aboutsummaryrefslogtreecommitdiff
path: root/src (unfollow)
Commit message (Expand)AuthorFilesLines
2010-02-16Split CALL/FUNC recording.Mike Pall2-447/+455
2010-02-15Improve error reporting for traces aborted in non-Lua functions.Mike Pall1-2/+13
2010-02-15Change record vmevent parameters. Dump function id on FUNC*.Mike Pall2-18/+18
2010-02-15Use a different marker for hot calls.Mike Pall5-1081/+1097
2010-02-15Add generic function handling for debug modules.Mike Pall4-48/+52
2010-02-15Minor cleanup of trace event handling.Mike Pall5-37/+33
2010-02-15Add missing FORI coercions in recorder.Mike Pall1-4/+13
2010-02-15Improve FOR loop const specialization and integerness checks.Mike Pall2-34/+68
2010-02-14Implement return hooks for Lua functions (zero-cost if disabled).Mike Pall9-1450/+1493
2010-02-14Implement call hooks (zero-cost if disabled).Mike Pall9-1583/+1693
2010-02-13Major redesign of function call handling.Mike Pall34-4178/+4324
2010-02-12Extend scope of local vars in debug info beyond final return.Mike Pall1-1/+1
2010-02-11Fix dump of bytecode instructions with missing operands.Mike Pall1-4/+5
2010-02-11Move dispatch tables out of GG_State struct.Mike Pall10-43/+45
2010-02-11Drop frame clearing in exit handling and JIT compiled code.Mike Pall2-21/+1
2010-02-11Switch to pre-initialized stacks. Drop frame clearing in interpreter.Mike Pall7-2686/+2697
2010-02-10Replace GCproto reference with bytecode PC in GCfuncL.Mike Pall3-30/+32
2010-02-08Drop bc field in GCproto since the bytecode is colocated.Mike Pall7-29/+23
2010-02-08Fix constructor bytecode generation for conditional values.Mike Pall1-1/+1
2010-02-08Reduce whitespace in lj_bcdef.h.Mike Pall1-4/+4
2010-02-08Redesign of prototype generation, part 5: colocation of protoype arrays.Mike Pall10-129/+90
2010-02-08Redesign of prototype generation, part 4: late creation of prototype.Mike Pall2-20/+20
2010-02-08Redesign of prototype generation, part 3: bc and lineinfo.Mike Pall3-88/+109
2010-02-08Redesign of prototype generation, part 2: late init of chunkname.Mike Pall2-5/+6
2010-02-08Redesign of prototype generation, part 1: varinfo and uvname.Mike Pall6-96/+109
2010-02-06Major cleanup of bytecode parser.Mike Pall1-1334/+1327
2010-02-06Update .gitignore.Mike Pall1-0/+1
2010-02-05Move bytecode offsets from lj_vm.* to generated header.Mike Pall11-62/+49
2010-02-0532/64 bit memory ref cleanup, part 4: GCproto ->varinfo.Mike Pall5-15/+21
2010-02-0532/64 bit memory ref cleanup, part 3: GCproto ->lineinfo.Mike Pall8-26/+38
2010-02-0532/64 bit memory ref cleanup, part 2: GCproto ->uvname and ->chunkname.Mike Pall8-30/+43
2010-02-0532/64 bit memory ref cleanup, part 1: GCproto ->bc and ->k.Mike Pall10-66/+84
2010-02-04RETF modifies BASE. Treat it like a store or it gets CSEd.Mike Pall2-1/+3
2010-02-04Reset the hotcount table after a JIT off to on transition.Mike Pall3-15/+18
2010-02-04Add shadow frame link stack for trace recorder.Mike Pall6-56/+82
2010-02-03Fix unroll limit checks and frame depth adjustment for pcall.Mike Pall1-13/+11
2010-02-03Clean up frame depth checks and loop detection.Mike Pall3-6/+12
2010-02-03Add extra check to suppress hotcall event during recording.Mike Pall1-2/+3
2010-02-03Another loop formation test must check for return to lower frame.Mike Pall1-1/+1
2010-02-01Improve coalescing of BASE register in side traces.Mike Pall1-21/+47
2010-01-30Revise hardcoded inlining in lj_asm.c. Saves 1-2K.Mike Pall1-4/+4
2010-01-30Add support for weak IR references to register allocator.Mike Pall1-18/+51
2010-01-29Loop formation test must check for return to lower frame.Mike Pall1-1/+2
2010-01-29Compile return to lower frame. Only for Lua frames right now.Mike Pall3-40/+75
2010-01-28Followup fix: set maxslot for continuation return.Mike Pall1-0/+1
2010-01-28Drop obsolete frame shrinking after continuation return.Mike Pall1-4/+1
2010-01-28Fix recording of metamethod result adjustment.Mike Pall1-1/+1
2010-01-28Don't modify jit_State and exit counters while in vmevent.Mike Pall1-15/+16
2010-01-28Fix snapshot dumps.Mike Pall1-1/+1
2010-01-27Avoid reuse of PHI registers, even for duplicate right PHIs.Mike Pall1-1/+1