[9fans] OS X

[9fans] OS X

[William Josephson]

I started porting the Unix bits to OS X
recently, but have only limited access
via ssh to a poorly configured OS X box
at the moment.  Can someone in the
audience tell me how I do a shared memory
fork? Something like FreeBSD's version of
rfork would do nicely, but the man page
not withstanding, rfork appears not to
exist...

Re: [9fans] OS X

[Jonathan Sergent]

On Tue, May 21, 2002 at 04:34:51AM -0400, William Josephson wrote:
> I started porting the Unix bits to OS X
> recently, but have only limited access
> via ssh to a poorly configured OS X box
> at the moment.  Can someone in the
> audience tell me how I do a shared memory
> fork? Something like FreeBSD's version of
> rfork would do nicely, but the man page
> not withstanding, rfork appears not to
> exist...

I think you want cthread_fork() from Mach.

Re: [9fans] OS X

[Ronald G Minnich]

On Tue, 21 May 2002, Jonathan Sergent wrote:
> Proof of nonexistence (okay, this isn't directly from the Apple
> repository, so it's probably a little out of date, but it doesn't need
> a password):


thanks, and bummer, since rfork (at least the RFMEM and files piece) is
really easy on xyzbsd (post-Mach Vm anyway).

ron

Re: [9fans] OS X

[Jonathan Sergent]

On Tue, May 21, 2002 at 02:40:43PM -0400, William Josephson wrote:
> On Tue, May 21, 2002 at 08:20:02AM -0600, Ronald G Minnich wrote:
>
> > I thinks that's wrong. I wrote the very first rfork for freebsd in 1995,
>
> That's FreeBSD, not OS X.  If OS X has rfork, it certainly
> doesn't have a prototype and isn't listed with the other
> syscalls.

Proof of nonexistence (okay, this isn't directly from the Apple
repository, so it's probably a little out of date, but it doesn't need
a password):

http://www.opendarwin.org/cgi-bin/cvsweb.cgi/xnu/bsd/kern/syscalls.c?rev=1.1.1.10&content-type=text/x-cvsweb-markup&cvsroot=Apple%20Darwin
http://www.opendarwin.org/cgi-bin/cvsweb.cgi/xnu/bsd/kern/kern_fork.c?rev=1.1.1.17&content-type=text/x-cvsweb-markup&cvsroot=Apple%20Darwin

Re: [9fans] OS X

[Geoff Collyer]

I now work in the Mac OS X group at Apple.  OS X is a stripped-down
FreeBSD kernel (to remove redundancy with Mach) on a Mach
``micro''-kernel; as far as I can tell, the name ``Darwin''
refers to the combination.  User mode stuff is increasingly
taken from FreeBSD, but there's also stuff from Apple and NeXT
(e.g., Netinfo).

I just got my bookshelf of Power PC and Altivec documents
back from reprographics, so once I find my way around the
open-source repositories, it will be interesting to see what
it would take to run Plan 9 on this hardware.

Re: [9fans] OS X

[Ronald G Minnich]

On Tue, 21 May 2002 rsc@plan9.bell-labs.com wrote:

> Isn't OS X actually FreeBSD user
> space on top of Mach?


oops, I stumbled us into a freebsd discussion on a plan9 list. Bad! Bad!

Anyway, not really. we can talk offline about it.

ron

Re: [9fans] OS X

[rsc]

> Look at the heritage of OSX and you'll see why I thought it should be in
> there.

Well, right, but it just wasn't
clear from your message which
you were talking about.

Isn't OS X actually FreeBSD user
space on top of Mach?

Re: [9fans] OS X

[Ronald G Minnich]

On Tue, 21 May 2002, William Josephson wrote:

> On Tue, May 21, 2002 at 08:20:02AM -0600, Ronald G Minnich wrote:
>
> > I thinks that's wrong. I wrote the very first rfork for freebsd in 1995,
>
> That's FreeBSD, not OS X.  If OS X has rfork, it certainly
> doesn't have a prototype and isn't listed with the other
> syscalls.
>


Look at the heritage of OSX and you'll see why I thought it should be in
there.

ron

Re: [9fans] OS X

[William Josephson]

On Tue, May 21, 2002 at 08:20:02AM -0600, Ronald G Minnich wrote:

> I thinks that's wrong. I wrote the very first rfork for freebsd in 1995,

That's FreeBSD, not OS X.  If OS X has rfork, it certainly
doesn't have a prototype and isn't listed with the other
syscalls.

Re: [9fans] OS X

[Ronald G Minnich]

    [ Part 2: "Included Message" ]

Date: Tue, 21 May 2002 04:34:51 -0400
Reply-To: 9fans@cse.psu.edu
To: 9fans@cse.psu.edu
Subject: [9fans] OS X

I started porting the Unix bits to OS X
recently, but have only limited access
via ssh to a poorly configured OS X box
at the moment.  Can someone in the
audience tell me how I do a shared memory
fork? Something like FreeBSD's version of
rfork would do nicely, but the man page
not withstanding, rfork appears not to
exist...

======

I thinks that's wrong. I wrote the very first rfork for freebsd in 1995,
based on the plan9 manual pages. Some things I could not do (notes, clean
name space) others were easy (shared copy of file descriptors, shared
address space after fork). They changed it a lot when they put it in the
distro, and in particular made it a bit harder to use (I made the stack
space always non-shared, they required you after the fork figure out your
thread IDs and move your stack pointers around so they didn't all have the
same %esp).

But unless they really broke something, RFMEM should work.

ron

Re: [9fans] OS X

[rog]

[-- Attachment #1: Type: text/plain, Size: 79 bytes --]

as far as i remember, you can use Mach threads to achieve the
same effect.


[-- Attachment #2: Type: message/rfc822, Size: 1765 bytes --]

To: 9fans@cse.psu.edu
Subject: [9fans] OS X
Date: Tue, 21 May 2002 04:34:51 -0400
Message-ID: <20020521043451.A32024@honk.eecs.harvard.edu>

I started porting the Unix bits to OS X
recently, but have only limited access
via ssh to a poorly configured OS X box
at the moment.  Can someone in the
audience tell me how I do a shared memory
fork? Something like FreeBSD's version of
rfork would do nicely, but the man page
not withstanding, rfork appears not to
exist...