[gentoo-catalyst] changing inittab

[gentoo-catalyst] changing inittab

[Joshua Hoblitt]

[-- Attachment #1: Type: text/plain, Size: 405 bytes --]

Hi Folks,

I'm attempting to generate an amd64 livecd that starts up a tty on a
serial port.  So I want to do is add the line
"s0:12345:respawn:/sbin/agetty 115200 ttyS0 vt100" to the /etc/inittab.
The trouble is, I can't figure out exactly where iniitab is coming from.
Where does catalyst get /etc/inittab from?  Is it direct from the stage1
seed?  How should I go about modifying it?

Thanks,


-J

--

[-- Attachment #2: Type: application/pgp-signature, Size: 189 bytes --]

Re: [gentoo-catalyst] changing inittab

[Nelson Batalha]

I believe you can overwrite it in the overlay or you can do "equery
belongs /etc/inittab" in your own system to discover it belongs to
sys-apps/sysvinit, so you can change that too.

Joshua Hoblitt escreveu:
> Hi Folks,
> 
> I'm attempting to generate an amd64 livecd that starts up a tty on a
> serial port.  So I want to do is add the line
> "s0:12345:respawn:/sbin/agetty 115200 ttyS0 vt100" to the /etc/inittab.
> The trouble is, I can't figure out exactly where iniitab is coming from.
> Where does catalyst get /etc/inittab from?  Is it direct from the stage1
> seed?  How should I go about modifying it?
> 
> Thanks,
> 
> 
> -J
> 
> --

-- 
gentoo-catalyst@gentoo.org mailing list

Re: [gentoo-catalyst] changing inittab

[Scott Biddlestone]

If you installed livecd-tools, the inittab is overwritten by 
/sbin/livecd-functions.sh when the cd boots ( it is backed up to 
inittab.old ).  One way to fix it is to  modify the script by creating 
a  new livecd-tools ebuild in your portage overlay - copy the current 
ebuild and just add the patch to it. 

-Scott

Nelson Batalha wrote:
> I believe you can overwrite it in the overlay or you can do "equery
> belongs /etc/inittab" in your own system to discover it belongs to
> sys-apps/sysvinit, so you can change that too.
>
> Joshua Hoblitt escreveu:
>   
>> Hi Folks,
>>
>> I'm attempting to generate an amd64 livecd that starts up a tty on a
>> serial port.  So I want to do is add the line
>> "s0:12345:respawn:/sbin/agetty 115200 ttyS0 vt100" to the /etc/inittab.
>> The trouble is, I can't figure out exactly where iniitab is coming from.
>> Where does catalyst get /etc/inittab from?  Is it direct from the stage1
>> seed?  How should I go about modifying it?
>>
>> Thanks,
>>
>>
>> -J
>>
>> --
>>     
>
>   

-- 
gentoo-catalyst@gentoo.org mailing list

Re: [gentoo-catalyst] changing inittab

[Chris Gianelloni]

[-- Attachment #1: Type: text/plain, Size: 1705 bytes --]

On Wed, 2007-08-01 at 10:00 -0400, Scott Biddlestone wrote:
> If you installed livecd-tools, the inittab is overwritten by 
> /sbin/livecd-functions.sh when the cd boots ( it is backed up to 
> inittab.old ).  One way to fix it is to  modify the script by creating 
> a  new livecd-tools ebuild in your portage overlay - copy the current 
> ebuild and just add the patch to it. 

Please don't touch livecd-tools.  Don't touch sysvinit.   Instead, do
the following in your livecd-stage2.spec file...

livecd/bootargs: console=ttyS0,115200

That's it.  When the CD is built, it'll have the console appended to
both the "normal" and "nofb" kernels.  Either kernel will work perfectly
fine for serial console, so you're done.

