]> wimlib.net Git - wimlib/blobdiff - src/integrity.c
--disable-verify-compression by default
[wimlib] / src / integrity.c
index bf02a6b108ecf58ebe6ca4a453cb6346fd2a1e7c..bbabaea289a2680180f227284c4b867ae9af3323 100644 (file)
@@ -1,13 +1,13 @@
 /*
  * integrity.c
  *
- * WIM files can optionally contain an array of SHA1 message digests at the end,
- * one digest for each 1 MB of the file.  This file implements the checking of
- * the digests, and the writing of the digests for new WIM files.
+ * WIM files can optionally contain a table of SHA1 message digests at the end,
+ * one digest for each chunk of the file of some specified size (often 10 MB).
+ * This file implements the checking and writing of this table.
  */
 
 /*
- * Copyright (C) 2012 Eric Biggers
+ * Copyright (C) 2012, 2013 Eric Biggers
  *
  * This file is part of wimlib, a library for working with WIM files.
  *
@@ -26,7 +26,7 @@
  */
 
 #include "wimlib_internal.h"
-#include "io.h"
+#include "buffer_io.h"
 #include "sha1.h"
 
 /* Size, in bytes, of each SHA1-summed chunk, when wimlib writes integrity
@@ -45,8 +45,9 @@ struct integrity_table {
        u8  sha1sums[0][20];
 };
 
-static int calculate_chunk_sha1(FILE *fp, size_t this_chunk_size,
-                               off_t offset, u8 sha1_md[])
+static int
+calculate_chunk_sha1(FILE *fp, size_t this_chunk_size,
+                    off_t offset, u8 sha1_md[])
 {
        int ret;
        u8 buf[BUFFER_SIZE];
@@ -86,7 +87,7 @@ static int calculate_chunk_sha1(FILE *fp, size_t this_chunk_size,
 
 
 /*
- * Reads the integrity table from a WIM file.
+ * read_integrity_table: -  Reads the integrity table from a WIM file.
  *
  * @res_entry:
  *     The resource entry that specifies the location of the integrity table.
@@ -109,46 +110,41 @@ static int calculate_chunk_sha1(FILE *fp, size_t this_chunk_size,
  *                                 data.
  *     * WIMLIB_ERR_READ:   Could not read the integrity data from the WIM file.
  */
-static int read_integrity_table(const struct resource_entry *res_entry,
-                               FILE *fp,
-                               u64 num_checked_bytes,
-                               struct integrity_table **table_ret)
+static int
+read_integrity_table(const struct resource_entry *res_entry,
+                    FILE *fp,
+                    u64 num_checked_bytes,
+                    struct integrity_table **table_ret)
 {
        struct integrity_table *table = NULL;
        int ret = 0;
        u64 expected_size;
        u64 expected_num_entries;
 
-       if (res_entry->original_size < 12) {
-               ERROR("Integrity table is too short (expected at least 12 bytes)");
-               ret = WIMLIB_ERR_INVALID_INTEGRITY_TABLE;
-               goto out;
+       if (resource_is_compressed(res_entry)) {
+               ERROR("Didn't expect a compressed integrity table");
+               return WIMLIB_ERR_INVALID_INTEGRITY_TABLE;
        }
 
-       if (res_entry->flags & WIM_RESHDR_FLAG_COMPRESSED) {
-               ERROR("Didn't expect a compressed integrity table");
-               ret = WIMLIB_ERR_INVALID_INTEGRITY_TABLE;
-               goto out;
+       if (res_entry->size < 8 || res_entry->size  > 0xffffffff) {
+               ERROR("Integrity table resource header is invalid");
+               return WIMLIB_ERR_INVALID_INTEGRITY_TABLE;
        }
 
        /* Read the integrity table into memory. */
-       if ((sizeof(size_t) < sizeof(u64)
-           && res_entry->size > ~(size_t)0)
-           || ((table = MALLOC(res_entry->size)) == NULL))
-       {
-               ERROR("Out of memory (needed %zu bytes for integrity table)",
-                     (size_t)res_entry->size);
-               ret = WIMLIB_ERR_NOMEM;
-               goto out;
+       if ((table = MALLOC(res_entry->size)) == NULL) {
+               ERROR("Can't allocate %"PRIu64" bytes for integrity table",
+                     (u64)res_entry->size);
+               return WIMLIB_ERR_NOMEM;
        }
 
        ret = read_uncompressed_resource(fp, res_entry->offset,
                                         res_entry->size, (void*)table);
 
        if (ret != 0) {
-               ERROR("Failed to read integrity table (size = %"PRIu64", "
+               ERROR("Failed to read integrity table (size = %u, "
                      " offset = %"PRIu64")",
-                     (u64)res_entry->size, res_entry->offset);
+                     (unsigned)res_entry->size, res_entry->offset);
                goto out;
        }
 
