Re: CVS commit: syssrc

[Jason Thorpe <thorpej%nas.nasa.gov@localhost>]

On Fri, 18 Jun 1999 10:01:53 -0700 (PDT) 
 "Perry E. Metzger" <perry%netbsd.org@localhost> wrote:

 > Module Name: syssrc
 > Committed By:        perry
 > Date:                Fri Jun 18 17:01:53 UTC 1999
 > 
 > Modified Files:
 >      syssrc/sys/uvm [netbsd-1-4]: uvm_map.c
 > 
 > Log Message:
 > pullup 1.39->1.40 (thorpej): fix the 1G RAM bug

FWIW, I'm a bit annoyed at your description of the change, here.

It is not a "fix for the 1G RAM bug".  It is a fix that does expanding
of the amount of KVA space available during EARLY bootstrapping.  The bug
this fixes happened to prevent i386 systems with 1G of RAM from
fully booting (they died in early VM initialization, because they couldn't
allocate more KVA space in which to map vm_page structures).  I actually
don't know what the threshold really was ... it could have been less than
1G ... but the bug was reported to exist on 1G i386 systems.

1G RAM (and more!) on Alpha systems worked just fine before this
change.  And calling it a "fix for the 1G RAM bug" on the _pc532_
is just silly.

To set the record straight, here is my original commit message:

revision 1.40
date: 1999/05/20 23:03:23;  author: thorpej;  state: Exp;  lines: +17 -10
Make a slight modification of pmap_growkernel() -- it now returns the
end of the mappable kernel virtual address space.  Previously, it would
get called more often than necessary, because the caller only new what
was requested.

Also, export uvm_maxkaddr so that uvm_pageboot_alloc() can grow the
kernel pmap if necessary, as well.  Note that pmap_growkernel() must
now be able to handle being called before pmap_init().

        -- Jason R. Thorpe <thorpej%nas.nasa.gov@localhost>