Configuration table's smp_enabled conditional on RTEMS_SMP

[Chris Johns]

The rtems_configuration_table has:

  #ifdef RTEMS_SMP
    bool                         smp_enabled;
  #endif

I would like the smp_enabled variable to always be defined for 4.12 and always set to false when RTEMS_SMP is not defined. It is impossible to parse the configuration table with external auditing tools with out this field always being present unless you examine the DWARF debug info.

I wonder if User_multiprocessing_table is the same so this means the members of the configuration table must always be defined.

Chris

[Sebastian Huber]

Yes, the top-level configuration tables should contain no pre-processor stuff.

[Joel Sherrill]

With per function/data element linking, will the smp enabled variable even make it to a linked uniprocessor exe if SMP is disabled and the variable is always false? There will be no code to reference it.

[Chris Johns]

Replying to joel.sherrill:

With per function/data element linking, will the smp enabled variable even make it to a linked uniprocessor exe if SMP is disabled and the variable is always false? There will be no code to reference it.

It is in the configuration table. All that is being suggested to always have the elements in the table. There is no code.

[Sebastian Huber]

The configuration table not longer exists. In particular the SMP enabled indicator is no longer used. The User_multiprocessing_table no longer exists. In MPCI configurations, there is not an _MPCI_Configuration.