28ee0944d8703c00d9208c19dd498a0bd552151e
2008-07-07 02:55 -0700
Brendan Eich - Indentation and comment wrapping (at 79, 99 looks too long and most comments still avoid going past 80+/-).
bc67eb0d7e8aa947e4549b5c352fc43646f93461
2008-07-07 02:51 -0700
Andreas Gal - Merge.
23d317c80d00330484a91f6bbc5afbc95fe3a588
2008-07-07 02:47 -0700
Andreas Gal - Sync nanojit with tamarin-tracing tip.
5b4529458a560d4fabcb625f3eaadd23bb151c39
2008-07-07 02:21 -0700
Brendan Eich - Use jsopcode.tbl for tracer JSOP_* method decls; trim trailing whitespace.
1aa7e11366a29c3fe26fed7639e140ead5ab5902
2008-07-07 01:37 -0700
Andreas Gal - Cleanup the native frame -> interpreter frame boxing code. We sometimes end up with numbers on our stack in double format that are really ints. We have to detect those and properly store them as in on the interpreter stack.
d88a5e56ec9382799f89b9a87d8b4f9a561ff1af
2008-07-07 01:05 -0700
Andreas Gal - Add support to demote stores of constants that are currently represented as float but are really integers. bitwise.js is now compiled complete fp-casts free.
0f7c4afcd5c83af511ec58c8798e0a4716524f96
2008-07-07 00:43 -0700
Andreas Gal - Demote floating point comparisons even if one side is constant (bug in nanojit, filed as #443884 against tamarin). Also demote add/sub/mul using the integer overflow detection side exit code that Ed adopted upstream. Tight loops (bitwise) are not emitted completely as integer code after the initial compilation using doubles triggered a speculative demotion of the context slots.
a051c0844e1ba0b0127db7aeccebebd09222e781
2008-07-07 00:10 -0700
Andreas Gal - Sink type conversions into the side exit by updating the map and seeing though the cast in the store (in ExitFilter). Add a whole bunch of asserts on the varios speculative type states to make sure we get the state machine right. Import speculated integer values as integers into the trace (indicate in entry map).
d268132c1bf643d4f98d139a5444468bbd87e5ee
2008-07-06 22:35 -0700
Andreas Gal - Added loop-tail driven context slot type demotion. Without all the buzzwords, what this essentially means is that we detect if the last value that the trace leaves in a slot (which is the value that flows along the backedge back to the loop header) is known to originate from an integer value via i2f, we flag the slot as type integer and recompile the trace. We do this where type stability is certain (result of an and, i.e.) or where its very likely (++operator). If the speculation fails, the same analysis flags the slot as blocked, which means it will always be double. The hope is that this analysis converges quickly (1-2 recompilations tops).
7f3e0b5b05dd1e749b772a6ca4357262b48b46eb
2008-07-06 20:27 -0700
Andreas Gal - Make sure we don't get the argument order in BoxDouble wrong again.
c3568285ff19a11ee557006da9b8e70553e69267
2008-07-06 20:25 -0700
Andreas Gal - Make typemap uint8_t and fix order of arguments in call to BoxDouble.
ce35d7c30a31f407fc779c1c40cc779c6b7e1b90
2008-07-06 23:11 -0400
shaver - remove outdated guard (now inferred via filter as needed)
513b636c199b467c3164ea1fbb319d0e6e84ad95
2008-07-06 23:01 -0400
shaver - we only speak double around here now
37dd888fcb4252ce63619cba74722c6640fe0154
2008-07-06 22:39 -0400
shaver - remove bogus NOT_REACHED
fdab515b3c8acd9f82177744961a47262e627aa9
2008-07-06 17:43 -0700
Andreas Gal - Strength reduce i2f(doubleToInt32(x)) and u2f(doubleToUint32(x)) to x. This eliminate most of the on-trace overhead, but we still need type peeling of loop variable into int to win back the performance loss casting introduced.
f6179396df600b85b1b6cda53ee18db24fa06576
2008-07-06 17:32 -0700
Andreas Gal - Reduce redundant doubleToInt32(i2f(x)) chains to simply x.
0b810d0bf2e7572dbc69b66fce1c3a8b323560bf
2008-07-06 16:38 -0700
Andreas Gal - Strength reduce BoxDouble(i2f(x)) to BoxInt32(x). Make sure loop indexes are actually integers. This guard will be eliminated further down in the pipeline if we decided to peel the type of the loop variable (index) down to integer.