> 
> -Scott
> 
> Nelson Batalha wrote:
> > I believe you can overwrite it in the overlay or you can do "equery
> > belongs /etc/inittab" in your own system to discover it belongs to
> > sys-apps/sysvinit, so you can change that too.
> >
> > Joshua Hoblitt escreveu:
> >   
> >> Hi Folks,
> >>
> >> I'm attempting to generate an amd64 livecd that starts up a tty on a
> >> serial port.  So I want to do is add the line
> >> "s0:12345:respawn:/sbin/agetty 115200 ttyS0 vt100" to the /etc/inittab.
> >> The trouble is, I can't figure out exactly where iniitab is coming from.
> >> Where does catalyst get /etc/inittab from?  Is it direct from the stage1
> >> seed?  How should I go about modifying it?
> >>
> >> Thanks,
> >>
> >>
> >> -J
> >>
> >> --
> >>     
> >
> >   
> 
-- 
Chris Gianelloni
Release Engineering Strategic Lead
Alpha/AMD64/x86 Architecture Teams
Games Developer/Council Member/Foundation Trustee
Gentoo Foundation

[-- Attachment #2: This is a digitally signed message part --]
[-- Type: application/pgp-signature, Size: 189 bytes --]

Re: [gentoo-catalyst] changing inittab

[Joshua Hoblitt]

[-- Attachment #1: Type: text/plain, Size: 961 bytes --]

On Wed, Aug 01, 2007 at 09:58:36AM -0700, Chris Gianelloni wrote:
> On Wed, 2007-08-01 at 10:00 -0400, Scott Biddlestone wrote:
> > If you installed livecd-tools, the inittab is overwritten by 
> > /sbin/livecd-functions.sh when the cd boots ( it is backed up to 
> > inittab.old ).  One way to fix it is to  modify the script by creating 
> > a  new livecd-tools ebuild in your portage overlay - copy the current 
> > ebuild and just add the patch to it. 
> 
> Please don't touch livecd-tools.  Don't touch sysvinit.   Instead, do
> the following in your livecd-stage2.spec file...
> 
> livecd/bootargs: console=ttyS0,115200
> 
> That's it.  When the CD is built, it'll have the console appended to
> both the "normal" and "nofb" kernels.  Either kernel will work perfectly
> fine for serial console, so you're done.

Eh, isn't that only good for kernel output?  I'd like to actually be
able to get a shell on the serial line...

-J

--

[-- Attachment #2: Type: application/pgp-signature, Size: 189 bytes --]

Re: [gentoo-catalyst] changing inittab

[Chris Gianelloni]

[-- Attachment #1: Type: text/plain, Size: 1840 bytes --]

On Wed, 2007-08-01 at 11:55 -1000, Joshua Hoblitt wrote:
> On Wed, Aug 01, 2007 at 09:58:36AM -0700, Chris Gianelloni wrote:
> > On Wed, 2007-08-01 at 10:00 -0400, Scott Biddlestone wrote:
> > > If you installed livecd-tools, the inittab is overwritten by 
> > > /sbin/livecd-functions.sh when the cd boots ( it is backed up to 
> > > inittab.old ).  One way to fix it is to  modify the script by creating 
> > > a  new livecd-tools ebuild in your portage overlay - copy the current 
> > > ebuild and just add the patch to it. 
> > 
> > Please don't touch livecd-tools.  Don't touch sysvinit.   Instead, do
> > the following in your livecd-stage2.spec file...
> > 
> > livecd/bootargs: console=ttyS0,115200
> > 
> > That's it.  When the CD is built, it'll have the console appended to
> > both the "normal" and "nofb" kernels.  Either kernel will work perfectly
> > fine for serial console, so you're done.
> 
> Eh, isn't that only good for kernel output?  I'd like to actually be
> able to get a shell on the serial line...

<rhetorical>
Why don't people trust that the guy who writes all this crap knows what
he's talking about?
</rhetorical>

:P

Normally, yes, it would only be good for kernel output, but as someone
else mentioned, you have livecd-functions.sh that re-writes the inittab.
Well, guess what triggers it?  That's right, a "console=" option on the
command line.  If you type (or via catalyst, add) console=something,
livecd-functions.sh (via functions.sh in baselayout) will interpret it
and modify inittab accordingly before we ever even hit runlevel 3, so it
starts a console on that serial line for login.

Try it.

