2.5.68 kernel no initrd

2.5.68 kernel no initrd

[Dave Mehler]

Hello,
    Ok, i should learn to leave well enough alone, but i don't. After
successfully installing a monolithic 2.5.67 kernel i decided i wanted
modules, so i made them, and what happened, it hung after the initrd
initialized. So, when 2.5.68 came out i of course grab it, compile/install
it without a hitch, but for one thing, as of now make install did not make
an initrd for that install. Does anyone know how to make this manually, it
won't boot without one?
Thanks.
Dave.

Re: 2.5.68 kernel no initrd

[Felipe Alfaro Solana]

On Sun, 2003-04-20 at 06:39, Dave Mehler wrote:
>     Ok, i should learn to leave well enough alone, but i don't. After
> successfully installing a monolithic 2.5.67 kernel i decided i wanted
> modules, so i made them, and what happened, it hung after the initrd
> initialized. So, when 2.5.68 came out i of course grab it, compile/install
> it without a hitch, but for one thing, as of now make install did not make
> an initrd for that install. Does anyone know how to make this manually, it
> won't boot without one?

I don't have experience with initrd, but why would you want a initrd?
Can't you simply build into the kernel the required pieces to mount the
root filesystem and leave the rest as loadable modules?
-- 
Please AVOID sending me WORD, EXCEL or POWERPOINT attachments.
See http://www.fsf.org/philosophy/no-word-attachments.html
Linux Registered User #287198

Re: 2.5.68 kernel no initrd

[Bill Davidsen]

On 20 Apr 2003, Felipe Alfaro Solana wrote:

> I don't have experience with initrd, but why would you want a initrd?
> Can't you simply build into the kernel the required pieces to mount the
> root filesystem and leave the rest as loadable modules?

As long as you are suporting or testing a small number of configurations,
say less than two, initrd buys not much. But using all modules allows you
to support a new config just by rebuilding initrd, vastly faster than
building a kernel on most machines. And for a vendor doing a distribution
it is a huge win.

Not to mention some drivers work differently when loaded as modules,
loading as modules allows control of the device scans, etc, etc. It's just
more flexible.

-- 
bill davidsen <davidsen@tmr.com>
  CTO, TMR Associates, Inc
Doing interesting things with little computers since 1979.

Re: 2.5.68 kernel no initrd

[Paweł Gołaszewski]

On Sun, 20 Apr 2003, Felipe Alfaro Solana wrote:
> > Ok, i should learn to leave well enough alone, but i don't. After
> > successfully installing a monolithic 2.5.67 kernel i decided i wanted
> > modules, so i made them, and what happened, it hung after the initrd
> > initialized. So, when 2.5.68 came out i of course grab it,
> > compile/install it without a hitch, but for one thing, as of now make
> > install did not make an initrd for that install. Does anyone know how
> > to make this manually, it won't boot without one?
> I don't have experience with initrd, but why would you want a initrd?
> Can't you simply build into the kernel the required pieces to mount the
> root filesystem and leave the rest as loadable modules?

initrd gives much more flexibility.
I can make one kernel and use it on _all_ of my mashines, just change 
initrd. quick, nice and flexible with proper initrd tools set.

-- 
pozdr.  Paweł Gołaszewski        
---------------------------------
worth to see: http://www.againsttcpa.com/
CPU not found - software emulation...

Re: 2.5.68 kernel no initrd

[Valdis.Kletnieks]

[-- Attachment #1: Type: text/plain, Size: 476 bytes --]

On Wed, 23 Apr 2003 15:49:54 +0200, =?ISO-8859-2?Q?Pawe=B3_Go=B3aszewski?= said:
> initrd gives much more flexibility.
> I can make one kernel and use it on _all_ of my mashines, just change 
> initrd. quick, nice and flexible with proper initrd tools set.

Amen.  initrd isn't just for modules - I'd not need an initrd at all if I could
figure out how to start up an LVM volume group from kernelspace - I suspect
people with / on a RAID disk have similar issues...


[-- Attachment #2: Type: application/pgp-signature, Size: 226 bytes --]

Re: 2.5.68 kernel no initrd

[Kevin P. Fleming]

Valdis.Kletnieks@vt.edu wrote:

> On Wed, 23 Apr 2003 15:49:54 +0200, =?ISO-8859-2?Q?Pawe=B3_Go=B3aszewski?= said:
> 
>>initrd gives much more flexibility.
>>I can make one kernel and use it on _all_ of my mashines, just change 
>>initrd. quick, nice and flexible with proper initrd tools set.
> 
> 
> Amen.  initrd isn't just for modules - I'd not need an initrd at all if I could
> figure out how to start up an LVM volume group from kernelspace - I suspect
> people with / on a RAID disk have similar issues...
> 

Well, even though I'm working on a solution to that, it still involves 
early userspace, just not the heavyweight "fake root" userspace that an 
initrd represents. This is what the initramfs technology in 2.5.X is 
for, so eventually (soon, hopefully) you'll be able to start md devices, 
LVM volume groups, etc. from early userspace and not have to have any 
autostart logic in the kernel nor will you have build and maintain an 
initrd separate from the kernel.

