From: Eric Biggers <ebiggers3@gmail.com>
Date: Mon, 5 Jul 2021 06:03:50 +0000 (-0700)
Subject: Warn rather than abort if SHA-1 is same but size is different
X-Git-Tag: v1.13.5~8
X-Git-Url: https://wimlib.net/git/?a=commitdiff_plain;h=ba8edbaf442ff0b30ca9ee1d16a4a646876ed94c;p=wimlib

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

Assertions should only be used for bugs in wimlib, but this scenario can
also happen if there is a SHA-1 collision, or if the SHA-1 hash provided
by the filesystem for a WIM-backed file on Windows is wrong.
---

diff --git a/include/wimlib/blob_table.h b/include/wimlib/blob_table.h
index f15c6117..d16f5a24 100644
--- a/include/wimlib/blob_table.h
+++ b/include/wimlib/blob_table.h
@@ -412,7 +412,7 @@ new_blob_from_data_buffer(const void *buffer, size_t size,
 extern 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);
 
 extern int
 hash_unhashed_blob(struct blob_descriptor *blob, struct blob_table *blob_table,
diff --git a/src/blob_table.c b/src/blob_table.c
index 3fcde011..0554a7a9 100644
--- a/src/blob_table.c
+++ b/src/blob_table.c
@@ -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"
@@ -1259,7 +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;
 
@@ -1273,7 +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;
@@ -1307,15 +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;
 }
 
diff --git a/src/template.c b/src/template.c
index e5413e5f..1c018df7 100644
--- a/src/template.c
+++ b/src/template.c
@@ -114,7 +114,8 @@ inode_copy_checksums(struct wim_inode *inode,
 
 		back_ptr = retrieve_pointer_to_unhashed_blob(blob);
 		copy_hash(blob->hash, template_blob->hash);
-		if (after_blob_hashed(blob, back_ptr, blob_table) != blob)
+		if (after_blob_hashed(blob, back_ptr, blob_table,
+				      inode) != blob)
 			free_blob_descriptor(blob);
 	}
 }
diff --git a/src/win32_capture.c b/src/win32_capture.c
index 0298cf00..8c9d726d 100644
--- a/src/win32_capture.c
+++ b/src/win32_capture.c
@@ -1604,7 +1604,8 @@ try_to_use_wimboot_hash(HANDLE h, struct wim_inode *inode,
 			return 0;
 		back_ptr = retrieve_pointer_to_unhashed_blob(blob);
 		copy_hash(blob->hash, hash);
-		if (after_blob_hashed(blob, back_ptr, blob_table) != blob)
+		if (after_blob_hashed(blob, back_ptr, blob_table,
+				      inode) != blob)
 			free_blob_descriptor(blob);
 	}