@@ -158,8 +154,8 @@ static int read_integrity_table(const struct resource_entry *res_entry,
 
        if (table->size != res_entry->size) {
                ERROR("Inconsistent integrity table sizes: Table header says "
-                     "%u bytes but resource entry says %"PRIu64" bytes",
-                     table->size, (u64)res_entry->size);
+                     "%u bytes but resource entry says %u bytes",
+                     table->size, (unsigned)res_entry->size);
                ret = WIMLIB_ERR_INVALID_INTEGRITY_TABLE;
                goto out;
        }
@@ -187,8 +183,9 @@ static int read_integrity_table(const struct resource_entry *res_entry,
        expected_num_entries = DIV_ROUND_UP(num_checked_bytes, table->chunk_size);
 
        if (table->num_entries != expected_num_entries) {
-               ERROR("%"PRIu64" entries would be required to checksum "
-                     "the %"PRIu64" bytes from the end of the header to the",
+               ERROR("%"PRIu64" integrity table entries would be required "
+                     "to checksum the %"PRIu64" bytes from the end of the "
+                     "header to the",
                      expected_num_entries, num_checked_bytes);
                ERROR("end of the lookup table with a chunk size of %u, but "
                      "there were only %u entries",
@@ -204,6 +201,8 @@ out:
 }
 
 /*
+ * calculate_integrity_table():
+ *
  * Calculates an integrity table for the data in a file beginning at offset 208
  * (WIM_HEADER_DISK_SIZE).
  *
@@ -215,16 +214,16 @@ out:
  *     Offset of byte after the last byte to be checked.
  *
  * @old_table:
- *     If non-NULL, a pointer to the table containing previously contained
+ *     If non-NULL, a pointer to the table containing the previously calculated
  *     integrity data for a prefix of this file.
  *
  * @old_check_end:
  *     If @old_table is non-NULL, the byte after the last byte that was checked
  *     in the old table.  Must be less than or equal to new_check_end.
  *
- * @show_progress:
- *     True if progress information is to be shown while calculating the
- *     integrity data.
+ * @progress_func:
+ *     If non-NULL, a progress function that will be called after every
+ *     calculated chunk.
  *
  * @integrity_table_ret:
  *     On success, a pointer to the calculated integrity table is written into
@@ -232,12 +231,13 @@ out:
  *
  * Returns 0 on success; nonzero on failure.
  */
-static int calculate_integrity_table(FILE *fp,
-                                    off_t new_check_end,
-                                    const struct integrity_table *old_table,
-                                    off_t old_check_end,
-                                    bool show_progress,
-                                    struct integrity_table **integrity_table_ret)
+static int
+calculate_integrity_table(FILE *fp,
+                         off_t new_check_end,
+                         const struct integrity_table *old_table,
+                         off_t old_check_end,
+                         wimlib_progress_func_t progress_func,
+                         struct integrity_table **integrity_table_ret)
 {
        int ret = 0;
        size_t chunk_size = INTEGRITY_CHUNK_SIZE;
@@ -245,7 +245,8 @@ static int calculate_integrity_table(FILE *fp,
        /* If an old table is provided, set the chunk size to be compatible with
         * the old chunk size, unless the old chunk size was weird. */
        if (old_table != NULL) {
-               if (old_table->chunk_size < INTEGRITY_MIN_CHUNK_SIZE ||
+               if (old_table->num_entries == 0 ||
+                   old_table->chunk_size < INTEGRITY_MIN_CHUNK_SIZE ||
                    old_table->chunk_size > INTEGRITY_MAX_CHUNK_SIZE)
                        old_table = NULL;
                else
@@ -272,6 +273,18 @@ static int calculate_integrity_table(FILE *fp,
        new_table->chunk_size = chunk_size;
 
        u64 offset = WIM_HEADER_DISK_SIZE;
+       union wimlib_progress_info progress;
+
+       if (progress_func) {
+               progress.integrity.total_bytes      = new_check_bytes;
+               progress.integrity.total_chunks     = new_num_chunks;
+               progress.integrity.completed_chunks = 0;
+               progress.integrity.completed_bytes  = 0;
+               progress.integrity.chunk_size       = chunk_size;
+               progress.integrity.filename         = NULL;
+               progress_func(WIMLIB_PROGRESS_MSG_CALC_INTEGRITY,
+                             &progress);
+       }
 
