A few questions

A few questions

[Hugo Silva]

Hi,

I've just heard about Xen, and I'm thinking about using NetBSD 2.0 + Xen 
as a virtual server solution. However, I have a few questions to ask:

- I see there is no port for x86_64 yet; The server I'll be running xen 
from will be a Dual AMD Opteron 64 bits. Will Xen correctly work in 
32-bit mode?

- No support for SMP on guest OS, right?

- Could you please enlighten me about the internals of xen, regarding 
network? How does it establishes the network with the host system? Can I 
run a firewall on a virtual server without any problems ?

- According to the site, a FreeBSD port should be close. Is there any 
date we can expect Xen ported to FreeBSD?

- Any plans to port to OpenBSD?

Xen seems like a great product; I'll love to try it out, even without 
SMP on my guest OSes! Very nicely done guys, I'll watch this project 
grow :-)

Best Regards


-------------------------------------------------------
This SF.Net email is sponsored by: IntelliVIEW -- Interactive Reporting
Tool for open source databases. Create drag-&-drop reports. Save time
by over 75%! Publish reports on the web. Export to DOC, XLS, RTF, etc.
Download a FREE copy at http://www.intelliview.com/go/osdn_nl

Re: A few questions

[Mark Williamson]

Ian's answered the rest, so I won't duplicate what he's said.  I wanted to 
expand on this, however:

> - Could you please enlighten me about the internals of xen, regarding
> network? How does it establishes the network with the host system? Can I
> run a firewall on a virtual server without any problems ?

From a configuration point of view, it's as if dom0 had a dedicated NIC for 
each guest NIC, connected by crossover cable.  Configuring guest networks is 
then just like setting up a Linux machine with multiple NICs.  Bridging, 
routing, firewalling, etc should all work fine.

> - Any plans to port to OpenBSD?

Lots of people have talked about it but AFAIK there's nobody working on it 
right now.

Cheers,
Mark

> Xen seems like a great product; I'll love to try it out, even without
> SMP on my guest OSes! Very nicely done guys, I'll watch this project
> grow :-)
>
> Best Regards
>
>
> -------------------------------------------------------
> This SF.Net email is sponsored by: IntelliVIEW -- Interactive Reporting
> Tool for open source databases. Create drag-&-drop reports. Save time
> by over 75%! Publish reports on the web. Export to DOC, XLS, RTF, etc.
> Download a FREE copy at http://www.intelliview.com/go/osdn_nl
> _______________________________________________
> Xen-devel mailing list
> Xen-devel@lists.sourceforge.net
> https://lists.sourceforge.net/lists/listinfo/xen-devel


-------------------------------------------------------
This SF.Net email is sponsored by: IntelliVIEW -- Interactive Reporting
Tool for open source databases. Create drag-&-drop reports. Save time
by over 75%! Publish reports on the web. Export to DOC, XLS, RTF, etc.
Download a FREE copy at http://www.intelliview.com/go/osdn_nl

A few questions

[Yao]

[-- Attachment #1: Type: text/plain, Size: 654 bytes --]

Hi,
Lately I'm studying SELinux and got some questions  which I want to be clear.


(1)I know SELinux is based on Flask architecture and I know where the SS is, but I'm not sure
where the OM locates, I guess the variable "security_ops" which belongs to LSM represents the OM, am I right?


(2)the struct "selinux_ops" in file hooks.c is declared as "static", why not add "const" qualifier so that 
the it will be put in read-only data section in the kernel?


(3)Is there any way to hack the SELinux, I mean, to disable it on the fly? For example, replace the policy db with a 
blank file so that any permission is allowed. Is it feasible?


Regards,
Yao

[-- Attachment #2: Type: text/html, Size: 974 bytes --]

Re: A few questions

[Eric Paris]

2012/3/7 Yao <yffbrave@163.com>:

> (2)the struct "selinux_ops" in file hooks.c is declared as "static", why not
> add "const" qualifier so that
> the it will be put in read-only data section in the kernel?

I probably can.  Thanks for the suggestion.

> (3)Is there any way to hack the SELinux, I mean, to disable it on the fly?
> For example, replace the policy db with a
> blank file so that any permission is allowed. Is it feasible?

There have been kernel attacks which disable SELinux.  If you can
write to arbitrary kernel memory you can win.  But I'm not going to
help tell you how to write your rootkit    :)

--
This message was distributed to subscribers of the selinux mailing list.
If you no longer wish to subscribe, send mail to majordomo@tycho.nsa.gov with
the words "unsubscribe selinux" without quotes as the message.

Re: A few questions

[David Quigley]

On 03/07/2012 10:15, Yao wrote:
> Hi,
> Lately I'm studying SELinux and got some questions which I want to be
> clear.
>
> (1)I know SELinux is based on Flask architecture and I know where the
> SS is, but I'm not sure
> where the OM locates, I guess the variable "security_ops" which
> belongs to LSM represents the OM, am I right?
>
> (2)the struct "selinux_ops" in file hooks.c is declared as "static",
> why not add "const" qualifier so that
> the it will be put in read-only data section in the kernel?
>
> (3)Is there any way to hack the SELinux, I mean, to disable it on the
> fly? For example, replace the policy db with a
> blank file so that any permission is allowed. Is it feasible?
>
> Regards,
> Yao

So In order.

I asked (1) when I first started as well and the answer I got was the 
kernel itself is the object manager. You'll notice a bunch of security_ 
calles through the kernel. These are the enforcement points which query 
the security server through the selinux specific hooks behind the LSM 
interface.

(2) I'm not sure if there is a reason not to. I don't know how this 
would effect runtime disabling since we unhook those functions if we are 
disabled.

