Binary translation using raw binary code with compiler produced metadata

A system for binary translation, the system comprising:

• a bus;
• a memory connected to the bus, having computer executable instructions stored thereon;
• and
• one or more processors connected to the bus, wherein at least one of the one or more processors executes the computer executable instructions of a method comprising:
• combining, by a binary translator, raw binary code and compiler-produced metadata associated with a compiled program module;
• reconciling, by the binary translator using the compiler-produced metadata, original compiler-produced control flow information with how lower-level machine instructions comprise a control flow in the raw binary code;
• reconciling, by the binary translator using the compiler-produced metadata, original compiler-produced aliasing information with how lower-level machine instructions access the memory locations described by the aliasing information according to predetermined criteria;
• preventing, by the binary translator using the compiler-produced metadata, copy propagation of values in temporary variables for decimal computations beyond offsets in the machine instructions where the temporary variables are killed;
• removing, by the binary translator using the compiler-produced metadata, of identified dead store instructions;
• and
• generating, by the binary translator, a new compiled program module in a form of an optimized version of the compiled program module having strict compatibility to an original version of the compiled program module.

A computer program product comprising a non-transitory computer readable storage medium having computer readable program instructions for binary translation thereon for causing a processor to perform a method comprising:

• combining, by a binary translator, raw binary code and compiler-produced metadata associated with a compiled program module;
• reconciling, by the binary translator using the compiler-produced metadata, original compiler-produced control flow information with how lower-level machine instructions comprise a control flow in the raw binary code;
• reconciling, by the binary translator using the compiler-produced metadata, original compiler-produced aliasing information with how lower-level machine instructions access the memory locations described by the aliasing information according to predetermined criteria;
• preventing, by the binary translator using the compiler-produced metadata, copy propagation of values in temporary variables for decimal computations beyond offsets in the machine instructions where the temporary variables are killed;
• removing, by the binary translator using the compiler-produced metadata, of identified dead store instructions;
• and
• generating, by the binary translator, a new compiled program module in a form of an optimized version of the compiled program module having strict compatibility to an original version of the compiled program module.

A computer-implemented method for binary translation combining raw binary code and compiler-produced metadata, the computer-implemented method comprising:

• in response to receiving original compiler-produced control flow information including descriptive binary data comprising basic blocks in an original intermediate representation (IR), with start and end original machine instruction offsets corresponding to each respective block, edges between these basic blocks in the original IR, out-of-line basic blocks for late created control flow, and information available to the original compiler on calls which never return, reconciling the original compiler-produced control flow information with how lower level machine instructions make up a control flow in raw binary code by:
• creating, by a binary translator using the descriptive binary data from original compiler-produced modules, progressively splitting basic blocks to be used during optimization and code generation phases of the binary translator by:
• splitting machine instructions using information about the basic blocks in the original IR and out-of-line basic blocks in the descriptive binary data;
• splitting the machine instructions further by analyzing branch instructions in the raw binary code in each IR-based basic block;
• and
• splitting the machine instructions further at calls which never return as determined from the descriptive binary data;
• and
• creating, by the binary translator control-flow edges between the basic blocks by:
• analyzing targets of direct branch instructions in the raw binary code, and
• adding edges in the IR of the binary translator using the descriptive binary data for targets of indirect branch instructions;
• in response to receiving the descriptive binary data further comprising aliasing information as it applies to the original IR and information including annotation/tracking of each original machine instruction that references memory, creating by the binary translator, aliasing information for memory references by combining analysis of the machine instructions in the raw binary code with original aliasing information in the descriptive binary data;
• in response to a determination a memory reference on an original machine instruction being processed by the binary translator is at least one of same as, and a part of, a reference in the aliasing information of the descriptive binary based on the original IR, duplicating aliasing information in an alias graph of the descriptive binary data for a corresponding reference in the IR;
• in response to a determination a memory reference consists of multiple references that are present in the aliasing information of the descriptive binary data, computing a union of aliases of all references in the aliasing information of the descriptive binary data corresponding to this single memory location;
• and
• in response to a determination the memory reference does not have a corresponding reference in the IR, creating the aliasing information that the memory reference is aliased with all other memory references, refined later by analyzing at least one of a data area and an offset of the memory reference in the raw binary code and comparing with other references in the aliasing information of the descriptive binary data;
• in response to receiving the descriptive binary data further comprising offsets of the machine instructions where temporary variables for decimal computations are killed, preventing, by the binary translator, copy propagation of values in the temporary variables for decimal computations beyond the offsets in the machine instructions where the temporary variables are killed;
• eliminating, by the binary translator, computations, and stores to memory of values in a middle of the decimal computations when the temporary variables to which the values are stored are killed after the decimal computations;
• and,
• generating, by the binary translator, a new compiled program module in a form of an optimized version of the compiled program module having strict compatibility to an original version of the compiled program module.

A computer-implemented method for binary translation, the computer-implemented method comprising:

• receiving as input by a tool an original compiler-produced module comprising raw binary code that does not have binary metadata;
• receiving further input by the tool a separate compiler-produced metadata in one of a binary form and a text form;
• combining by the tool the original compiler-produced module with the separate compiler-produced metadata to create a new module containing the compiler-produced metadata in the one of the binary form;
• passing the new module to a binary translator;
• reconciling, by the binary translator using the new module with compiler-produced metadata, original compiler-produced control flow information with how lower-level machine instructions comprise a control flow in the raw binary code;
• reconciling, by the binary translator using the new module with the compiler-produced metadata, original compiler-produced aliasing information with how lower-level machine instructions access the memory locations described by the aliasing information according to predetermined criteria;
• preventing, by the binary translator using the new module with the compiler-produced metadata, copy propagation of values in temporary variables for decimal computations beyond offsets in the machine instructions where the temporary variables are killed;
• removing, by the binary translator using the compiler-produced metadata, identified dead store instructions;
• and
• generating, by the binary translator, optimized version of the new module having strict compatibility to an original version of the compiled program module.