-- 
Chris Gianelloni
Release Engineering Strategic Lead
Alpha/AMD64/x86 Architecture Teams
Games Developer/Council Member/Foundation Trustee
Gentoo Foundation

[-- Attachment #2: This is a digitally signed message part --]
[-- Type: application/pgp-signature, Size: 189 bytes --]

Re: [gentoo-catalyst] changing inittab

[Joshua Hoblitt]

[-- Attachment #1: Type: text/plain, Size: 2061 bytes --]

On Wed, Aug 01, 2007 at 05:26:03PM -0700, Chris Gianelloni wrote:
> On Wed, 2007-08-01 at 11:55 -1000, Joshua Hoblitt wrote:
> > On Wed, Aug 01, 2007 at 09:58:36AM -0700, Chris Gianelloni wrote:
> > > On Wed, 2007-08-01 at 10:00 -0400, Scott Biddlestone wrote:
> > > > If you installed livecd-tools, the inittab is overwritten by 
> > > > /sbin/livecd-functions.sh when the cd boots ( it is backed up to 
> > > > inittab.old ).  One way to fix it is to  modify the script by creating 
> > > > a  new livecd-tools ebuild in your portage overlay - copy the current 
> > > > ebuild and just add the patch to it. 
> > > 
> > > Please don't touch livecd-tools.  Don't touch sysvinit.   Instead, do
> > > the following in your livecd-stage2.spec file...
> > > 
> > > livecd/bootargs: console=ttyS0,115200
> > > 
> > > That's it.  When the CD is built, it'll have the console appended to
> > > both the "normal" and "nofb" kernels.  Either kernel will work perfectly
> > > fine for serial console, so you're done.
> > 
> > Eh, isn't that only good for kernel output?  I'd like to actually be
> > able to get a shell on the serial line...
> 
> <rhetorical>
> Why don't people trust that the guy who writes all this crap knows what
> he's talking about?
> </rhetorical>
> 
> :P

Your assertion is that it's completely obvous that a variable named
"bootargs", that passes options to the kernel, would have other magical
side effects?

> Normally, yes, it would only be good for kernel output, but as someone
> else mentioned, you have livecd-functions.sh that re-writes the inittab.
> Well, guess what triggers it?  That's right, a "console=" option on the
> command line.  If you type (or via catalyst, add) console=something,
> livecd-functions.sh (via functions.sh in baselayout) will interpret it
> and modify inittab accordingly before we ever even hit runlevel 3, so it
> starts a console on that serial line for login.

Seriously, can you cut and paste that paragraph into the manual?

Thanks for your help.

-J

--

[-- Attachment #2: Type: application/pgp-signature, Size: 189 bytes --]

Re: [gentoo-catalyst] changing inittab

[Joshua Hoblitt]

[-- Attachment #1: Type: text/plain, Size: 1175 bytes --]

On Wed, Aug 01, 2007 at 05:26:03PM -0700, Chris Gianelloni wrote:
> Normally, yes, it would only be good for kernel output, but as someone
> else mentioned, you have livecd-functions.sh that re-writes the inittab.
> Well, guess what triggers it?  That's right, a "console=" option on the
> command line.  If you type (or via catalyst, add) console=something,
> livecd-functions.sh (via functions.sh in baselayout) will interpret it
> and modify inittab accordingly before we ever even hit runlevel 3, so it
> starts a console on that serial line for login.

So I tried setting 'livecd/bootargs: console=ttyS0,115200' in
installcd-stage2-minimal.spec and indeed it does setup a tty on the
serial line.  However, it doesn't actually seem to pass "bootargs" to
the kernel or there is some other wierdness going on as the dmesg
information AND all of the sysvinit output remains only on the vga
console.  The last output on the fbconsole is some wierd error message
about bootsplash right before attaching a tty to the serial line.  This
is an improvement for me but I would really like to see the kernel
output/etc. so I can see if the boot processes has hung.

Cheers,

-J

--

[-- Attachment #2: Type: application/pgp-signature, Size: 189 bytes --]

Re: [gentoo-catalyst] changing inittab

[Andrew Gaffney]

