mar->buffer) { + fprintf(stderr, "ERROR: MAR buffer could not be allocated\n"); + free(mar); + return MAR_MEM_ERROR; + } + mar->data_len = fread(mar->buffer, 1, buffer_size, fp); + if (fgetc(fp) != EOF) { + fprintf(stderr, "ERROR: File is larger than buffer (%lli)\n", + (long long int)buffer_size); + free(mar->buffer); + free(mar); + return MAR_IO_ERROR; + } + if (ferror(fp)) { + fprintf(stderr, "ERROR: Failed to read MAR\n"); + free(mar->buffer); + free(mar); + return MAR_IO_ERROR; + } + + mar->item_table_is_valid = 0; + memset(mar->item_table, 0, sizeof(mar->item_table)); + mar->index_list = NULL; + + *out_mar = mar; + return MAR_READ_SUCCESS; +} + +MarReadResult mar_open(const char* path, MarFile** out_mar) { + *out_mar = NULL; + + FILE* fp; + + fp = fopen(path, "rb"); + if (!fp) { + fprintf(stderr, "ERROR: could not open file in mar_open()\n"); + perror(path); + return MAR_IO_ERROR; + } + + MarReadResult result = mar_fpopen(fp, out_mar); + fclose(fp); + return result; +} + +#ifdef XP_WIN +MarReadResult mar_wopen(const wchar_t* path, MarFile** out_mar) { + *out_mar = NULL; + + FILE* fp; + + _wfopen_s(&fp, path, L"rb"); + if (!fp) { + fprintf(stderr, "ERROR: could not open file in mar_wopen()\n"); + _wperror(path); + return MAR_IO_ERROR; + } + + MarReadResult result = mar_fpopen(fp, out_mar); + fclose(fp); + return result; +} +#endif + +void mar_close(MarFile* mar) { + MarItem* item; + SeenIndex* index; + int i; + + free(mar->buffer); + + for (i = 0; i < TABLESIZE; ++i) { + item = mar->item_table[i]; + while (item) { + MarItem* temp = item; + item = item->next; + free(temp); + } + } + + while (mar->index_list != NULL) { + index = mar->index_list; + mar->index_list = index->next; + free(index); + } + + free(mar); +} + +int mar_read_buffer(MarFile* mar, void* dest, size_t* position, size_t size) { + // size may be provided by the MAR, which we may not have finished validating + // the signature on yet. Make sure not to trust it in a way that could + // cause an overflow. + if (size > mar->data_len) { + return -1; + } + if (*position > mar->data_len - size) { + return -1; + } + memcpy(dest, mar->buffer + *position, size); + *position += size; + return 0; +} + +int mar_read_buffer_max(MarFile* mar, void* dest, size_t* position, + size_t size) { + // size may be provided by the MAR, which we may not have finished validating + // the signature on yet. Make sure not to trust it in a way that could + // cause an overflow. + if (mar->data_len <= *position) { + return 0; + } + size_t read_count = mar->data_len - *position; + if (read_count > size) { + read_count = size; + } + memcpy(dest, mar->buffer + *position, read_count); + *position += read_count; + return read_count; +} + +int mar_buffer_seek(MarFile* mar, size_t* position, size_t distance) { + // distance may be provided by the MAR, which we may not have finished + // validating the signature on yet. Make sure not to trust it in a way that + // could cause an overflow. + if (distance > mar->data_len) { + return -1; + } + if (*position > mar->data_len - distance) { + return -1; + } + *position += distance; + return 0; +} + +/** + * Determines the MAR file information. + * + * @param mar An open MAR file. + * @param mar_position The current position in the MAR. + * Its value will be updated to the current + * position in the MAR after the function exits. + * Since its initial value will never actually be + * used, this is effectively an outparam. + * @param hasSignatureBlock Optional out parameter specifying if the MAR + * file has a signature block or not. + * @param numSignatures Optional out parameter for storing the number + * of signatures in the MAR file. + * @param hasAdditionalBlocks Optional out parameter specifying if the MAR + * file has additional