 Jonas Maebe
|
04d295f186
* first step towards supporting 32 bit targets with the LLVM code generator:
|
vor 6 Jahren |
 Jeppe Johansen
|
ceb38833f2
Added RiscV32/64 target, from a cleaned up version of branches/laksen/riscv/trunk.
|
vor 7 Jahren |
|
Jonas Maebe
|
4686f61002
* keep track of the temp position separately from the offset in references,
|
vor 7 Jahren |
 nickysn
|
518cdf9674
* replaced the saved_XXX_registers arrays with virtual methods inside
|
vor 7 Jahren |
|
nickysn
|
ddba821561
* GetNextReg(), used by 16-bit and 8-bit code generators (i8086 and avr) moved
|
vor 8 Jahren |
|
Jonas Maebe
|
a25ebbba3e
+ added volatility information to all memory references
|
vor 8 Jahren |
 florian
|
5595651816
* define avr => cpu8bitalu
|
vor 10 Jahren |
|
Jonas Maebe
|
0fc1fd6ac1
* replaced current_procinfo.currtrue/falselabel with storing the true/false
|
vor 10 Jahren |
 pierre
|
21fe7616a4
Avoid line longer than 255 characters, to be able to edit inside FP IDE
|
vor 10 Jahren |
|
Jonas Maebe
|
f4ada6cda0
+ aarch64-specific treference fields
|
vor 10 Jahren |
 sergei
|
d6c6c84f76
* calc_divconst_magic_unsigned: explicitly limit the shift amount to number of bits of target platform. This happens implicitly only on x86, on other 32-bit targets shifting by more than 31 bits calculates as zero.
|
vor 11 Jahren |
|
sergei
|
15bf176bc0
* Fixed building on x86_64 with 2.6.4, which generates unsigned division with sign-extended left for "(-qword) mod qword". Mantis #26336.
|
vor 11 Jahren |
|
sergei
|
d83fb7790e
* Fixed masking high bits when calculating 32-bit magic numbers for unsigned division on 64-bit CPU, namely on x86_64. Original powerpc64 code did calculations always in 64 bits, that's why incorrect masking went unnoticed.
|
vor 11 Jahren |
|
sergei
|
90d2009a31
* Moved procedures calculating "magic" numbers for division by constants from powerpc64/cgcpu.pas to cgutils.pas, so they can be reused for all targets.
|
vor 11 Jahren |
|
Jonas Maebe
|
908bbed312
* moved jvm-specific tarrayreftype from cgutils to jvm/cpubase
|
vor 11 Jahren |
 Károly Balogh
|
70f9b098e0
no longer define a segment field for m68k
|
vor 11 Jahren |
|
nickysn
|
af49e50a3c
* do not cause an internal error when location_reg2string is called on a
|
vor 11 Jahren |
|
nickysn
|
338d72f564
* extracted the code from psub.translate_registers for generating the name of
|
vor 11 Jahren |
|
nickysn
|
70649932de
* tlocation.value reverted back to longint on non 64-bit archs. This fixes the tset5.pp and tset5a.pp regressions on powerpc and probably other 32-bit platforms.
|
vor 12 Jahren |
|
nickysn
|
11f0b0b5aa
tlocation.value (for LOC_CONSTANTs) changed to TCGInt to avoid trouble on < 32-bit cpus
|
vor 12 Jahren |
|
florian
|
7361e19799
+ support for handling OS_128/OS_S128 on 64 Bit CPUs as far as needed for method pointers in registers
|
vor 13 Jahren |
|
Jonas Maebe
|
0fc422f244
* moved definition of maxcpuregister and tcpuregisterset from cgbase to
|
vor 13 Jahren |
|
Jonas Maebe
|
992cc352c6
* generalized handling of pointers to non-implicit pointer types:
|
vor 14 Jahren |
|
Jonas Maebe
|
2a6caa539c
+ support for constant array indices in treference for the JVM target
|
vor 14 Jahren |
|
Jonas Maebe
|
742f4ee12e
+ first (partial) implementation of thlcg for the jvm
|
vor 14 Jahren |
|
florian
|
4f82149dbd
* aint -> asizeint change
|
vor 14 Jahren |
|
Jonas Maebe
|
f13f6627c4
* moved use_fixed_stack from cgutils to a method in paramgr so it can
|
vor 15 Jahren |
|
florian
|
e5ebb2a26e
* started to integrate the code of David Zhang's MIPS code from fpc-mips on sf into fpc trunk
|
vor 15 Jahren |
|
Jonas Maebe
|
0eec337a00
- removed SUPPORT_UNALIGNED, FPC_SUPPORTS_UNALIGNED and FPC_UNALIGNED_FIXED
|
vor 16 Jahren |
|
florian
|
67cb961ed3
* only x86 and m68k need treference.segment
|
vor 16 Jahren |
