Flow - roybaer/sdcc-wiki GitHub Wiki

Table of Contents Flow of compilation in SDCC some details on the "top level" source what does it do

1. SDCPP traditional C preprocessing (macro expansion, ...?)
2. compiling (after reaching the end of the source file, the initializers for any non-const global or static variables also go through these same steps)
   1. parsing: preprocessed code -> AST
   2. AST -> icode conversion
   3. target independent icode optimisation
   4. target dependent icode optimisation and register allocation
   5. assembly code generation: icode -> target-specific asm
   6. peephole optimisation
3. assembling
4. linking

---

main = SDCC.EXE

• sdcc/sdcc/src/SDCCmain.c

See the tree on top for links to pages with info on each step:

• based on command line establishes which port is to be used
• initialises this port
• parses command line options
• passes the relevant options to the port
• calls the preprocessor SDCPP.EXE
• feeds the output into the parser "character by character"
  • via a pipe - preprocessing is completely separate from other parts, there are no common data - it could be run standalone, output into a temporary file and feed that into the lexer part of compiler)
• parser calls the subsequent steps - function by function
  • once the parser recognizes that a function is complete, createFunction() (in src/SDCCast.c) is called, which has the AST annotated (probably via decorateType()), has it turned it into iCode (iCodeFromAst() calling ast2iCode()), and lets the backend emit appropriate code (via eBBlockFromiCode() in port->assignRegisters, which maps to, e.g.src/mcs51/ralloc.c, mcs51_assignRegisters())
  • eBBlockFromiCode(): cseAllBlocks(), killDeadCode(), loopOptimizations(), all src/SDCCopt.c
• calls the assembler
• calls the linker

Run SDCC with both "-V --verbose" to see some of these steps.