Protecting secure software in a multi-security-CPU system

1. Cross Reference to Related Applications

This application claims the benefit of priority to <patcit id="pcit0001" dnum="US61684479A"><text>U.S. Provisional Application No. 61/684,479, filed August 17, 2012</text></patcit>, and <patcit id="pcit0002" dnum="US61729169A"><text>U.S. Provisional Application No. 61/729,169, filed November 21, 2012</text></patcit>, which are incorporated herein by reference.

2. Technical Field

This disclosure relates to system security performed by a security central processing unit (SCPU), and more particularly, to security functions performed in a system by multiple SCPUs operating at different security levels.

3. Background

Rapid advances in electronics and communications technologies, driven by consumer demand, have resulted in the widespread adoption of data-driven devices, including those for handling and converting third-party media content. Third-party customers or clients want their content handled with security such that the content is not copied or used outside of certain permission levels. Systems that stream content digitally for multiple media providers to consumers seek to include higher levels of security so that competitor providers cannot access each other's secrets. In a large system on a chip (SOC), a single security central processing unit (SCPU) may perform security functions.

BRIEF SUMMARY OF THE INVENTION

According to an aspect, a computing system comprises: <ul id="ul0001" list-style="none" compact="compact"><li>a first central processing unit (CPU); and</li><li>a second CPU coupled with the first CPU and with a host processor,</li><li>where, in response to a request by the host processor to boot the second CPU, the first CPU is configured to execute secure booting of the second CPU comprising decrypting encrypted code to generate decrypted code executable by the second CPU but that is inaccessible by the host processor.</li></ul>

Advantageously, the host processor writes the encrypted code into dynamic random access memory (DRAM) from flash memory, the encrypted code including a first stage code and a second stage code for executing a two-stage boot process.

Advantageously, the second CPU includes a private local memory, the first and second CPUs are located on a system on a chip (SOC), and where the host processor configures a first stage boot loader executable to: <ul id="ul0002" list-style="none" compact="compact"><li>read the first stage code from DRAM;</li><li>decrypt the first stage code on the SOC; and</li><li>write the decrypted first stage code into the private local memory.</li></ul>

Advantageously, the first CPU is further configured to: <ul id="ul0003" list-style="none" compact="compact"><li>authenticate the decrypted first stage code for execution by the second CPU; and</li><li>authorize the second CPU to execute the decrypted first stage code from the private local memory.</li></ul>

Advantageously, the decrypted first stage code is executable to make decisions regarding how to load, decrypt and execute the second stage code.

Advantageously, in executing the decrypted first stage code, the second CPU is configured to request the first CPU to generate a key usable to decrypt the second stage code, in response to which the first CPU is further configured to: <ul id="ul0004" list-style="none" compact="compact"><li>generate the key with specified access rights for use by the second CPU, the access rights including a first region of the DRAM from which to decrypt the second stage code and a second region of the DRAM to which to write the decrypted second stage code; and</li><li>aid the second CPU in creation of a memory checker configured to ensure decryption occurs only from the first region to the second region of the DRAM.</li></ul>

Advantageously, the host processor can access the first region of the DRAM but not the second region of the DRAM.

Advantageously, the second CPU is further configured to: <ul id="ul0005" list-style="none" compact="compact"><li>decrypt the second stage code stored in the first region of memory using the key; and</li><li>write the decrypted second stage code to the second region of the DRAM.</li></ul>

Advantageously, the computing system further comprises an instruction checker configured to: <ul id="ul0006" list-style="none" compact="compact"><li>monitor execution of instructions of the decrypted second stage code by the second CPU; and</li><li>prevent execution of an instruction the second CPU attempts to execute outside of the second region of the DRAM, or force execution of the instruction within the second region.</li></ul>

Advantageously, the instruction checker is further configured to reset the computing system in response to detecting an attempt to execute the instruction outside of the second region of the DRAM, and where the first CPU operates at a first security level and the second CPU operates at a second security level less secure than the first security level.

