Jeff asks:

> This "patch" doesn't back out the other fork() enhancements, does it?

Good question...I remember discussing the desirability of doing something
like:

   try new code (the one that fails sometimes, but is faster);

   if it fails, revert to the old code.

the text quoted by Goetz doesn't reveal if that's what they're doing.

However, we'll be able to tell (probably) ... when the patch is installed,
put a breakpoint in copy_non_virgin_pages_of_object, and do a fork() ...
if we hit the breakpoint, it's probably an indication that they're doing
the new-first/old-second strategy.  If we don't, then it implies only
the old code is in place.

--
Stan Sieler                                          [log in to unmask]
                                     http://www.allegro.com/sieler.html