regalloc: improve spilling and cleanup

Instead of spilling current interval when running out of registers,
spill the longest-lived active interval with a lower spill cost than
current. Spill costs are estimated based on multiplicative loop depth.
Also cleanup regalloc.c somewhat.
Update todo.txt too.

1 files changed, 5 insertions, 5 deletions

diff --git a/todo.txt b/todo.txt
index 48deffe..d0618a7 100644
--- a/todo.txt
+++ b/todo.txt
@@ -4,10 +4,10 @@
 - DWARF debug information
 - implement GNU extensions: __attribute__, __builtin_*, ...
 - compiler optimizations:
-    - dead store/load optimization,
-    - inlining
-    - loop inversion
+    - dead store/load optimization, forwarding
+    - all kinds of loop optimizations
+    - switch statements (bsearch, maybe jumptables?)
     - instruction scheduling
-    - better regalloc spilling heuristics
+    - use of memory operands in arith instrs in x86
 
-- testing on real world codebases
+- more testing on real world codebases