Jump to content

Philipp A Hartmann

Members
  • Content Count

    524
  • Joined

  • Last visited

  • Days Won

    126

Posts posted by Philipp A Hartmann


  1. Actually, it is expected that you don't have the QuickThreads package built on MinGW.  As you can see from your config.log, the threading package to use is

    Quote

    ---------------------------------------------------------------------
    Configuration summary of SystemC 2.3.3 for x86_64-pc-mingw64
    ---------------------------------------------------------------------

    ...

     Build settings:
       Enable compiler optimizations  : yes
       Include debugging symbols      : no
       Coroutine package for processes: WinFiber
       Enable VCD scopes by default   : yes
       Disable async_request_update   : no
       Phase callbacks (experimental) : no
    --------------------------------------------------------------------

    But it seems that the source code is missing an explicit check for _WIN64 to catch the MinGW-64 case. Can you try compiling with WIN64 defined (without leading underscore):

    ../configure CXXFLAGS="-DWIN64"

    Hope that helps,
      Philipp


  2. This compiler warning is a false positive.

    There is a loop in sc_fifo<T>::read(T&) ensuring that the fifo is not empty (and the success of the nb_read(T&) is even guarded by an sc_assert😞

        while( num_available() == 0 ) {
    	sc_core::wait( m_data_written_event );
        }
        bool read_success = sc_fifo<T>::nb_read(val_);
        sc_assert( read_success );

    The check for num_available() is even stricter than the check in buf_read, but I can imagine that some compilers might not be able to prove this invariant.  Therefore, unconditionally initializing the local variable to silence the warning might be an acceptable trade-off.


  3. sim_input.h:344 and sim_sync.h:29 points to your code and the issue reported from sc_spawn_object is very likely caused via inlining from your Tasker::MethodFunction<> class.  So you will need to look into your code to address these particular reports.

     For SystemC 2.3.1, the are some known Valgrind reports, most of which should have been addressed in SystemC 2.3.3.

    Greetings from Duisburg,
      Philipp


  4. Hi Wim,

    On 11/25/2019 at 5:41 PM, wvandamm said:

    This seemed to imply to me that it is possible to have a cci_param which contains a cci_value_list or cci_value_map with entries that are cci_values and hence have individual handles

    Currently, cci_param<T> indeed does not support any of cci_value, cci_value_list, cci_value_map as value type T. The main reason is, that it would otherwise cause ambiguities in the API of cci_value(_list,_map) itself, if the generic conversion functions (which are used by cci_param internally) would support this.  However, it might be possible to extend cci_param<T> to support this scenario explicitly.

    Greetings from Duisburg,
      Philipp


  5. As the error message says, you have a failing assertion in your testbench code in adder_testbench.cpp, line 30 (emphasis mine):

    20 hours ago, Avnita said:

    simv: /home/avnita/Workspace/SYSTEM_C_FILES/adder_testbench.cpp:30: void testbench::process(): Assertion `COUT == SC_LOGIC_1' failed.

    There is not much we can do about this without knowing your testbench and your model.

    Hope that helps,
       Philipp


  6. Thanks for the report.  I don't fully understand the summary, though.  Some questions:

    Are the changes to the API check and/or the exception handling in sc_simcontext.cpp really needed? I would hope that removing/skipping the assert in sc_cor_qt.cpp is sufficient to work around the mprotect restrictions on CentOS 7+?

    Your instructions do not include --enable-shared=no, but your description says that you only(?) see it on a static SystemC library. Can you please clarify?

    To better understand the failing mprotect call in your environment, can you please provide the value of errno after the call?  This can be obtained by adding something like:

    #include <errno.h>
    // ...
    
    ret = mprotect( ... );
    if( ret != 0 ) {
      int mprotect_errno = errno;
      printf( "mprotect errno: %d, %s\n", mprotect_errno, strerror(mprotect_errno) );
    }

     

    IIRC, Linux generally allows calling mprotect on allocated memory.  The memory needs to be properly aligned at a page boundary, of course.  One option might be to allocate the stack memory via posix_memalign (if available) instead of new.  We can also change the implementation to gracefully ignore a failing protection and only restore the permissions if the mprotect(...,PROT_NONE) call was successful earlier.

    This brings me to my remaining question: Which one of the mprotect calls actually fails: Is it the one removing the protection (PROT_NONE) or the one trying to restore them?

    Thanks,
      Philipp


  7. Hi Rainer,

    thanks for your detailed report.  The issue seems indeed an incomplete implementation of the CCI 1.0 LRM.  I'll forward your finding to the Accellera CCI working group.

    To achieve the expected behavior, you can locally change the from_json implementation to something like:

    inline cci_value
    cci_value::from_json( std::string const & json )
    {
      cci_value v;
      bool ok = v.json_deserialize( json );
      if( !ok ) {
        // report CCI_VALUE_ERROR instead of using sc_assert
        v.report_error( "JSON conversion failed", __FILE__, __LINE__ );
      }
      return v;
    }

    Local error handling on the calling side then looks like:

    cci::cci_value v;
    try {
      v = cci::cci_value::from_json(string);
    } catch (... ) {
      // catch CCI value errors only, others are re-thrown
      cci::cci_handle_exception( cci::CCI_VALUE_FAILURE );
      // customize error behavior, e.g. via SC_REPORT_WARNING( ... );   
    }

    Thanks and Greetings from Duisburg,
      Philipp


  8. Please be aware, that an sc_and_event_list does not imply that the events in the list are triggered at the same time. I would suggest to keep the only the clock sensitivity and act on the triggers in the body of the method instead:
     

    SC_METHOD(func2);
      sensitive << clk.pos();
      dont_initialize();
    
    // ...
    
    void func2() {
      if( nreset.posedge() ) { // nreset went high in this clock cycle
        // ... 
      }
    }

    Alternatively, you can be sensitive to nreset.pos()  and check for clk.posedge() (as a consistency check), if you don't have anything else to do in the body of the method.  With this approach, you might be able to avoid unnecessary triggers of the method.

    Side note to Eyck: There's a small typo in the example above, which should should use "&=" to append to an sc_event_and_list.

    ev_list &= nreset;

     


  9. Port-to-port binding is not tracked by SystemC after elaboration is completed.  You can try to restore (or at least approximate) the mapping yourself by checking "overlap" in the get_interface() results across ports and the hierarchical relationship between the ports' parent objects.


  10. Changing the default writer policy requires to be set consistently throughout the whole application, which might be difficult with third-party infeeds.

    Since SystemC 2.3.2, you can also set the environment variable SC_SIGNAL_WRITE_CHECK=CONFLICT, which effectively enables SC_MANY_WRITERS at runtime (for all signals, though). Quoting the release notes:

      - [2.3.2] Add support for SC_SIGNAL_WRITE_CHECK=CONFLICT
    
        Instead of disabling all runtime signal write checks via the
        environment setting SC_SIGNAL_WRITE_CHECK=DISABLE, setting the
        variable to SC_SIGNAL_WRITE_CHECK=CONFLICT allows detecting
        conflicting writes to a signal within a single evaluation phase
        (see INSTALL).

     

×
×
  • Create New...