 |
Claims  |
|
|
Having thus described our invention, what we claim as new, and desire to
secure by Letters Patent is:
1. A software copy-protection apparatus, which is operable with a host
computer system, comprising:
a magnetic medium having tracks formed thereon which are divided into
sectors, with each sector being comprised of a plurality of bit storage
locations, with indicia being formed in at least one portion of at least
one sector, with said indicia not being modifiable by the medium write
process;
a product program stored on said medium, at least a portion of which is in
an encrypted form, and at least a portion of which may be in an
unencrypted form;
a support computing system operable with said host computer system,
including a decryption key for use in executing said product programs on
said host computer system;
means for ascertaining said indicia are present on said medium;
means included in said support computing system for utilizing said
decryption key to decrypt said encrypted portion of said program; and
means responsive to the ascertaining that said indicia is on said medium,
and said encrypted portion of said program has been decrypted, to permit
said support computing system to execute the encrypted portion of said
program product and said host computer system to execute said unencrypted
portion of said program product, if any.
2. The combination claimed in claim 1, including means for modifying the
information on said medium by said host computer system to identify said
program for use on said system.
3. The combination claimed in claim 2, including means for utilizing the
modified information to prohibit the use of the program on said medium on
any other computing system.
4. In a computing system, the combination comprising:
a host computer connected to a host system bus;
an interface system connected to said host system bus;
a support computer connected to a support system bus, which support system
bus is connected to said interface system, with at least a portion of said
support computer being logically inaccessible to said host computer with
said support computer system having stored therein a first decryption key
for use in decrypting programs which may be run on said computing system;
an original magnetic medium having tracks formed thereon which are divided
into sectors, with each sector being comprised of a plurality of bit
storage locations, with indicia in at least one bit storage location of at
least one sector of at least one track that are not modifiable by the
medium write process, with magnetic domain pattern marks overlapping said
indicia, with said domain pattern marks not being createable by the medium
write process;
information stored on said magnetic medium which includes a second
encrypted decryption key which can be decrypted with said first decryption
key, an encrypted part of a program which can be decrypted with said
second decryption key, possibly an unencrypted part of said program, and
an encrypted description of said indicia and said magnetic domain marks,
which can be decrypted with said second decryption key;
means for determining first use of said medium by ascertaining the presence
of said indicia and said domain pattern marks on said medium, with said
magnetic domain pattern mark being destroyed by the determination, whereby
said medium is made inoperable on a different computing system;
means included in said support computer for utilizing said first decryption
key to decrypt said second decryption key;
means responsive to the decryption of said second decryption key to decrypt
the encrypted description of said indicia and said magnetic domain pattern
marks;
means for comparing the decrypted indicia and magnetic domain pattern marks
with the actual indicia and the actual magnetic domain pattern marks to
ascertain the authenticity of said magnetic medium;
means responsive to said magnetic medium being identified as authentic for
storing said decrypted second key in said support computer, including
means for storing an identifier on said magnetic medium which identifies
the storage location in said support computer where said encrypted second
key is stored;
means responsive to determining the existence of said identifier on said
medium, including means to locate the storage location of the decrypted
second key in said support computer;
means for retrieving said decrypted second key from the storage location in
said support computer and for using said key to decrypt and run the
encrypted portions of said program on said support computer; and means for
running the unencrypted portions, if any, of said program on said host
computer.
5. In a computing system, the combination comprising:
a host computer system connected to a host computer bus;
a processor connected to a support computer bus, which communicates with
said host computing system, and which executes a particular set of
instructions, with the execution and results of predetermined ones of said
particular set of instructions being inaccessible to said host computer
system;
a read-only memory, connected to said support computer bus, addressable by
said processor, and 13 not addressable by said host computer system,
wherein said read-only memory includes data representing a private
decryption key, which is to be used in conjunction with a particular
public-key encryption/decryption algorithm; a first read-write memory,
connected to said support computer bus, addressable by said processor, and
not addressable by said host computer system;
a second read-write memory, connected to each of said host computer bus and
said support computer bus, addressable by each of said processor and said
host computer system;
a set of communicating registers, connected to each of said host computer
bus and said support computer bus, addressable by each of said processor
and said host computer system for transferring data between each other;
and
a set of bus receivers, connected from said host computer bus to said
support computer bus, which enable the state of said host system bus to be
monitored from said support computer bus.
