Warn rather than abort if SHA-1 is same but size is different

[wimlib] / src / blob_table.c

diff --git a/src/blob_table.c b/src/blob_table.c
index 54eca2c9f9d753514056cd082aeaae2673d47af2..0554a7a942e96897679cf65026734a101cc5cdc7 100644 (file)
--- a/src/blob_table.c
+++ b/src/blob_table.c
@@ -9,7 +9,7 @@
  */
 
 /*
- * Copyright (C) 2012, 2013, 2014, 2015 Eric Biggers
+ * Copyright (C) 2012-2016 Eric Biggers
  *
  * This file is free software; you can redistribute it and/or modify it under
  * the terms of the GNU Lesser General Public License as published by the Free
@@ -36,6 +36,7 @@
 #include "wimlib/assert.h"
 #include "wimlib/bitops.h"
 #include "wimlib/blob_table.h"
+#include "wimlib/dentry.h"
 #include "wimlib/encoding.h"
 #include "wimlib/endianness.h"
 #include "wimlib/error.h"
@@ -900,7 +901,7 @@ read_blob_table(WIMStruct *wim)
        if (!table)
                goto oom;
 
-       /* Allocate and initalize blob descriptors from the raw blob table
+       /* Allocate and initialize blob descriptors from the raw blob table
         * buffer.  */
        for (size_t i = 0; i < num_entries; i++) {
                const struct blob_descriptor_disk *disk_entry =
@@ -1007,9 +1008,6 @@ read_blob_table(WIMStruct *wim)
                }
 
                if (reshdr.flags & WIM_RESHDR_FLAG_METADATA) {
-
-                       cur_blob->is_metadata = 1;
-
                        /* Blob table entry for a metadata resource.  */
 
                        /* Metadata entries with no references must be ignored.
@@ -1056,7 +1054,10 @@ read_blob_table(WIMStruct *wim)
                         * this overrides the actual locations of the metadata
                         * resources themselves in the WIM file as well as any
                         * information written in the XML data.  */
-                       wim->image_metadata[image_index++]->metadata_blob = cur_blob;
+                       wim->image_metadata[image_index] = new_unloaded_image_metadata(cur_blob);
+                       if (!wim->image_metadata[image_index])
+                               goto oom;
+                       image_index++;
                } else {
                        /* Blob table entry for a non-metadata blob.  */
 
@@ -1091,8 +1092,6 @@ read_blob_table(WIMStruct *wim)
 
        if (wim->hdr.part_number == 1 && image_index != wim->hdr.image_count) {
                WARNING("Could not find metadata resources for all images");
-               for (u32 i = image_index; i < wim->hdr.image_count; i++)
-                       put_image_metadata(wim->image_metadata[i], NULL);
                wim->hdr.image_count = image_index;
        }
 
@@ -1261,7 +1260,7 @@ new_blob_from_data_buffer(const void *buffer, size_t size,
 struct blob_descriptor *
 after_blob_hashed(struct blob_descriptor *blob,
                  struct blob_descriptor **back_ptr,
-                 struct blob_table *blob_table)
+                 struct blob_table *blob_table, struct wim_inode *inode)
 {
        struct blob_descriptor *duplicate_blob;
 
@@ -1275,7 +1274,16 @@ after_blob_hashed(struct blob_descriptor *blob,
                 * this blob to the duplicate and update the reference to this
                 * blob (from a stream) to point to the duplicate.  The caller
                 * is responsible for freeing @blob if needed.  */
-               wimlib_assert(duplicate_blob->size == blob->size);
+               if (duplicate_blob->size != blob->size) {
+                       tchar hash_str[SHA1_HASH_SIZE * 2 + 1];
+
+                       sprint_hash(blob->hash, hash_str);
+                       WARNING("SHA-1 collision at \"%"TS"\"\n"
+                               "          (hash=%"TS", size=%"PRIu64", other_size=%"PRIu64").\n"
+                               "          File will be corrupted!",
+                               inode_any_full_path(inode), hash_str,
+                               blob->size, duplicate_blob->size);
+               }
                duplicate_blob->refcnt += blob->refcnt;
                blob->refcnt = 0;
                *back_ptr = duplicate_blob;
@@ -1309,15 +1317,17 @@ hash_unhashed_blob(struct blob_descriptor *blob, struct blob_table *blob_table,
                   struct blob_descriptor **blob_ret)
 {
        struct blob_descriptor **back_ptr;
+       struct wim_inode *inode;
        int ret;
 
        back_ptr = retrieve_pointer_to_unhashed_blob(blob);
+       inode = blob->back_inode;
 
        ret = sha1_blob(blob);
        if (ret)
                return ret;
 
-       *blob_ret = after_blob_hashed(blob, back_ptr, blob_table);
+       *blob_ret = after_blob_hashed(blob, back_ptr, blob_table, inode);
        return 0;
 }