Re: [why is kernel recompilation necessary?]

[Pollywog]

On 28-Jul-2000 Noah L. Meyerhans wrote:
> C'mon, you don't have to recompile MS Windows in order to install a
> "module".  Do you really think the Linux community would let Windows
> actually do something *better* than Linux?  ;^)

Well let's just say faxing was easy in Windows, but that is one thing many
of us have difficulty with in Linux.

--
Andrew

Re: [why is kernel recompilation necessary?]

[Noah L. Meyerhans]

-BEGIN PGP SIGNED MESSAGE-

On Fri, 28 Jul 2000, Pollywog wrote:

> On 28-Jul-2000 Noah L. Meyerhans wrote:
> > That's not the case at all.  As some people have mentioned in this thread,
> > certain proprietary drivers are distrubuted as binary-only kernel
> > modules.  This would obviously not be possible if the whole kernel had to
> > be rebuilt to use a module.  The commercial version of the Open Sound
> > System is one example of this (or was; I haven't used it since I built a
> > new machine with an ALSA supported sound card).
> 
> I use the OSS commercial driver and was not aware that it is a kernel module.
> It seems that whenever I have installed a kernel module, I had to recompile
> the kernel; that is how I was told to do it.

There are other examples, too, of course.  The userlink module, part of
some IPSec package is distributed separately from the kernel.  The
installation instructions consist of ./configure && make && make install.  
The 3dfx module needed to run glide apps as non-root users is another
example.

C'mon, you don't have to recompile MS Windows in order to install a
"module".  Do you really think the Linux community would let Windows
actually do something *better* than Linux?  ;^)

noah
 ___
| Web: http://web.morgul.net/~frodo/
| PGP Public Key: http://web.morgul.net/~frodo/mail.html 

-BEGIN PGP SIGNATURE-
Version: PGPfreeware 5.0i for non-commercial use
Charset: noconv

iQCVAwUBOYHUm4dCcpBjGWoFAQF4AwP/SkBFqY4qd0FECWdi4OoXUOEt6FbFwjBd
qvclkkmUttLUWLVO62t8TIzSx2Hgi0J5aIj4JBI7XTkpoGn8Jh5HCTv5Tx759zvi
EEyX/+/B84UPKIand610v/8GV+j3RRVD+AtwVbWL+/q3zDSkPNA9revjyS5R3XfS
RQZ7/At2Xhw=
=BoHR
-END PGP SIGNATURE-

Re: [why is kernel recompilation necessary?]

[Pollywog]

On 28-Jul-2000 Noah L. Meyerhans wrote:
> That's not the case at all.  As some people have mentioned in this thread,
> certain proprietary drivers are distrubuted as binary-only kernel
> modules.  This would obviously not be possible if the whole kernel had to
> be rebuilt to use a module.  The commercial version of the Open Sound
> System is one example of this (or was; I haven't used it since I built a
> new machine with an ALSA supported sound card).

I use the OSS commercial driver and was not aware that it is a kernel module.
It seems that whenever I have installed a kernel module, I had to recompile
the kernel; that is how I was told to do it.

*still a newbie*

--
Andrew

Re: [why is kernel recompilation necessary?]

[Noah L. Meyerhans]

-BEGIN PGP SIGNED MESSAGE-

On Fri, 28 Jul 2000, Pollywog wrote:

> On 28-Jul-2000 Kent West wrote:
> > 
> > In still other words, can you use "make menuconfig" to compile a minimal
> > kernel and then add modules later from whatever source even though you
> > didn't tell the kernel to expect these modules when you did the "make
> > menuconfig".
> 
> I don't think that would work, because you have to tell the kernel which
> modules you want, or they won't load when you need them. If you compile
> the just modules, the kernel won't know what to do with them.

That's not the case at all.  As some people have mentioned in this thread,
certain proprietary drivers are distrubuted as binary-only kernel
modules.  This would obviously not be possible if the whole kernel had to
be rebuilt to use a module.  The commercial version of the Open Sound
System is one example of this (or was; I haven't used it since I built a
new machine with an ALSA supported sound card).

noah
 ___
| Web: http://web.morgul.net/~frodo/
| PGP Public Key: http://web.morgul.net/~frodo/mail.html 

-BEGIN PGP SIGNATURE-
Version: PGPfreeware 5.0i for non-commercial use
Charset: noconv

iQCVAwUBOYHD5IdCcpBjGWoFAQE2EgP/axBAqQeI4BYi9xH5t6sKlebM1xJa9uGB
CDkC9Ygybu2nE7c8/Zm+hc42yQFqj6gALkPEjGTqubEZs9+0qqLHmtnkiEao06IJ
quGEV6Pr56c/2IKaF+3VfVB8iGch66tiWiBchKXhKNk8aWuePkoD1OFgzBoO/5qt
+qUiwECLkSs=
=Zrm8
-END PGP SIGNATURE-

Re: [why is kernel recompilation necessary?]

