Commit message (Collapse) | Author | Age | Files | Lines | ||
---|---|---|---|---|---|---|
... | ||||||
* | Refactoring of conversion ops, part 4: use CONV instead of TOINT/TONUM. | Mike Pall | 2010-12-31 | 1 | -6/+7 | |
| | | | | Also narrow CONV.int.num and CONV.i64.num. | |||||
* | Refactoring of conversion ops, part 3: add FOLD rules for IR_CONV. | Mike Pall | 2010-12-31 | 1 | -31/+222 | |
| | ||||||
* | Copy destination type for CONV from ir->t to op2, too. | Mike Pall | 2010-12-31 | 1 | -6/+4 | |
| | ||||||
* | Refactoring of conversion ops, part 1: add IR_CONV. | Mike Pall | 2010-12-30 | 1 | -0/+22 | |
| | ||||||
* | FFI: Drop IR_CNEWI. Add IR_CNEWP only for pointers/refs. | Mike Pall | 2010-12-17 | 1 | -18/+9 | |
| | ||||||
* | FFI: FOLD load of initializers even across PHIs. | Mike Pall | 2010-12-17 | 1 | -1/+1 | |
| | ||||||
* | Add alias analysis for XLOAD/XSTORE. Add DSE for XSTORE. | Mike Pall | 2010-12-12 | 1 | -0/+3 | |
| | ||||||
* | Add missing PHI barrier to strength reduction of widening. | Mike Pall | 2010-12-12 | 1 | -0/+2 | |
| | ||||||
* | Extend all FOLD rules to work on 64 bit integers. | Mike Pall | 2010-12-11 | 1 | -32/+130 | |
| | ||||||
* | Regroup FOLD rules for constant folding. | Mike Pall | 2010-12-11 | 1 | -104/+112 | |
| | ||||||
* | FFI: Add IR_CNEW/IR_CNEWI to allocate/init cdata objects. | Mike Pall | 2010-12-11 | 1 | -2/+33 | |
| | ||||||
* | Strength-reduce 32 to 64 bit widening using scalar evolution analysis. | Mike Pall | 2010-12-09 | 1 | -2/+43 | |
| | ||||||
* | Apply narrowing optimization to IR_TOI64, too. | Mike Pall | 2010-12-08 | 1 | -0/+2 | |
| | ||||||
* | Add FOLD rule to reassociate 64 bit (x+k1)+k2. | Mike Pall | 2010-12-08 | 1 | -0/+14 | |
| | ||||||
* | Add IR_TOI64. | Mike Pall | 2010-12-06 | 1 | -0/+26 | |
| | ||||||
* | Add FOLD rule to turn i << 1 into i + i. | Mike Pall | 2010-12-06 | 1 | -0/+5 | |
| | ||||||
* | Add support for integer IR_MUL. | Mike Pall | 2010-12-06 | 1 | -12/+74 | |
| | ||||||
* | Avoid compiler warnings. | Mike Pall | 2010-12-06 | 1 | -1/+1 | |
| | ||||||
* | Add IR_XSTORE. | Mike Pall | 2010-12-06 | 1 | -12/+1 | |
| | ||||||
* | Add minimal set of fold rules for KINT64. | Mike Pall | 2010-12-06 | 1 | -0/+19 | |
| | ||||||
* | Rename IRT_PTR to IRT_P32. | Mike Pall | 2010-12-05 | 1 | -1/+1 | |
| | ||||||
* | Drop redundant (UGE any +0). | Mike Pall | 2010-09-21 | 1 | -0/+8 | |
| | ||||||
* | Improve FOLD/CSE of field loads and array/hash refs across NEWREF. | Mike Pall | 2010-09-21 | 1 | -15/+8 | |
| | ||||||
* | Add IR_VLOAD for vararg loads. | Mike Pall | 2010-09-14 | 1 | -0/+1 | |
| | | | | Also fixes the broken AA improvement in the last commit. | |||||
* | Record vararg expressions with varargs defined off-trace. | Mike Pall | 2010-09-13 | 1 | -2/+7 | |
| | | | | Add SLOAD variant to access the frame type/size. | |||||
* | PPC: Clean up masked shift/rotate target settings. | Mike Pall | 2010-08-29 | 1 | -3/+2 | |
| | ||||||
* | No longer let the GC replace dead keys with the LJ_TDEADKEY tag. | Mike Pall | 2010-04-21 | 1 | -2/+3 | |
| | | | | | | | | Important: this changes the semantics of the write barrier! Carefully read the big comment block in lj_obj.h This helps HREFK key slot specialization and allows safely hoisting HREF/HREFK across GC steps, too (fix for a barely reproducible bug). Dead keys are only removed during a table resize (as before). | |||||
* | Fold HREF of TNEW/TDUP to niltv. Fold HLOAD of niltv to nil. | Mike Pall | 2010-03-28 | 1 | -0/+31 | |
| | ||||||
* | Add array bounds check elimination (-Oabc, on by default). | Mike Pall | 2010-03-15 | 1 | -14/+53 | |
| | ||||||
* | Fix folding of (comparison x x). | Mike Pall | 2010-03-14 | 1 | -1/+1 | |
| | ||||||
* | RETF modifies BASE. Treat it like a store or it gets CSEd. | Mike Pall | 2010-02-04 | 1 | -0/+1 | |
| | ||||||
* | Improve alias analysis of upvalues using a disambiguation hash value. | Mike Pall | 2010-01-09 | 1 | -2/+2 | |
| | | | | | | | | All upvalue objects hold a disambiguation hash value now. It's built from the parent prototype and the slot number. Different hash values imply the upvalues cannot alias. Same hash values don't imply anything (collision or different closures). Upvalue disambiguation makes use of a reduced hash due to IR contraints. | |||||
* | Bump all copyright dates to 2010. | Mike Pall | 2010-01-09 | 1 | -1/+1 | |
| | ||||||
* | Fast forward to sync public repo. | Mike Pall | 2009-12-08 | 1 | -35/+92 | |
| | | | | | | | | Compile math.sinh(), math.cosh(), math.tanh() and math.random(). Compile various io.*() functions. Drive the GC forward on string allocations in the parser. Improve KNUM fuse vs. load heuristics. Add abstract C call handling to IR. | |||||
* | RELEASE LuaJIT-2.0.0-beta2v2.0.0-beta2 | Mike Pall | 2009-12-08 | 1 | -4/+4 | |
| | ||||||
* | RELEASE LuaJIT-2.0.0-beta1v2.0.0-beta1 | Mike Pall | 2009-12-08 | 1 | -0/+1415 | |