blocks or not. + * @param offsetAdditionalBlocks Optional out parameter for the offset to the + * first additional block. Value is only valid if + * hasAdditionalBlocks is not equal to 0. + * @param numAdditionalBlocks Optional out parameter for the number of + * additional blocks. Value is only valid if + * hasAdditionalBlocks is not equal to 0. + * @return 0 on success and non-zero on failure. + */ +int get_open_mar_file_info(MarFile* mar, size_t* mar_position, + int* hasSignatureBlock, uint32_t* numSignatures, + int* hasAdditionalBlocks, + uint32_t* offsetAdditionalBlocks, + uint32_t* numAdditionalBlocks) { + uint32_t offsetToIndex, offsetToContent, signatureCount, signatureLen, i; + + /* One of hasSignatureBlock or hasAdditionalBlocks must be non NULL */ + if (!hasSignatureBlock && !hasAdditionalBlocks) { + return -1; + } + + /* Skip to the start of the offset index */ + *mar_position = 0; + if (mar_buffer_seek(mar, mar_position, MAR_ID_SIZE) != 0) { + return -1; + } + + /* Read the offset to the index. */ + if (mar_read_buffer(mar, &offsetToIndex, mar_position, + sizeof(offsetToIndex)) != 0) { + return -1; + } + offsetToIndex = ntohl(offsetToIndex); + + if (numSignatures) { + /* Skip past the MAR file size field */ + if (mar_buffer_seek(mar, mar_position, sizeof(uint64_t)) != 0) { + return -1; + } + + /* Read the number of signatures field */ + if (mar_read_buffer(mar, numSignatures, mar_position, + sizeof(*numSignatures)) != 0) { + return -1; + } + *numSignatures = ntohl(*numSignatures); + } + + /* Skip to the first index entry past the index size field + We do it in 2 calls because offsetToIndex + sizeof(uint32_t) + could overflow in theory. */ + *mar_position = 0; + if (mar_buffer_seek(mar, mar_position, offsetToIndex) != 0) { + return -1; + } + + if (mar_buffer_seek(mar, mar_position, sizeof(uint32_t)) != 0) { + return -1; + } + + /* Read the first offset to content field. */ + if (mar_read_buffer(mar, &offsetToContent, mar_position, + sizeof(offsetToContent)) != 0) { + return -1; + } + offsetToContent = ntohl(offsetToContent); + + /* Check if we have a new or old MAR file */ + if (hasSignatureBlock) { + if (offsetToContent == MAR_ID_SIZE + sizeof(uint32_t)) { + *hasSignatureBlock = 0; + } else { + *hasSignatureBlock = 1; + } + } + + /* If the caller doesn't care about the product info block + value, then just return */ + if (!hasAdditionalBlocks) { + return 0; + } + + /* Skip to the start of the signature block */ + *mar_position = 0; + if (mar_buffer_seek(mar, mar_position, SIGNATURE_BLOCK_OFFSET) != 0) { + return -1; + } + + /* Get the number of signatures */ + if (mar_read_buffer(mar, &signatureCount, mar_position, + sizeof(signatureCount)) != 0) { + return -1; + } + signatureCount = ntohl(signatureCount); + + /* Check that we have less than the max amount of signatures so we don't + waste too much of either updater's or signmar's time. */ + if (signatureCount > MAX_SIGNATURES) { + return -1; + } + + /* Skip past the whole signature block */ + for (i = 0; i < signatureCount; i++) { + /* Skip past the signature algorithm ID */ + if (mar_buffer_seek(mar, mar_position, sizeof(uint32_t)) != 0) { + return -1; + } + + /* Read the signature length and skip past the signature */ + if (mar_read_buffer(mar, &signatureLen, mar_position, sizeof(uint32_t)) != + 0) { + return -1; + } + signatureLen = ntohl(signatureLen); + if (mar_buffer_seek(mar, mar_position, signatureLen) != 0) { + return -1; + } + } + + if (*mar_position <= (size_t)INT64_MAX && + (int64_t)mar_position == (int64_t)offsetToContent) { + *hasAdditionalBlocks = 0; + } else { + if (numAdditionalBlocks) { + /* We have an additional block, so read in the number of additional blocks + and set the offset. */ + *hasAdditionalBlocks = 1; + if (mar_read_buffer(mar, numAdditionalBlocks, mar_position, + sizeof(uint32_t)) != 0) { + return -1; + } + *numAdditionalBlocks = ntohl(*numAdditionalBlocks); + if (offsetAdditionalBlocks) { + if (*mar_position > (size_t)UINT32_MAX) { + return -1; + } + *offsetAdditionalBlocks = (uint32_t)*mar_position; + } + } else if (offsetAdditionalBlocks) { + /* numAdditionalBlocks is not specified but offsetAdditionalBlocks + is, so fill it! */ + if (mar_buffer_seek(mar, mar_position, sizeof(uint32_t)) != 0) { + return -1; + } + if (*mar_position > (size_t)UINT32_MAX) { + return -1; + } + *offsetAdditionalBlocks = (uint32_t)*mar_position; + } + } + + return 0; +} + +/** + * Reads the product info block from the MAR file's additional block section. + * The caller is responsible for freeing the fields in infoBlock + * if the return is successful. + * + * @param infoBlock Out parameter for where to store the result to + * @return 0 on success, -1 on failure + */ +int read_product_info_block(char* path, + struct ProductInformationBlock* infoBlock) { + int rv; + MarFile* mar; + MarReadResult result = mar_open(path, &mar); + if (result != MAR_READ_SUCCESS) { + fprintf(stderr, + "ERROR: could not open file in read_product_info_block()\n"); + return -1; + } + rv = mar_read_product_info_block(mar, infoBlock); + mar_close(mar); + return rv; +} + +/** + * Reads the product info block from the MAR file's additional block section. + * The caller is responsible for freeing the fields in infoBlock + * if the return is successful. + * + * @param infoBlock Out parameter for where to store the result to + * @return 0 on success, -1 on failure + */ +int mar_read_product_info_block(MarFile* mar, + struct ProductInformationBlock* infoBlock) { + uint32_t offsetAdditionalBlocks, numAdditionalBlocks, additionalBlockSize, + additionalBlockID; + int hasAdditionalBlocks; + size_t mar_position = 0; + + /* The buffer size is 97 bytes because the MAR channel name < 64 bytes, and + product version < 32 bytes + 3 NULL terminator bytes. */ + char buf[MAXADDITIONALBLOCKSIZE + 1] = {'\0'}; + if (get_open_mar_file_info(mar, &mar_position, NULL, NULL, + &hasAdditionalBlocks, &offsetAdditionalBlocks, + &numAdditionalBlocks) != 0) { + return -1; + } + + /* We only have the one additional block type and only one is expected to be + in a MAR file so check if any exist and process the first found */ + if (numAdditionalBlocks > 0) { + /* Read the additional block size */ + if (mar_read_buffer(mar, &additionalBlockSize, &mar_position, + sizeof(additionalBlockSize)) != 0) { + return -1; + } + additionalBlockSize = ntohl(additionalBlockSize) - + sizeof(additionalBlockSize) - + sizeof(additionalBlockID); + + /* Additional Block sizes should only be 96 bytes long */ + if (additionalBlockSize > MAXADDITIONALBLOCKSIZE) { + return -1; + } + + /* Read the additional block ID */ + if (mar_read_buffer(mar, &additionalBlockID, &mar_position, + sizeof(additionalBlockID)) != 0) { + return -1; + } + additionalBlockID = ntohl(additionalBlockID); + + if (PRODUCT_INFO_BLOCK_ID == additionalBlockID) { + const char* location; + int len; + + if (mar_read_buffer(mar, buf, &mar_position, additionalBlockSize) != 0) { + return -1; + } + + /* Extract the MAR channel name from the buffer. For now we + point to the stack allocated buffer but we strdup this + if we are within bounds of each field's max length. */ + location = buf; + len = strlen(location); + infoBlock->MARChannelID = location; + location += len + 1; + if (len >= 64) { + infoBlock->MARChannelID = NULL; + return -1; + } + + /* Extract the version from the buffer */ + len = strlen(location); + infoBlock->productVersion = location; + if (len >= 32) { + infoBlock->MARChannelID = NULL; + infoBlock->productVersion = NULL; + return -1; + } + infoBlock->MARChannelID = strdup(infoBlock->MARChannelID); + infoBlock->productVersion = strdup(infoBlock->productVersion); + return 0; + } else { + /* This is not the additional block you're looking for. Move along. */ + if (mar_buffer_seek(mar, &mar_position, additionalBlockSize) != 0) { + return -1; + } + } + } + + /* If we had a product info block we would have already returned */ + return -1; +} + +const MarItem* mar_find_item(MarFile* mar, const char* name) { + uint32_t hash; + const MarItem* item; + + if (!mar->item_table_is_valid) { + if (mar_read_index(mar)) { + return NULL; + } else { + mar->item_table_is_valid = 1; + } + } + + hash = mar_hash_name(name); + + item = mar->item_table[hash]; + while (item && strcmp(item->name, name) != 0) { + item = item->next; + } + + /* If this is the first time seeing this item's indexes, return it */ + if (mar_insert_offset(mar, item->offset, item->length) == 1) { + return item; + } else { + fprintf(stderr, "ERROR: file content collision in mar_find_item()\n"); + return NULL; + } +} + +int mar_enum_items(MarFile* mar, MarItemCallback callback, void* closure) { + MarItem* item; + int i, rv; + + if (!mar->item_table_is_valid) { + if (mar_read_index(mar)) { + return -1; + } else { + mar->item_table_is_valid = 1; + } + } + + for (i = 0; i < TABLESIZE; ++i) { + item = mar->item_table[i]; + while (item) { + /* if this is the first time seeing this item's indexes, process it */ + if (mar_insert_offset(mar, item->offset, item->length) == 1) { + rv = callback(mar, item, closure); + if (rv) { + return rv; + } + } else { + fprintf(stderr, "ERROR: file content collision in mar_enum_items()\n"); + return 1; + } + item = item->next; + } + } + + return 0; +} + +int mar_read(MarFile* mar, const MarItem* item, int offset, uint8_t* buf, + int bufsize) { + int nr; + size_t mar_position = 0; + + if (offset == (int)item->length) { + return 0; + } + if (offset > (int)item->length) { + return -1; + } + + nr = item->length - offset; + if (nr > bufsize) { + nr = bufsize; + } + + // Avoid adding item->offset and offset directly, just in case of overflow. + if (mar_buffer_seek(mar, &mar_position, item->offset)) { + return -1; + } + if (mar_buffer_seek(mar, &mar_position, offset)) { + return -1; + } + + return mar_read_buffer_max(mar, buf, &mar_position, nr); +} + +/** + * Determines the MAR file information. + * + * @param path The path of the MAR file to check. + * @param hasSignatureBlock Optional out parameter specifying if the MAR + * file has a signature block or not. + * @param numSignatures Optional out parameter for storing the number + * of signatures in the MAR file. + * @param hasAdditionalBlocks Optional out parameter specifying if the MAR + * file has additional blocks or not. + * @param offsetAdditionalBlocks Optional out parameter for the offset to the + * first additional block. Value is only valid if + * hasAdditionalBlocks is not equal to 0. + * @param numAdditionalBlocks Optional out parameter for the number of + * additional blocks. Value is only valid if + * has_additional_blocks is not equal to 0. + * @return 0 on success and non-zero on failure. + */ +int get_mar_file_info(const char* path, int* hasSignatureBlock, + uint32_t* numSignatures, int* hasAdditionalBlocks, + uint32_t* offsetAdditionalBlocks, + uint32_t* numAdditionalBlocks) { + int rv; + MarFile* mar; + size_t mar_position = 0; + MarReadResult result = mar_open(path, &mar); + if (result != MAR_READ_SUCCESS) { + fprintf(stderr, "ERROR: could not read file in get_mar_file_info()\n"); + return -1; + } + + rv = get_open_mar_file_info(mar, &mar_position, hasSignatureBlock, + numSignatures, hasAdditionalBlocks, + offsetAdditionalBlocks, numAdditionalBlocks); + + mar_close(mar); + return rv; +} |