~cnx/roux - Alternative QBE compiler

Age	Commit message (Collapse)	Author
2015-09-15	change references representation	Quentin Carbonneaux

2015-09-15	try writing a parser, painful	Quentin Carbonneaux

2015-09-15	change name of test file	Quentin Carbonneaux

2015-09-15	some new C	Quentin Carbonneaux

2015-09-15	add an untracked directory	Quentin Carbonneaux

2015-09-15	bugfix in regalloc (Div,Rem) and in peucl	Quentin Carbonneaux

2015-09-15	attempt to implement euclide algorithm	Quentin Carbonneaux

2015-09-15	restore addition in test	Quentin Carbonneaux

2015-09-15	generate code for the division	Quentin Carbonneaux

2015-09-15	reserve edx in all cases for divisions	Quentin Carbonneaux

2015-09-15	more cleaning	Quentin Carbonneaux

2015-09-15	add gitignore	Quentin Carbonneaux

2015-09-15	fix code emitting for x = y op x	Quentin Carbonneaux

2015-09-15	fix addition opcodes	Quentin Carbonneaux

2015-09-15	do a bigger test computation	Quentin Carbonneaux

2015-09-15	compile tmain with optimizations	Quentin Carbonneaux

2015-09-15	run multiple rounds for stable timing	Quentin Carbonneaux

2015-09-15	quiet	Quentin Carbonneaux

2015-09-15	time execution	Quentin Carbonneaux

2015-09-15	cosmetics	Quentin Carbonneaux

2015-09-15	cleanup overflow computation	Quentin Carbonneaux

2015-09-15	fix spill code	Quentin Carbonneaux

2015-09-15	add `Ret jump and dump elf files	Quentin Carbonneaux

2015-09-15	remove default tests in Elf	Quentin Carbonneaux

2015-09-15	add popd in the Heap module	Quentin Carbonneaux

2015-09-15	add simple Makefile	Quentin Carbonneaux

2015-04-04	write down some more ideas	Quentin Carbonneaux

2015-04-01	do not freeze r when allocating for r2	Quentin Carbonneaux

2015-03-31	update TODO with fix commit	Quentin Carbonneaux

2015-03-31	tentative fix of messed up spills for phis	Quentin Carbonneaux

2015-03-31	add bugs	Quentin Carbonneaux

2015-03-31	avoid evicting arguments	Quentin Carbonneaux

2015-03-31	add move constant to spill	Quentin Carbonneaux

2015-03-31	better support for constants	Quentin Carbonneaux

2015-03-27	refine TODO	Quentin Carbonneaux

2015-03-27	add a scratch TODO	Quentin Carbonneaux

2015-03-25	expect reg in brz, might not be true	Quentin Carbonneaux

2015-03-25	simplify label logic	Quentin Carbonneaux

2015-03-25	branch code	Quentin Carbonneaux

2015-03-25	more codegen	Quentin Carbonneaux

2015-03-25	start writing some codegen	Quentin Carbonneaux

2015-02-20	improvements to elf.ml	Quentin Carbonneaux

2015-02-20	fix bugs in movgen	Quentin Carbonneaux

2015-02-20	start work on movgen	Quentin Carbonneaux

2015-02-18	cosmetics	Quentin Carbonneaux

2015-02-18	add spill test	Quentin Carbonneaux

2015-02-18	kind of working!	Quentin Carbonneaux

2015-02-18	wip	Quentin Carbonneaux

2015-02-18	wip	Quentin Carbonneaux

2015-02-15	add notes	Quentin Carbonneaux
	I start to believe that it is possible to simply reuse the regalloc I wrote. All the data structures seem fine. Here is a list modifications to make for sure: - Keep track of the mapping for live in variables for each block. - Make sure 'loc' (or a new function) returns registers for the arguments of base instructions. Spilling -------- When we need a register and none are left. If we are at the end of a block, simply spill anybody (the resolving will take care of adding the moves). If we are in a block, find somebody to spill and its next use, if the next use is in the same block, emit reload code, otherwise, decide to emit the reload at the end of the cheapest block (this means that we have to patch some mappings that were already stored). We let the resolve handle the insertion of the reload code. Example: +----------------------+ \| \| \| \| \| +----+ +----+ \| +----+ -+-> \| b1 \| --> \| b2 \| -+-> \| b3 \| --> +----+ +----+ +----+ ^ ^ Spill x First use of x (Was in r1) The cheapest position for reload code is at the beginning of b3, so we have to modify the mapping at the end of b1, beginning of b2, and end of b2 to change it from (x -> r1) to (x -> spill). Reloading --------- When a spilled variable is needed in register. In this case, we are necessarily inside a block (or at a branch) because that is the only place we require a variable to be in register. Here again we are constrained to insert the spill code before the next use of the variable, but more importantly, we must do it before the register chosen is in use next.