Patent Grant
7841010
This disclosure relates to integrity verification in the computer field, and more particularly to integrity verification using checksums or hashes.
Integrity verification is well known in the computer/cryptographic fields. Typical applications are to ensure that a message that is transmitted is received without being corrupted accidentally (due to errors caused by the communications channel) or maliciously such as by a hacker. The same verification techniques are used to make sure that the message (or other data including software programs) that has been stored, such as in computer memory, has not been similarly corrupted. Hackers corrupt computer data or software for two reasons: one is mere vandalism, to render the corrupted material worthless to its owner. The other is more subtle and important, in that hackers tend to tamper with the software programs (‘code’) of others by planting therein their own bits of code that allow the hackers to illicitly manipulate or use for their own purposes the tampered-with software.
An example of such tampering is that distributors of music and video content using the Internet or other computer networks do so using a Digital Rights Management system (DRM) to protect the content from illicit copying and use. DRM is used to protect digital content transferred over a network and transferred from a computer to an associated playback device. The DRM is implemented by software resident in the audio/video player or associated computer, and hence is available for tampering by a hacker, who thereby can “crack” the DRM and render it useless. For instance, the DRM tends to use cryptographic techniques to restrict usage to those who have been granted a right to the content, typically pieces of music or video programs, and cracking it may make the deciphering keys available to the hacker for unauthorized use of the content.
For many digital file transfer situations, a hacker who legitimately receives the content may try thereby to break the DRM protection scheme, so as to give illicit access to third parties. An identified weak link in DRM security is the overall process, rather than the encryption scheme itself. Thus hackers may try to corrupt the DRM software to gain illicit access to the content so it is for instance stripped of the encryption or other protective mechanisms, using the above described tampering approach.
Such corruption of computer/data files is a well known problem with a well known solution, known as integrity verification. This involves taking the file as a whole, or a fixed length part thereof, and computing a checksum or hash value for that part. A checksum is a form of redundancy check, and is a simple measure to protecting the integrity of data by allowing one to later detect errors in data sent through space (a communication) or time (stored and later retrieved.) It operates by adding up the values of the basic components of the file or part thereof (which are already in numerical, typically binary, form) and storing the resulting total value. Later, anyone can perform the same checksum operation on the data, compare the result to the authentic checksum, and if the totals match, conclude that the data was likely not corrupted. Simple checksums cannot detect many types of errors, such as reordering of portions of the data, inserting or deleting zero values, or multiple errors which sum to zero.
More sophisticated types of redundancy checks are known to overcome these weaknesses. Even they are not of much use against hackers, since their simple and known operations are easy to circumvent. To provide a higher level of integrity against hackers, the use of a cryptographic hash function (one way function) known as an HMAC (hash message authentication code) is well known in the field. Examples are MD5 (Message Digest version 5) and Secure Hash Algorithm 1 (SHA 1) generally referred to hereinafter as hash or hash functions. These are generally secure in a mathematical sense against hackers due to the complexity of the calculation of the hash value, which is used instead of the simpler checksum. However, like most advanced cryptographic techniques, such hash functions are computationally intensive thus being slow or requiring substantial computer performance, and while suitable for transmission and protection of high value material, may not be suitable for the commercial distribution of audio/video material which must be played on a compact and inexpensive dedicated player.
This disclosure is directed to a method and apparatus for integrity verification, suitable for protection of computer software and other types of data files from tampering by hackers, and suitable for use in devices including computers but also including audio/video players, cell phones, PDAs, etc. which may have limited computing power. One application, as stated above, is to protect the integrity of the DRM code in such a device against having malicious code planted therein by a hacker. Hence when the protected code is executed, the verification process takes place and if the code has been tampered with, it stops executing. Of course, this requires that the code be accompanied by the requisite verification data.
