After GIMPLE, the source code is converted into the static single assignment (SSA) representation. The central idea of this form is the fact that each variable is assigned to only once, but can be used at the right hand side of an expression many times. Every time the same variable of a tree in the GIMPLE form is reassigned, the compiler creates a new version of that variable and stores the new value into it. When the same variable is assigned to in both branches of a conditional expression, one needs to merge the two possible values of the variable into a single variable. This operation is denoted as PHI function in the SSA form. The SSA form is also used for optimizations. GCC performs more than 20 different optimizations on SSA trees. After the SSA optimization pass, the tree is converted back to the GIMPLE form which is then used to generate a register-transfer language (RTL) form of a tree. RTL is a hardware-based representation that corresponds to an abstract target architecture with an infinite number of registers. An RTL optimization pass optimizes the tree in the RTL form. Finally, a GCC back-end generates the assembly code for the target architecture using the RTL representation. Examples of back-ends are x86 back end, mips back end, etc. In the next sections we describe the internals of the C front end and the x86 back end. The compiler starts with its initialization and command line options processing. After that the C front end preprocesses the source file, parses it and performs a number of optimizations. The back end then generates the assembly code for the target platform and saves it to a file. |
说点什么...