Jordan Hubbard <[EMAIL PROTECTED]> writes:
> Well, that single 4GB of address space is divided up into kernel data
> structures, which are in the address space of the process but subject
> to various levels of MMU-provided memory protection, and the process'
> own "user data." I believe the break is currently set in the middle
> at 2GB, and various attempts to adjust it more aggressively (in user
> data's favor) have been interesting but ultimately also proved to
> break things like BSD/OS binaries, which have their own assumptions
> about the setting of the break.
Ahem.
The other way around, actually (we increased KVM space from 256 MB to
1 GB - not 2 GB as you claim). And the problem with legacy BSDI
binaries (newer ones don't have this problem) was fixed a long time
ago, in 3.0 (before 3.0-RELEASE).
DES
--
Dag-Erling Smorgrav - [EMAIL PROTECTED]
To Unsubscribe: send mail to [EMAIL PROTECTED]
with "unsubscribe freebsd-hackers" in the body of the message