Joshua Hoblitt wrote:
> On Wed, Aug 01, 2007 at 05:26:03PM -0700, Chris Gianelloni wrote:
>> Normally, yes, it would only be good for kernel output, but as someone
>> else mentioned, you have livecd-functions.sh that re-writes the inittab.
>> Well, guess what triggers it?  That's right, a "console=" option on the
>> command line.  If you type (or via catalyst, add) console=something,
>> livecd-functions.sh (via functions.sh in baselayout) will interpret it
>> and modify inittab accordingly before we ever even hit runlevel 3, so it
>> starts a console on that serial line for login.
> 
> So I tried setting 'livecd/bootargs: console=ttyS0,115200' in
> installcd-stage2-minimal.spec and indeed it does setup a tty on the
> serial line.  However, it doesn't actually seem to pass "bootargs" to
> the kernel or there is some other wierdness going on as the dmesg
> information AND all of the sysvinit output remains only on the vga
> console.  The last output on the fbconsole is some wierd error message
> about bootsplash right before attaching a tty to the serial line.  This
> is an improvement for me but I would really like to see the kernel
> output/etc. so I can see if the boot processes has hung.

You're positive that it's not being passed to the kernel? I'm pretty sure that 
livecd-tools would read it from /proc/cmdline, which is the arguments passed to 
the kernel. If it shows up in /proc/cmdline, then it's the kernel that isn't 
honoring it, not catalyst/genkernel/isolinux that is doing something wrong.

-- 
Andrew Gaffney                                 http://dev.gentoo.org/~agaffney/
Gentoo Linux Developer             Catalyst/Installer + x86 release coordinator
-- 
gentoo-catalyst@gentoo.org mailing list

Re: [gentoo-catalyst] changing inittab

[Chris Gianelloni]

[-- Attachment #1: Type: text/plain, Size: 2136 bytes --]

On Thu, 2007-08-02 at 10:28 -1000, Joshua Hoblitt wrote:
> > <rhetorical>
> > Why don't people trust that the guy who writes all this crap knows what
> > he's talking about?
> > </rhetorical>
> > 
> > :P
> 
> Your assertion is that it's completely obvous that a variable named
> "bootargs", that passes options to the kernel, would have other magical
> side effects?

No, my assertion is that when I provide a solution to someone's query
that I might know what I'm talking about.  ;]

> > Normally, yes, it would only be good for kernel output, but as someone
> > else mentioned, you have livecd-functions.sh that re-writes the inittab.
> > Well, guess what triggers it?  That's right, a "console=" option on the
> > command line.  If you type (or via catalyst, add) console=something,
> > livecd-functions.sh (via functions.sh in baselayout) will interpret it
> > and modify inittab accordingly before we ever even hit runlevel 3, so it
> > starts a console on that serial line for login.
> 
> Seriously, can you cut and paste that paragraph into the manual?

Ehh... What manual?

There's no manual for catalyst that is current.  Also, I don't have the
time (or the desire) to document all of the idiosyncrasies of the
packages that make up a Gentoo LiveCD build.  We use quite a few
packages that other people don't need to use on their own media.
Remember that everything written by Release Engineering is designed for
Release Engineering consumption.  No guarantees are made to anyone else
on anything.  We guarantee that it will build a release the way we want
it given a particular package set with specific versions, the exact ones
used by us for our builds.  Also, realize that livecd-tools is *not*
catalyst and shouldn't be documented with catalyst.

> Thanks for your help.

You're quite welcome.  This list is pretty much here just so people can
get help using catalyst, so feel free to ask anything that you need.

-- 
Chris Gianelloni
Release Engineering Strategic Lead
Alpha/AMD64/x86 Architecture Teams
Games Developer/Council Member/Foundation Trustee
Gentoo Foundation

[-- Attachment #2: This is a digitally signed message part --]
[-- Type: application/pgp-signature, Size: 189 bytes --]

Re: [gentoo-catalyst] changing inittab

[Chris Gianelloni]

[-- Attachment #1: Type: text/plain, Size: 1861 bytes --]

On Thu, 2007-08-02 at 13:05 -1000, Joshua Hoblitt wrote:
> On Wed, Aug 01, 2007 at 05:26:03PM -0700, Chris Gianelloni wrote:
> > Normally, yes, it would only be good for kernel output, but as someone
> > else mentioned, you have livecd-functions.sh that re-writes the inittab.
> > Well, guess what triggers it?  That's right, a "console=" option on the
> > command line.  If you type (or via catalyst, add) console=something,
> > livecd-functions.sh (via functions.sh in baselayout) will interpret it
> > and modify inittab accordingly before we ever even hit runlevel 3, so it
> > starts a console on that serial line for login.
> 
> So I tried setting 'livecd/bootargs: console=ttyS0,115200' in
> installcd-stage2-minimal.spec and indeed it does setup a tty on the
> serial line.  However, it doesn't actually seem to pass "bootargs" to
> the kernel or there is some other wierdness going on as the dmesg
> information AND all of the sysvinit output remains only on the vga
> console.  The last output on the fbconsole is some wierd error message
> about bootsplash right before attaching a tty to the serial line.  This
> is an improvement for me but I would really like to see the kernel
> output/etc. so I can see if the boot processes has hung.

