[llvm-dev] Function - replaceAllUsesWith

[Stefan Gränitz via llvm-dev]

Hi Björn

What does the IR code look like when you dump your module?
llvm::outs() << mainModule;

I guess for the external function "puts" it will have a declaration like
this:
declare i32 @puts(i8*)

When loading your module in the JIT, RuntimeDyld will resolve it to an
actual address. This lookup is done by name and it happens in a stage
called external symbol resolution. You should be able to inspect how
this works by setting a breakpoint here:
https://github.com/llvm/llvm-project/blob/cce1c2eb0e87e84be6345c870d3334f748bdc5bb/llvm/lib/ExecutionEngine/RuntimeDyld/RuntimeDyld.cpp#L1084

Hope it helps.
Stefan

Am 15.01.19 um 08:02 schrieb Gaier, Bjoern via llvm-dev:
>
> On Mon, Jan 14, 2019 at 3:21 AM Gaier, Bjoern via llvm-dev
> <llvm-dev at lists.llvm.org <mailto:llvm-dev at lists.llvm.org>> wrote:
>
>     Hello LLVM-World,
>
>      
>
>     currently I play around with a „llvm::Module” and an external
>     function defined there “puts”. Normally this function gets
>     resolved in the JIT-Process but I wonder about two things:
>
>      
>
>      1. Can I resolve the function already in this step? I used
>         “replaceAllUsesWith” and passed a “llvm::ConstantInt” to the
>         function. But this didn’t worked.
>
>  
>
> I'm not sure if I followed correctly, but you can't replace a function
> declaration with a constant int. The value for the function will be
> used (mostly) as an argument to the call instruction, which won't
> really be able to operate on an int. It isn't clear to me what you're
> trying to do exactly, replace the call to puts with an indirect call
> to the pointer value that correspond to puts in your current process?
> You would need to at least cast this pointer value to the right type,
> and use an indirect call.
>
>  
>
> Hey Mehdi,
>
>  
>
> I simply try overloading “puts” with a custom address before I start
> the JIT process. I want to replace the call to “puts” with my own
> address. For a test purpose I tried this now:
>               mainModue->getFunction("puts")->replaceAllUsesWith(
>
>                      llvm::ConstantExpr::getPointerCast(
>
>                           
> llvm::ConstantInt::get(llvm::IntegerType::get(context, 64), 0xF),
>
>                            mainModue->getFunction("puts")->getType()
>
>                      )
>
>               );
>
> This code will crash the application though. I know that 0xF is not a
> correct address and that there can be stuff optimized, but currently I
> just want to ‘see’ that “puts” was overloaded with 0xF.
>
>  
>
> Kind greetings
>
> Björn
>
>  
>
> Best,
>
>  
>
> -- 
>
> Mehdi
>
>  
>
>  
>
>  
>
>      1.  
>      2. What might happen if I have two modules and use
>         “replaceAllUsesWith” on a function of Module A, passing a
>         function of Module B.
>
>      
>
>     Thank you for the help in advance!
>
>      
>
>     Kind greetings
>
>     Björn
>
>     Als GmbH eingetragen im Handelsregister Bad Homburg v.d.H. HRB
>     9816, USt.ID-Nr. DE 114 165 789 Geschäftsführer: Dr. Hiroshi
>     Nakamura, Dr. Robert Plank, Markus Bode, Heiko Lampert, Takashi
>     Nagano, Takeshi Fukushima. Junichi Tajika
>
>     _______________________________________________
>     LLVM Developers mailing list
>     llvm-dev at lists.llvm.org <mailto:llvm-dev at lists.llvm.org>
>     http://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-dev
>
> Als GmbH eingetragen im Handelsregister Bad Homburg v.d.H. HRB 9816,
> USt.ID-Nr. DE 114 165 789 Geschäftsführer: Dr. Hiroshi Nakamura, Dr.
> Robert Plank, Markus Bode, Heiko Lampert, Takashi Nagano, Takeshi
> Fukushima. Junichi Tajika
>
> _______________________________________________
> LLVM Developers mailing list
> llvm-dev at lists.llvm.org
> http://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-dev

-- 
https://weliveindetail.github.io/blog/
https://cryptup.org/pub/stefan.graenitz@gmail.com

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.llvm.org/pipermail/llvm-dev/attachments/20190115/a5319a77/attachment-0001.html>