6. The combination claimed in claim 5, wherein said processor includes
means for examining the execution of instructions and the results of
instruction execution by said host computer, utilizing said set of bus
receivers.
7. The combination claimed in claim 6, wherein said processor, said
read-only memory, said first and second read-write memories, said set of
communicating registers and said set of bus receivers are physically
enclosed in a tamper-proof package.
8. The combination claimed in claim 7, including:
an original magnetic medium having tracks formed thereon which are divided
into sectors, with each sector being comprised of a plurality of bit
storage locations, with doubly-marked regions comprised of a first mark
comprised of indicia in at least one bit storage location of at least one
sector of at least one track that are not modifiable by the medium write
process, and a second mark comprised of magnetic domain pattern marks
overlapping said indicia, with said magnetic domain pattern marks not
being createable by the medium write process;
information stored on said magnetic medium including an encrypted
decryption key and encrypted mark descriptors, and further including an
application program, at least a portion of which is encrypted;
means for reading said encrypted decryption key into said first read-write
memory; means for utilizing said private decryption key stored in said
read-only memory and said encrypted decryption key stored in said first
read-write memory to decrypt said encrypted decryption key, and storing
the decrypted key in said first read-write memory;
means for reading said encrypted mark descriptors from said magnetic medium
and storing same in said first read-write memory;
means for utilizing said decrypted key stored in said first-read write
memory to decrypt the encrypted descriptors stored in said first
read-write memory, including means for storing the decrypted descriptors
back in said first read-write memory;
means for one of detecting the existence of, and measuring the properties
of, said doubly marked regions on said magnetic medium to produce mark
descriptors, including means for storing said mark descriptors in said
first read-write memory;
means for producing a first comparison by comparing said decrypted
descriptors with said mark descriptors;
means responsive to the first comparison to indicate one of said magnetic
medium is a copy if there is no comparison, and indicating said magnetic
medium is an original if there is a comparison;
a portion of said first read-write memory which is persistent, in the sense
that information written into said persistent portion is retained when
electrical power is not supplied to said persistent portion;
means responsive to the determination that there is a comparison, for
storing said decrypted key in the persistent portion of said first
read-write memory, and writing said original magnetic medium with the
address where said decrypted key is stored in said first read-write
memory. |
|
 |
|
 |
Claims |
|
|
|
Description |
|
|
DESCRIPTION
1. Technical Field
The invention is in the field of data processing, and specifically is
directed to a software copy protection mechanism. In particular, a
mechanism is provided which restricts software, distributed on disks or
other magnetic medium, to use on a single computing machine, while
allowing the creation of "backup" copies without compromising the
protection.
2. Background Art
Copy-protection mechanisms are utilized to inhibit software piracy, which
is the creation of unauthorized copies of commercial software. As the
market for personal computers, home computers, work stations and
intelligent products grows, piracy increasingly becomes a problem. The
purpose of a copy-protection mechanism is to deter piracy by making
copying of software as difficult as possible. Two basic classes of copy
protection mechanisms have evolved to deter piracy, namely, software-based
and hardware-key.
Software-based methods encode information on a disk so that conventional
copying facilities available in most operating systems cannot accurately
copy the information to another disk. The program on the disk checks for
this encoded information, and fails to function unless it is there.
Copying programs are now commercially available which can successfully
copy most disks protected in this manner.