The present method is in part conventional as explained above. The code (or other data) to be protected, before being stored in the host device (such as the media player) which will later execute it, is processed to provide the protection. This involves partitioning the code (typically at this point already in binary form, also known as object code) in portions for which a checksum or preferably a hash value is computed and stored. However, unlike in the usual checksum and hash techniques which are applied to standard length blocks of data, the portions here are not of equal length. Also, unlike prior techniques, not all portions are subject to verification. Instead, the user who is the programmer or designer of the DRM system sets criteria for deciding which files or portions of each file are subject to verification. The process thereby creates an integrity verification file associated with each data/code file, where the integrity verification file includes a list of entries, each entry including the beginning address in main memory of each verified portion, the length of the portion, and the checksum/hash value.
Thus each time a content or code file is accessed or a code file is executed, the system reads the associated integrity verification file and performs the integrity check on the actual file contents, to test for corruption, as is conventional for integrity verification.
This process is applied to the compiled code, in its binary form. The tool (program) used to generate the integrity information takes the binary code file as input, and some debugging information or a link map to know where each part of the alphanumeric code is located in the binary code.
For example, assume a media player program is composed of the following source code files:
The compiler/linker takes all those source code files as input and generates files player.exe and player.map. File player.exe contains the generated program in its binary form, player.map contains a link map, generated by the linker which tells where each part of the code is located in memory or in the binary. Note that instead of the map file, one could also use debugging information generated by the compiler, embedded in the binary usually. Now, assume the designer does not care if the input/output code or the GUI (graphical user interface) code is modified, but does care if the DRM code or the decoder code is, since that is much more important from a security standpoint.
The designer creates two rules, the first one is to checksum (or hash) any code that belongs to the file drm.c, and the second one is to checksum (or hash) any code that belongs to the file decoder.c. Here the rule is based on the file name, but it could also be based on the function name. For example, a rule could be to checksum only the function “foo” in the file drm.c. Rules are based on the filename, or the function names. A rule can also be a regular expression that will match several filenames of several function names. The tool generating the Integrity information parses the player.exe file, and helped by the map file or the debugging information, it identifies all the part of the code that need to be checksummed or hashed. For example, if the map files says that function “foo” in file drm.c is mapped at address 0x10000 and is 144 bytes long, the generating tool will parse the player.exe file locate those 144 bytes and will checksum them. Another useful rule is to limit the maximum length of the portion to be subject to the checksum or hash, to enhance performance (reducing processing time).
At 16, the selected binary information (e.g., a file) is partitioned into portions (blocks, but not of fixed length), and a first portion here designated Pi is selected. The length of each portion is set to vary. Hence both the selection of which portions (or files) are to be subject to the remainder of the process, and the length of each portion as determined at 18, are decided by the person performing the process (typically a computer software engineer), according to some predetermined criteria or rules. Note that the length of each portion varies, in some hard to predict way, to make it difficult to “crack” by hackers. The length of each portion is defined as some number of bits (or bytes or other convenient unit), here designated Li.
Next at 20, those Li bits of portion Pi are accessed. In 24, the predetermined checksum or hash function is applied to all the Li bits of Pi, to calculate the checksum or hash value Ci. Next at 26, an integrity verification (IV) file is created with a number of entries, one for each portion Pi. Each entry (or ‘tuple’ or vector in computer nomenclature) has three parts (fields): the address in the memory of the beginning of Pi, the length Li of Pi, and the checksum or hash value Ci. Then at 28, the value of i is incremented, and the process reentered at step 12, until the entire message/code file has been so processed. Thus the integrity verification file consists of all the concatenated IV vectors, which file then is stored.
Note that software (code) takes two main forms: source code, which is the alpha-numeric form as written by programmers and which is intelligible to people, and binary or object code, which is the compiled form of the source code and actually executed by a computer or processor. Since as described above, the human programmer decides which code portions are to be provided with integrity verification, he makes that decision based on the source code version, but the integrity verification process then must be applied to the binary version. Thereby, in accordance with the invention, a mapping between the source and binary code is provided, as part of the above process. There are several ways to do this, but exemplary well known methods use a link map or debugging symbols, which show explicitly where each part of the source code has been linked in the binary code.