According to an aspect, a method is provided for securing software in a multi-security central processing unit (CPU) system that includes a first and a second CPU coupled together, comprising: <ul id="ul0007" list-style="none" compact="compact"><li>receiving a request from a host processor to boot the second CPU;</li><li>writing, by the host processor, encrypted code into a dynamic random access memory (DRAM) from flash memory, the encrypted code including a first stage code and a second stage code for executing a two-stage boot process; and</li><li>decrypting the encrypted first and second stage codes by the first and second CPUs such that the host processor cannot access either the decrypted first stage code or the decrypted second stage code after decryption.</li></ul>

Advantageously, the method further comprises: <ul id="ul0008" list-style="none" compact="compact"><li>reading, by a first stage boot loader of the host processor, the first stage code from the DRAM;</li><li>decrypting the first stage code;</li><li>writing the decrypted first stage code into a private local memory of the second CPU;</li><li>authenticating, by the first CPU, the decrypted first stage code for execution by the second CPU; and</li><li>authorizing, by the first CPU, the second CPU to execute the decrypted first stage code from the private local memory.</li></ul>

Advantageously, the method further comprises: <ul id="ul0009" list-style="none" compact="compact"><li>requesting, by the second CPU, that the first CPU generate a key;</li><li>generating the key by the first CPU, the key including access rights that include a first region of the DRAM from which to decrypt the second stage code and a second region of the DRAM to which to write the decrypted second stage code; and</li><li>creating, by the first and second CPUs, a memory checker.</li></ul>

Advantageously, the method further comprises the memory checker: <ul id="ul0010" list-style="none" compact="compact"><li>detecting decryption of the encrypted second stage code attempted by the host processor;</li><li>determining whether decryption is from the first region to the second region; and</li><li>in response to determining that decryption is attempted from other than the first region or to other than the second region, alerting the second CPU of a decryption violation.</li></ul>

Advantageously, the method further comprises the second CPU: <ul id="ul0011" list-style="none" compact="compact"><li>preventing the decryption of the encrypted second stage code or forcing the decrypted second stage code to be written to the second region.</li></ul>

Advantageously, the method further comprises the second CPU: <ul id="ul0012" list-style="none" compact="compact"><li>decrypting the second stage code stored in the first region of memory using the key;</li><li>writing the decrypted second stage code to the second region of the DRAM;</li><li>executing the decrypted first second stage code; and</li><li>alerting the second CPU, by an instruction checker, when the host processor attempts to execute an instruction outside of the second region.</li></ul>

According to an aspect, a computing system comprises: <ul id="ul0013" list-style="none" compact="compact"><li>a first central processing unit (CPU);</li><li>a dynamic random access memory (DRAM) connected with the first CPU;</li><li>a flash memory;</li><li>a second CPU coupled with the first CPU and with the DRAM;</li><li>a host processor coupled with the first and second CPUs, with the DRAM and with the flash memory, the host processor configured to write encrypted code into the DRAM from the flash memory, the encrypted code including a first stage code and a second stage code for executing a two-stage boot process;</li><li>where the first CPU is configured to: <ul id="ul0014" list-style="none" compact="compact"><li>receive a request form the host processor to boot the second CPU; and</li><li>decrypt the encrypted first and second stage codes by the first and second CPUs such that the host processor cannot access either the decrypted first stage code or the decrypted second stage code after decryption.</li></ul></li></ul>

Advantageously, the computing system further comprises a first stage boot loader executable to read the first stage code from DRAM, where the second CPU comprises a private local memory, and where the first CPU is further configured to: <ul id="ul0015" list-style="none" compact="compact"><li>decrypt the first stage code;</li><li>write the decrypted first stage code into the private local memory;</li><li>authenticate the decrypted first stage code for execution by the second CPU; and</li><li>authorize the second CPU to execute the decrypted first stage code from the private local memory.</li></ul>

Advantageously, the second CPU is further configured to request the first CPU generate a key, in response to which the fist CPU is further configured to generate the key including access rights comprising a first region of the DRAM from which to decrypt the second stage code and a second region of the DRAM to which to write the decrypted second stage code.

Advantageously, the first and second CPUs are further configured to create a memory checker configured to: <ul id="ul0016" list-style="none" compact="compact"><li>detect decryption of the encrypted second stage code attempted by the host processor;</li><li>determine whether decryption is from the first region to the second region; and response to determining that decryption is attempted from other than the first region or to other than the second region, preventing the decryption of the encrypted second stage code or forcing the decrypted second stage code to be written to the second region.</li></ul>

BRIEF DESCRIPTION OF THE DRAWINGS

The systems and methods may be better understood with reference to the following drawings and description. In the figures, like reference numerals designate corresponding parts throughout the different views.

