[9fans] 9atom-nix

[9fans] 9atom-nix

[erik quanstrom]

9atom-nix (as nemo calls it) just replaced the old 9atom image.
you can find the old image @ 0201-9atom.iso.bz2

key additions to sources
- iplpxe (from cinap), a bios based pxe loader
- /sys/src/9/pcpae (from charles), a pentium kernel with pae.
- /sys/src/9/nix, a nix kernel that includes a terminal.
- rpi (richard miller), a raspberry pi kernel.
- intro(9) (forsyth), abbreviated set of kernel man pages
- kernel qmalloc (more below)
- and as always, "proper" rune support.

the pae kernel deserves a word or so.  the pae kernel brings
with it a streamlined kernel, supporting only very modern
machines, such as the pentium pro.  it supports segments
of differing page sizes.  so there are some changes to the
page and segment interfaces.  also, since this makes paging
out hard, it's just not done.  (this trait is inherited by all
kernels.)  it supports up to ~48G of memory, since mapping
more Page structures in kernel space becomes annoying
after that.  also note that the kernel allocator is no longer
from pool(2).  it's a hybrid quickfit/k&r allocator done by
charles.  see types(2), http://www.quanstro.net/magic/man2html/2/types
for more information on navigating the new thicket of types.

so many thanks to charles.

i hope to get a better source repo set up soon.  in the
mean while:
	http://ftp.quanstro.net/other/9atom.iso.bz2

- erik

Re: [9fans] 9atom-nix

[Anthony Martin]

This is really great. Thanks for all your work, Erik.

  Anthony

Re: [9fans] 9atom-nix

[Charles Forsyth]

It's based on my qmalloc but with many changes for the kernel
environment by jmk.

On 9 January 2013 03:03, erik quanstrom <quanstro@quanstro.net> wrote:
> also note that the kernel allocator is no longer
> from pool(2).  it's a hybrid quickfit/k&r allocator done by
> charles.

Re: [9fans] 9atom-nix

[Venkatesh Srinivas]

> the pae kernel deserves a word or so.  the pae kernel brings
> with it a streamlined kernel, supporting only very modern
> machines, such as the pentium pro.

If it requires PAE, it will not run on Banias and Dothan B-stepping
and C-stepping (on 400MHz FSB) CPUs (most Pentium M's).

Re: [9fans] 9atom-nix

[erik quanstrom]

On Wed Jan  9 16:27:59 EST 2013, me@endeavour.zapto.org wrote:
> > the pae kernel deserves a word or so.  the pae kernel brings
> > with it a streamlined kernel, supporting only very modern
> > machines, such as the pentium pro.
>
> If it requires PAE, it will not run on Banias and Dothan B-stepping
> and C-stepping (on 400MHz FSB) CPUs (most Pentium M's).

the pae kernel requires pae.

don't worry, there's a regular old-skool pc port (no pae) port
in the same source tree.  we run it on our ancient via terminals.

- erik

Re: [9fans] 9atom-nix

[Steve Simon]

> If it requires PAE, it will not run on Banias and Dothan B-stepping
> and C-stepping (on 400MHz FSB) CPUs (most Pentium M's).

one question, is the kernel able to print a message if it is started
on incompatible machine:

	panic: this kernel requires PAE which is not supported by this hardware.

Could save a lot of time messing about with the wrong kernel.

just my 2cents worth.

-Steve

Re: [9fans] 9atom-nix

[erik quanstrom]

On Wed Jan  9 16:37:17 EST 2013, steve@quintile.net wrote:
> > If it requires PAE, it will not run on Banias and Dothan B-stepping
> > and C-stepping (on 400MHz FSB) CPUs (most Pentium M's).
>
> one question, is the kernel able to print a message if it is started
> on incompatible machine:
>
> 	panic: this kernel requires PAE which is not supported by this hardware.
>
> Could save a lot of time messing about with the wrong kernel.

no.  you're welcome to add the test, but it would have to be
some fiddly assembly before the page tables are set up.

non-pae machines are unlikely to have >4gb of memory.

if unsure, with the 386 kernel, you can do this

	; aux/cpuid -f|grep pae

- erik