        for (u32 i = 0; i < new_num_chunks; i++) {
                size_t this_chunk_size;
@@ -279,18 +292,6 @@ static int calculate_integrity_table(FILE *fp,
                        this_chunk_size = new_last_chunk_size;
                else
                        this_chunk_size = chunk_size;
-               if (show_progress) {
-                       unsigned percent_done;
-                       u64 checked_bytes = offset - WIM_HEADER_DISK_SIZE;
-                       percent_done = checked_bytes * 100 / new_check_bytes;
-                       printf("\rCalculating integrity checksums for WIM: "
-                              "%"PRIu64" MiB of %"PRIu64" MiB (%u%%) done",
-                              checked_bytes >> 20,
-                              new_check_bytes >> 20,
-                              percent_done);
-                       fflush(stdout);
-               }
-
                if (old_table &&
                    ((this_chunk_size == chunk_size && i < old_num_chunks - 1) ||
                      (i == old_num_chunks - 1 && this_chunk_size == old_last_chunk_size)))
@@ -306,21 +307,23 @@ static int calculate_integrity_table(FILE *fp,
                                break;
                }
                offset += this_chunk_size;
+               if (progress_func) {
+                       progress.integrity.completed_chunks++;
+                       progress.integrity.completed_bytes += this_chunk_size;
+                       progress_func(WIMLIB_PROGRESS_MSG_CALC_INTEGRITY,
+                                     &progress);
+               }
        }
-       if (ret != 0) {
-               FREE(new_table);
-       } else {
-               printf("\rCalculating integrity checksums for WIM: "
-                      "%"PRIu64" MiB of %"PRIu64" MiB (100%%) done\n",
-                      new_check_bytes >> 20,
-                      new_check_bytes >> 20);
-               fflush(stdout);
+       if (ret == 0)
                *integrity_table_ret = new_table;
-       }
+       else
+               FREE(new_table);
        return ret;
 }
 
 /*
+ * write_integrity_table():
+ *
  * Writes a WIM integrity table (a list of SHA1 message digests of raw 10 MiB
  * chunks of the file).
  *
@@ -350,9 +353,9 @@ static int calculate_integrity_table(FILE *fp,
  *     If nonzero, the offset of the byte directly following the old lookup
  *     table in the WIM.
  *
- * @show_progress:
- *     True if progress information is to be shown while writing the integrity
- *     table.
+ * @progress_func
+ *     If non-NULL, a progress function that will be called after every
+ *     calculated chunk.
  *
  * Returns:
  *     0 on success, nonzero on failure.  The possible error codes are:
@@ -360,11 +363,12 @@ static int calculate_integrity_table(FILE *fp,
  *        * WIMLIB_ERR_READ:   Could not read a chunk of data that needed
  *                             to be checked.
  */
-int write_integrity_table(FILE *fp,
-                         struct resource_entry *integrity_res_entry,
-                         off_t new_lookup_table_end,
-                         off_t old_lookup_table_end,
-                         bool show_progress)
+int
+write_integrity_table(FILE *fp,
+                     struct resource_entry *integrity_res_entry,
+                     off_t new_lookup_table_end,
+                     off_t old_lookup_table_end,
+                     wimlib_progress_func_t progress_func)
 {
        struct integrity_table *old_table;
        struct integrity_table *new_table;
@@ -372,11 +376,11 @@ int write_integrity_table(FILE *fp,
        off_t cur_offset;
        u32 new_table_size;
 
+       wimlib_assert(old_lookup_table_end <= new_lookup_table_end);
+
        cur_offset = ftello(fp);
-       if (cur_offset == -1) {
-               ERROR_WITH_ERRNO("Failed to get offset in WIM");
+       if (cur_offset == -1)
                return WIMLIB_ERR_WRITE;
-       }
 
        if (integrity_res_entry->offset == 0 || old_lookup_table_end == 0) {
                old_table = NULL;
@@ -395,7 +399,7 @@ int write_integrity_table(FILE *fp,
 
        ret = calculate_integrity_table(fp, new_lookup_table_end,
                                        old_table, old_lookup_table_end,
-                                       show_progress, &new_table);
+                                       progress_func, &new_table);
        if (ret != 0)
                goto out_free_old_table;
 