[I. Tura]

>The whole point of my post was to not have to read 17 web pages!
>chris

Oh man, nothing is perfect! But let me tell ou something. When some
hardware companies are such bastards that they do such indecent drivers
that collide with Windoze and at my left I have a scanner that I know that
it will only run under Linux, where all processes are clear and
well-documented... Or when actually I have to tell with the disk drivers to
W98 _SE_ that the modem it has is this one (a model of _1996_), not
another... The very cheap Plug & Play.

Under Debian I've always solved all my problems. Not in Windoze, 9X or 
NT.

Debian is some effort... But it gives you lots of rewards.


I.T.


___
Do You Yahoo!?
Achetez, vendez! À votre prix! Sur http://encheres.yahoo.fr

Re: [why is kernel recompilation necessary?]

[Pollywog]

On 28-Jul-2000 Kent West wrote:
> 
> In still other words, can you use "make menuconfig" to compile a minimal
> kernel and then add modules later from whatever source even though you
> didn't tell the kernel to expect these modules when you did the "make
> menuconfig".

I don't think that would work, because you have to tell the kernel which
modules you want, or they won't load when you need them. If you compile
the just modules, the kernel won't know what to do with them.

--
Andrew

Re: why is kernel recompilation necessary?

[Ethan Pierce]

If 128 mb of ram is considered a beast of a machine?  What class does my ½ gig 
of ram box fall in :)

>>> Daniel Reuter <[EMAIL PROTECTED]> 07/28/00 06:22AM >>>
Hello there,

On Fri, 28 Jul 2000, Preben Randhol wrote:

> Krzys Majewski <[EMAIL PROTECTED]> wrote on 28/07/2000 (00:29) :
> > Why is it that under Windows or whatever I don't have to recompile
> > the kernel just to add a new driver? Is it a protection thing? 
> > Or an optimization thing? Or something else? -chris
> 
> Usually you don't have to recompile your kernel under Linux. Just use
> the kernel-package that contains the kernel with all the modules you
> need. 

You don't HAVE to, but if you want a really fast, memory saving kernel,
you SHOULD do it and exclude everything, you don't need. (Installation
kernel was almost twice as large as the kernel I compiled by myself, as I
could exclude SCSI-support and a few other things). Sure not a point of
much interest on those beasts with 128 Mb RAM sold today, but on a
computer with 8 Mb RAM, a large kernel eats up your vital memory.
So in some way, really an optimization thing.

> I guess you could recompile your Windows kernel too _if_ you have
> access to the Windows source files and own a compiler.

Right.

Regards,
Daniel



-- 
Unsubscribe?  mail -s unsubscribe [EMAIL PROTECTED] < /dev/null

Re: why is kernel recompilation necessary?

[David Wright]

Quoting Krzys Majewski ([EMAIL PROTECTED]):
> Really? So the kernel doesn't compile any hooks for itself to enable
> loading latesthardwaredevice.o as a module? -chris

Yes, it has to do that, but the installation kernel has those
hooks compiled because it uses modules itself (e.g. for network
cards, sound etc.).

BUT the installation kernels have to have built-in all the drivers
that anyone might need to get at their root filing system, plus
anything else that still needs to be compiled as built-in (because
there are people who wish to continue to use these kernels for good).

> > >Why is it that under Windows or whatever I don't have to recompile
> > >the kernel just to add a new driver? Is it a protection thing? 
> > >Or an optimization thing? Or something else?

It's a "protection" thing. Protection of their secrets. Only they
can recompile from the source.

Here's one of the advantages of recompilation:

potato installation:
-rwxr-xr-x1 root root  1046612 Jun  7 17:57 vmlinuz-2.2.15

the kernel that I use on my nine machines, from old 486s
to 350MHz SCSI Pentiums:
-rw-r--r--1 root root   501346 Jul  7 11:40 vmlinuz-2.2.15

Cheers,

-- 
Email:  [EMAIL PROTECTED]   Tel: +44 1908 653 739  Fax: +44 1908 655 151
Snail:  David Wright, Earth Science Dept., Milton Keynes, England, MK7 6AA
Disclaimer:   These addresses are only for reaching me, and do not signify
official stationery. Views expressed here are either my own or plagiarised.

Re: why is kernel recompilation necessary?

[Preben Randhol]

Daniel Reuter <[EMAIL PROTECTED]> wrote on 28/07/2000 (12:20) :
> You don't HAVE to, but if you want a really fast, memory saving kernel,
> you SHOULD do it and exclude everything, you don't need. (Installation

But of course. :-)

-- 
Preben Randhol -- [EMAIL PROTECTED] -- 
+---+ "There was, I think, never any reason to  believe in any innate
| ! |  superiority of the male, except his superior muscle."
+---+  -- Bertrand Russell, Ideas That Have Harmed Mankind (1950)

Re: why is kernel recompilation necessary?