Re: 2.5.68 kernel no initrd

[Felipe Alfaro Solana]

On Wed, 2003-04-23 at 15:49, Paweł Gołaszewski wrote:
> initrd gives much more flexibility.
> I can make one kernel and use it on _all_ of my mashines, just change 
> initrd. quick, nice and flexible with proper initrd tools set.

I don't have any doubts that initrd is a very flexible solution and
provides for a generic kernel. However, in the end (I'm talking about my
experiences), initrd has caused me more troubles than problems it
solved. I always keep all "config" file for every kernel I use on my
machines.

-- 
Please AVOID sending me WORD, EXCEL or POWERPOINT attachments.
See http://www.fsf.org/philosophy/no-word-attachments.html
Linux Registered User #287198

Re: 2.5.68 kernel no initrd

[Bill Davidsen]

[-- Warning: decoded text below may be mangled, UTF-8 assumed --]
[-- Attachment #1: Type: TEXT/PLAIN; charset=US-ASCII, Size: 2076 bytes --]

On Wed, 23 Apr 2003, Kevin P. Fleming wrote:

> Valdis.Kletnieks@vt.edu wrote:
> 
> > On Wed, 23 Apr 2003 15:49:54 +0200, =?ISO-8859-2?Q?Pawe=B3_Go=B3aszewski?= said:
> > 
> >>initrd gives much more flexibility.
> >>I can make one kernel and use it on _all_ of my mashines, just change 
> >>initrd. quick, nice and flexible with proper initrd tools set.
> > 
> > 
> > Amen.  initrd isn't just for modules - I'd not need an initrd at all if I could
> > figure out how to start up an LVM volume group from kernelspace - I suspect
> > people with / on a RAID disk have similar issues...
> > 
> 
> Well, even though I'm working on a solution to that, it still involves 
> early userspace, just not the heavyweight "fake root" userspace that an 
> initrd represents. This is what the initramfs technology in 2.5.X is 
> for, so eventually (soon, hopefully) you'll be able to start md devices, 
> LVM volume groups, etc. from early userspace and not have to have any 
> autostart logic in the kernel nor will you have build and maintain an 
> initrd separate from the kernel.

It isn't too important where the setup resides in terms of flexibility,
the benefit comes from avoiding building one kernel for each
configuration.

On 23 Apr 2003, Felipe Alfaro Solana wrote:

> On Wed, 2003-04-23 at 15:49, Pawe³ Go³aszewski wrote:
> > initrd gives much more flexibility.
> > I can make one kernel and use it on _all_ of my mashines, just change 
> > initrd. quick, nice and flexible with proper initrd tools set.
> 
> I don't have any doubts that initrd is a very flexible solution and
> provides for a generic kernel. However, in the end (I'm talking about my
> experiences), initrd has caused me more troubles than problems it
> solved. I always keep all "config" file for every kernel I use on my
> machines.

Other than needing to build and maintain all those kernels, what does it
gain you over installing the  modules you need and having a single kernel?

-- 
bill davidsen <davidsen@tmr.com>
  CTO, TMR Associates, Inc
Doing interesting things with little computers since 1979.

Re: 2.5.68 kernel no initrd

[Felipe Alfaro Solana]

On Thu, 2003-04-24 at 22:19, Bill Davidsen wrote:
> > I don't have any doubts that initrd is a very flexible solution and
> > provides for a generic kernel. However, in the end (I'm talking about my
> > experiences), initrd has caused me more troubles than problems it
> > solved. I always keep all "config" file for every kernel I use on my
> > machines.
> 
> Other than needing to build and maintain all those kernels, what does it
> gain you over installing the  modules you need and having a single kernel?

Simply said, I don't like "initrd". I have still to figure out how to
make an "initrd" that includes the newer modutils.

Anyways, I don't mantain so much kernels (only 3 or 4 different
versions), and have a mix of AMD/P4/P3/P2/PI machines that need things
that can't be configured/don't yet work as modules, so I prefer to pass
initrd by...

-- 
Please AVOID sending me WORD, EXCEL or POWERPOINT attachments.
See http://www.fsf.org/philosophy/no-word-attachments.html
Linux Registered User #287198