Subject: Re: new pid allocation code
From: David Laight <firstname.lastname@example.org>
Date: 03/12/2003 08:11:11
> Hmmm.... I was thinking that'd cause a problem because when you grow the
> table, sequence # bits now are slot ID bits. But if you're careful when
> you split, it should be ok (i.e. some existing processes might end up in
> the new part of the table).
Yes - the table split code 'unzips' the odd and even seq numbers, leaves
empty slots alone in the low half and puts the created empty slots onto
the head of the free chain (they can't be added to the tail...).
The code collapsed down to something surprisingly small.
David Laight: email@example.com