llvm dev - Apr 2013 - [LLVMdev] code gen variable mapping

Since variables in the source code are renamed in the IR, I wanted to ask you
how and where this mapping is done in Clang.I am interested in converting the
variable names in some C strings to the one in the IR.For example if at a
certain point of the program I have a string  like "x>0" I want to
change it in "%x>0" (I already implemented a parser to recognize
the identifiers so I only need to translate the name).If you have any suggestion
in how to do it easily (like if there is already a data structure or a method to
call) it will be really appreciated.
this is essentially hopeless.  Most names in LLVM IR have no meaning, they are
just there to make debugging easier: they can be removed altogether and all will
still work fine.  Even if the front-end is kind enough to set a name, optimizers
can change them.  Why do you want this name mapping?  Maybe you should be using
debug info instead?

I am interested in using the information a programmer can put in some annotation
comments in order to improve the optimization on the source code.Since the
programmer will write those annotations using the source code names, I need to
find a way to carry on these comments in the IR generation in order to make them
available for later optimization passes. I thought about attaching those comment
as a metadata string to the next instruction after the comment.I am new to LLVM
and by reading the .s output files I thought that clang does some kind of
renaming to the variables. Do you think there is an easier way to do that?
Thank you again,
Niko
 		 	   		  
-------------- next part --------------
An HTML attachment was scrubbed...
URL:
<http://lists.llvm.org/pipermail/llvm-dev/attachments/20130401/eb4eb499/attachment.html>

On Mon, Apr 1, 2013 at 1:21 PM, Niko Zarzani <koni10 at hotmail.it>
wrote:
>
> Since variables in the source code are renamed in the IR, I wanted to ask
> you how and where this mapping is done in Clang.
>
> I am interested in converting the variable names in some C strings to the
> one in the IR.
>
> For example if at a certain point of the program I have a string  like
"x>0"
> I want to change it in "%x>0" (I already implemented a parser
to recognize
> the identifiers so I only need to translate the name).
>
> If you have any suggestion in how to do it easily (like if there is already
> a data structure or a method to call) it will be really appreciated.
>
>
> this is essentially hopeless.  Most names in LLVM IR have no meaning, they
> are
> just there to make debugging easier: they can be removed altogether and all
> will
> still work fine.  Even if the front-end is kind enough to set a name,
> optimizers
> can change them.  Why do you want this name mapping?  Maybe you should be
> using
> debug info instead?
>
>
> I am interested in using the information a programmer can put in some
> annotation comments in order to improve the optimization on the source
code.
> Since the programmer will write those annotations using the source code
> names, I need to find a way to carry on these comments in the IR generation
> in order to make them available for later optimization passes.
If you parse the annotations & actually form references to the
original variables, you may be able to maintain that mapping during
Clang's IRGen, lowering your annotations into something LLVM can work
with (probably in some sort of LLVM metadata).
> I thought about attaching those comment as a metadata string to the next
> instruction after the comment.
> I am new to LLVM and by reading the .s output files I thought that clang
> does some kind of renaming to the variables.
> Do you think there is an easier way to do that?
>
> Thank you again,
>
> Niko
>
> _______________________________________________
> LLVM Developers mailing list
> LLVMdev at cs.uiuc.edu         http://llvm.cs.uiuc.edu
> http://lists.cs.uiuc.edu/mailman/listinfo/llvmdev
>