245 commits

Author	SHA1	Message	Date
Magnus Norddahl	53b11dd0d8	- fix missing zero extend	2019-04-15 14:32:36 +02:00
Magnus Norddahl	29f6258d33	- fix that the code for META and CLSS was swapped	2019-04-15 14:32:34 +02:00
Magnus Norddahl	c2669d3a03	- some jump fixes	2019-04-15 14:32:33 +02:00
Magnus Norddahl	f7dd6a1313	- fix jump direction in EmitComparisonOpcode	2019-04-15 14:32:31 +02:00
Jonathan Russell	9530a2e43c	- more spelling mistakes...	2019-04-15 14:32:29 +02:00
Jonathan Russell	a770b3af16	- fix minor spelling error	2019-04-15 14:32:28 +02:00
Jonathan Russell	821c336c5e	- add vector comparison opcodes	2019-04-15 14:32:26 +02:00
Magnus Norddahl	14d6aa8cac	- remove some macros and add functions checking if a register needs to be copied	2019-04-15 14:32:25 +02:00
Magnus Norddahl	35a8162022	- implemented LBIT, SBIT, VTBL, SCOPE, NEW, NEW_K, THROW, EQA_R and EQA_K	2019-04-15 14:32:23 +02:00
Magnus Norddahl	4cab630871	- implement emitter functions for LKP_R, META, CLSS, LO, LO_R, LP, LP_R, SO, SO_R, SP, SP_R ... - add emit code for throwing exceptions on null pointers	2019-04-15 14:32:22 +02:00
Magnus Norddahl	2efaa7846d	- improve jit debug output slightly	2019-04-15 14:32:20 +02:00
Magnus Norddahl	f0dbc8414a	- fix offsetting bug when loading address registers	2019-04-15 14:32:19 +02:00
Magnus Norddahl	db78d02d79	- fix swapped operands in EmitSUBF_KR	2019-04-15 14:32:17 +02:00
Magnus Norddahl	ede8c6d86b	- implement MODF_RR, MODF_RK and MODF_KR	2019-04-15 14:32:16 +02:00
Magnus Norddahl	708362b31f	- fix mul using wrong instruction	2019-04-15 14:32:14 +02:00
Magnus Norddahl	c46a4d4483	- implement BOUND opcodes and add logging the resulting assembly code if asmjit throws an exception	2019-04-15 14:32:13 +02:00
Magnus Norddahl	c2938926a7	- implement a few more opcodes and fix some bugs	2019-04-15 14:32:11 +02:00
Magnus Norddahl	9427e22d7c	- implemented pow, min, max	2019-04-15 14:32:10 +02:00
Magnus Norddahl	f4181be6b8	- implement ATAN2 and FLOP	2019-04-15 14:32:08 +02:00
Magnus Norddahl	77fe376eca	- implement DYNCAST*, TEST, TESTN and JMP opcodes	2019-04-15 14:32:07 +02:00
Magnus Norddahl	9474a2554d	- convert JitCompile's switch into a class to enable easier code reuse	2019-04-15 14:32:05 +02:00
Magnus Norddahl	9889d2fbb6	- fix clang compile error	2019-04-15 14:32:03 +02:00
Magnus Norddahl	976a7fab3a	- implement throwing by storing exception information in a struct, then return from the jitted function and throw from c++	2019-04-15 14:32:02 +02:00
Jonathan Russell	45bc0d266e	- removed the ability to use printf-like formatting in emitAbortExceptionCall, it barely worked	2019-04-15 14:32:00 +02:00
Jonathan Russell	197b20af5d	- add first draft of abort exceptions	2019-04-15 14:31:58 +02:00
Magnus Norddahl	c65b7076eb	- call play sqrt when calculating lengths	2019-04-15 14:31:57 +02:00
Magnus Norddahl	50d0049460	- fix 8 and 16 bit store	2019-04-15 14:31:55 +02:00
Magnus Norddahl	76c08bf725	- destroy the jit runtime when all script functions are destroyed	2019-04-15 14:31:54 +02:00
Magnus Norddahl	4820c32794	- disable all unimplemented opcodes in the switch to ensure 'Unknown VM opcode' fatal error is generated if CanJit has a bug	2019-04-15 14:31:52 +02:00
Magnus Norddahl	e9914f38af	Fix compile errors on macOS	2019-04-15 14:31:51 +02:00
Jonathan Russell	601c492cf2	- added remaining int arithmetic opcodes (and fixed float loading offset issue)	2019-04-15 14:31:49 +02:00
Jonathan Russell	e771064657	- forgot to turn off debug switch in last commit	2019-04-15 14:31:47 +02:00
Jonathan Russell	4228891da2	- made absMaskInt const	2019-04-15 14:31:46 +02:00
Jonathan Russell	8306815292	- implemented CMP_APPROX for OP_EQF_*, making ~== work for doubles	2019-04-15 14:31:44 +02:00
Chronos Ouroboros	e07af1428c	Updated CanJit's supported opcodes list.	2019-04-15 14:31:43 +02:00
Chronos Ouroboros	63cda1c13d	Pointer addition opcodes must leave NULL pointers as they are.	2019-04-15 14:31:41 +02:00
Chronos Ouroboros	3c8aa9f641	Changed binary int ops to use a temporary register, fixed some opcodes. ... For some reason, binary ops might reuse one of the input registers as the output register. This is a problem for very obvious reasons.	2019-04-15 14:31:39 +02:00
Jonathan Russell	62ece01c0a	- made CMP_CHECK checking compile time	2019-04-15 14:31:37 +02:00
Jonathan Russell	f897f2bfd3	- added exact floating point comparison opcodes	2019-04-15 14:31:36 +02:00
Jonathan Russell	5846357268	- implemented all integer comparison operators	2019-04-15 14:31:34 +02:00
Jonathan Russell	166ce87a86	- adds OP_EQ_R, the first instruction using jumps	2019-04-15 14:31:33 +02:00
Magnus Norddahl	5c3ef8ac15	- add misc float opcodes	2019-04-15 14:31:31 +02:00
Jonathan Russell	b242ffdbbe	- added most vector maths opcodes	2019-04-15 14:31:29 +02:00
Chronos Ouroboros	73892f4a6e	Fix floats and implement some more opcodes.	2019-04-15 14:31:27 +02:00
Magnus Norddahl	113714990e	- copy initial register values from the VM frame	2019-04-15 14:31:26 +02:00
Magnus Norddahl	10a6ab4d56	- implement most of the integer math opcodes	2019-04-15 14:31:24 +02:00
Magnus Norddahl	d651c02b17	- embed and use asmjit to JIT ZScript VM functions	2019-04-15 14:31:22 +02:00
Christoph Oelckers	53c87fd5eb	- extended the register limit for 'param'. ... The instruction one free instruction byte so it's now using that to extend its argument's register range to 65535. For param this is needed because it passes strings by reference and creating an implicit temporary copy for string constants does not work here.	2018-10-14 10:31:48 +02:00
Christoph Oelckers	608e5f5bcf	- fixed bad code for OP_SRL_KR.	2018-09-02 18:47:42 +02:00
alexey.lysiuk	c55cf00c8e	- added missing null pointer checks to VM ops	2018-07-16 19:11:20 +02:00