<figref idref="f0001">Figure 1</figref> is a block diagram of an exemplary multi-security central processing unit (CPU) system on a chip.

<figref idref="f0002">Figure 2</figref> is a flow chart of an exemplary method for implementing the multi-security CPU system on a chip of <figref idref="f0001">Figure 1</figref>.

<figref idref="f0003">Figure 3</figref> is a block diagram of an exemplary first of multi-stages used to boot a security CPU-B by a security CPU-A.

<figref idref="f0004">Figure 4</figref> is a block diagram of a second of two stages used to boot the security CPU-B by the security CPU-A.

<figref idref="f0004">Figure 5</figref> is a block diagram of exemplary running of protected second stage code from memory after security CPU-B has been booted.

<figref idref="f0005">Figures 6</figref> and <figref idref="f0006">7</figref> are a flow chart of an example method for protecting software in a multi-security-CPU system by securely booting security CPU-B by security CPU-A, where <figref idref="f0005">Figure 6</figref> shows the first stage and <figref idref="f0006">Figure 7</figref> shows the second stage of a multi-stage boot process.

<figref idref="f0007">Figure 8</figref> is an exemplary block diagram of the exemplary system of <figref idref="f0001">Figure 1</figref> in which key separation between the security CPU-B and the host processor protects sensitive files from host access or decryption.

<figref idref="f0008">Figure 9</figref> is a flow chart of an exemplary method for executing key separation between the security CPU-B and the host processor that protects sensitive files from host access or decryption.

DETAILED DESCRIPTION

The discussion below makes reference to system security performed by a security central processing unit (SCPU), and more particularly, to security functions performed in a system by multiple SCPUs operating at different security levels. For purposes of explanation, two SCPUs are described, but more SCPUs can be implemented. The SCPUs may be instantiated on a system on a chip (SOC) such as may be used in a set top box (STB) for streaming media to consumers. Such media may include audio or audio/video content viewable by consumers on media devices.

In a large system on a chip (SOC) with a single security central processing unit (SCPU) that performs security functions, the system trust level may be binary: either the operation is highly secure, or it is completely untrusted. For operations that require an intermediate level of security, there exist two choices: (1) perform these operations in the SCPU; or (2) perform them in the host.

The first option may not be ideal because the SCPU may be responsible for highly sensitive tasks, like managing one time password (OTP) authentication, routing customer secrets, and the like. Mixing these highly sensitive tasks with lower security functions may create risk and detract from the main mission of the SCPU. The second option may not be ideal because performing intermediate security tasks in the host may leave the SOC unsecured, as the host CPU may not be trusted.

Executing security functions by a host on the chip may open up security too wide to provide sufficient security to chip operation. And, mixing highly secure and less secure functions as executed by a single SCPU may open system security up to some level of risk and detract from the main mission of a SCPU to secure the most sensitive functions of chip operation.

Furthermore, because the SCPU manages proprietary information to the SOC vendor, allowing the SCPU to be programmed by an end user customer may be problematic. Yet some of the intermediate level security tasks are better performed by end user code, and thus the SOC vendor may want to allow the customer to program aspects of SCPU functionality. To allow the user to program functions by the only SCPU on the chip may open up the security operation of the chip to additional risk and attack.

In an SOC, having a single SCPU may no longer be sufficient. Using a multi-security CPU approach, such as with a first SCPU dedicated for highly secure functions and a second SCPU used for lower security tasks, like digital rights management (DRM), managing transcoding, watermarking, etc. For purposes of explanation, the first SCPU is labeled security CPU-A herein, and the second security level SCPU is labeled security CPU-B.

<figref idref="f0001">Figure 1</figref> is an example diagram of a multi-security central processing unit (SCPU) system on a chip (SOC) 100. The SOC 100 may be integrated within or coupled with a media device 10 when deployed into operation. The system 100 may include first SCPU 102 also referred to as security CPU-A, second SCPU 104 also referred to as CPU-B, and a host processor 110, the general-purpose processor that runs the majority of the chip operation. The security CPU-A may be smaller than the security CPU-B and be configured to operate at a first security level higher than a second security level at which security CPU-B operates. The security CPU-B may be provided a higher trust level than the host processor 110 and the host processor may be denied execution of the second security level functions. The host processor 110 may also, at least in part, be located outside of the SOC 100.