The programmer can use a map file or debugging information. The map file is generated by the linker (Id or link.exe for example) and is stored on the computer hard disk usually. It is not created by the integrity verification tool. The map is only used for debugging purposes usually, and is not used by the compiler. Most of the time it is not used at all, and linkers usually do not generate one by default. Using it, software engineers can identify where in the source code a fault occurred for example.
Here is an example of map file content:
01:09f86e0_initSharedGlobals 6e9f96e0 StdPlayer: StdPlayerRoutines.obj
01:09f9470_drawStatusStuff 6e9fa470 StdPlayer: StdPlayerRoutines.obj
01:09f9a90_getPseudoChapterMenuWidth 6e9faa90 StdPlayer:StdPlayerRoutines.obj
01:09f9ba0_CalculateChapterCheckSum 6e9faba0 StdPlayer:StdPlayerRoutines.obj
Reading this file, one can deduce:
If debugging information is used instead of the link map, then the same kind of deductions can be performed by parsing it. The debugging information can be embedded in the binary code or in a separate file (for respectively, for instance, the Apple Mac and other operating systems or the Windows operating system). For each platform, the debugging information has a well known format (PDB for Windows, DWARF or STABS for the Mac OS). The debugging information is generated by the compiler, whereas the link map is generated by the linker.
The entity that carries out the process of
The associated apparatus which carries out the
The relevant portion Pi, of length Li, is then held in portion buffer 40, and loaded into checksum/hash calculator 46 to provide the checksum/hash value Ci. Value Ci is then loaded into the Ci field of IV entry (tuple or vector) buffer 48, the other fields of which are similarly loaded with values Li and the beginning address in storage 30 of portion Pi. Then, the entire contents of buffer 48 are loaded in the IV file store 50, concatenated with all such earlier determined IV vectors.
In
This downloaded code is then loaded in to portion buffer 70, and transferred to checksum/hash calculator 74 which carries out the same operation as calculator 46 in
As well known in the computer field, a binary file may be subject to relocation in memory. For example, assume that software libraries A and B are both designed to load at particular address, and are both position dependent. As well known, position dependent means that the addresses are absolute in computer memory, not relative to a start address. A library in this context means a collection of subprograms used in modular fashion. Typically a library includes helper code and data. If a particular program is using both libraries A and B at the same time, then one of them must be moved in terms of its memory address. In order to move it, the host device's operating system must modify the code of the moved library dynamically. This of course would alter the computed checksum/hash value, hence it would cause the IV match to fail. To avoid this, the verification process of
This disclosure is illustrative and not limiting; further modifications will be apparent to one skilled in the art in light of this disclosure, and are intended to fall within the scope of the appended claims.
| Number | Name | Date | Kind |
|---|---|---|---|
| 5382955 | Knierim | Jan 1995 | A |
| 5454000 | Dorfman | Sep 1995 | A |
| 6738932 | Price | May 2004 | B1 |
| 7225341 | Yoshino et al. | May 2007 | B2 |
| 20030018878 | Dorward et al. | Jan 2003 | A1 |
| 20030188117 | Yoshino et al. | Oct 2003 | A1 |
| 20030188231 | Cronce | Oct 2003 | A1 |
| 20030200207 | Dickinson | Oct 2003 | A1 |
| 20030221116 | Futoransky et al. | Nov 2003 | A1 |
| 20040117616 | Silvester | Jun 2004 | A1 |
| 20070156638 | Vadekar et al. | Jul 2007 | A1 |
| Number | Date | Country |
|---|---|---|
| WO-9927450 | Jun 1999 | WO |
| WO-2006116871 | Nov 2006 | WO |
| WO-2006116871 | Nov 2006 | WO |
| Number | Date | Country | |
|---|---|---|---|
| 20080168564 A1 | Jul 2008 | US |
