Issue with circular buffer

[Radoslaw Nawrot]

Hi All,

I'm new at this forum. 

I encounter problem with circular_buffer.

Let me explain it on simple example (code is from circular_buffer.h):

 

 

I have a circular_buffer with few elements .

After read one of them element is destroyed:

 

template < typename T >
T
circular_buffer<T>::read()
{
  T t = read_data();

  buf_clear( m_buf, m_ri );
  increment_read_pos();

  return t;
}

 

template < typename T >

inline void

circular_buffer<T>::buf_clear( void* buf, int n )

    { (static_cast<T*>(buf) + n)->~T(); }

 

} // namespace tlm

 

... but

 

In dectructor  all elements are destroyed ... again

 

template < typename T >
circular_buffer<T>::~circular_buffer()
{
  for( int i=0; i < used(); i++ ) {
    buf_clear( m_buf, i ); //!!!!
  }
  buf_free( m_buf );
}
 

 

 

This case GPF's randomly.

I notice that when I change desctructor code 

to 

 

for( int i=m_ri; i < used() % m_size; i++ ) {

 

problem disappears

 

I'm using SystemC 2.3.0 Is this a bug in TLM sources or I'm doing something wrong ?

 

Radek

[Philipp A Hartmann]

AFAICS, this is another symptom of a known issue and fixed in 2.3.1. You should upgrade.

Quoting the RELEASENOTES:
 

•  The implementation-defined tlm::circular_buffer class has been updated with the following changes
  •     now provides a "clear()" member function to drop the current contents,
  •     fix a segmentation fault due to freeing invalid memory in "resize()", which could happen in some cases,
  •     work around a parsing error on some EDG-based C++ frontends.

 

 

Greetings from Duisburg,
  Philipp

[Philipp A Hartmann]

Oh, there seems to be still an issue in 2.3.1, you should change the circular_buffer::clear function to

template < typename T >
void
circular_buffer<T>::clear()
{
  for( int i=0; i < used(); i++ ) {
    // buf_clear( m_buf, i ); // << BUG HERE
    buf_clear( m_buf, (m_ri + i) % m_size ); // This should(tm) be correct
  }
  m_free = m_size;
  m_used = m_ri = m_wi = 0;
}

Thanks for reporting. I'll take this to the Language Working Group to make sure a fix will be included in the next release of the proof-of-concept simulator.

(I would still recommend to upgrade to 2.3.1... )

Greetings from Duisburg,
  Philipp

[Radoslaw Nawrot]

Philipp,

Thanks for fast answer. I 'll upgrade to  2.3.1 and make a correction in  sources ,

Thanks,

Radek