Try it with nofb and let me know if it works.  It looks like the problem
is the order in which the command line options are passed.  If it does
work, file a bug asking me to reverse the order than custom arguments
get added with the splash arguments.  Currently, the splash stuff is
always last.  If the kernel has more than one "console=" line, the last
one gets the kernel output, so the splash one is getting it.

-- 
Chris Gianelloni
Release Engineering Strategic Lead
Alpha/AMD64/x86 Architecture Teams
Games Developer/Council Member/Foundation Trustee
Gentoo Foundation

[-- Attachment #2: This is a digitally signed message part --]
[-- Type: application/pgp-signature, Size: 189 bytes --]

Re: [gentoo-catalyst] changing inittab

[Chris Gianelloni]

[-- Attachment #1: Type: text/plain, Size: 2019 bytes --]

On Thu, 2007-08-02 at 19:01 -0500, Andrew Gaffney wrote:
> Joshua Hoblitt wrote:
> > On Wed, Aug 01, 2007 at 05:26:03PM -0700, Chris Gianelloni wrote:
> >> Normally, yes, it would only be good for kernel output, but as someone
> >> else mentioned, you have livecd-functions.sh that re-writes the inittab.
> >> Well, guess what triggers it?  That's right, a "console=" option on the
> >> command line.  If you type (or via catalyst, add) console=something,
> >> livecd-functions.sh (via functions.sh in baselayout) will interpret it
> >> and modify inittab accordingly before we ever even hit runlevel 3, so it
> >> starts a console on that serial line for login.
> > 
> > So I tried setting 'livecd/bootargs: console=ttyS0,115200' in
> > installcd-stage2-minimal.spec and indeed it does setup a tty on the
> > serial line.  However, it doesn't actually seem to pass "bootargs" to
> > the kernel or there is some other wierdness going on as the dmesg
> > information AND all of the sysvinit output remains only on the vga
> > console.  The last output on the fbconsole is some wierd error message
> > about bootsplash right before attaching a tty to the serial line.  This
> > is an improvement for me but I would really like to see the kernel
> > output/etc. so I can see if the boot processes has hung.
> 
> You're positive that it's not being passed to the kernel? I'm pretty sure that 
> livecd-tools would read it from /proc/cmdline, which is the arguments passed to 
> the kernel. If it shows up in /proc/cmdline, then it's the kernel that isn't 
> honoring it, not catalyst/genkernel/isolinux that is doing something wrong.

I think it is an ordering issue, rather than it not being passed.  I
just checked the code and the splash code appends to the very end of the
kernel command line, meaning it takes precedence.

-- 
Chris Gianelloni
Release Engineering Strategic Lead
Alpha/AMD64/x86 Architecture Teams
Games Developer/Council Member/Foundation Trustee
Gentoo Foundation

[-- Attachment #2: This is a digitally signed message part --]
[-- Type: application/pgp-signature, Size: 189 bytes --]

Re: [gentoo-catalyst] changing inittab

[Joshua Hoblitt]

[-- Attachment #1: Type: text/plain, Size: 2105 bytes --]