(3) So if you're in the kernel there are techniques to disable SELinux 
and make it appear like it is still running. Replacing the policy db 
with an empty one would be the opposite of the effect you want. Since 
SELinux is deny by default it would give you a system with no access not 
all access. The MDP program in the kernel tree makes a policy with one 
type that has access to everything but that would be obvious and not 
useful. Especially when all of the domains you're programs are currently 
running in suddenly disappear.

Dave


--
This message was distributed to subscribers of the selinux mailing list.
If you no longer wish to subscribe, send mail to majordomo@tycho.nsa.gov with
the words "unsubscribe selinux" without quotes as the message.

Re: A few questions

[Stephen Smalley]

On Wed, 2012-03-07 at 10:57 -0500, David Quigley wrote:
> On 03/07/2012 10:15, Yao wrote:
> > Hi,
> > Lately I'm studying SELinux and got some questions which I want to be
> > clear.
> >
> > (1)I know SELinux is based on Flask architecture and I know where the
> > SS is, but I'm not sure
> > where the OM locates, I guess the variable "security_ops" which
> > belongs to LSM represents the OM, am I right?
> >
> > (2)the struct "selinux_ops" in file hooks.c is declared as "static",
> > why not add "const" qualifier so that
> > the it will be put in read-only data section in the kernel?
> >
> > (3)Is there any way to hack the SELinux, I mean, to disable it on the
> > fly? For example, replace the policy db with a
> > blank file so that any permission is allowed. Is it feasible?
> >
> > Regards,
> > Yao
> 
> So In order.
> 
> I asked (1) when I first started as well and the answer I got was the 
> kernel itself is the object manager. You'll notice a bunch of security_ 
> calles through the kernel. These are the enforcement points which query 
> the security server through the selinux specific hooks behind the LSM 
> interface.

In general, the "object manager" is the component that implements the
object abstraction and operations and is responsible for enforcement of
the policy decisions.  The kernel can either be viewed as a single
object manager or as a collection of object managers (e.g. the process
management subsystem, the vfs and filesystem implementations, the
networking subsystem, the ipc subsystem, ...).  In the case of the Flask
microkernel-based OS, the various subsystems were in fact separate tasks
running on the microkernel.
 
-- 
Stephen Smalley
National Security Agency


--
This message was distributed to subscribers of the selinux mailing list.
If you no longer wish to subscribe, send mail to majordomo@tycho.nsa.gov with
the words "unsubscribe selinux" without quotes as the message.

A few questions

[Mat]

Toby, Alex, Christopher & the others

firstly: Thanks a lot !

I just thought the Internet Wayback Machine's Archive (besides google cache)
might come in handy:

http://web.archive.org/web/*/http://www.namesys.com

(if you're not already using it)

Cheers

Mat

A few questions

[Alex Caudill]

Hi,

With namesys.com being down, I can't find reiser4 documentation, and,
frankly, that sucks.

I'm prepared to:

1.) Provide free hosting for ReiserFS version control and documentation.

2.) Provide an erudified.com subdomain or purchase a top-level domain
for the same.

3.) Make a reasonable donation via PayPal to anyone willing to
transfer Reiser4 documentation and code onto a new machine. I'm also
willing to help with this.

#3 applies even if you guys opt to use SourceForge or some other
service. I don't particularly care where or how the info gets put
online, I'd just like to see it put there.

I'm not subscribed to this list, so please CC me with responses if
you're interested.

Thanks in advance!

P.S. I don't care, even slightly, about any aspect of Hans' personal
life. My lack of concern is so complete that I don't even have an
opinion. I'm constitutionally unwilling to discuss any aspect of it,
except to say that I don't give a shit. My interest is the
continuation of reiser4 development.

Re: A few questions

[Alex Caudill]

Thanks to all who followed up, and my apologies for top-posting (if I
did). I'll subscribe after I send this.

Christopher Sawtell was kind enough to send me a copy of the documentation.

I've made it available at:

http://reiser4.erudified.com

Toby Thain went ahead and registered reiser4.{com,net,org} (see below)
and I've added ServerAlias entries for all of them, so if/when he
points DNS to 72.249.66.100, there should be a decent TLD source for
reiserfs information. Still, it isnt exactly authoritative, and it
certainly doesn't contain any of the wiki entries from namesys.com.
It's late, but I'll try to go through the google cache and add them
this week, as well as fix any broken links.

Toby also volunteered to provide hosting, and I'm sure there are
plenty of others who'd be willing to do so. I currently use the
machine for other things, but I can provide root access on another
one, or simply donate some cash for those developing reiser4 to make
their own arrangements.

I think it's obviously in the project's best interests to have a
single, authoritative source for documentation, etc, even if it's just
until Namesys changes hands. It'd also show the community at large
that reiser4 is still kicking, which can only be a good thing.

So, again, a couple questions:

1.) Are the ReiserFS developers willing to make a public statement
about the future of the project? Will it continue as an open source
project even if future Namesys management opts to make it proprietary?
Is anyone even in a position to make such a statement?

2.) Who's in the best position to provide hosting? My machine is a
pretty basic dual core, 2gb ram, as much bandwidth as you can shake
your stick at. I can donate root access to another machine with the
same specs. Toby offered to provide hosting as well, so maybe he could
chime in with the particulars of his arrangements. I'm willing to
donate a fairly substantial amount of cash to any developer with a
proven track record of ReiserFS contributions who'd like to set up
some sort of authoritative source for ReiserFS docs/releases. If
that's you, please drop me a line.

I'd like to emphasize that I'm not at all interested in taking control
of anything or butting heads with anyone. I'd simply like to see the
reiser4 community come to a concensus about where and how
documentation and releases will be made available in the absence of
Hans' management. I'm happy to provide any kind of support I can,
financial or otherwise.

