Added -v-cg switch, which right now just prints "codegen: module.name (module/name.d)" to stdout, this can really help figuring out where, in some complex build command, things go wrong.
In particular, %.nest_arg is always aligned even though it's bitcast from i8*.
Pointers in vtables are also guaranteed to be stored at aligned addresses.
- Moved main() into its own file gen/main.cpp
- Fixed basic cross compilation
- removed the option for setting OS
- added support for llc's mattr, mcpu and mtriple switches
- added basic ABI abstraction for return value rewrites, it's not perfect and will probably be completely rewritten once I get to handling parameter rewrites as well.
- x86-64 extern(C) abi for cfloat returns now match (llvm-)gcc.
Applied patch from ticket #129 to compile against latest LLVM. Thanks Frits van Bommel.
Fixed implicit return by asm block at the end of a function on x86-32. Other architectures will produce an error at the moment. Adding support for new targets is fairly simple.
Fixed return calling convention for complex numbers, ST and ST(1) were switched around.
Added some testcases.
I've run a dstress test and there are no regressions. However, the runtime does not seem to compile with symbolic debug information. -O3 -release -inline works well and is what I used for the dstress run. Tango does not compile, a small workaround is needed in tango.io.digest.Digest.Digest.hexDigest. See ticket #206 .
Changed -g switch to emit DW_LANG_D debug info, make demangling work with a patched GDB, still more work to do for full support of D's Dwarf extensions.
Added getNullValue helper method.
Fixed function literals in static initializers.
Changed alignment of delegates from 2*PTRSIZE to just PTRSIZE.
Changed errors to go to stderr instead of stdout.
Fairly major rewriting of struct/union/class handling, STILL A BIT BUGGY !!!
Changed runtime functions taking opaque[] to void[].
Implemented proper type painting, to avoid "resizing" array casts in runtime calls that previously took opaque[].
Implemented dynamic arrays as first class types, this implements proper ABI for these types on x86.
Added dwarf region end after call to assert function, fixes some problems with llvm not allowing this to be missing.
Reverted change to WithStatement from rev [704] it breaks MiniD, mini/with2.d needs to be fixed some other way...
Fixed tango bug 1339 in runtime, problem with _adReverseChar on invalid UTF-8.
Disabled .bc generation in the compiler runtime part, genobj.d triggers some llvm bug when using debug info. the .o seems to work fine.
I've implemented it this way since it doesn't require any changes in the
frontend. However, I think having TypeTuple expressed as LLVM struct types
would make much more sense and open the door to tuple lvalues.
Reimplemented support for nested functions/class using a new approach.
Added error on taking address of intrinsic.
Fixed problems with the ->syntaxCopy of TypeFunction delegate exp.
Removed DtoDType and replaced all uses with ->toBasetype() instead.
Removed unused inplace stuff.
Fixed a bunch of issues in the runtime unittests, not complete yet.
Added mini tests.
Removed the -fp80 option and made real be 80bit floats on X86, this is what the D spec really says it should be and fixes a bunch of issues.
Changed the handling of parameter attributes to a bit more generalized approach.
Added sext/zext attributes for byte/short/ubyte/ushort parameters, fixes#60 .
Parameter attribs now properly set for intrinsic calls if necessary.
Made the tango.math.Math patch less intrusive.
Fixed/added some mini tests.
Fixed issues with DMD generated assert statements when using class invariants, generally due to incomplete ASTs.
Removed some dead code.
Added a few comments.
Changed the way moduleinfo is registered to use the same approach as DMD, this eliminates the need for correct linking order and should make the way for using a natively compiled runtime library. This should speed up linking tremendously and should now be possible.
Fixed the llvm.used array to only be emitted if really necessary.
Changed the tryfinally handlers to a more generalized EnclosingHandler.
Changed ClassInfoS to be mutable so they can be used as locks.
Added new BB after throw ala return/break etc.
Added: started support for x86 80bit floating point.
Changed: aggregates passed by value now use the llvm 'byval' parameter attribute, also lays ground work for
using other attributes.
Changed: eliminated a lot more std::vectorS, these showed up pretty much at the top when profiling!
Changed: performed other misc. cleanups.
Changed: halt expression now call the new llvm trap intrinsic instead of an assert(0).
Changed: dstress suite now passes -O0 by default, this only eliminates unreferenced globals, which speeds up
linking quite a bit.
Fixed a bunch of the built-in typeinfos for arrays, they did not inherit TypeInfo_Array.
Applied patch to tango/text/convert/Layout.d by fvbommel, closes#47 .
Cleaned up some type code.
Replaced uses of llvm::Type with LLType (a typedef), same for Value and Constant.
Fixed a few cases where typeinfo for user structs could be emitted multiple times, seems to still be some cases of this :/
Changed: removed the crappy realloc based dynamic memory runtime and started moving over to DMD style runtime support, part of moving to real GC.
Fixed: dynamic arrays now use GC runtime for allocating memory.
Fixed: new expression now use GC for allocating memory.
Changed: revamped the dynamic array support routines related to dynamic memory.
Fixed: assertions no longer create exsessive allocas.
Changed: misc. minor cleanups.
* Fixed a minor linking order mishap *
* Added an command line option -annotate *
* Fixed some problems with running optimizations *
* Added std.stdio and dependencies to lphobos (still not 100% working, but compiles and links) *
* Fixed problems with passing aggregate types to variadic functions *
* Added initial code towards full GC support, currently based on malloc and friends, not all the runtime calls the GC yet for memory *
* Fixed problems with resolving nested function context pointers for some heavily nested cases *
* Redid function argument passing + other minor code cleanups, still lots to do on this end... *
Fixed some problems with the string runtime support functions.
Fixed initialization of array of structs.
Fixed slice assignment where LHS is slice but RHS is dynamic array.
Fixed problems with result of assignment expressions.
Fixed foreach problems with key type mismatches.
Basically it tries to do the following in order: Resolve types, Declare symbols, Create constant initializers, Apply initializers, Generate functions bodies.
ClassInfo is now has the most useful(biased?) members working.
Probably other stuf...
Split up declaration, constant initializer gen and definition for globals, structs, classes and functions.
Improved ClassInfo support (not complete), not in vtable yet.
Fixed a bunch of forward reference problems.
Much more. Major commit! :)
Added support for array .sort and .reverse properties.
Fixed some bugs with pointer arithmetic.
Disabled some DMD AST optimizations that was messing things up, destroying valuable information.
Added a KDevelop project file, this is what I use for coding LLVMDC now :)
Other minor stuff.
