I'm trying to run a program from the LLVM bitcode generated by my compiler, but when I run the lli
command it returns an error
lli-3.6: test2.ll:9:1: error: expected instruction opcode
When I use the lli
with the .ll generated by clang -S -emit-llvm
it works. There are many optimizations in this code, though. I tried to insert some of them manually, but it didn't work.
My problem is that I don't know if the structure of my code is correct, or if it is just missing something particular for the interpreter to work properly. Originally, I tried to use the JIT in the code, but it was giving me more errors with the libraries and the documention was not helpful.
My llvm bitcode is the following:
%struct.test = type { i32, i32 }
define internal void @test_program() {
entry:
%a = alloca i32
store i32 5, i32* %a
call void @printf(i32 3)
%bar = alloca %struct.test
}
define internal void @f(i32 %x) {
entry:
%b = alloca i32
%mul = mul i32 6, 2
%add = add i32 %mul, 3
%add1 = add i32 10, %add
store i32 %add1, i32* %b
%tmp_eq = icmp eq i32* %b, i32 25
br i1 %tmp_eq, label %cond_true, label %cond_false
cond_true: ; preds = %entry
store i32 40, i32* %b
cond_false: ; preds = %entry
store i32 50, i32* %b
}
declare void @printf()
The LLVM IR file you provided is malformed - it's missing terminator instructions on its basic blocks (except on %entry
in @f
). It looks like you have a bug in your custom optimizations.