[llvm-dev] [RFC] Nios II backend

[Hal Finkel via llvm-dev]

On 04/12/2017 07:49 AM, Joerg Sonnenberger via llvm-dev wrote:
> On Wed, Apr 12, 2017 at 10:20:30AM +0000, Grischenko, Andrei L via llvm-dev wrote:
>> I am proposing the integration of a backend targeting Nios II processor architecture.
> Nice to hear that.

+1

  -Hal

>
>> Plan for initial implementation corresponding to patches (obviously the plan can be changed after comments)
>> - Target registration
>> 	a)  Nios2.td, Nios2.h, Nios2TargetMachine.h, Nios2TargetMachine.cpp, Nios2Subtarget.h, Nios2Subtarget.cpp, TargetInfo/Nios2TargetInfo.cpp
>> 	b) MCTargetDesc/Nios2MCTargetDesc.h, MCTargetDesc/Nios2MCTargetDesc.cpp...
>> - Register information (Nios2RegisterInfo.cpp, Nios2RegisterInfo.h, Nios2RegisterInfo.td)
>> - ISA specification (Nios2InstrFormats.td, Nios2InstrInfo.cpp, Nios2InstrInfo.h, Nios2InstrInfo.td)
>> - GlobalISel (At the moment I use DAG selector, but work on porting to GlobalISel is already started)
>> - Assembly printer (InstPrinter/Nios2InstPrinter.h, InstPrinter/Nios2InstPrinter.cpp, Nios2AsmPrinter.cpp, Nios2AsmPrinter.h, MCTargetDesc/Nios2MCAsmInfo.cpp, MCTargetDesc/Nios2MCAsmInfo.h, .)
>> - Target triple registration in clang
>> - Built-in functions (clang and codegen part)
>> - New options support (clang driver and codegen part)
>> - Direct ELF streaming (binary emission)
> As discussed in the recent target review threads, I would strongly
> recommend doing the assembly printer and even possibly the ELF emission
> as early as possible. Most of the other changes like the ISA
> specification and register information are more difficult to test
> without them and LLVM has a very strong preference for testing for all
> new code.
>
> Joerg
> _______________________________________________
> LLVM Developers mailing list
> llvm-dev at lists.llvm.org
> http://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-dev

-- 
Hal Finkel
Lead, Compiler Technology and Programming Languages
Leadership Computing Facility
Argonne National Laboratory