Hardware-key methods rely on the existence of information known as the
"key", available to the program, but resident in the system hardware
rather than on changeable magnetic medium like the disk. The program
checks for the "key" information, and fails to function if the key is not
found. Hardware duplication facilities are not commonly found in personal
computers, while disk duplication facilities are. This makes hardware keys
more costly to duplicate than software, so these methods can be more
effective than software-based methods in detering piracy.
One proposed hardware-key method requires that the computer manufacturer
install a hardware serial number in each machine as the hardware-key. This
method requires every piece of software to be customized to a particular
machine. This limits the availability and the interchangeability of
software. A second method, currently in use, requires the software user to
buy a special piece of hardware with each software product. This piece of
hardware provides the key, and it has to be attached to the machine
whenever the corresponding software is run, which makes the method
unattractive.
In a large part each of the above protection methods is vulnerable to
copying of a binary image of the running application from the system
working memory, after such key checks have been made.
A number of patents have issued directed to software copy-protection
mechanisms, each having certain advantages and disadvantages. Two such
patents, U.S. Pat. No. 4,246,638 to Thomas and U.S. Pat. No. 4,168,396 to
Best provide a protection mechanism by means which are essentially similar
to each other. In both instances, the software package which is to be run
on some particular personal computer is customized by the manufacturer to
be compatible with the decryption keys and systems built into that
particular computer. This is extremely cumbersome and places a large
burden on users and vendors alike. The software used in the operation of
the apparatus described in these patents causes no changes in structure to
take place either in the user's computer or in the distribution disk.
U.S. Pat. No. 4,238,854 to Ehrsam et al addresses itself to a different
problem, namely a means by which an encryption/decryption engine may be
integrated into a multi-user mainframe computer system to protect a user's
files from access by other users. It does not address itself either to the
implementation of a distribution channel between software vendors and
users of personal computers or to the execution environment for such
protected software. It assumes that an operating system or hardware
mechanism for provision of levels of privilege already exist in the
machine in which this engine is to be installed. This is not the common
case in personal computers.
U.S. Pat. No. 3,996,449 to Attanasio et al addresses itself to a problem
faced by the operators of large mainframe computers. This problem is
"penetration of security" meaning "a successful subversion of the file
management component to change the backup copy of the operating system" or
"counterfeiting the computer manufacturer's packing and delivery
procedure" (for software). Such subversion or counterfeiting is
accomplished by a third party who seeks to gain access to confidential
files, payroll programs, or other potentially lucrative information or
processors by means of features one has installed in the operating system.
U.S. patent application Ser. No. 06-567,294 entitled, "A Hardware
Key-on-Disk System for Copy-Protecting Magnetic Storage Media", filed Dec.
30, 1983, which application is assigned to the assignee of the subject
invention, incorporates the best features of both software based and
hardware key methods. A hardware key is encoded directly onto a magnetic
medium such as a floppy disk. This key consists of indicia in at least one
subsection of at least one section of the disk that are not modifiable by
the conventional medium write process. The data read from a section
containing indicia differs in a predictable way from the data written to
that section. The disk can be authenticated as the original disk by
comparing a read-following-write with the expected results of such an
operation. The software functions only in the presence of this key, as the
key indicates the original medium is present. Software use is thus
restricted to those users possessing an original distribution disk. Backup
copies are not allowed by this system.
According to the subject invention, a method is set forth which restricts
software, distributed on disks, to use on a single machine and allows
backup copying. This mechanism involves making the distribution disks
functionally uncopyable, until it is modified by the execution of a
procedure which requires the cooperation of a co-processor. Upon
modification, the software may be copied but can only be used on the
machine containing the co-processor which participated in the modification
procedure.
