felixf
|
03ee1f7202
patched stack alignment bugs
|
7 éve |
felixf
|
c5d3f8fcd4
added missing variable
|
7 éve |
felixf
|
deee03e015
support for conversion UNSIGNED --> FLOAT
|
7 éve |
eth.negelef
|
780c02bf0d
Added support for comparing complex unsigned values
|
7 éve |
eth.negelef
|
1c05ad7c45
Use volatile register in 64-bit case for initialising stack frame
|
7 éve |
eth.negelef
|
68506774db
Fixed direct adressing in 64-bit mode for dynamic module loading
|
7 éve |
negelef
|
9e591b1329
Fixed hard-coded register width in copy and fill instruction
|
7 éve |
felixf
|
dd3d5477c3
patched all method visibility conflicts: extending methods keep visibility of super method.
|
7 éve |
felixf
|
3b3e542b91
Corrected 64bit memory size for pop float
|
7 éve |
felixf
|
5b35c0cf7a
removed unnecessary register shift by one (XMM0 is the first floating point register passed in A64bit WINAPI calling convention)
|
7 éve |
felixf
|
81fc79f11e
Improved and simplified parameter register handling:
|
7 éve |
felixf
|
4e4e439576
Removed priority from generic sections (replaced by section types)
|
7 éve |
felixf
|
dad7129d9a
Added back in support for inline assembler using CODE .. WITH IN .. OUT .. END
|
7 éve |
felixf
|
2a4118ee58
patch in register allocator (64 bit)
|
7 éve |
felixf
|
d0805e8e34
Use register D for initialiuation (instead of register C : conflict with coop kernel !)
|
7 éve |
felixf
|
9dc297a69d
Finally: 64-bit A2 on Windows boots and works with the Garbage Collector
|
7 éve |
felixf
|
b38157c723
corrected push of floating point values less wide than address size
|
7 éve |
felixf
|
90d4e0f347
64 bit A2 conversions:
|
7 éve |
felixf
|
4f19b4ed93
Same register allocation scheme for 32 and 64 Bit.
|
7 éve |
felixf
|
ddba57bf50
Return (WinAPI 64 bit) needs a 0
|
7 éve |
felixf
|
5ba63086f9
WinAPI calling convention (shadow registers to stack)
|
7 éve |
felixf
|
d362c8f14e
Minor patch: avoid a trap when finding PC and compilation cancels
|
7 éve |
felixf
|
c982c62f41
revert optimistaion of plus
|
8 éve |
felixf
|
64a9e39e25
Allow array 4 of char in XMM registers (for color blending)
|
8 éve |
felixf
|
af37ecfd9a
Added support for 4-vectors with 8/16 bit entries.
|
8 éve |
felixf
|
6849ac77e4
Improved error handling: line/column position now available everywhere
|
8 éve |
felixf
|
e0864e2384
Also treated line numbers correctly in assembler code
|
8 éve |
felixf
|
66b9af235a
Patched copy&paste error
|
8 éve |
felixf
|
f16b38f79e
Improved emission of MOD when power of two
|
8 éve |
felixf
|
a3d159abe8
Added optimization: replace mul / div by shift if possible
|
8 éve |