Introduce a hidden `-dwarf-version=<uint>` switch to support old gdb versions
used by Travis. Older versions only supporting DWARF-2 and/or not supporting
TLS correctly are detected in dmd-testsuite's CMakefile.
- byte,ubyte should be encoded as DW_ATE_signed_char/DW_ATE_unsigned_char, not DW_ATE_signed/DW_ATE_unsigned
- encode wchar,dchar as UTF
- do not declare struct/class definitions as FlagFwdDecl
- pass deco as unique identifier to UDTs
- dynamic arrays: use type name, don't leave it empty
- use proper type name for delegates
- add support for type "void" and "typeof(null)"
- Fix the value of closure variables in the parent function. Previously the passed value was the GEP instruction into .frame, and that instruction corresponds to a register so resulted in "DWARF-2 expression error: DW_OP_reg operations must be used either alone(...)" (also the address was incorrect anyway because the offset would get doubled by 'addr')
- When emitting debug info for a parent function parameter that is accessed from a nested function, call DIBuilder.createAutoVariable instead of createParameterVariable. Two different variables for the same parameter index was triggering an assert in llvm::DbgVariable::addMMIEntry. This fixes the compilation of druntime and phobos when building against LLVM 3.8.
Also adds the CMake infrastructure to compile and link the D source files.
The build is partially broken:
- A few files in Phobos and druntime do not build
- MSVC build is broken because of unresolved symbols involving reals
Now with *almost* working EH codegen. Does not compile Phobos yet
because we run into the "instruction does not dominate all uses"
issue when an r-value result of toElemDtor is used and we need to
run cleanups in between. Should easily be fixed by promoting those
values to allocas.
Most of the changes outside of ir/irfunction.{h, cpp} are just
because CreateCallOrInvoke moved locations. I took the
opportunity to also make use of the different arg count
overloads where possible.
This is likely an LLVM bug: The code in InlineFunctions.cpp
only updates the source location metadata for the instructions
in the inlined callee when the call site has a source location
set. When the caller has no location metadata, the scope for
the inlined instructions thus still points to the DISubprogram
for the original callee. This then leads to an assertion during
codegen.
GitHub: Fixes#998.