Let's get the ball rolling!

And, last but not least, thanks for the great work!! =)

On 3/3/08, Toby Thain <toby@telegraphics.com.au> wrote:

> I would love to help. This may seem a little opportunistic, but I
>  registered reiser4.{com,net,org} just now. I can run DNS for these
>  pointing to whatever host you use (assuming that your idea progresses).
>
>  I can also provide hosting of whatever services are required (have
>  colo servers in 2 continents).
>
>  --Toby
>

Re: A few questions

[Alex Caudill]

I absolutely agree about compromisng Hans' ability to make bank, and
should have been more clear. I'm NOT AT ALL proposing any sort of
rename or transfer of 'ownership' of the project. It's Hans' baby, and
regardless of what happens, it will remain his baby in my eyes.

I'm proposing that, in his absence, the community comes together and
commits to continue developing and releasing it (and working towards
mainline kernel inclusion). I oppose arbitrary renames and am
absolutely in favor of giving back the proverbial reins (which, again,
to be clear: I have no interest in holding or ability to hold) when/if
he returns, regardless of the trial outcome. I want to use and sponsor
ReiserFS, not some politically correct renamed fork. ReiserFS is what
it is because of Hans, and I don't want it to be something else.

Note that I didn't register the domains. I actually did check them,
but I decided against it. I don't want to do anything that doesn't
have the full, unanimous support of the folks actually doing the
development. Anything I do pertaining to ReiserFS will be posted on
list, and if there's a halfway reasonable problem with any of it, I'll
stop immediately. I want to further Hans' vision. His personal life
has zero effect on the merit of his filesystem design.

I'd happily consider sponsoring further development of the repacker -
if you'd like to hack on it, let me know, and we can work something
out. I'm not able to fund a full-time development effort, but I can
definitely help.

I'll work on getting the source code comments into HTML.

On 3/4/08, Christopher Sawtell <csawtell@gmail.com> wrote:
> > Christopher Sawtell was kind enough to send me a copy of the documentation.
>
>
> Please note that the image files for the diagrams which are be part of
>  the whitepaper are missing, and that there is a lot of more up to date
>  detailed doco in the source code files.
>
>  If there if more work that would be well worth funding, it would be
>  the proposed real-time re-packer, and the re-integration of the
>  ability to access extended attributes as sub-files of the main
>  data-file. Yes, I do know that 'The Powers-That-Be' don't approve, but
>  it's a brilliant idea.
>
>  P.S. I think we should not yet do anything which would compromise
>  Hans' ability to earn a crust if the current situation turns out to be
>  just a ghastly nightmare.
>
>
>  --
>  Sincerely etc.
>
> Christopher Sawtell
>

Re: A few questions

[Toby Thain]

On 4-Mar-08, at 2:53 AM, Alex Caudill wrote:

> Thanks to all who followed up, and my apologies for top-posting (if I
> did). I'll subscribe after I send this.
>
> Christopher Sawtell was kind enough to send me a copy of the  
> documentation.
>
> I've made it available at:
>
> http://reiser4.erudified.com
>
> Toby Thain went ahead and registered reiser4.{com,net,org} (see below)

Partly because I wanted to prevent a squatter or hostile party from  
getting them first...

> and I've added ServerAlias entries for all of them, so if/when he
> points DNS to 72.249.66.100,

I decided to HTTP redirect to your site instead, so that crawlers see  
only one web site instead of copies under different domains.

> there should be a decent TLD source for
> reiserfs information. Still, it isnt exactly authoritative, and it
> certainly doesn't contain any of the wiki entries from namesys.com.

namesys.com isn't down for me?

> It's late, but I'll try to go through the google cache and add them
> this week, as well as fix any broken links.
>
> Toby also volunteered to provide hosting, ... maybe he could
> chime in with the particulars of his arrangements.

I have 2 colo machines that could be used, one in Australia and one  
in Canada. I don't think running services for the project is going to  
be particularly demanding of resources. Sounds like you have that  
part covered anyway, Alex.

> I'm willing to
> donate a fairly substantial amount of cash to any developer with a
> proven track record of ReiserFS contributions who'd like to set up
> some sort of authoritative source for ReiserFS docs/releases. If
> that's you, please drop me a line.
>
> I'd like to emphasize that I'm not at all interested in taking control
> of anything or butting heads with anyone. I'd simply like to see the
> reiser4 community come to a concensus about where and how
> documentation and releases will be made available in the absence of
> Hans' management. I'm happy to provide any kind of support I can,
> financial or otherwise.

I can't chip in financially but certainly hosting or services can be  
done as needed.

--Toby

>
> Let's get the ball rolling!
>
> And, last but not least, thanks for the great work!! =)
>
> On 3/3/08, Toby Thain <toby@telegraphics.com.au> wrote:
>
>> I would love to help. This may seem a little opportunistic, but I
>>  registered reiser4.{com,net,org} just now. I can run DNS for these
>>  pointing to whatever host you use (assuming that your idea  
>> progresses).
>>
>>  I can also provide hosting of whatever services are required (have
>>  colo servers in 2 continents).
>>
>>  --Toby
>>
> --
> To unsubscribe from this list: send the line "unsubscribe reiserfs- 
> devel" in
> the body of a message to majordomo@vger.kernel.org
> More majordomo info at  http://vger.kernel.org/majordomo-info.html

Re: A few questions

[Jeff Mahoney]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Alex Caudill wrote:
> Hi,
> 
> With namesys.com being down, I can't find reiser4 documentation, and,
> frankly, that sucks.
> 
> I'm prepared to:
> 
> 1.) Provide free hosting for ReiserFS version control and documentation.
> 
> 2.) Provide an erudified.com subdomain or purchase a top-level domain
> for the same.
> 
> 3.) Make a reasonable donation via PayPal to anyone willing to
> transfer Reiser4 documentation and code onto a new machine. I'm also
> willing to help with this.

