Memory leak in scheduling simulation

[N. Vienne]

It looks like a memory leak : each time I ran the scheduling simulation an amount of
memory is taken but not released. It is about 80 Mb for each simulation. I start Cheddar
with 1.12 Go of pagefile and 1.64 Go of free physical memory and after 15 runs I have
2.27 Go of pagefile and only 430 Mo of free RAM.

This behavior can be obtained with the example file
project_examples/tests/test_buffer2.xml for instance.

I'm using Cheddar 2.0 (march the 5th 2007) under Windows. I have not checked if this
behavior occurs under Linux.

[jerome]

After discussion with Frank it appears scheduling table is not freed if the simulation is launched again.

Scheduling table is a 2 dimensional table (event table for each processor).
The scheduling table has to be deallocated before launching another simulation by :

• Deallocate for each index (processor), the event table.
• Deallocate the scheduling table itself.