DISCLOSURE OF THE INVENTION
Method and apparatus are disclosed for the copy-protection of software,
distributed on magnetic medium such as floppy disks, and used on a
computing system. The apparatus which comprise the copy protection system
consists of structures or marks imposed on the distribution magnetic
medium, and a hardware subsystem installed in the intended recipient
computer system. The hardware subsystem is a computing system the
components of which include; a CPU, read only memory (ROM) which is
logically inaccessible from the host system and containing software in the
form of a monitor (which begins execution at power-on-time), random access
memory (RAM) a portion of which is logically inaccessible from the host
system and a portion of which may, under the control of the subsystem CPU
be read from or written to by the host CPU, a memory in the form of
"nonvolatile" RAM such as EEPROM, which is logically inaccessible from the
host system, a timer and a real time clock which are logically
inaccessible from the host system, a register which may, under the control
of the subsystem CPU be read from or written to by the host CPU, and a set
of bus receivers by means of which the subsystem CPU may "observe" the
state of the host system bus. All of the above mentioned components of the
hardware subsystem are logically accessible to the subsystem CPU, and are
packaged in a manner which makes them physically inaccessible to the user
of the host computing system.
The portion of the apparatus which consists of structures imposed on the
distribution magnetic medium consists of two sorts of structures. One of
these kinds is purely a pattern of magnetic domains on the medium which
are not within the repertoire of domain patterns which can be created by
the medium read/write apparatus of the target computer. The other kind of
structure consists of regions on the medium on which boundaries between
magnetic domains cannot be imposed by the medium read/write apparatus of
the target computer. Very large magnetic domains are an example of the
first kind of structure. Media voids are an example of the second kind of
structure. The apparatus which consists of the two kinds of structures
imposed on the distribution magnetic medium is so configured that the
structures overlap each other.
While the medium read/write apparatus of the host system cannot create
these structures, it can, through operations which can be performed by the
read/write apparatus of the host system, detect and measure these
structures. The operations required to measure the structures which won't
support domain transitions are precisely the operations which will destroy
the structure which consists of a domain pattern.
The destruction of one kind of structure by the write-read operation
performed on the other kind of structure provides a means to the hardware
subsystem to determine whether or not a particular piece of magnetic
medium has been accessed by a subsystem of its own kind. The procedures
performed by the subsystem allow the transfer of a certain critical piece
of information from the medium only if this transfer has not ever been
performed from this medium in the past. The procedure of transferring the
information thus changes the structures on the medium so that no apparatus
of this kind will perform the critical information transfer in the future.
The hardware subsystem is supplied with a piece of information built in
which is critical to the use of the information transferred from the
magnetic medium, thus, the subsystem cannot be replicated by the user. The
subsystem can "observe" the portion of the transfer process mediated by
the host system. Thus, the transfer cannot be mimicked by software run on
the host. The magnetic medium cannot be replicated by the user or used
more than once for the transfer, thus, the medium cannot be copied in a
form useful for the transfer operation.
The critical information which is transferred is a decryption key needed to
run a portion of the application software on the subsystem. The decryption
key is itself encrypted. The critical information built into the subsystem
is the decryption key needed to restore the transferred decryption key to
useful form, thus, the user cannot use this information without the
cooperation of a subsystem of this kind.
Means are thus provided to bind a particular software distribution package,
some part of which is in encrypted form, to a particular hardware
subsystem and means are provided to make repeating this binding with the
same particular software distribution package or replica thereof to
another system, exceedingly difficult. Software distributed on magnetic
medium is, by these means, protected from copying.
After such a binding has taken place, the support hardware may be called
upon to execute some portion of the protected software. The support
hardware which experienced the binding to that software package alone has
the means to fill this call, as it alone has the key to decrypt the
software. Both decryption and execution take place in memory which is
logically and physically inaccessible to the user. Thus, the software is
protected from copying by never being exposed to the user in a useable
form.
After the transfer process is complete, when the support hardware is called
upon to decrypt and execute some portion of the software, the apparatus
comprised of structures on the distribution medium are not accessed. Since
all other parts of the distribution medium are reproducible, the medium
read/write apparatus of the host system can reproduce them. Thus, "backup"
copies can be made after transfer but only the original system has the
means to use the "backups".
