Microprocessor with private microcode RAM
A microprocessor includes a private RAM (PRAM), for use by microcode, which is non-user-accessible and within its own distinct address space from the system memory address space. The PRAM is denser and slower than user-accessible registers of the microprocessor macroarchitecture, thereby enabling it to provide significantly more storage for microcode. The microinstruction set includes a microinstruction for loading data from the PRAM into the user-accessible registers, and a microinstruction for storing data from user-accessible registers to the PRAM. The microcode may also use the two microinstructions to load/store between the PRAM and non-user-accessible registers of the microarchitecture. Examples of PRAM uses include: computational temporary storage area; storage of x86 VMX VMCS in response to VMREAD and VMWRITE macroinstructions; instantiation of non-user-accessible storage, such as the x86 SMBASE register; and instantiation of x86 MSRs that tolerate the additional access latency of the PRAM, such as the IA32_SYSENTER_CS MSR.
1. A microprocessor, comprising:
a user-accessible register set;
a random access memory (RAM), within a non-user-accessible address space of the microprocessor, wherein said RAM is denser than said user-accessible register set and is slower to access than said user-accessible register set; and
microcode, comprising a plurality of microinstructions of a microinstruction set of the microprocessor, said microinstruction set comprising:
a first microinstruction, executable for storing data from said user-accessible register set to said RAM; and
a second microinstruction, executable for loading data from said RAM to said user-accessible register set.
2. The microprocessor as recited in claim 1 , wherein said user-accessible register set comprises an x86 architecture user-accessible register set.
3. The microprocessor as recited in claim 1 , further comprising:
a non-user-accessible register set, wherein said non-user-accessible register set is less dense than said RAM and is faster to access than said RAM;
wherein said first microinstruction is further executable for storing data from said non-user-accessible register set to said RAM, and said second microinstruction is further executable for loading data from said RAM to said non-user-accessible register set.
4. The microprocessor as recited in claim 3 , wherein said microcode is configured to execute one or more of said first microinstruction to store a portion of an x86 architecture virtual-machine extensions (VMX) virtual machine control structure (VMCS) to said RAM in response to a VMX VMWRITE instruction.
5. The microprocessor as recited in claim 3 , wherein said microcode is configured to execute one or more of said second microinstruction to load a portion of an x86 architecture virtual-machine extensions (VMX) virtual machine control structure (VMCS) from said RAM in response to a VMX VMREAD instruction.
6. The microprocessor as recited in claim 3 , wherein said microcode is configured to execute one or more of said first microinstruction to store a system management mode base address (SMBASE) to said RAM in response to an x86 architecture RSM instruction.
7. The microprocessor as recited in claim 3 , wherein said microcode is configured to execute one or more of said second microinstruction to load a system management mode base address (SMBASE) from said RAM in response to an x86 architecture system management interrupt (SMI).
8. The microprocessor as recited in claim 1 , wherein said microcode is configured to execute one or more of said first microinstruction to store a segment selector for a privilege level 0 code segment value to said RAM in response to an x86 architecture WRMSR instruction specifying an x86 architecture IA32_SYSENTER_CS Model Specific Register (MSR) as a destination MSR.
9. The microprocessor as recited in claim 1 , wherein said microcode is configured to execute one or more of said second microinstruction to load a segment selector for a privilege level 0 code segment value from said RAM in response to an x86 architecture RDMSR instruction specifying an x86 architecture IA32_SYSENTER_CS Model Specific Register (MSR) as a source MSR.
10. The microprocessor as recited in claim 1 , wherein said microcode is configured to execute said first and second microinstructions to use said RAM as a temporary storage area for computations performed by said microcode.
11. The microprocessor as recited in claim 1 , wherein said non-user-accessible address space is a non-translated address space.
12. The microprocessor as recited in claim 1 , wherein said non-user-accessible address space is a non-cached address space.
13. The microprocessor as recited in claim 1 , wherein the microprocessor generates no accesses on a bus external to the microprocessor in response to accesses by said microcode instructions to said non-user-accessible address space.
14. The microprocessor as recited in claim 1 , wherein said RAM is integrated onto a same semiconductor substrate as other circuits of the microprocessor.