Functions of security CPU-A at the first security level may include, for example, managing root keys, performing an initial secure boot, and routing secrets of a third-party content provider. Functions of security CPU-B at the second security level include, for example, digital rights management, license management, transcoder management, watermarking, and manipulation of data in secure memory. The security CPU-A may be configured with software code that treats commands from the security CPU-B as privileged, and that generates multiple unique commands for execution by the security CPU-B that are not executable by the host processor 110.

As configured to perform the majority of the processor-intensive security functions, the security CPU-B may be nearly as powerful as the host processor 110 in some implementations, and may, for instance, at the time of filing, perform up to 1,000 or 1,500 or more Dhrystone million instructions per second (DMIPS). As such, the security CPU-B may be focused on lower security functions. Security CPU-A may need a fraction of the power and operate at less than 1,000 DMIPS.

The system 100 may further include on-chip sensors 113, memory such as dynamic random access memory (DRAM) 115, a local checker 118 coupled with the security CPU-A and multiple peripherals 120 of the security CPU-B. "Coupled with" herein can mean to be directly connected to a component or indirectly connected through one or more components. The DRAM 115 may include a portion of which is protected code 116 stored in secure memory 117. The secure memory 117 may be partitioned into specific or identified regions of the DRAM.

The security CPU-A may include a host interface 122 to communicate with the host processor 110 and an SCPU-B interface 124 with which to communicate with the security CPU-B. The security CPU-B may include a CPU-B 130, a local static random access memory (SRAM) 132, an instruction cache (i-cache) 134, a data cache (d-cache) 136, an instruction checker 138 and an interrupt controller 140. The local SRAM 132 may be private local memory for access by the security CPU-B in which instructions may be saved and data temporarily stored that is not accessible to the host processor 110 or to other on-chip entities or clients.

The security CPU-B and the security CPU-A may be coupled together with a dedicated, secure communications bus 142 that operates as a private channel between the CPU-B 130 and the security CPU-A. The secure communications bus 142 may be inaccessible by the host processor and inaccessible by third-party clients that exist on the SOC 100. The secure communications bus 142 may be configured with a combination of hardware and firmware and perform a master slave relationship in which, in some operations, the security CPU-A is either the master of or slave to the security CPU-B. For instance, the security CPU-A may be the master of booting CPU-B securely in memory. But, the security CPU-A may also receive commands from security CPU-B or the local checker 118, for instance.

A third party that exists on the SOC 100 may have its own CPU, its own block of logic, or exist on the chip as a combination of hardware and software having access to capabilities of the SOC 100. The third party CPU may include a secure interface managed by the CPU-A.

The system 100 may further include a common register bus 144 to which third-party clients that exist on the SOC 100 have access. The common register bus 144 may be used to write into registers of the memory 115. The security CPU-A may be configured to prevent certain on-chip clients from stopping security CPU-B from operating as intended, as disclosed herein.

The local checker 118, which may be coupled with the security CPU-A and security CPU-B, may be a piece of hardware configured to prevent some clients or hardware existing on the SOC 100 from accessing certain regions of the DRAM. Likewise, the local checker 118 may prevent blocking access by the security CPU-B to the common register bus 144 and/or to reading from and writing to the DRAM of the SOC 100.

The security CPU-A may also program the local checker 118 to ensure internal peripherals 120 of the security CPU-B are inaccessible by third-party entities having access to the host processor 110. The peripherals may include, but not be limited to, a Universal Asynchronous Receiver/Transmitter (UART), a timer, an interrupt, memory, data storage, a media device, or a combination thereof.

The instructions checker 138 may monitor instructions to be executed out of the DRAM and determine whether instructions called for execution from the secure memory by a component are cleared for execution by that component. To clear instructions for execution, the instruction checker 138 may ensure that the security CPU-B does not operate outside of a region of the secure memory 117 that has been authenticated for secure operation or conditional access by the host processor 110. For example, the instruction checker may monitor reads and writes to the DRAM 115 and compare the DRAM addresses of those memory accesses with an address range set up by the security CPU-A as a pre-authenticated region for execution of the instructions. If the security CPU-B attempts to execute instructions outside of the region of memory, the security CPU-B may be rebooted or the entire SOC be reset.