BRIEF DESCRIPTION OF THE DRAWINGS
FIG. 1 is a block diagram representation of a computing system, including
support hardware, according to the invention;
FIGS. 2.0-2.3 are sectional diagrams of a disk, illustrating how doubly
marked regions are created thereon;
FIG. 3 is a section of a sector on a disk, illustrating the placement of
doubly marked regions relative to a data record;
FIGS. 4.0-4.3 are sections of a sector on a disk, illustrating the four
step process for reading and writing data from the sector to verify the
existence of a doubly marked region on the disk;
FIG. 5 is a flow chart illustrating the support software which executes in
the host computer;
FIG. 6 is a flow chart illustrating the monitor software which executes in
the support hardware;
FIGS. 7.1, 7.2 and 7.3, when viewed with FIG. 7.1 on top, FIG. 7.2 in the
middle and FIG. 7.3 on the bottom, constitute a flow chart illustrating
how the first-use-initialization (FUI) software which executes in the
support hardware;
FIG. 8 is a detailed flow chart of flow chart element 88 in FIG. 7.2;
FIGS. 9.1 and 9.2 when viewed with 9.1 on the top and 9.2 on the bottom,
constitute a detailed flow chart of element 102 in FIG. 7.2;
FIG. 10 is a detailed flow chart of the elements 78 and 136 in FIGS. 7.1
and 7.2, respectively; and
FIG. 11 is a flow chart of the load-decrypt-run (LDR) program according to
the invention.
BEST MODE OF CARRYING OUT THE INVENTION
The name Personal Computer, or Single User System or Individual Work
Station given to a class of small computers is misleading. Unless the user
has also written all the programs including the operating system used in
the computer, these machines are better titled Single Operator Systems.
From the point of view of classical operating systems, this places the
user/operator in a position of trust in which he has access to all the
system code and system facilities. In common Single Operator Systems, this
exposure of the system code and the system hardware facilities is an
opportunity to replicate and distribute code, which on classical large
computing systems, with separate operators and users, would be
unavailable. The means by which this security is achieve on large systems
is the implementation of a system in which users are given a privilege
status level which is tested by the system to determine whether or not the
user may execute certain instructions or access certain data for reading
or modification.
It is the purpose of this invention to teach how a system utilizing
privilege may be implemented on Single Operator Systems to the advantage
of the hardware manufacturer, the software vendor, and the scrupulous
Single Operator System user. It is called a "Shared Higher Level of
Privilege System" because it can be viewed as providing each software
vendor with an instanciation of a higher privilege level than the user,
without giving any vendor access to other vendors' privileged information.
By using this system, a portion of the hardware and software of the system
is hidden in a co-processor subsystem (hereafter called the Support
Hardware) which is installed in the Single Operator System (hereafter
called the host) so that some portion of the vendor software can be made
inaccessible to the user. In addition a method for transporting software
on floppy disks or other magnetic medium is provided which allows the
software vendor to hide some fraction of the software from the user in
spite of the user being able to examine it with the resources available to
him on the system. This is not to say that if the user makes a sufficient
investment of time and money, by adding resources to his system beyond
those needed for computing (e.g., logic analyzers and digital recorders),
that he will still be unable to expose the code, rather that without such
things, the resources of the Single Operator System which are available to
the user are insufficient to obtain the code for piracy. Use of the
resources of the Single Operator System are the overwhelmingly common
means piracy, thus this system can dramatically reduce piracy in its
common form and make piracy unprofitable due to its cost in time and tools
in other cases.
For purposes of description, the host hardware is assumed to be an IBM
Personal Computer (PC) the operation of which is described in the IBM
Personal Computer Technical Reference Manual, 1981, and the host disk
operating system (DOS) is assumed to be IBM PC DOS. This is done for the
sake of clarity, and because the operations and DOS services of this
combination are typical of a class of machine in which this system is
useful. It should be understood that these DOS and hardware operations are
intended to be representative of all analogous operations on this or other
possible host systems under this or other operating systems.