15. The microprocessor as recited in claim 1 , wherein said RAM requires at least twice as much time to access as said user-accessible register set.
16. The microprocessor as recited in claim 1 , wherein said RAM is at least twice as dense as said user-accessible register set.
17. A method for efficiently storing data within a microprocessor, the method comprising:
executing a first microcode instruction, by storing data from a user-accessible register set of the microprocessor to a random access memory (RAM) of the microprocessor, wherein said RAM is within a non-user-accessible address space of the microprocessor, wherein said RAM is denser than said user-accessible register set and is slower to access than said user-accessible register set; and
executing a second microcode instruction, by loading data from said RAM to said user-accessible register set, wherein said first and second microinstructions are two of a plurality of microinstructions of a microinstruction set of the microprocessor.
18. The method as recited in claim 17 , wherein said user-accessible register set comprises an x86 architecture user-accessible register set.
19. The method as recited in claim 17 , further comprising:
executing said first microinstruction, by storing data from a non-user-accessible register set to said RAM, wherein said non-user-accessible register set is less dense than said RAM and is faster to access than said RAM; and
executing said second microinstruction, by loading data from said RAM to said non-user-accessible register set.
20. The method as recited in claim 19 , further comprising:
decoding an x86 architecture virtual-machine extensions (VMX) VMWRITE instruction; and
executing one or more of said first microinstruction to store a portion of a VMX virtual machine control structure (VMCS) to said RAM, in response to said decoding.
21. The method as recited in claim 19 , further comprising:
decoding an x86 architecture virtual-machine extensions (VMX) VMX VMREAD instruction; and
executing one or more of said second microinstruction to load a portion of a VMX virtual machine control structure (VMCS) from said RAM, in response to said decoding.
22. The method as recited in claim 19 , further comprising:
decoding an x86 architecture RSM instruction; and
executing one or more of said first microinstruction to store a system management mode base address (SMBASE) to said RAM, in response to said decoding.
23. The method as recited in claim 19 , further comprising:
receiving an x86 architecture system management interrupt (SMI); and
executing one or more of said second microinstruction to load a system management mode base address (SMBASE) from said RAM, in response to said receiving.
24. The method as recited in claim 17 , further comprising:
decoding an x86 architecture WRMSR instruction specifying an x86 architecture IA32_SYSENTER_CS Model Specific Register (MSR) as a destination MSR; and
executing one or more of said first microinstruction to store a segment selector for a privilege level 0 code segment value to said RAM, in response to said decoding.
25. The method as recited in claim 17 , further comprising:
decoding an x86 architecture RDMSR instruction specifying an x86 architecture IA32_SYSENTER_CS Model Specific Register (MSR) as a source MSR; and
executing one or more of said second microinstruction to load a segment selector for a privilege level 0 code segment value from said RAM, in response to said decoding.
26. The method as recited in claim 17 , further comprising:
executing said first and second microinstructions to use said RAM as a temporary storage area for microcode computations.
27. A computer program product for use with a computing device, the computer program product comprising:
a computer usable storage medium, having computer readable program code embodied in said medium, for providing a microprocessor, said computer readable program code comprising:
first program code for providing a user-accessible register set;
second program code for providing a random access memory (RAM),
within a non-user-accessible address space of the microprocessor,
wherein said RAM is denser than said user-accessible register set and is slower to access than said user-accessible register set; and
third program code for providing microcode, comprising a plurality of microinstructions of a microinstruction set of the microprocessor,
said microinstruction set comprising:
a first microinstruction, executable for storing data from said user-accessible register set to said RAM; and
a second microinstruction, executable for loading data from said RAM to said user-accessible register set.
28. The computer program product as recited in claim 27 , said computer readable program code further comprising:
fourth program code for providing a non-user-accessible register set, wherein said non-user-accessible register set is less dense than said RAM and is faster to access than said RAM;
wherein said first microinstruction is further executable for storing data from said non-user-accessible register set to said RAM, and said second microinstruction is further executable for loading data from said RAM to said non-user-accessible register set.
29. The computer program product as recited in claim 28 , wherein said microcode is configured to execute one or more of said first microinstruction to store a portion of an x86 architecture virtual-machine extensions (VMX) virtual machine control structure (VMCS) to said RAM in response to a VMX VMWRITE instruction.