steve
/
factor
1
0
Fork 0
Code Issues Pull Requests Releases Wiki Activity
factor/basis/compiler/cfg/linear-scan/numbering/numbering.factor

24 lines
703 B
Factor
Raw Normal View History

Various codegen improvements: - new-insn word to construct instructions - cache RPO in the CFG - re-organize low-level optimizer so that MR is built after register allocation - register allocation now stores instruction numbers in the instructions themselves - split defs-vregs into defs-vregs and temp-vregs
2009-05-29 14:11:34 -04:00
! Copyright (C) 2009 Slava Pestov.
! See http://factorcode.org/license.txt for BSD license.
compiler.cfg: Major restructuring -- do not compute liveness before local optimization, and instead change local optimizations to be more permissive of undefined values. Now, liveness is only computed once, after phi elimination and before register allocation. This means liveness analysis does not need to take phi nodes into account and can now use the new compiler.cfg.dataflow-analysis framework
2009-07-22 04:08:28 -04:00
USING: kernel accessors math sequences grouping namespaces
compiler.cfg: new system to track when results of analyses need to be recomputed (reverse post order, linear order, predecessors, dominance, loops). Passes can now call needs-predecessors, needs-dominance, needs-loops at the beginning, and cfg-changed, predecessors-changd at the end. Linearization order now takes loop nesting into account, and linear scan now uses linearization order instead of RPO.
2009-08-08 21:02:56 -04:00
compiler.cfg.linearization.order ;
Various codegen improvements: - new-insn word to construct instructions - cache RPO in the CFG - re-organize low-level optimizer so that MR is built after register allocation - register allocation now stores instruction numbers in the instructions themselves - split defs-vregs into defs-vregs and temp-vregs
2009-05-29 14:11:34 -04:00
IN: compiler.cfg.linear-scan.numbering
: number-instructions ( rpo -- )
compiler.cfg: new system to track when results of analyses need to be recomputed (reverse post order, linear order, predecessors, dominance, loops). Passes can now call needs-predecessors, needs-dominance, needs-loops at the beginning, and cfg-changed, predecessors-changd at the end. Linearization order now takes loop nesting into account, and linear scan now uses linearization order instead of RPO.
2009-08-08 21:02:56 -04:00
linearization-order 0 [
Various codegen improvements: - new-insn word to construct instructions - cache RPO in the CFG - re-organize low-level optimizer so that MR is built after register allocation - register allocation now stores instruction numbers in the instructions themselves - split defs-vregs into defs-vregs and temp-vregs
2009-05-29 14:11:34 -04:00
instructions>> [
[ (>>insn#) ] [ drop 2 + ] 2bi
] each
compiler.cfg: new system to track when results of analyses need to be recomputed (reverse post order, linear order, predecessors, dominance, loops). Passes can now call needs-predecessors, needs-dominance, needs-loops at the beginning, and cfg-changed, predecessors-changd at the end. Linearization order now takes loop nesting into account, and linear scan now uses linearization order instead of RPO.
2009-08-08 21:02:56 -04:00
] reduce drop ;
compiler.cfg.linear-scan: debugging spilling, add more assertions
2009-07-07 14:01:27 -04:00
SYMBOL: check-numbering?
ERROR: bad-numbering bb ;
: check-block-numbering ( bb -- )
dup instructions>> [ insn#>> ] map sift [ <= ] monotonic?
[ drop ] [ bad-numbering ] if ;
compiler.cfg: Major restructuring -- do not compute liveness before local optimization, and instead change local optimizations to be more permissive of undefined values. Now, liveness is only computed once, after phi elimination and before register allocation. This means liveness analysis does not need to take phi nodes into account and can now use the new compiler.cfg.dataflow-analysis framework
2009-07-22 04:08:28 -04:00
: check-numbering ( cfg -- )
compiler.cfg: new system to track when results of analyses need to be recomputed (reverse post order, linear order, predecessors, dominance, loops). Passes can now call needs-predecessors, needs-dominance, needs-loops at the beginning, and cfg-changed, predecessors-changd at the end. Linearization order now takes loop nesting into account, and linear scan now uses linearization order instead of RPO.
2009-08-08 21:02:56 -04:00
check-numbering? get
[ linearization-order [ check-block-numbering ] each ] [ drop ] if ;