A software copy protection system employing a shared higher level of
privilege is composed of two parts; the hardware privilege support system
installed in the work station, and the floppy disks or other magnetic
medium which are used to transport the software from the vendor to the
user. The floppy disk and the support hardware are modified by the first
attempt to use the software on the disk. These modifications make the fact
that the disk has been loaded detectable.
Disks which are used with this system are prepared by creating two kinds of
marks on the disk which are not producible by conventional disk drives but
which can be detected by them. These marks are in the form of the absence
of material which can have its magnetization changed by a disk drive write
head (the medium coating is either absent or replaced with higher
coercivity material) and in the form of domains which cannot be created by
conventional disk drive write heads (for instance a region in which the
orientation of the domain remains unchanged for distances large enough to
cause a loss of synchronization in the disk system).
For the remainder of this description, marks which are made by modifying
the medium will be referred to as MM marks, and marks which are made by
the creation of a domain pattern will be referred to as DP marks.
Both these marks have properties such as location and extent, which can be
used to encode information. The location and extent of the DP marks and
the location and extent of the MM marks can be detected by use of
appropriate procedures on conventional disk drives.
The location within a given sector of the DP type of mark may be found by
reading a sector twice, and comparing the results of the two read
operations. Since DP marked sectors do not contain the transitions needed
by the disk control system to keep its clock synchronized, the two read
operation will reliably show different data in the portion of the sector
which follows the beginning of the DP mark. The location of the DP mark
found by this method is approximately reproducible within limits set by
the hardware in the disk system.
The marks made by modifying the magnetic medium are detected by a sequence
of operations. First, a pattern of domains such that some domain
transitions coincide with the MM marks is written to the sector containing
the marks. The sector is then read, and the results of the read operation
are compared with the result expected given the write operation if the
sector had been unmarked. The location and extent of the marks can be
derived from the results of this comparison.
The MM marks can be "written" on the disk by laser photodecomposition
ablation as described in previously referenced patent application Ser. No.
06-567,294. DP marks may then be made over MM marks by moving a formatted
disk through a uniform magnetic field of width approximately equal to the
width of a sector, so that sectors with MM marks are swept by the magnetic
field. A band of large magnetic domains is thus created across the disk.
The disk is then reformatted on all the tracks except those containing MM
marks.
A disk treated this way would then contain both kinds of marks, with the DP
marks covering the MM marks.
It is important to note that the operations required to "read" the MM marks
are exactly the kind which will destroy the DP kind of marks. That is, the
act of writing a pattern onto a sector will create domains on the disk
which will support the synchronization in the disk system.
On a protected disk, the DP marks are made over a sector containing the MM
marks. This is done in order to insure that the DP mark will be destroyed
by reading the MM marks which lie "below" it. Any domain structure which
both cannot be made by a conventional drive and which can be measured in
some way by a read operations can be used for this purpose.
The location and extent of both kinds of mark are recorded in a file on the
disk. This information can be used in the preparation of other files on
the disk but it is always encrypted before the disk preparation is
complete.
In addition to the marks there are files stored on the disk. The files fall
into two categories: (1) the protected application software, and (2) the
information needed by the support hardware to load and run the encrypted
part of the application.
The application software must consist of at least one file of encrypted
program. This part of the application is encrypted with a key provided by
the software vendor. The decryption key for that file is itself encrypted
with an RSA public key provided by the manufacturer of the support
hardware. This encrypted decryption key (EDK) is also recorded in a file
on the disk. It is in the best interest of the software vendor to encrypt
those fractions of the application software which he considers
proprietary, as it is the encrypted fraction of the software which will be
protected from redistribution by the user. A complete, prepared disk which
is ready to be sold or released consists of at least:
1. A doubly marked region which could be unique to the disk.
2. An application which consists of at least one file of program encrypted
with a key selected by the software vendor which may include the disk
marking parameters in the key.
3. The decryption key in encrypted form where the encryption is by the RSA
public key provided by the support hardware manufacturer.