5ebea36d79ade1ab78789c7d79a0bbb3fbfeb1cf
2008-07-06 16:18 -0700
Andreas Gal - Since numbers are now always represented by doubles at the recorder/type level, using BoxDouble and UnboxDouble only when moving numbers from or to memory. A filter will then turn f2i(UnboxDouble) into UnboxInt where appropriate.
8374e34d597e56684f05628edb4f5c0bf8752e44
2008-07-06 15:55 -0700
Andreas Gal - Move the type level from int/double to number. All traces start out as double in all slots, and denote and promote to/from int as needed. The FuncFilter optimizes on-trace casting and elininates redundant f->i-> chains. More optimization needed on this of course, and this code is now a bit slower than the previous integer-register use. However, this does solve the q += 2.5 issues. The heap access code does not properly cast yet and is likely unstable.
360a6ce57d28f76b5f739770c81c1d0ff57926ce
2008-07-06 16:59 -0400
shaver - Trace GETGVAR and INCGVAR, though not yet correctly.
d84c7e0344ed2c6bd7c28ba6a3eb286558c490e4
2008-07-06 13:16 -0700
Andreas Gal - Add unary and binary helpers that automatically demote and promote when dealing with integer operations (not used yet, need loop typemap peeling in place first.)
8e62c283774efb8a96b0d6e0aa9043f71061a69a
2008-07-06 12:16 -0700
Andreas Gal - JSVAL_IS_BOOLEAN does what isTrueOrFalse was trying to do.
0ce416b5a3df5c324d78859a3d91ddf78678a3b5
2008-07-06 12:03 -0700
Andreas Gal - Fix template, should use typename, not class.
f89cf130bd52037f006c1b950abfebe0e1c6df7f
2008-07-06 11:48 -0700
Andreas Gal - Add ExitFilter, which builds side exit typemaps for us. This had to be moved into a filter, because this has to happen after we eliminate redundant i2f-f2i chains as we can sink those into the side exit by simply storing the unpromoted/undemoted value and just flip the type in the exit map.
fcd3450f9272bba03d987ed0fe90040c2253a801
2008-07-06 11:19 -0700
Andreas Gal - Merge.
a53c3daea0e8bce5abd06e13a810b60177cf5ba5
2008-07-06 10:59 -0700
Andreas Gal - Added doubleToUint32 builtin and make tracker a template.
a2e3ea92df6516cb0f4f4f0dd04f49ee0cba5218
2008-07-06 13:58 -0400
shaver - fix builtin_UnboxInt32 signature and name shape_ins for debugging
ba845ab9dff1c7f57f002bd5de9a3a247b43b15e
2008-07-06 13:38 -0400
shaver - Trace JSOP_NAME, and refactor out jsval unboxing.
b83911cb6d173cd2e7300554810276c3bfccebc1
2008-07-06 12:15 -0400
shaver - begin on JSOP_NAME, refactor stobj_get_slot to chain better
2e8512ba0fb481206e3f7d5850888bd370ee7f37
2008-07-06 10:24 -0400
shaver - Fix optimizer flags for interp and builtins in debug mode, and clean out old
e9367ede4167ffedad12590489df214df230dd24
2008-07-06 09:56 -0400
shaver - actually record at HOTLOOP1 (fencepost)
f4cec2d38f041826b410393e653496db3efa2e31
2008-07-06 09:52 -0400
shaver - some remaining low-hanging ops
e4fae864396eb19a24de8d3c48fd9bfe3fe01a17
2008-07-06 09:33 -0400
shaver - move JSOP_DOUBLE impl to, er, JSOP_DOUBLE (wtf?)
6b5d40cc17fed5881b74cf387a56cf121a8ae22a
2008-07-06 07:28 -0400
shaver - Implement JSOP_DOUBLE.
859e409ef3796a8abd3e636595d06bb481e4d13d
2008-07-05 23:21 -0700
Andreas Gal - Added support for semi-stable loop variables. Compiling for(...) q += 2.5; is ridiculously difficult because it flip-flops between int and double. Add support to promote integer values to doubles at the loop tail if at loop entry we expect a double. Since this isn't possible the other way around, we have to get luck that we catch a path into the loop where q is already double. For this we add 3 trigger points (10, 13, 37). We will try three times to record a trace at those iteration counts of a loop. If none succeed the loop is blacklisted. This probably needs more tuning down the road.
774553375994ae0990e57d95af34880e8fc6b67f
2008-07-05 22:23 -0700
Andreas Gal - Merge and rename jsIf to ifop.