I agree something needs to be done here, and appreciate the offer, but
I'm afraid it's one that we can't accept. I'm sure your hosting is
perfectly adequate, but hosting for tools that millions of people use
(or at least have on their systems) requires a high level of trust. The
trust just isn't there. I'm sure you're perfectly trustworthy, but we
don't know you. I've been hoping to get space on kernel.org or, failing
that, sourceforge. These sites are accepted nearly universally as
trustworthy, and have gobs of bandwidth as well as extensive mirroring
and backup facilities.

I have space on opensuse.org that I could offer up, but I don't think
that's the right solution since the hosting site should be distribution
agnostic. Unfortunately, with namesys.com now, I fear that opensuse.org
may have become the de facto location to obtain reiserfsprogs (v3) sources.

- -Jeff

- --
Jeff Mahoney
SUSE Labs
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v2.0.4-svn0 (GNU/Linux)
Comment: Using GnuPG with SUSE - http://enigmail.mozdev.org

iD8DBQFHza1aLPWxlyuTD7IRAjJuAJ4pZnu7Gf8p+Sv7sSVYyPz+jQEiYwCdEO3P
wJwe0qoW5Eg74wRMM6k18Z8=
=8YzR
-----END PGP SIGNATURE-----

Re: A few questions

[Alex Caudill]

On 3/4/08, Jeff Mahoney <jeffm@suse.com> wrote:
[snip]
> I agree something needs to be done here, and appreciate the offer, but
>  I'm afraid it's one that we can't accept. I'm sure your hosting is
>  perfectly adequate, but hosting for tools that millions of people use
>  (or at least have on their systems) requires a high level of trust. The
>  trust just isn't there. I'm sure you're perfectly trustworthy, but we
>  don't know you. I've been hoping to get space on kernel.org or, failing
>  that, sourceforge. These sites are accepted nearly universally as
>  trustworthy, and have gobs of bandwidth as well as extensive mirroring
>  and backup facilities.
>
>  I have space on opensuse.org that I could offer up, but I don't think
>  that's the right solution since the hosting site should be distribution
>  agnostic. Unfortunately, with namesys.com now, I fear that opensuse.org
>  may have become the de facto location to obtain reiserfsprogs (v3) sources.

I can completely understand this, re: trustworthiness. It's critical
on many levels for any project, especially a filesystem. I believe
we'd all like to see the project made easily available and hosted by
the most capable/trustworthy/etc party that can be found, and I
wholeheartedly agree that this isn't me. I'm willing to give it a
shot, but there are certainly better options on the table.

I probably have no right to say this, but I think there's an
obligation here to the community and to Hans himself to continue in
the most professional way possible. Can we agree on this?

If you (the reiserfs developers) come to a consensus about where to
host it, I'll send $1000 USD to whoever's taking the lead in getting
it done. I'm willing to sponsor ongoing development on a
feature-by-feature basis as well. I'm really, honestly, seriously just
interested in seeing Reiser4 become a first-class filesystem citizen.
It deserves it.

Re: A few questions

[Edward Shishkin]

Jeff Mahoney wrote:

> Alex Caudill wrote:
>
> >Hi,
>
> >With namesys.com being down, I can't find reiser4 documentation, and,
> >frankly, that sucks.
>
> >I'm prepared to:
>
> >1.) Provide free hosting for ReiserFS version control and documentation.
>
> >2.) Provide an erudified.com subdomain or purchase a top-level domain
> >for the same.
>
> >3.) Make a reasonable donation via PayPal to anyone willing to
> >transfer Reiser4 documentation and code onto a new machine. I'm also
> >willing to help with this.
>
>
> I agree something needs to be done here, and appreciate the offer, but
> I'm afraid it's one that we can't accept. I'm sure your hosting is
> perfectly adequate, but hosting for tools that millions of people use
> (or at least have on their systems) requires a high level of trust. The
> trust just isn't there. I'm sure you're perfectly trustworthy, but we
> don't know you. I've been hoping to get space on kernel.org


Eventually we have got it.
Btw, it seems you are lucky.. :)

> or, failing
> that, sourceforge. These sites are accepted nearly universally as
> trustworthy, and have gobs of bandwidth as well as extensive mirroring
> and backup facilities.
>
> I have space on opensuse.org that I could offer up, but I don't think
> that's the right solution since the hosting site should be distribution
> agnostic. Unfortunately, with namesys.com now, I fear that opensuse.org
> may have become the de facto location to obtain reiserfsprogs (v3) 
> sources.
>
> -Jeff
>
> --
> Jeff Mahoney
> SUSE Labs

--
To unsubscribe from this list: send the line "unsubscribe reiserfs-devel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

RE: A few questions

[Ian Pratt]

> I've just heard about Xen, and I'm thinking about using 
> NetBSD 2.0 + Xen 
> as a virtual server solution. However, I have a few questions to ask:

As far as I'm aware, NetBSD is only supported as a guest, not domain 0
('host' in vmware speak).

> - I see there is no port for x86_64 yet; The server I'll be 
> running xen 
> from will be a Dual AMD Opteron 64 bits. Will Xen correctly work in 
> 32-bit mode?

Yes.
 
> - No support for SMP on guest OS, right?

There is prototype support for SMP linux 2.6 guests.

> - Could you please enlighten me about the internals of xen, regarding 
> network? How does it establishes the network with the host 
> system? 

You can use bridging or routing in domain0.

> Can I run a firewall on a virtual server without any problems ?

Yes.

> - According to the site, a FreeBSD port should be close. Is there any 
> date we can expect Xen ported to FreeBSD?

FreeBSD is available for testing -- see the list archive (Kip Macy).

> - Any plans to port to OpenBSD?

Not as yet, but its certainly possible.

Ian


-------------------------------------------------------
This SF.Net email is sponsored by: IntelliVIEW -- Interactive Reporting
Tool for open source databases. Create drag-&-drop reports. Save time
by over 75%! Publish reports on the web. Export to DOC, XLS, RTF, etc.
Download a FREE copy at http://www.intelliview.com/go/osdn_nl

A few questions.

[Stefan Berndtsson]

First question:

What is the status of m3mirror:ing with an ibook2 (600MHz from 2001)
using a 2.6 kernel? There was some patch sent out a long time ago, but
I couldn't get it to apply in any sensible way.


Second question:

What is the status of sound output on the G5?


Third, and actually most interesting question for me:

I want some sort of video recording hardware for my G5. What should I get?
My demands (or at least strong wishes) are:

1. Works on the G5 with Linux (ppc32 or ppc64 doesn't matter).
2. If PCI, works in the PCI-X slots.
3. Has a tuner (preferable, but if all else works, this can be skipped).
4. Has S-Video input.
5. Can handle full framesize in full framerate.
6. At least handle PAL, although handling NTSC is a plus.
7. Doesn't cost a fortune.

Possible to buy in Sweden is another plus.


** Sent via the linuxppc-dev mail list. See http://lists.linuxppc.org/

Re: A few questions.

[Benjamin Herrenschmidt]

On Sat, 2004-01-31 at 21:06, Stefan Berndtsson wrote:
> First question:
>
> What is the status of m3mirror:ing with an ibook2 (600MHz from 2001)
> using a 2.6 kernel? There was some patch sent out a long time ago, but
> I couldn't get it to apply in any sensible way.

That doesn't quite work yet. Someday, somebody (me ?) has to
spend the time & dedication to setup the second output properly
once for all =P

> Second question:
>
> What is the status of sound output on the G5?

No driver for it yet.


** Sent via the linuxppc-dev mail list. See http://lists.linuxppc.org/

Re: A few questions.

[Stefan Berndtsson]

[-- Attachment #1: Type: text/plain, Size: 759 bytes --]

Benjamin Herrenschmidt <benh@kernel.crashing.org> writes:

> On Sat, 2004-01-31 at 21:06, Stefan Berndtsson wrote:
>> First question:
>>
>> What is the status of m3mirror:ing with an ibook2 (600MHz from 2001)
>> using a 2.6 kernel? There was some patch sent out a long time ago, but
>> I couldn't get it to apply in any sensible way.
>
> That doesn't quite work yet. Someday, somebody (me ?) has to
> spend the time & dedication to setup the second output properly
> once for all =P

Ok, in the meantime, I took a patch for 2.4 and brutally stuffed it into
a 2.6.2-rc1-ben1. With some minor tweaking it now seems it actually works.

I assume the patch is much too specific and unfit to go in a proper kernel
release, but it's included below anyway.

/Stefan


[-- Attachment #2: M3-Mirror patch for 2.6 --]
[-- Type: text/plain, Size: 5211 bytes --]

diff -ur linux-2.6.2-rc1-ben1-orig/drivers/video/aty/aty128fb.c linux-2.6.2-rc1-ben1/drivers/video/aty/aty128fb.c
--- linux-2.6.2-rc1-ben1-orig/drivers/video/aty/aty128fb.c	2003-10-14 09:28:07.000000000 +0200
+++ linux-2.6.2-rc1-ben1/drivers/video/aty/aty128fb.c	2004-01-31 16:29:58.000000000 +0100
@@ -312,6 +312,11 @@
 	struct aty128_crtc crtc;
 	struct aty128_pll pll;
 	struct aty128_ddafifo fifo_reg;
+#ifdef CONFIG_PMAC_PBOOK
+        struct aty128_crtc crtc2;
+        struct aty128_pll pll2;
+        struct aty128_ddafifo fifo_reg2;
+#endif
 	u32 accel_flags;
 	struct aty128_constants constants;  /* PLL and others      */
 	void *regbase;                      /* remapped mmio       */
@@ -740,6 +745,28 @@
 	aty_st_pll(PPLL_CNTL, aty_ld_pll(PPLL_CNTL) & ~(0x00030000));
 }

+#ifdef CONFIG_PMAC_PBOOK
+/* Program the CRTC2 registers */
+static void
+aty128_set_crtc2(const struct aty128_crtc *crtc,
+		const struct aty128fb_par *par)
+{
+	aty_st_le32(CRTC2_GEN_CNTL, crtc->gen_cntl);
+
+	/* FIXME - Hardcodec */
+	aty_st_le32(CRTC2_H_TOTAL_DISP, crtc->h_total & ~0xf | 0xa);
+	aty_st_le32(CRTC2_H_SYNC_STRT_WID,
+		    crtc->h_sync_strt_wid & ~0xff | 0x10);
+
+	aty_st_le32(CRTC2_V_TOTAL_DISP, crtc->v_total);
+	aty_st_le32(CRTC2_V_SYNC_STRT_WID, crtc->v_sync_strt_wid);
+	aty_st_le32(CRTC2_PITCH, crtc->pitch);
+	aty_st_le32(CRTC2_OFFSET, crtc->offset);
+	aty_st_le32(CRTC2_OFFSET_CNTL, crtc->offset_cntl);
+	/* Disable ATOMIC updating.  Is this the right place? */
+	aty_st_pll(PPLL_CNTL, aty_ld_pll(PPLL_CNTL) & ~(0x00030000));
+}
+#endif

