Closed martin-fink closed 3 years ago
Previously the value from memory would be loaded using the immediate's type. This would result in the following instruction
cmp [rdi + 8], 0
being raised to the following:
%0 = inttoptr i64 %arg1 to i8* %1 = load i8, i8* %0, align 1 ; loads only one of four bytes ...
After this commit, the raised bitcode looks like this:
%0 = inttoptr i64 %arg1 to i32* %1 = load i32, i32* %0, align 1 ...
Previously the value from memory would be loaded using the immediate's type. This would result in the following instruction
being raised to the following:
After this commit, the raised bitcode looks like this: