 florian
|
0dc7bfde42
* patch by Christo Crause: support for megaAVR 0 and tinyAVR 0/1, #36616, part 1/3
|
5 years ago |
|
florian
|
3ee32648c0
* system unit can be build for subarch avrtiny
|
5 years ago |
|
florian
|
4f5b872098
* patch by Christo Crause: avrtiny26 controller is not subarch avr25 but avr2, resolves #36620
|
5 years ago |
|
florian
|
0a66f8f1dd
* patch by Christo Crause: Correct subarch for avrtiny series in compiler, resolves #36619
|
5 years ago |
|
florian
|
8917c66151
* AVR: use CP ...,r1 instead of CPI ...,0 to enable all registers being used as first operand
|
5 years ago |
|
florian
|
ce336a08c5
* omit first not, helps the reg. allocator to find a workable colouring, resolves #36611
|
5 years ago |
|
florian
|
23d26d495f
* AVR: attiny can do post increment operations, so remove code working around this
|
5 years ago |
|
florian
|
6d2880e809
* AVR: do not insert jmp if not supported
|
5 years ago |
|
florian
|
0add128d7b
+ AVR: initial support of the avrtiny architecture in the rtl
|
5 years ago |
|
florian
|
bee41253bd
+ AVR: initial support for the avrtiny architecture
|
5 years ago |
|
florian
|
150fac54cb
* changes shift parameter of shift helpers into ALUUInt, as there is an "and" used anyways, this does not change semantics but reduces register pressure on 8 bit targets
|
6 years ago |
|
florian
|
4a73c31c82
* AVR: tcgavr.normalize_ref must always load the reference into tmpreg
|
5 years ago |
|
florian
|
a9bbcccb5d
+ AVR: GetDefaultZeroReg and GetDefaultTmpReg
|
5 years ago |
|
florian
|
cda78c398f
+ AVR: cputype cpu_avrtiny
|
5 years ago |
|
florian
|
3328f991c7
* AVR: fixes pushing and handling of stack passed parameters
|
5 years ago |
|
florian
|
fbb3fe56a9
* AVR: fixed TDwarfAsmCFILowLevel.generate_initial_instructions
|
5 years ago |
|
florian
|
201a89504e
* patch by Christo Crause: reg. allocation information if non-volatile registers
|
5 years ago |
|
florian
|
b04f398d27
* avr: help the compiler with constant loading to avoid ie 200309041
|
5 years ago |
|
florian
|
3b5dfbf79c
* avr: more precise execution weigth for shift loops
|
5 years ago |
|
florian
|
b241fec17b
* avr directly encodes constant shifts of 64 bit values
|
5 years ago |
|
florian
|
647c5889de
* simplified fpc_mul_qword so it compiles again on avr, fixes #36236 but does not solve the root of the problem
|
5 years ago |
|
florian
|
63375effda
* patch by Christo Crause: fix for r43136
|
6 years ago |
|
florian
|
8e68ea69cb
* (modified/re-formatted) patch by Christo Crause: AVR: Optimizing code generation for shift with compile time constant
|
6 years ago |
|
florian
|
bf85e51962
* AVR: patch by Christo Crause: fixes tcgavr.g_concatcopy for data with size>255, resolves #36066
|
6 years ago |
 Jeppe Johansen
|
821a83f735
* Fix exitcode handling for avrsim target. Patch by Christo Crause
|
6 years ago |
|
florian
|
97b0bda68d
* fix tcgavr.g_flags2reg for sizes > 1 byte
|
6 years ago |
|
Jeppe Johansen
|
d14e3e96a7
- Fix bug in software overflow checking for longint's.
|
6 years ago |
|
Jeppe Johansen
|
0c4d299514
AVR: Add support for using nostackframe in interrupt routines.
|
6 years ago |
|
Jeppe Johansen
|
49ac4e0ed3
AVR: Add optimizations for sign testing, and a better Abs() implementation.
|
6 years ago |
|
Jeppe Johansen
|
bca0d5774f
- Adds a number of optimizations for 64bit integer operations on AVR. Patch from Christo Crause in issue #35691.
|
6 years ago |