On Fri, Aug 03, 2007 at 12:01:23PM -0700, Chris Gianelloni wrote:
> On Thu, 2007-08-02 at 19:01 -0500, Andrew Gaffney wrote:
> > Joshua Hoblitt wrote:
> > > On Wed, Aug 01, 2007 at 05:26:03PM -0700, Chris Gianelloni wrote:
> > >> Normally, yes, it would only be good for kernel output, but as someone
> > >> else mentioned, you have livecd-functions.sh that re-writes the inittab.
> > >> Well, guess what triggers it?  That's right, a "console=" option on the
> > >> command line.  If you type (or via catalyst, add) console=something,
> > >> livecd-functions.sh (via functions.sh in baselayout) will interpret it
> > >> and modify inittab accordingly before we ever even hit runlevel 3, so it
> > >> starts a console on that serial line for login.
> > > 
> > > So I tried setting 'livecd/bootargs: console=ttyS0,115200' in
> > > installcd-stage2-minimal.spec and indeed it does setup a tty on the
> > > serial line.  However, it doesn't actually seem to pass "bootargs" to
> > > the kernel or there is some other wierdness going on as the dmesg
> > > information AND all of the sysvinit output remains only on the vga
> > > console.  The last output on the fbconsole is some wierd error message
> > > about bootsplash right before attaching a tty to the serial line.  This
> > > is an improvement for me but I would really like to see the kernel
> > > output/etc. so I can see if the boot processes has hung.
> > 
> > You're positive that it's not being passed to the kernel? I'm pretty sure that 
> > livecd-tools would read it from /proc/cmdline, which is the arguments passed to 
> > the kernel. If it shows up in /proc/cmdline, then it's the kernel that isn't 
> > honoring it, not catalyst/genkernel/isolinux that is doing something wrong.
> 
> I think it is an ordering issue, rather than it not being passed.  I
> just checked the code and the splash code appends to the very end of the
> kernel command line, meaning it takes precedence.

I won't be able to test anything until Monday.  Would an option to
completely disable the splash be reasonable?

-J

--

[-- Attachment #2: Type: application/pgp-signature, Size: 189 bytes --]

Re: [gentoo-catalyst] changing inittab

[Andrew Gaffney]

Joshua Hoblitt wrote:
> On Fri, Aug 03, 2007 at 12:01:23PM -0700, Chris Gianelloni wrote:
>> On Thu, 2007-08-02 at 19:01 -0500, Andrew Gaffney wrote:
>>> Joshua Hoblitt wrote:
>>>> On Wed, Aug 01, 2007 at 05:26:03PM -0700, Chris Gianelloni wrote:
>>>>> Normally, yes, it would only be good for kernel output, but as someone
>>>>> else mentioned, you have livecd-functions.sh that re-writes the inittab.
>>>>> Well, guess what triggers it?  That's right, a "console=" option on the
>>>>> command line.  If you type (or via catalyst, add) console=something,
>>>>> livecd-functions.sh (via functions.sh in baselayout) will interpret it
>>>>> and modify inittab accordingly before we ever even hit runlevel 3, so it
>>>>> starts a console on that serial line for login.
>>>> So I tried setting 'livecd/bootargs: console=ttyS0,115200' in
>>>> installcd-stage2-minimal.spec and indeed it does setup a tty on the
>>>> serial line.  However, it doesn't actually seem to pass "bootargs" to
>>>> the kernel or there is some other wierdness going on as the dmesg
>>>> information AND all of the sysvinit output remains only on the vga
>>>> console.  The last output on the fbconsole is some wierd error message
>>>> about bootsplash right before attaching a tty to the serial line.  This
>>>> is an improvement for me but I would really like to see the kernel
>>>> output/etc. so I can see if the boot processes has hung.
>>> You're positive that it's not being passed to the kernel? I'm pretty sure that 
>>> livecd-tools would read it from /proc/cmdline, which is the arguments passed to 
>>> the kernel. If it shows up in /proc/cmdline, then it's the kernel that isn't 
>>> honoring it, not catalyst/genkernel/isolinux that is doing something wrong.
>> I think it is an ordering issue, rather than it not being passed.  I
>> just checked the code and the splash code appends to the very end of the
>> kernel command line, meaning it takes precedence.
> 
> I won't be able to test anything until Monday.  Would an option to
> completely disable the splash be reasonable?

That's what the -nofb kernel does. However, there isn't currently an option to 
have a framebuffer without splash. I'm not sure if it would still pass the 
splash options if you don't enable splash stuff in livecd-stage2.spec, but I bet 
it would.

-- 
Andrew Gaffney                                 http://dev.gentoo.org/~agaffney/
Gentoo Linux Developer             Catalyst/Installer + x86 release coordinator
-- 
gentoo-catalyst@gentoo.org mailing list