ed5cea57ea
compiler.cfg: Reading the return value of a float-returning function on x86-32 had a side effect of popping the x87 stack, so it was not correct for DCE to just eliminate this if the return value was not used. Fix this by adding a new dead-outputs slot to alien-call-insns and having DCE move dead returns there
2010-08-13 23:19:56 -07:00
f70b01e6f2
compiler.cfg.save-contexts: don't insert ##save-context in front of ##phi
2010-07-19 09:27:10 -04:00
f133a5f2be
compiler: re-architect low-level optimizer to allow more than one output value per instruction
2010-07-13 07:40:14 -04:00
8fd5d3c86a
FFI rewrite part 6: deconcatenatize
2010-07-02 15:44:12 -04:00
fb4de47abf
Add GC maps to ##box, ##box-long-long, ##alien-invoke, ##alien-indirect and ##call-gc; remove ##gc-map instruction
2010-06-14 19:39:46 -04:00
d8fc595383
compiler.cfg: clean up ##phi literals in tests
2010-05-03 17:34:24 -04:00
8849e370ca
Register allocation now uses SSA properties to coalesce values with different representations
2010-05-03 17:34:20 -04:00
fc7a1ad8b3
New GC checks work in progress
2010-05-03 17:34:16 -04:00
0f0571e48a
Moving new-sets to sets
2010-02-26 16:01:01 -05:00
f80416b40e
Fixing write-barrier elimination; adding bb as a parameter to join-sets in dataflow analysis
2009-08-12 23:52:29 -05:00
88d15dfbea
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 20:02:56 -05:00
2d575d7ec9
compiler.cfg: virtual registers are integers now, and representations are stored off to the side. Fix bug in representation selection that would manifest if a value was used as a float and a fixnum in different branches; cannot globally unbox float in this case
2009-08-08 04:02:18 -05:00
4d2160799f
Split off the notion of a register representation from a register class
2009-08-07 17:44:50 -05:00
b3786fb6eb
Revert "compiler.cfg.liveness.ssa: remove"
...
This reverts commit 6144fee101
2009-08-02 18:41:16 -05:00
6144fee101
compiler.cfg.liveness.ssa: remove
2009-08-02 17:19:01 -05:00
64b719269f
compiler.cfg.utilities: add each-phi combinator to iterate over all ##phi instructions in a basic block
2009-08-02 06:16:58 -05:00
8028eb5776
compiler.cfg: clean up unit tests using some new utilities
2009-08-02 03:49:25 -05:00
7ee7346505
compiler.cfg.liveness.ssa: fix bad bug discovered by littledan
2009-08-01 18:54:24 -05:00
0d42e0fc1e
compiler.cfg.ssa.destruction: tweak in preparation for landing Dan's new SSA liveness analysis
2009-07-29 23:35:51 -05:00
bbd4e4af3e
compiler.cfg: Minor optimization. Instructions can now only ever produce a single value; this eliminates 1array constructions and some iterations
2009-07-28 12:29:07 -05:00
2fe741568c
compiler.cfg.liveness: fix liveness computation in case where instruction uses a register it defines
2009-07-28 06:45:02 -05:00
d7f12feb64
compiler.cfg.coalescing: more work done
2009-07-27 00:31:21 -05:00
c355c21de6
Move liveness back to compiler.cfg.liveness, fix a corner-case where it didn't work on some forms of non-SSA code
2009-07-22 06:07:28 -05:00
5d9f7b0ed2
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 03:08:28 -05:00
b801022472
Move conjoin-at from compiler.cfg.liveness to sets
2009-07-21 03:02:06 -05:00
fc029299d8
compiler.cfg.liveness: update for new ##phi node representation
2009-07-02 17:28:02 -05:00
57d9d9f961
Split off local-optimization combinator into compiler.cfg.local, factor out CFG -> MR into compiler.cfg.mr, split off GC check insertion into a new compiler.cfg.gc-checks pass
2009-05-31 12:20:46 -05:00
968a9bb666
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 13:11:34 -05:00
6b81b3e055
compiler.cfg.liveness: correct handling of phi nodes
2009-05-28 17:54:27 -05:00
1242a3a411
New local-optimization combinator removes some boilerplate
2009-05-26 19:56:56 -05:00
25805f9bfe
Refactoring low-level optimizer to support stack analysis pass
2009-05-26 19:31:19 -05:00
Slava Pestov
Slava Pestov
Daniel Ehrenberg
Slava Pestov