6ab468ffe7f4b34721667d3eb85259d037f9eeca
2008-07-05 22:20 -0700
Andreas Gal - Added FuncFilter from tamarin core (not part of nanojit yet since its slightly VM dependant).
b5556367172da3d22140e614f878a8c05eee47fa
2008-07-05 22:04 -0700
Andreas Gal - Add helpers for i->f and f->i conversion. The f->i path goes via builtin calls.
46aabcce02752fce4ae10eda1cfc9a6371337474
2008-07-05 21:51 -0700
Andreas Gal - Add type primitives for numbers (isNumber and asNumber).
8bff19dbf023f7acfb312941c8ccb149d1c4f388
2008-07-05 23:47 -0400
shaver - Implement JSOP_IFEQ, JSOP_IFNE, JSOP_DUP, JSOP_DUP2.
3d66327238969be22f6d92dfca5fa7bde24dc8af
2008-07-05 20:30 -0700
Andreas Gal - Add casting for objects and cleanup casting code.
b1d8fd0005e9ce0d1d0387ad1f2a2122d19bf336
2008-07-05 23:04 -0400
shaver - let JSOP_GETELEM handle boolean values too
6272876588dca9a03a207db551c0d1104c56df6c
2008-07-05 22:39 -0400
shaver - Handle boolean lval in SETELEM; now runs access-nsieve unmodified.
355d79be4adfe120ebcb0ee2f61ee463106f10e5
2008-07-05 19:25 -0700
Andreas Gal - Believe it or not NEG can actually overflow the int32 range, so add an overflow bailout. This will only trap for -0x80000000.
f4c96a566c4c826d41a6249d784b90e465c66cff
2008-07-05 19:18 -0700
Andreas Gal - Merge.
138fc2d7f4fe744306e5244549030edfb3d9469c
2008-07-05 19:15 -0700
Andreas Gal - Added blacklisting of recording points where we failed to complete a trace (overly aggressive at this point, needs tuning, we want to try several times for each point.)
1a4d440c60e4d2dbc9fd3c6c50ab7c9dc6e0b0b7
2008-07-05 21:50 -0400
shaver - Allow hole-filling JSOP_SETELEM to remain on trace.
04f3c8fdf238f14566efd1fa27c716d3e486c4d8
2008-07-05 18:36 -0700
Andreas Gal - Fixed boolean boxing.
0523b54c86d82a7aa4dc531290557630c07a4c70
2008-07-05 18:34 -0700
Andreas Gal - Added boolean boxing/unboxing code.
e3af49ee7476edbcc11fab2097190f52c66933e9
2008-07-05 18:18 -0700
Andreas Gal - Use default parameter in LSH/RSH/URSH to indicate we don't care for the overflow.
d23078399de1315e772b009814d1680b3edc51dc
2008-07-05 17:29 -0700
Andreas Gal - Added missing jsbuiltins.cpp
6ffdcbdb066c1963380a2e12e4e99f601da2f8ba
2008-07-05 17:26 -0700
Andreas Gal - Add support for some misc opcodes including binary and arithmetic and/or/not.
639f37f76fa0a16a05e95a5b05c2fb81026914a5
2008-07-05 17:10 -0700
Andreas Gal - Track trace entry/exit in debug mode and count cycles.
f5a30a112d06eacf7a701fcf27123fd4907e97cf
2008-07-05 16:28 -0700
Andreas Gal - Signal error from the boxing/unboxing using magic cookies since gcc seems to very seriously object to the use of uint64 return values during a fastcall (horribly inefficient code).
8c6d36837b228e0b633f5e0dbc9cfaa01ad7f798
2008-07-05 15:46 -0700
Andreas Gal - Help the branch predictor in the builtins.
9edc74143be669124e0589c0636e026bd9d7519b
2008-07-05 17:37 -0400
shaver - use FASTCALL for builtins
ac9c1fa73e6f3abb9bd736025dfbdaaf3652c400
2008-07-05 14:05 -0700
Andreas Gal - Style nitpicking. Fix overlong lines.
c0ee4caf332f33b3e992a3e7087b2e874edee87c
2008-07-05 14:00 -0700
Andreas Gal - Finish SETELEM for int and double values.
977464d95d02128f8a5071cb41c4d5518ea48e41
2008-07-05 13:44 -0700
Andreas Gal - Merge unboxing code into GETELEM code.
788d1cebba153024a3dd5f932495339fa6d888ae
2008-07-05 11:46 -0700
Andreas Gal - Merge.
(0) -10000 -3000 -1000 -300 -100 -60 +60 +100 +300 +1000 +3000 +10000 +30000 +100000 +300000 tip