4. A program which calls for the services support hardware
(first-use-initialization and load-decrypt-run), and which obtains
services from the host for the support hardware at its request.
5. The descriptions of the doubly marked region(s) in an encrypted form
where the encryption key is the same key used with the application
software.
It should be noted here that the RSA private decryption key must be kept a
secret by the hardware manufacturer, and that the software vendor is
better protected if the encryption key used to protect his software
includes the marking information. The encryption key can be made unique to
each disk by this method. It should be clear that the degree of protection
offered by this system depends on the fraction of the total protection
system utilized by an implementer.
If piracy of a protected disk is attempted at this point, then the pirate
could be attempting to make copies of the disk on which the vendor has
supplied the protected software which will work without the support
hardware, or which can be transported to systems with the support
hardware. Each case will be discussed separately.
If the pirate wanted to make copies of the disk which will operate on
systems which contain the support hardware, then he must duplicate all the
features of the ready-for-distribution disk. Any conventional copying
program should be able to copy the encrypted application and the encrypted
decryption key in the encrypted non-executable form, but no conventional
disk drive could make the doubly marked region. No copy program running on
a personal computer has the hardware facilities needed to copy that part
of the distribution disk. As will be seen later, the support hardware
utilizes and changes the mark at the first-use-initialization. It will not
accept the transfer of the data need to run the protected program if at
the first use of a protected program it does not find a doubly marked
region whose marks correspond correctly to the descriptions of the marks
stored in an encrypted file. Piracy by copying the protected disk for use
on a support-hardware-equipped system is thus inhibited by the
difficulties of duplicating the doubly marked region(s).
If the pirate wishes to make copies of the disk which will run on systems
without support hardware, he must first decrypt those parts of the
application program which have been encrypted. Since there are two
processors in support hardware equipped system, it is possible that the
application may be written to operate concurrently on the two processors,
or use special facilities on the support hardware. If so, the application
must be drastically modified to be operational.
Piracy by copying a decrypted version of the application is thus inhibited
by the strength of the encryption method used to render the application
and the EDK unreadable. It is practical to make this a virtually
insurmountable task. Even if this were accomplished, the software could
still be useless unless it were rewritten to run without a coprocessor.
In order for the copy protection system to be useful, it is necessary not
only that no useful copy can be made of the distribution disk, but also
that:
1. The distribution disk be able to be used in one and only one system.
2. The user be able to make unlimited quantities of backup copies of the
disk, all of which are useless on other systems.
3. The software never reside in system memory in a form which allows the
user to make a binary image of the system memory with a complete working
version of the application which could be loaded in other systems.
This part of the protection system is implemented with the support
hardware. The support hardware is itself a computing system. It has its
own processor, firmware in read only memory (ROM), hardware timers, a real
time clock (or other hardware means for obtaining a "random" number), and
RAM. It could be installed in a personal computer as a card set. It
communicates with the system in which it resides through a region of
common memory, and through a set of registers which reside in the port
address space of the host system. It is important to note that the common
memory is part of the support hardware system. The support hardware
controls its bus transceivers and can cause this memory to be unavailable
to the host for read operations. Other configurations are possible, but
all require that only a portion of the support hardware memory be
addressable by the host system. It is also necessary that the portion of
the support hardware memory which is not addressable by the host system be
physically inaccessible to the user. It is this memory in which the
support hardware will decrypt and run the encrypted portion of the
application software.
In addition to the processor, common memory, hardware timers, and port
addressed registers, the support hardware has physically and logically
secure memory space which contains ROM and EEPROM memory devices.
The ROM devices contain the system firmware. It is in the form of a monitor
whose commands are the services which the host system may request from the
support hardware. A complete set of such services would include as a
minimum set:
1. Perform first-use initialization.
2. Load, decrypt, and run application.
The EEPROM device is used by the support hardware as a secure, nonvolatile
memory in which decryption keys of initialized applications are stored.