 static int
 aty128_var_to_crtc(const struct fb_var_screeninfo *var,
@@ -997,7 +1024,7 @@
 {
 	if (on) {
 		aty_st_le32(CRTC_EXT_CNTL, aty_ld_le32(CRTC_EXT_CNTL) | CRT_CRTC_ON);
-		aty_st_le32(DAC_CNTL, (aty_ld_le32(DAC_CNTL) | DAC_PALETTE2_SNOOP_EN));
+		aty_st_le32(DAC_CNTL, (aty_ld_le32(DAC_CNTL) | DAC_PALETTE2_SNOOP_EN | DAC_CLK_SEL));
 	} else
 		aty_st_le32(CRTC_EXT_CNTL, aty_ld_le32(CRTC_EXT_CNTL) & ~CRT_CRTC_ON);
 }
@@ -1069,6 +1096,45 @@
 	aty_st_pll(PPLL_CNTL, aty_ld_pll(PPLL_CNTL) & ~PPLL_RESET);
 }

+#ifdef CONFIG_PMAC_PBOOK
+static void
+aty128_set_pll2(struct aty128_pll *pll, const struct aty128fb_par *par)
+{
+	u32 div;
+
+	unsigned char post_conv[] =	/* register values for post dividers */
+        { 2, 0, 1, 4, 2, 2, 6, 2, 3, 2, 2, 2, 7 };
+
+	/* select PPLL_DIV_3 */
+	aty_st_le32(CLOCK_CNTL_INDEX, aty_ld_le32(CLOCK_CNTL_INDEX) | (3 << 8));
+
+	/* reset PLL */
+	aty_st_pll(P2PLL_CNTL,
+		   aty_ld_pll(P2PLL_CNTL) | PPLL_RESET | PPLL_ATOMIC_UPDATE_EN);
+
+	/* write the reference divider */
+	aty_pll_wait_readupdate(par);
+	aty_st_pll(P2PLL_REF_DIV, par->constants.ref_divider & 0x3ff);
+	aty_pll_writeupdate(par);
+
+	div = aty_ld_pll(P2PLL_DIV_0);
+	div &= ~XPLL_FB_DIV_MASK;
+	div |= pll->feedback_divider;
+	div |= post_conv[pll->post_divider] << 16;
+
+	/* write feedback and post dividers */
+	aty_pll_wait_readupdate(par);
+	aty_st_pll(P2PLL_DIV_0, div);
+	aty_pll_writeupdate(par);
+
+	aty_pll_wait_readupdate(par);
+	aty_st_pll(HTOTAL_CNTL, 0);	/* no horiz crtc adjustment */
+	aty_pll_writeupdate(par);
+
+	/* clear the reset, just in case */
+	aty_st_pll(P2PLL_CNTL, aty_ld_pll(P2PLL_CNTL) & ~PPLL_RESET);
+}
+#endif

 static int
 aty128_var_to_pll(u32 period_in_ps, struct aty128_pll *pll,
@@ -1130,6 +1196,16 @@
 	aty_st_le32(DDA_ON_OFF, dsp->dda_on_off);
 }

+#ifdef CONFIG_PMAC_PBOOK
+static void
+aty128_set_fifo2(const struct aty128_ddafifo *dsp,
+		 const struct aty128fb_par *par)
+{
+        /* FIXME - Hardcoded */
+	aty_st_le32(DDA2_CONFIG, 0x010502aa);
+	aty_st_le32(DDA2_ON_OFF, 0x11805a74);
+ }
+#endif

 static int
 aty128_ddafifo(struct aty128_ddafifo *dsp,
@@ -1245,6 +1321,9 @@
 	if (par->chip_gen == rage_M3) {
 		aty128_set_crt_enable(par, par->crt_on);
 		aty128_set_lcd_enable(par, par->lcd_on);
+		aty128_set_crtc2(&par->crtc2, par);
+		aty128_set_pll2(&par->pll2, par);
+		aty128_set_fifo2(&par->fifo_reg2, par);
 	}
 #endif
 	if (par->accel_flags & FB_ACCELF_TEXT)
@@ -1282,6 +1361,17 @@
 	if ((err = aty128_ddafifo(&fifo_reg, &pll, crtc.depth, par)))
 		return err;

+#ifdef CONFIG_PMAC_PBOOK
+	if ((err = aty128_var_to_crtc(var, &par->crtc2, par)))
+		return err;
+
+	if ((err = aty128_var_to_pll(var->pixclock, &par->pll2, par)))
+		return err;
+
+	if ((err = aty128_ddafifo(&par->fifo_reg2, &par->pll2, par->crtc2.depth, par)))
+		return err;
+#endif
+
 	par->crtc = crtc;
 	par->pll = pll;
 	par->fifo_reg = fifo_reg;
@@ -1370,7 +1460,7 @@
 	      struct aty128fb_par *par)
 {
 	if (par->chip_gen == rage_M3) {
-#if 0
+#if CONFIG_PMAC_PBOOK
 		/* Note: For now, on M3, we set palette on both heads, which may
 		 * be useless. Can someone with a M3 check this ?
 		 *
diff -ur linux-2.6.2-rc1-ben1-orig/include/video/aty128.h linux-2.6.2-rc1-ben1/include/video/aty128.h
--- linux-2.6.2-rc1-ben1-orig/include/video/aty128.h	2002-07-24 22:04:54.000000000 +0200
+++ linux-2.6.2-rc1-ben1/include/video/aty128.h	2004-01-31 15:19:02.000000000 +0100
@@ -258,7 +258,7 @@
 #define PLL_TEST_CNTL				0x0013
 #define P2PLL_CNTL				0x002a
 #define P2PLL_REF_DIV				0x002b
-#define P2PLL_DIV_0				0x002b
+#define P2PLL_DIV_0				0x002c
 #define POWER_MANAGEMENT			0x002f

 #define PPLL_RESET				0x01

Re: A few questions.

[Stefan Berndtsson]

Stefan Berndtsson <stefan@nocrew.org> writes:

> Benjamin Herrenschmidt <benh@kernel.crashing.org> writes:
>
>> On Sat, 2004-01-31 at 21:06, Stefan Berndtsson wrote:
>>> First question:
>>>
>>> What is the status of m3mirror:ing with an ibook2 (600MHz from 2001)
>>> using a 2.6 kernel? There was some patch sent out a long time ago, but
>>> I couldn't get it to apply in any sensible way.
>>
>> That doesn't quite work yet. Someday, somebody (me ?) has to
>> spend the time & dedication to setup the second output properly
>> once for all =P
>
> Ok, in the meantime, I took a patch for 2.4 and brutally stuffed it into
> a 2.6.2-rc1-ben1. With some minor tweaking it now seems it actually works.

After some more brief tests, it seems that, although the mirroring works,
changing the resolution on the machine stops the working mirror until a
reboot takes place and reinitialises something. I don't know if this was
the case with the patch used on 2.4, but I suspect it was.

/Stefan

** Sent via the linuxppc-dev mail list. See http://lists.linuxppc.org/

A few questions

[Adam Luchjenbroers]

A few questions, two of which because I simply realise I don't know the 
answers when I probably should, and a third because I'll need to know how.

1) What is the intended method to edit the path for a user-account or 
globally? I know hacking together a shell script to add a directory and 
putting it in startup works, but I suspect there is a much more elegant and 
effective solution.

2) How do I remount devfs?

3) How do I backup all user accounts and passwords. I want to be able to 
re-install and have all the old user accounts + passwords restored, and also 
just for general backup purposes.

-- 
"Do not expose your LaserWriter to naked flame..." - Apple LaserWriter Manual

Please avoid sending me Word or PowerPoint attachments.
See http://www.fsf.org/philosophy/no-word-attachments.html


-
To unsubscribe from this list: send the line "unsubscribe linux-newbie" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.linux-learn.org/faqs

Re: A few questions

[dashielljt]

By way of answering question 3, have a look at the sysback script below.
Understand I'm going to patch it a little so another version that puts
dates in file names for the .log file and the actual backup files as well
so another version will appear.

cut here.
#!/bin/sh
# system backup script
# uses tar to do a system backup to zip disks in /dev/hdd4
# excluded directories are: /proc/ and /tmp/
echo "put a zip disk in drive for system backup"
read
cd /
echo "system backup log started:" >/tmp/sysback.log
date >>/tmp/sysback.log
mount -t vfat /dev/hdd4 /zip
tar zcMpPv --exclude=/proc,/tmp --file=/zip/sysback.tar / 2>&1 >>/root/sysback.log
mv /tmp/sysback.log /root/sysback.log
cp /root/sysback.log /var/adm/archive/sysback.log
umount /zip




cut here.
Jude <dashielljt(at)gmpexpress-dot-net>

-
To unsubscribe from this list: send the line "unsubscribe linux-newbie" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.linux-learn.org/faqs

Re: A few questions

[Adam Luchjenbroers]

On Sunday 01 Dec 2002 10:36 pm, dashielljt wrote:
> By way of answering question 3, have a look at the sysback script below.
> Understand I'm going to patch it a little so another version that puts
> dates in file names for the .log file and the actual backup files as well
> so another version will appear.

Nice, but I only wanted to back-up the specific files that held the user 
accounts. I suspect right now that alot of my disk-space is being used up by 
assorted junk and I figure backing up the important stuff and re-installing 
will help me recover it.

-- 
"There is no such thing as sleep deprivation, only caffeine deficiency." - 
Simmons

Please avoid sending me Word or PowerPoint attachments.
See http://www.fsf.org/philosophy/no-word-attachments.html


-
To unsubscribe from this list: send the line "unsubscribe linux-newbie" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.linux-learn.org/faqs

Re: A few questions

[dashielljt]

You'd want to do more than one tar command of the form:
tar zcvf user1.tar.gz /home/user1/
tar zcvf user2.tar.gz /home/user2/
and so on.  You'd have to specify the path of user1.tar.gz like where it
was going to go and the same for the other archive files too.


Jude <dashielljt(at)gmpexpress-dot-net>

-
To unsubscribe from this list: send the line "unsubscribe linux-newbie" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.linux-learn.org/faqs

Re: A few questions

[Adam Luchjenbroers]

On Monday 02 Dec 2002 11:56 pm, dashielljt wrote:
> You'd want to do more than one tar command of the form:
> tar zcvf user1.tar.gz /home/user1/
> tar zcvf user2.tar.gz /home/user2/
> and so on.  You'd have to specify the path of user1.tar.gz like where it
> was going to go and the same for the other archive files too.

Home directories are on a seperate partition, what I want to back up is the 
actual password/account list.

/etc/passwd seems to hold some of this, but it doesn't appear to show the 
actual password hashes (unless cat is programmed to replace them all with a 
'x').