In one example, the content saved to the secure memory 117 may include media content that a client does not want distributed in an unauthorized manner. The security CPU-B may decrypt the content for viewing on a consumer device, but not allow other peripherals to access or distribute the content outside of the system 100. The security CPU-B may ensure a consumer can view the content but not be directly accessed by the host. The security CPU-A and CPU-B may set up hardware that limits certain chip components in their access to the secure memory 117. For example, the security CPU-A and security CPU-B may make a region of memory off limits to the host processor. Furthermore, the security CPU-B may perform watermarking or manipulate time stamps with regards to content that a consumer may view.

More particularly, the secure memory 117 may be accessible exclusively to the security CPU-A and CPU-B and to local decompression and rendering engines. Thus, the security CPU-B may decrypt content into this memory region and then local display processes may read the decrypted content for local rendering. None of these steps requires the host processor 110 to have access to the secure memory. The security CPU-B may ensure a secure data flow by decrypting the content into this restricted region, out of reach of the host processor.

The interrupt controller 140 may be coupled with the security CPU-A and be configured to monitor the on-chip conditions from data generated by the sensors 113. The on-chip sensors 113 may generate data related to attributes of the chip such as, for example, temperature, voltage levels at specific points on the chip, clock speed and the like. If one of these attributes varies too much or in the wrong way, it may be indicative of a potential intrusion or hacker trying to disrupt normal operation of the SOC that would allow access to secure data and/or secure operation. The interrupt controller 140 may aggregate and mask interrupts from other functional blocks of the SOC 100 that may include interrogation of the sensors to detect predetermined threshold values of those sensors for use in determining whether or not the interrupt controller 140 masks an interrupt.

Furthermore, the interrupt controller 140 may generate an interrupt or a hookup in response to detecting conditions indicative of an intrusion. The interrupt or hookup may adjust operation of the host processor 110 or the security CPU-B in real time to ensure secure system operation. The host processor may also have a separate memory buffer than the memory buffer the security CPU-B uses, where the memory buffer for the security CPU-B may be configured as not accessible by the host processor 110 and configured to provide control logic to the second security CPU-B.

The system 100 may communicate over a network 15 such as the Internet or any wide or local area network to communicate with a Web server 20 and media clients 30. The clients 30 may be the customers that procure the SOC 100 for use by consumers to stream media content to consumer media devices 10. The security CPU-B may obtain the time and day over the Internet from the Web server 20 using a secure protocol. The time and day may then be considered secure time and be stored by the security CPU-B in either the local SRAM 132 or the secure memory 117 in DRAM 115. So stored, the security CPU-B may prevent access by the host processor or other on-chip programmed components to the secure time, which may be used by digital rights management and other forms of lower level security functions during execution of the secure functions.

<figref idref="f0002">Figure 2</figref> is an example flow chart of a method for implementing the multi-security CPU system of <figref idref="f0001">Figure 1</figref>. A system on a chip (SOC) may include a first security central processing unit (SCPU) and a second SCPU (202, 210). The SOC may operate the first SCPU at a first security level (206). The SOC may operate the second SCPU at a second security level less secure than the first security level, but higher than a level at which a host processor operates (214). The first SCPU may create commands only executable by the second SCPU (218).

The second SCPU may determine whether an instruction called for execution from secure memory is cleared for execution by a component of the SOC calling the instruction (222). If the component is not cleared to execute the instruction, neither the first nor the second SCPU are permitted to execute the instruction (226). If the component is cleared to execute the instruction, the second SCPU may determine whether the instruction executes a first or the second security level function (230). If the component requests to execute a first security function, the first SCPU executes the instruction (234). If the component requests to execute a second security function, the second SCPU executes the instruction (238). The first SCPU may also be the requestor component of either the first or second security functions and the second SCPU may be the requestor component of either first or second security functions, depending on the function.

