minor style fixes to eliminate clang's warnings
extras: missing semicolon in sed command broke myjit-unique script
arm32: fixed issues when dealing with large immediate values
raspi: forced clear cache after moving code into its final buffer
documentation: updates reflecting recent API changes.
arm64: fixed calling convention to be compliant with Aapcs64 on non-apple hardware
arm64: fixed calling convention on Darwin/Apple Silicon
putarg[ir] has an additional operand to determine the size of the argument
win64: debugging features
win64: support for variadic functions
new operation 'jit_prepare_optional' signaling variadic arguments
fixed handling of caller/callee-saved registers
win64: initial support
win64: tests has been adapted to support Win64's LLP64 data model
arm64: fixed missing header files on apple silicon
fixed position of a macro defining a source type
win32: makefiles and test update
fixed preprocessor directives for unixes
win32: initial support
aarch64: fixed memory protection on Apple Silicon
initial changes that will allow better support for multiple operating systems
personal note
refactoring towards compiling MyJIT from multiple object files
i386: single precision FP arguments are converted to doubles in the prolog; misc. bugfixes
AMD64: single precision float arguments converted to doubles in the prolog of the function
ARM64: more efficient handling single precision FP arguments
arm64 port
RET ops considered as leaf-nodes again in the flow analysis
correct (unoptimized) register mapping for branch ops
correct stack space allocation when mixing floats and integers
fixed FP tests and simplified FRET on amd64
fixed macOS compilation issue
new tests for floats
buffer overflow test for DATA_BYTES
introduced output buffer for debug outputs
fixed buffer overflow caused by large data_bytes
jitlib.h: explicit casts for some branch-ops
You're right. Thank you for pointing this out and for the bugfix.
documentation update w.r.t. to new release
AMD64: fixed bugs discovered with new tests
documentation update w.r.t. to new release
AMD64: fixed bugs discovered with new tests
removed unused code
ARM32: more tests
README file update
ARM32: minor bugfixes
documentation update
ARM32: complete disassembler - partial vfp support
ARM32: codegen cleanup, Adv SIMD instructions r...
ARM32: extended tests
ARM32: unsigned variants of div and mod, fixed ...
ARM32: more tests
ARM32: fixed FST, FLD, FSTX, FLDX operations wi...
ARM32: removed unused code
ARM32: various bugfixes and optimizations, new ...
ARM32: passing single precision values via stack
ARM32: complete function calls and argument pro...
ARM32: unified algorithm for scheduling input a...
ARM32: adjusted tests, w.r.t. variadic function...
ARM32: new register scheduling algorithm
ARM32: stack alignment
ARM32: new tests for argument passing
ARM32: rounding and minor fixes: stack + reg. a...
ARM32: initial support for processing floating ...
ARM32: extr
ARM32: loads and stores, branches, complete ari...
ARM32: initial support for FP values
ARM32: MUL optimizations
documentation: correct pseudocode for unary ope...
ARM32: mostly complete disassembler
ARM32: initial implementation of the integrated...
ARM32: transfer operations
ARM32: memset
ARM32: memcpy polishing
ARM32: memcpy
ARM32: data referecnes, jmpr
tests: new test computing factorial recursively
ARM32: argument passing and function calls
fixed allocator bug causing unnecessary loads (2)
fixed allocator bug causing unnecesary loads
ARM32: addressing optimizations and refactoring...
ARM32: improved processor detection
ARM32: function pointer to value conversion
ARM32: cleanup and bugfixes in code generator
ARM32: bugfixes and some code polishing
ARM32: fixed race condition in mprotect
ARM32: registers, stack allocation, modulo
ARM32: debugging, i.e., msg and trace
ARM32: assignments, loads and stores, branches
ARM32: conditionals, multipliction + division, ...
ARM32: immediate values, word loads, more arith...
ARM32: assginments, basic arithmetics, ineffici...
ARM32: new code generation macros
ARM32: fixed code listing
Initial infrastructure for ARM32
TOUCH op. forces register association
all operations are marked as 'in_use' by default
x86: fixed bug in modi optimization
dead-code elimination switch
aliases for common forms of jit_add_op