d79b462f75
compiler: add intrinsic for PMOVMSKB/MOVMSKP[SD]
...
Combined with a fast bit-count this will let us rice byte-counting.
2011-11-13 16:10:20 -08:00
0c022f2454
compiler: add ##safepoint IR insn
2011-10-27 21:14:46 -07:00
64252dbdbc
32 and 64 bit Linux PPC support
2011-05-23 23:36:14 -04:00
6b5fffc026
compiler: preliminary implementation of tracking derived pointers in GC maps
2010-09-27 20:12:44 -07:00
1985705413
compiler: now that FFI has been deconcatenatized, we no longer need the special ##unary-float-function and ##binary-float-function fastpaths
2010-08-13 23:59:19 -07:00
b5fc39c198
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
168dd1f825
FFI rewrite part 7: compile callback bodies with the optimizing compiler
2010-07-28 00:49:26 -04:00
b23aac1beb
compiler.cfg: open-code parameter boxing and unboxing for certain C types
2010-07-19 10:25:13 -04:00
e5dd21b0b9
compiler.cfg: nuke ##allot-byte-array instruction
2010-07-16 19:57:45 -04:00
0fd636b4b9
compiler.cfg: ##unbox-long-long can have multiple outputs now, clean up long long parameter passing code using this
2010-07-15 19:49:29 -04:00
e27adb2830
compiler: re-architect low-level optimizer to allow more than one output value per instruction
2010-07-13 07:40:14 -04:00
a55c8ee671
FFI rewrite part 6: deconcatenatize
2010-07-02 15:44:12 -04:00
e86f434f26
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
806e54630a
GC maps for more compact inline GC checks
2010-06-11 20:06:00 -04:00
7fe04fa8ee
compiler.cfg.intrinsics.simd: intrinsic support for double-2>float-4 conversion
2010-06-01 00:34:50 -07:00
ba7cb61133
Stack allocation improvements
...
- New with-out-parameters combinator
- Inhibit tail call optimization in frames with local allocation, to ensure that passing a stack allocated value to the last word in the quotation works
- local allocations are now aligned properly
- spill slots are now aligned properly aligned in frames which have parameter and local allocation areas
2010-05-22 02:37:00 -04:00
91cd3b854d
Use ##local-allot to simplify longlong unboxing
2010-05-19 01:07:22 -04:00
77516c6932
compiler.cfg: generalize ##prepare-struct-caller instruction to alloca()-like ##local-allot
2010-05-19 00:33:15 -04:00
6eaf2d7534
cpu.x86: add x87 codegen for Model T enthusiasts
2010-05-17 11:35:47 -04:00
7ff309d50e
cpu.x86: get rid of this extra-stack-space nonsense
2010-05-17 02:14:44 -04:00
3f8e13bf66
compiler.cfg: refactor stack frame code and fix frame pointer usage in callbacks
2010-05-16 21:04:40 -04:00
a7b5957f6d
compiler: hack to make XCreateIC() work on x86-64
2010-05-16 04:49:53 -04:00
645164188d
compiler.cfg.instructions: ##select-vector instruction mapping to SSE4 PEXTR*
2010-05-16 00:50:19 -07:00
5b48cd2a63
FFI rewrite part 5: return value boxing and callback parameter boxing now uses vregs; simplify return value unboxing
2010-05-16 03:43:23 -04:00
8c878ada17
compiler.cfg.instructions: ##gather-int-vector-2/4 instructions that map to SSE4 PINSR/PEXTR
2010-05-15 23:48:22 -07:00
7fec53b509
compiler.cfg.instructions: ##bit-count insn
2010-05-15 13:57:35 -07:00
5a67711bfd
compiler.cfg: use x86 TEST instruction to optimize 'bitand 0 ='
2010-05-14 18:37:06 -04:00
3766abd65e
compiler.cfg.instructions: ##shuffle-vector-halves-imm insn to map to SSE's two-input SHUFPS/SHUFPD
2010-05-14 00:20:21 -07:00
42b0d456cd
Refactor x86-32 stack cleanup logic
2010-05-12 02:09:11 -04:00
f89b85db7b
Temporary fixes for x86-32 until FFI boxing is rewritten
2010-05-12 01:41:49 -04:00
ee0640f176
Move flatten-c-type to death row so that it can be executed in part 5
2010-05-11 22:26:31 -04:00
eb802208d1
FFI rewrite part 4: parameter and return value unboxing redesign
2010-05-11 19:11:44 -04:00
4478c3a51a
FFI rewrite part 3: eliminate ##push-context-stack and ##pop-context-stack
2010-05-10 00:46:29 -04:00
7316d41226
FFI rewrite part 2: use ##peek and ##replace instructions to access stack
2010-05-09 23:26:43 -04:00
c211c3e84e
FFI rewrite part 1: split up ##alien-invoke and friends into smaller instructions
2010-05-09 21:36:52 -04:00
f988dad79c
compiler.cfg: add ##load-float instruction for single precision floating point constants
2010-05-07 18:26:00 -04:00
6c35652666
cpu: cleanups
2010-05-03 23:08:24 -04:00
58e4381eb1
compiler: new "binary literal area" at the end of a word's machine code stores constant floats and SIMD vectors; this allows ##load-reference/##load-memory fusion to be performed on x86-64, with a RIP-relative address reaching the data; also simplifies VM since custom relocation types used by the previous 32-bit-only optimization are no longer needed
2010-05-03 17:34:35 -04:00
428ba9fd0f
compiler.cfg: fuse ##load-integer/##load-reference into ##replace to form ##replace-imm
2010-05-03 17:34:28 -04:00
9788323d25
compiler: add ##load-vector instruction to avoid wasting a temporary register on x86-32
2010-05-03 17:34:28 -04:00
95ff5ffe51
New GC checks work in progress
2010-05-03 17:34:16 -04:00
a9ccfa8f05
strings: move string-nth primitive out of the VM and into the library
2010-05-03 17:34:12 -04:00
0f5d9974a0
compiler.cfg.representations: fix various bugs
2010-05-03 17:34:11 -04:00
378c2b2a46
compiler.cfg: add ##load-memory and ##store-memory instructions implementing complex addressing modes, and associated value numbering optimizations
2010-05-03 17:34:07 -04:00
2475699736
compiler.cfg: more flexible addressing for ##slot and ##set-slot
2010-05-03 17:34:06 -04:00
913b95192e
compiler.cfg: merge all alien accessors into ##load-memory-imm and ##store-memory-imm
2010-05-03 17:34:06 -04:00
279eb461b1
compiler.cfg: remove ##set-string-nth-fast instruction since it can be expressed just as efficiently using other instructions
2010-05-03 17:34:04 -04:00
2ce926594d
compiler.cfg.value-numbering: new optimizations; reassociation for shifts and redistribution for shifts/multiplies over additions/subtractions
2010-05-03 17:34:02 -04:00
82fb1879af
Debugging untagged fixnums
2010-05-03 17:34:02 -04:00
2517b2fc2b
compiler: combine ##load-constant followed by ##alien-double into a ##load-double on x86-32, saving an integer register
2010-04-18 21:42:45 -05:00
Joe Groff
Erik Charlebois
Slava Pestov
Slava Pestov