<figref idref="f0003">Figures 3</figref> and <figref idref="f0004">4</figref> are block diagrams of an exemplary multi-stage boot process by which the security CPU-A securely boots the security CPU-A. <figref idref="f0003">Figure 3</figref> shows execution of the first stage and <figref idref="f0004">Figure 4</figref> shows execution of the second stage. These stages may ensure that decrypted (or clear) code in the DRAM that is executed is not visible, and thus not accessible, to other than the SCPUs 102 and 104. To do so, the security CPU-A securely boots the security CPU-B by decrypting encrypted code to generate decrypted code in the DRAM 115 executable by the security CPU-B but that is inaccessible by the host processor.

More particularly, an encrypted first stage code 317 and an encrypted second stage code 319 are saved in flash memory 315. The host processor 110 may attempt to boot security CPU-A, and thus access the first and second stage encrypted codes. In response, the host processor 110 may be allowed to write the first and second stage codes 317 and 319 into the DRAM 115 from flash memory 315, but the first and second stage codes remain encrypted. The host processor 110 may then configure a first stage boot loader (FSBL) 320 that includes a DRAM interface 324. The DRAM interface reads the first stage code 317 from memory and the FSBL 320 decrypts the first stage code 317 and writes the decrypted code 327 into the private local SRAM 132 of the security CPU-B.

The security CPU-A may authenticate the decrypted code 328, either in place in the local SRAM 132 or on the fly as the decrypted code 327 is being written into the local SRAM 132. The security CPU-A may be further configured to authorize the security CPU-B to execute the decrypted code with the CPU-B 130 once authenticated. The decrypted first stage code 327 may be executed to make decisions with regards to how to load, decrypt and execute the second stage code.

With further reference to <figref idref="f0004">Figure 4</figref>, in executing the decrypted first stage code 327 as a trusted kernel within the local SRAM 132, the security CPU-B may be configured to request the security CPU-A to generate a key usable to decrypt the second stage code. In response to the request, the security CPU-A may generate the key with specified access rights for use by the security CPU-B. The access rights may include a first region 321 of the DRAM 115 from which to decrypt the second stage code and a second region 421 of the DRAM to which to write the decrypted second stage code 419. For instance, the first region may be the region 321 where the second stage encrypted code 319 is stored, and the second region 421 may be where the decrypted second stage code 419 is designated for storage in <figref idref="f0004">Figure 4</figref> by the access rights created in the generated key.

The security CPU-B may work with the high-security CPU-A to create a memory checker 428, a special piece of hardware. The memory checker 428 may be configured to ensure decryption of the second-stage encrypted code 319 occurs only from the first region 321 to the second region 421 of memory in the DRAM 115. If the host processor attempts to write the decrypted second stage code 419 into memory outside of the second region 421, the memory checker 428 may alert the security CPU-B so that the code is not written outside of the second region or may force the decrypted second stage code to be written to the second region.

The security CPU-B may then decrypt the second stage code 319 using the key generated by the security CPU-A and write the decrypted second stage code 419 into the DRAM 115 at the second region 421. While the host processor 110 may access the first region it cannot access the second region. The security CPU-B can, however, access the second region to execute the decrypted second stage code. Accordingly, the second stage of the boot process ensures that when the decrypted second stage code 419 is saved to memory, the decrypted second stage code is executable upon request by the host processor, but without allowing the host processor to access the decrypted second stage code. As third-party clients of the SOC 100 have access to the host processor, the two-stage boot process ensures that third-party entities are not able to snoop on the decrypted second stage code 419 through pin snooping or the like.

<figref idref="f0004">Figure 5</figref> is an example diagram of running the protected second stage code 419 from memory after the security CPU-B has been booted. The instruction checker 138 referred to earlier may also be configured to ensure execution of the decrypted second stage code 419 does not spread outside of the second region 421 of memory. The instruction checker 138 may monitor instructions of the decrypted second stage code 419 being called for execution by the host processor 110, and ensure that the execution remains within the second region 421 of secure memory of the DRAM 115.

When the host processor 110 attempts to execute an instruction from the decrypted second stage code outside of the second region 421 of the DRAM 115, the instruction checker 138 may block the execution of the instruction, or reset the SOC 100, causing a reboot of the security CPU-B. In the alternative, or in addition, the instructions checker 138 may alert the security CPU-B of the attempt so that the security CPU-B may perform the block or force execution of the instruction to the second region 421 of DRAM.