The point should be made that the processor in the support hardware must
have at least two levels of privilege itself so that the memory occupied
by the EEPROM and the ROM, and the clock and hardware timers, can be
properly secured from the user.
All applications software decrypted and run on the support hardware is at a
lower level of privilege than the ROM resident firmware which controls
EEPROM access, loading, decrypting, and running operations. This structure
is needed in the support hardware to prevent the user from writing a
monitor which would run on the support hardware which would access the
firmware and the EEPROM and dump the contents of these into common memory.
As was noted earlier, the support hardware must be physically as well as
logically secure. This security is required in order to prevent the user
from using logic analyzers or other digital control and recording devices
to gain a record of the content of the secure memory space. It is worth
noting that, given the present state of the art of semiconductor
technology, physical security for the support hardware could be obtained
by packaging the complete support system in a single chip package. This
package could be built so that any effort at physical access (to probe the
memory content to obtain a set of decryption keys and algorithms) would
destroy the information in the ROM and EEPROM. This could be accomplished
with a combination of piezo-electric drives (to destroy the MOS gates in
the memory devices if the package were stressed sufficiently or if stress
in the package were released) and conducting lines on the IC or package
which would oxidize rapidly if the package were opened in the air.
While a single IC package is the preferred packaging technique for the
support hardware, the system could be built by at least two other methods:
1. As a set of chips which communicate with each other over a proprietary
encrypted bus.
2. As a conventional chip set assembled on a board and encapsulated with a
tamper protection system.
The support hardware is an addition made to a "Host" Individual Work
Station. This work station is a single common bus microprocessor based
computer system. The IBM PC is typical of this class of machines. Such
systems use the bus (which can be an array of transmission lines with
sockets at intervals) as a communications medium between logically
separate subsystems. Some of these subsystems may reside on the same
packaging element (in this case a printed circuit board called the "System
Board") as supports the bus. Subsystems which are necessary to the
function of the system or for expansion of the function of the system are
added by attachment to the bus through the sockets. It should be noted
that the components which constitute a subsystem may be made so that parts
of the subsystem may reside on different packaging elements.
The complement of subsystems which are shown in FIG. 1, in the region
labeled "Conventional Computing System" as indicated at 2, is an example
of a common, nearly minimal host system. The host CPU 4 is a single chip
microprocessor and a group of support chips. The host CPU 4 is supplied
with a periodic signal called a clock and with connection to the bus by
the support chips. The microprocessor is commonly supplied with more
support than this, but all support is aimed at executing a repeating cycle
of fetches of instructions from memory, fetches of data from some selected
element of the system (such as Random Access Memory), execution of
instructions, and when needed, storage of resulting data in a selected
element of the system. The host CPU 4 may have support supplied to it
called direct memory access (DMA) which allows the microprocessor to be
relieved of tasks which involve the movement of data from one addressable
element to another.
The microprocessor controls the type of bus operation performed (fetch,
store, etc.) and the type of element selected (RAM, Port Addressed
Register, etc.) by which of the control lines in the bus is "asserted"
(changed to the appropriate potential according to a protocol called the
bus definition). By these means, the microprocessor is able to obtain a
collection of instructions (called a program), execute the instructions on
a set of data, and cause the data stored in other elements of the system
to change as a consequence of the execution of the instructions.
The RAM 6 is a subsystem from which data can be fetched from or written to
by the host CPU 4. It is the subsystem used to store data and instructions
which are loaded from some other source. If it has meaningful content,
then that content has been written to it by the host CPU 4. At the time
that the computer is powered on, the RAM 6 contents are, for practical
purposes, meaningless.
The ROM 8 is a subsystem from which data can only be fetched. It may
contain a collection of programs which are needed to start useful
operation of the computer and which are useful for controlling the
remaining subsystems.
The remaining subsystems, terminal control unit 9, display 11, input device
13, disk system control unit 15 and disk drive 17 can be characterized as
having both addressable elements and mechanical, optical, or
electromagnetic (or other) elements which can affect human se | | |
