llvm dev - Jun 2011 - [LLVMdev] Function name changes in .bc on running llvm-g++

Hi

I am running llvm-g++ on a .cpp file to generate the llvm bitcode.

The function definition in the .cpp file is:

float pFL(Points *points, int *feasible, int numfeasible, float z, long *k,
      double cost, long iter, float e, int pid, pthread_barrier_t* barrier)
{

However, in the bit code I see that the function name has been modified to
the following

define float @_Z3pFLP6PointsPiifPldlfiP17pthread_barrier_t(%struct.Points*
%points, i32* %feasible, i32 %numfeasible, float %z, i32* %k, double %cost,
i32 %iter, float %e, i32 %pid, %union.pthread_barrier_t* %barrier) nounwind
{

What is the reason for the function renaming and is there a way to avoid so
that the function name remains unchanged.

This is how I am invoking llvm-g++
llvm-g++ -emit-llvm -c -O0 streamcluster.cpp -o streamcluster.bc

Thanks!
Malveeka
-------------- next part --------------
An HTML attachment was scrubbed...
URL:
<http://lists.llvm.org/pipermail/llvm-dev/attachments/20110603/c36224d8/attachment.html>

On 6/3/11 9:55 PM, Malveeka Tewari wrote:
> Hi
> 
> I am running llvm-g++ on a .cpp file to generate the llvm bitcode.
> 
> The function definition in the .cpp file is:
> 
> float pFL(Points *points, int *feasible, int numfeasible, float z, long *k,
>       double cost, long iter, float e, int pid, pthread_barrier_t*
> barrier) {
> 
> However, in the bit code I see that the function name has been modified
> to the following
> 
> define float
> @_Z3pFLP6PointsPiifPldlfiP17pthread_barrier_t(%struct.Points* %points,
> i32* %feasible, i32 %numfeasible, float %z, i32* %k, double %cost, i32
> %iter, float %e, i32 %pid, %union.pthread_barrier_t* %barrier) nounwind {
> 
> What is the reason for the function renaming 
You wrote it in C++, that's why! You see, C++ supports something called
"function overloading", where you can define two or more functions
with
the same name but different sets of arguments. That name--which is said
to be "mangled" for obvious reasons--is so overloaded functions
don't
conflict with each other.
>and is there a way to avoid
> so that the function name remains unchanged.
Yeah... don't write it in C++.

Chip

Charles Davis wrote:
> On 6/3/11 9:55 PM, Malveeka Tewari wrote:
>> Hi
>>
>> I am running llvm-g++ on a .cpp file to generate the llvm bitcode.
>>
>> The function definition in the .cpp file is:
>>
>> float pFL(Points *points, int *feasible, int numfeasible, float z, long
*k,
>>        double cost, long iter, float e, int pid, pthread_barrier_t*
>> barrier) {
>>
>> However, in the bit code I see that the function name has been modified
>> to the following
>>
>> define float
>> @_Z3pFLP6PointsPiifPldlfiP17pthread_barrier_t(%struct.Points* %points,
>> i32* %feasible, i32 %numfeasible, float %z, i32* %k, double %cost, i32
>> %iter, float %e, i32 %pid, %union.pthread_barrier_t* %barrier) nounwind
{
>>
>> What is the reason for the function renaming
> You wrote it in C++, that's why! You see, C++ supports something called
> "function overloading", where you can define two or more
functions with
> the same name but different sets of arguments. That name--which is said
> to be "mangled" for obvious reasons--is so overloaded functions
don't
> conflict with each other.
>> and is there a way to avoid
>> so that the function name remains unchanged.
> Yeah... don't write it in C++.
Wrapping the function in an extern "C" { ... } block also works.

Nick