<figref idref="f0005">Figures 6</figref> and <figref idref="f0006">7</figref> are a flow chart of an example method for protecting software in a multi-security-CPU system by securely booting security CPU-B by security CPU-A, where <figref idref="f0005">Figure 6</figref> shows a first stage and <figref idref="f0006">Figure 7</figref> shows a second stage of a two-stage boot process. The host processor begins the process when the SOC 100 boots up by requesting the security CPU-B to also boot up (602). The host processor writes encrypted SCPU code into the DRAM from flash memory (604), the encrypted SCPU code including encrypted first stage and second stage code.

The first stage boot loader (FSBL) reads the first stage code from DRAM (608). The FSBL decrypts the first stage code (612). The FSBL writes the decrypted first stage code to private local SRAM of the security CPU-B (616). The security CPU-A then authenticates the decrypted code, whether on the fly as it is decrypted or after being stored in local SRAM (620). When the decrypted code is not authenticated, the security CPU-A does not authorize the security CPU-B from executing the decrypted first stage code (624). When the decrypted code is authenticated, the security CPU-A authorizes the security CPU-B to execute the decrypted first stage code (628). The security CPU-B then executes the decrypted first stage code, when authorized to do so (630).

With reference to <figref idref="f0006">Figure 7</figref>, while executing the decrypted first stage code, the security CPU-B requests the security CPU-A to generate a decryption key (702). The security CPU-A generates the key including specific access rights affiliated with the key (704). Those access rights may include a first region (Region 1) of DRAM from which the encrypted second stage code may be decrypted and a second region (Region 2) of DRAM to which the decrypted second stage code may be written and from which may be executed. The security CPU-B may work with security CPU-A to create a memory checker, a special piece of hardware that ensure decryption goes from Region 1 to Region 2 when the encrypted second stage code is decrypted (708).

When the host attempts decryption of the second stage code, the memory checker determines whether decryption is from Region 1 (716). When decryption is not from Region 1, the memory checker and the security CPU-B do not allow the decryption to take place (720). When decryption is from Region 1, the security CPU-B decrypts the encrypted second stage code with the key (724).

The memory checker then determines whether a request is received to write the decrypted code to Region 2 (728). When the memory checker determines the host processor attempts to write the decrypted second stage code to other than Region 2, the memory checker or the security CPU-B prevents the decrypted second stage code from be written outside of Region 2 of DRAM (732). When the memory checker determines the host processor requests to write the decrypted second stage code to Region 2, the security CPU-B writes the decrypted second stage code to Region 2 of the DRAM (734).

The security CPU-B may then execute the decrypted second stage code on behalf of the host processor (738). An instruction checker may determine whether the executed instructions are being executed within Region 2 of DRAM, one instruction at a time (742). When an instruction is to be executed outside of Region 2, the instruction checker may alert the security CPU-B so that the security CPU-B does not execute the instruction as requested or may force the instruction to be executed within Region 2 (746). When an instruction is to be executed inside of Region 2, the instruction checker may inform the security CPU-B to continue with executing instructions (750).

<figref idref="f0007">Figure 8</figref> is an example diagram of the system of <figref idref="f0001">Figure 1</figref> in which key separation between the security CPU-B 104 and the host processor 110 protects sensitive files from host access. Both the host processor 110 and the security CPU-B may request the security CPU-A to generate decryption keys for access to or decryption of the sensitive files, for use by software that would not otherwise be given access or the ability to decrypt the files.

These sensitive files may include those related to digital rights management (DRM) of propriety media such as audio or audio/video files or to other sensitive files. For example, in a DRM system, files may be protected from unsecure software by using special keys that are not known to the software. However, although the keys are not known, the commands used to decrypt the files are available to the host. This allows unsecure software to later send the command to decrypt the data in memory. Furthermore, it may be desirable to transmit commands from the security CPU-B to the security CPU-A to create a completely different key structure than the same commands issued from the host processor 110, to secure security CPU-A. Issuing separate commands may be done, as will be explained, in a way that the host cannot emulate the same keys.

