I would like to know LLVM IR under different optimization options (e.g. -O0 -O1 …) What is the difference under different optimization options (e.g. -O0 -O1 …)?
Is it possible to use only some of the fixed instructions under the -O0 option?
I want to construct a VM to execute IR under the -O0 optimization, but I don’t want to analyze the full IR instructions.
Looking forward to your reply!
LLVM IR does not have a built-in concept of IR levels, the way some other IRs do (GEM, Open64).
There is already an IR interpreter, look at the lli tool, which might be helpful for your project.
Thank you for your reply.
I will try to use lli, also I would like to ask if there is a way to control the kind of instructions generated by IR?
I’m not sure I understand the question. Do you mean, is there a way to control the kind of IR instructions generated by Clang? In general, Clang will translate C/C++ code into IR instructions in a straightforward way, and the only way to affect what it produces is to give it different C/C++ input.
For example, if you give Clang a source file that does not use exceptions, you can expect that it will not emit any instructions related to exception handling (invoke, landingpad).
If you are trying to ask a different question, please give more details about your project.
Thank you for solving my problem.