[llvm-dev] Dumping compiled module as LLVM IR bitcode after certain optimization passes

[Mehdi Amini via llvm-dev]

Hi,

Usually we disable the llvm optimizations (clang -mllvm disable-llvm-optzns ) and then use opt to run the passes till the point of interest. I don’t know of a shortcut in clang (the new PassManager will make it possible to pass an arbitrary pipeline with a string, but not there yet).

— 
Mehdi

> On Dec 31, 2016, at 11:43 AM, Alex Susu via llvm-dev <llvm-dev at lists.llvm.org> wrote:
> 
>  Hello.
>    Is there a possibility to instruct at command line the clang or the opt tools to emit LLVM bitcode file after a certain optimization pass such as LoopVectorize (or all optimization passes)? (This question was asked also at http://stackoverflow.com/questions/17904440/dump-ir-after-each-llvm-optimization-each-pass-both-llvm-ir-passes-and-backen .)
>    As far as I can see giving:
>      clang -save-temps ...
>    generates a .bc file for the compiled module just before passing the module to the optimization passes (opt).
> 
>    An alternative I have is to programmatically generate the corresponding bitcode file at the right time after finishing the pass in discussion - see, for example, http://stackoverflow.com/questions/13903686/writing-module-to-bc-bitcode-file, methods WriteBitcodeToFile() or WriteModule() .
> 
>  Thank you,
>    Alex
> _______________________________________________
> LLVM Developers mailing list
> llvm-dev at lists.llvm.org
> http://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-dev