Because the security CPU-B operates at a higher security level than the host processor, <figref idref="f0007">Figure 8</figref> shows how the system 100 may provide unique keys to the security CPU-B for exclusive access to files not provided to the host processor. The key generation described below may also be applied to root key generation, so that the security CPU-B may request the security CPU-A to generate a unique root key to decrypt a system or software file for execution by the system 100.

In one example, the security CPU-A includes a variant module 800, a key generator 805 and a filter 115. The key generator 805 generates keys 820 such as the key 820 displayed in <figref idref="f0007">Figure 8</figref>. The key includes access rights 824, which include a designation of access to a specific, secure region of memory, such as Region 1 or Region 2 in the DRAM 115. Such designation of access may also include the right to write decrypted data to Region 1 or Region 2. A key table 828, which may also be stored in the DRAM 115 or other memory, provides links between keys 820 and corresponding access rights accessible by the security CPU-A. Once the security CPU-A generates a key and provides access to the security CPU-B or the host processor, then the security CPU-B (or the host processor) may access the region of memory containing the decrypted content. In the case of generation of a root key, the security CPU-B (or the host processor) may be able to decrypt a system or software file and write the decrypted file to the protected region of memory from where it may be executed.

In some cases, the security CPU-B needs a unique key generated with exclusive access that the host processor is not provided. As previously discussed, the security CPU-B may communicate with the security CPU-A through the secure, dedicate communications bus 142. This bus 142 may be physically and/or logically separate from a bus 840 through which the host processor communicates with the security CPU-A. The security CPU-B may issue a special command or request of the security CPU-A over the dedicated bus 142 to the variant module 800 for generation of a unique key. The variant module 800 may make multiple possible changes or variants available for selection by the security CPU-B. Upon selection of a change or variant, the key generator 805 may then apply the selected variant(s) to an algorithm used to generate the key. The key generator may apply the variant as done in a key ladder or hash algorithm so that if any step in the decryption is not carried out, then access is denied.

The filter 810 may authenticate access to the variant module 800 by the security CPU-B and deny access to the variant module 800 by the host processor 110. The filter 810 may also distinguish between commands or requests issued by the host processor 110 and the security CPU-B, e.g., that the commands/requests are respectively correctly indicated as coming from the host processor 100 and the security CPU-B for purposes of key generation and access to the secured regions of memory. With authenticated access using a request with which the security CPU-A generates a unique key, the security CPU-B may obtain exclusive access to sensitive files, including DRM-protected media content. In some cases, access may also be provided to a different region of memory for writing the content elsewhere in the DRAM 115. Where the key generated is a root key, the security CPU-B may gain exclusive decryption rights of system or software files for execution by the system 100 from protected regions in DRAM 115, regions inaccessible by the host processor 110.

<figref idref="f0008">Figure 9</figref> is an example flow chart of a method for executing key separation between the security CPU-B and the host processor 110 that protects sensitive files from host access or decryption. The host processor and the security CPU-B both may request the security CPU-A to generate keys for access to and decryption of sensitive files as discussed with reference to <figref idref="f0007">Figure 8</figref> (902). The security CPU-B may request the security CPU-A to generate a unique key, to which the host processor does not have access (906).

The security CPU-A receives a request from the host or the security CPU-B to generate a key. The filter of the security CPU-A may determine whether the request is from the host or the security CPU-B (910). If the request is from the host, the security CPU-A may generate a key as requested for use by the host processor (914). The security CPU-A provides access or decryption rights to the host processor corresponding to the key (918).

If the request is from the security CPU-B, the security CPU-A may also receive selection of an algorithm variant through the variant module (922). The key generator of the security CPU-A may generate a unique key with use of the algorithm variant (930). The security CPU-A may provide access or decryption rights to the security CPU-B corresponding to the unique key exclusive of the host processor (934).

The methods, devices, and logic described above may be implemented in many different ways in many different combinations of hardware, software or both hardware and software. For example, all or parts of the system may include circuitry in a controller, a microprocessor, or an application specific integrated circuit (ASIC), or may be implemented with discrete logic or components, or a combination of other types of analog or digital circuitry, combined on a single integrated circuit or distributed among multiple integrated circuits.

While various embodiments of the invention have been described, it will be apparent to those of ordinary skill in the art that many more embodiments and implementations are possible within the scope of the invention. Accordingly, the invention is not to be restricted except in light of the attached claims and their equivalents.