-- 
"Not Hercules could have knock'd out his brains, for he had none." - 
Shakespeare

Please avoid sending me Word or PowerPoint attachments.
See http://www.fsf.org/philosophy/no-word-attachments.html


-
To unsubscribe from this list: send the line "unsubscribe linux-newbie" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.linux-learn.org/faqs

Re: A few questions

[Jim Reimer]

Adam Luchjenbroers wrote:
> On Monday 02 Dec 2002 11:56 pm, dashielljt wrote:
> 
>>You'd want to do more than one tar command of the form:
>>tar zcvf user1.tar.gz /home/user1/
>>tar zcvf user2.tar.gz /home/user2/
>>and so on.  You'd have to specify the path of user1.tar.gz like where it
>>was going to go and the same for the other archive files too.
> 
> 
> Home directories are on a seperate partition, what I want to back up is the 
> actual password/account list.
> 
> /etc/passwd seems to hold some of this, but it doesn't appear to show the 
> actual password hashes (unless cat is programmed to replace them all with a 
> 'x').

The 'x' indicates that you have passwords shadowed (they are held in a
seperate file readable only by root).

You can either unshadow the passwords while you do the backup
(man pwconv) or just backup the shadow file along with everything
else, which is probably less trouble.  You need to save these files:

/etc/passwd
/etc/group
/etc/shadow
/etc/gshadow

Since the shadow files are readable only by root, you'll obviously
have to be logged in as root to do the backup.

-jdr-





-
To unsubscribe from this list: send the line "unsubscribe linux-newbie" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.linux-learn.org/faqs

A few questions

[Rocco Stanzione]

I have a few questions about iptables, and about some of the traffic I've seen 
on this list.

1) I've successfully set up DNAT so that I can, say, request port 81 on the 
firewall/gateway box and hit the webserver on port 80 on a box behind the 
gateway.  No problem.  But it doesn't work on the LAN.  I understand why, 
after reading some docs, but I've been trying to find a way around it.  
Problem is, I don't know what all iptables will auto-reverse for me and what 
I need to specify.  For example, I suppose I could SNAT local connections 
through the gateway --to-source $GATEWAYIP, but would returning packets be 
taken care of, or must I come up with a rule for that?  I've tried both 
without success.

2) Quite a few people are using my iptables script, and the most common 
complaint I get is dcc failure.  ip_conntrack_irc (and everything else) is 
compiled into my (monolithic) kernel.  Most people don't have this setup.  
Dcc of course works for me.  I have the user modprobe ip_conntrack_irc, and 
typically it still doesn't work for them.  Logs show it failing at a 
catch-all rule, which tells me conntrack isn't working.  However if we add 
ip_conntrack_irc to /etc/modules and reboot, all is well.  So, 2a) why is 
this? and 2b) surely there's something less drastic than a reboot that will 
put ip_conntrack_irc to work?

3) I see quite a few messages on the list about people going to a great deal 
of trouble to get a VPN connection to NAT through an iptables firewall.  I 
have an always-on PPTP connection to the M$ VPN server at work, and I have 
never had to do anything special to get it to work.  I have to assume, then, 
that I have some liberal rule or policy that is likely insecure, and I wonder 
what it could be.  I'm afraid I don't have a pasteable rule set, as it's 
spread out across config files and such.  If anyone cares to pore over it, 
it's at http://www.linuxkungfu.org/ipkungfu-0.2.0.tgz

Thanks!

Rocco

Re: A few questions

[Nix N. Nix]

On Fri, 2002-11-15 at 03:20, Rocco Stanzione wrote:
> I have a few questions about iptables, and about some of the traffic I've seen 
> on this list.
> 
> 1) I've successfully set up DNAT so that I can, say, request port 81 on the 
> firewall/gateway box and hit the webserver on port 80 on a box behind the 
> gateway.  No problem.  But it doesn't work on the LAN.  I understand why, 
> after reading some docs, but I've been trying to find a way around it.  
> Problem is, I don't know what all iptables will auto-reverse for me and what 
> I need to specify.  For example, I suppose I could SNAT local connections 
> through the gateway --to-source $GATEWAYIP, but would returning packets be 
> taken care of, or must I come up with a rule for that?  I've tried both 
> without success.

AFAIK, DNAT and SNAT take care of entire connections, not just
individual packets. For example:

iptables -t nat --destination ${OUTSIDE_IP} --dport ${OUTSIDE_PORT} -j
DNAT --to-destination ${INSIDE_IP}:${INSIDE_PORT}

takes care of both packets of the form 

src:${OUTSIDE_IP}:${OUTSIDE_PORT}
dst:${INSIDE_IP}:${INSIDE_PORT}

as well as the return packets

src:${INSIDE_IP}:${INSIDE_PORT}
dst:${OUTSIDE_IP}:${OUTSIDE_PORT}

> 
> 2) Quite a few people are using my iptables script, and the most common 
> complaint I get is dcc failure.  ip_conntrack_irc (and everything else) is 
> compiled into my (monolithic) kernel.  Most people don't have this setup.  
> Dcc of course works for me.  I have the user modprobe ip_conntrack_irc, and 
> typically it still doesn't work for them.  Logs show it failing at a 
> catch-all rule, which tells me conntrack isn't working.  However if we add 
> ip_conntrack_irc to /etc/modules and reboot, all is well.  So, 2a) why is 
> this? and 2b) surely there's something less drastic than a reboot that will 
> put ip_conntrack_irc to work?

I think you need both ip_conntrack_irc.o and ip_nat_irc.o to get DCC
working.  I run a DCC fserve behind my firewall.

> 
> 3) I see quite a few messages on the list about people going to a great deal 
> of trouble to get a VPN connection to NAT through an iptables firewall.  I 
> have an always-on PPTP connection to the M$ VPN server at work, and I have 
> never had to do anything special to get it to work.  I have to assume, then, 
> that I have some liberal rule or policy that is likely insecure, and I wonder 
> what it could be.  I'm afraid I don't have a pasteable rule set, as it's 
> spread out across config files and such.  If anyone cares to pore over it, 
> it's at http://www.linuxkungfu.org/ipkungfu-0.2.0.tgz
> 
> Thanks!
> 
> Rocco