Re: A few warning patches ...

[Jeffrey A Law <law at cygnus dot com>]

  In message <19981027074438.A7067@tiktok.cygnus.com>you write:
  > Sorry, it must have gotten lost.  I believe I originally intended to
  > release it at the end of function handling, and then realized I didn't
  > have to.  For the sake of working with malloc leak detectors, we probably
  > should free it at the end of gcc (maybe if --enable-checking is done).
I've put a call in at the end of compile_file.

Note that we may still want to deallocate at the end of a function.  Consider
cases where funtions early in the file have lots of regs and functions later
do not.  We'll keep holding onto the larger data even though it's not needed.

Maybe we should consider free-ing it at the end of a function if  the register
count hit some magic number.  Just thinking out loud.

Thanks,
jeff