@@ -430,6 +434,8 @@ out_free_old_table:
 }
 
 /*
+ * verify_integrity():
+ *
  * Checks a WIM for consistency with the integrity table.
  *
  * @fp:
@@ -442,9 +448,9 @@ out_free_old_table:
  *     Number of bytes in the WIM that need to be checked (offset of end of the
  *     lookup table minus offset of end of the header).
  *
- * @show_progress:
- *     True if progress information is to be shown while checking the
- *     integrity.
+ * @progress_func
+ *     If non-NULL, a progress function that will be called after every
+ *     verified chunk.
  *
  * Returns:
  *     > 0 (WIMLIB_ERR_*) on error
@@ -452,12 +458,27 @@ out_free_old_table:
  *     were no inconsistencies.
  *     -1 (WIM_INTEGRITY_NOT_OK) if the WIM failed the integrity check.
  */
-static int verify_integrity(FILE *fp, const struct integrity_table *table,
-                           u64 bytes_to_check, bool show_progress)
+static int
+verify_integrity(FILE *fp, const tchar *filename,
+                const struct integrity_table *table,
+                u64 bytes_to_check,
+                wimlib_progress_func_t progress_func)
 {
        int ret;
        u64 offset = WIM_HEADER_DISK_SIZE;
        u8 sha1_md[SHA1_HASH_SIZE];
+       union wimlib_progress_info progress;
+
+       if (progress_func) {
+               progress.integrity.total_bytes      = bytes_to_check;
+               progress.integrity.total_chunks     = table->num_entries;
+               progress.integrity.completed_chunks = 0;
+               progress.integrity.completed_bytes  = 0;
+               progress.integrity.chunk_size       = table->chunk_size;
+               progress.integrity.filename         = filename;
+               progress_func(WIMLIB_PROGRESS_MSG_VERIFY_INTEGRITY,
+                             &progress);
+       }
        for (u32 i = 0; i < table->num_entries; i++) {
                size_t this_chunk_size;
                if (i == table->num_entries - 1)
@@ -473,38 +494,31 @@ static int verify_integrity(FILE *fp, const struct integrity_table *table,
                if (!hashes_equal(sha1_md, table->sha1sums[i]))
                        return WIM_INTEGRITY_NOT_OK;
 
-               if (show_progress) {
-                       u64 checked_bytes = offset - WIM_HEADER_DISK_SIZE;
-                       unsigned percent_done = checked_bytes * 100 / bytes_to_check;
-                       printf("\rVerifying integrity of WIM: "
-                              "%"PRIu64" MiB of %"PRIu64" MiB (%u%%) done",
-                              checked_bytes >> 20,
-                              bytes_to_check >> 20,
-                              percent_done);
-                       fflush(stdout);
-               }
                offset += this_chunk_size;
+               if (progress_func) {
+                       progress.integrity.completed_chunks++;
+                       progress.integrity.completed_bytes += this_chunk_size;
+                       progress_func(WIMLIB_PROGRESS_MSG_VERIFY_INTEGRITY,
+                                     &progress);
+               }
        }
-       printf("\rVerifying integrity of WIM: "
-              "%"PRIu64" MiB of %"PRIu64" MiB (100%%) done\n",
-              bytes_to_check >> 20,
-              bytes_to_check >> 20);
-       fflush(stdout);
        return WIM_INTEGRITY_OK;
 }
 
 
 /*
+ * check_wim_integrity():
+ *
  * Verifies the integrity of the WIM by making sure the SHA1 message digests of
  * ~10 MiB chunks of the WIM match up with the values given in the integrity
- * tabel.
+ * table.
  *
  * @w:
  *     The WIM, opened for reading, and with the header already read.
  *
- * @show_progress:
- *     True if progress information is to be shown while checking the
- *     integrity.
+ * @progress_func
+ *     If non-NULL, a progress function that will be called after every
+ *     verified chunk.
  *
  * Returns:
  *     > 0 (WIMLIB_ERR_*) on error
@@ -514,7 +528,8 @@ static int verify_integrity(FILE *fp, const struct integrity_table *table,
  *     -2 (WIM_INTEGRITY_NONEXISTENT) if the WIM contains no integrity
  *     information.
  */
-int check_wim_integrity(WIMStruct *w, bool show_progress)
+int
+check_wim_integrity(WIMStruct *w, wimlib_progress_func_t progress_func)
 {
        int ret;
        u64 bytes_to_check;
@@ -540,7 +555,8 @@ int check_wim_integrity(WIMStruct *w, bool show_progress)
                                   bytes_to_check, &table);
        if (ret != 0)
                return ret;
-       ret = verify_integrity(w->fp, table, bytes_to_check, show_progress);
+       ret = verify_integrity(w->fp, w->filename, table,
+                              bytes_to_check, progress_func);
        FREE(table);
        return ret;
 }