[Daniel Reuter]

Hello there,

On Fri, 28 Jul 2000, Preben Randhol wrote:

> Krzys Majewski <[EMAIL PROTECTED]> wrote on 28/07/2000 (00:29) :
> > Why is it that under Windows or whatever I don't have to recompile
> > the kernel just to add a new driver? Is it a protection thing? 
> > Or an optimization thing? Or something else? -chris
> 
> Usually you don't have to recompile your kernel under Linux. Just use
> the kernel-package that contains the kernel with all the modules you
> need. 

You don't HAVE to, but if you want a really fast, memory saving kernel,
you SHOULD do it and exclude everything, you don't need. (Installation
kernel was almost twice as large as the kernel I compiled by myself, as I
could exclude SCSI-support and a few other things). Sure not a point of
much interest on those beasts with 128 Mb RAM sold today, but on a
computer with 8 Mb RAM, a large kernel eats up your vital memory.
So in some way, really an optimization thing.

> I guess you could recompile your Windows kernel too _if_ you have
> access to the Windows source files and own a compiler.

Right.

Regards,
Daniel

Re: [why is kernel recompilation necessary?]

[Lehel Bernadt]

On 28-Jul-2000 Krzys Majewski wrote:
> Yes but what I'm wondering is not why linux users recompile their kernels,
> or why windows users can't, but how is it that windows users get away with
> not having to? The closest answer I got is that windows kernels have 
> a bunch of drivers already compiled in, and any additional drivers
> compiled as modules. What I'm still not clear on is whether either 
> windows or linux kernels (or both) need to have some sort of hooks 
> enabling them to expect whatever modules at runtime, or not. 
> For example, if I build a new device and write a driver for it,
> can I add support for this device to both windows and linux without
> having to modify either kernel? -chris
> 

First, I don't think you can compare the windows and linux kernels. Linux was
designed as a monolithic unix kernel, while the windows kernel...uhm...I really
don't know if it was designed at all... Second, what do you mean by modifying
the windows kernel ? Disassembling and binary patching ?
More seriously, the linux kernel has a module interface defined in module.h.
Shortly, in your code you have to use init_module, cleanup_module,
module_register_chrdev etc. As long as this interface doesn't change, you can
compile your module separately and use it with any kernel version (unless you
use something in your code that's specific to a certain version).
If you want to know more, I suggest to read the docs dealing with the linux
kernel at linuxdoc.org.

Re: [why is kernel recompilation necessary?]

[Morten Liebach]

On 27, jul, 2000 at 09:18:07 -0500, John Hasler wrote:
> Krzys Majewski writes:
> > For example, if I build a new device and write a driver for it, can I add
> > support for this device to both windows and linux without having to
> > modify either kernel?
> 
> You can for Linux, and probably for Windows as well.  In a sense you are
> modifying the kernel when you load a module.
> 
In windows you have to reboot to load new modules of course ...

Regards
Morten

-- 
UNIX, reach out and grep someone!

Re: [why is kernel recompilation necessary?]

[Matthew Dalton]

Well, I *think* so. But I'm not speaking from experience, just reason.

Try it and see.

Krzys Majewski wrote:
> 
> Woo, so I can just do a
> 
> make xconfig && make modules && make modules_install
> 
> instead of
> 
> make xconfig && make dep && make clean && make modules && make modules_install
> 
> ??
> 
> -chris

Re: why is kernel recompilation necessary?

[Preben Randhol]

Krzys Majewski <[EMAIL PROTECTED]> wrote on 28/07/2000 (00:29) :
> Why is it that under Windows or whatever I don't have to recompile
> the kernel just to add a new driver? Is it a protection thing? 
> Or an optimization thing? Or something else? -chris

Usually you don't have to recompile your kernel under Linux. Just use
the kernel-package that contains the kernel with all the modules you
need. 

I guess you could recompile your Windows kernel too _if_ you have
access to the Windows source files and own a compiler.

-- 
Preben Randhol -- [EMAIL PROTECTED] -- 
+---+ "There was, I think, never any reason to  believe in any innate
| ! |  superiority of the male, except his superior muscle."
+---+  -- Bertrand Russell, Ideas That Have Harmed Mankind (1950)

Re: [why is kernel recompilation necessary?]

[Krzys Majewski]

Yes!
-chris


On 27 Jul 2000, John Hasler wrote:

> Krzys Majewski writes:
> > The whole point of my post was to not have to read 17 web pages!
> 
> Why?  Do you find learning painful?
> -- 
> John Hasler
> [EMAIL PROTECTED] (John Hasler)
> Dancing Horse Hill
> Elmwood, WI
> 
> 
> -- 
> Unsubscribe?  mail -s unsubscribe [EMAIL PROTECTED] < /dev/null
> 

Re: [why is kernel recompilation necessary?]

[Krzys Majewski]

Woo, so I can just do a 

make xconfig && make modules && make modules_install 

instead of 

make xconfig && make dep && make clean && make modules && make modules_install

??

-chris


On Fri, 28 Jul 2000, Matthew Dalton wrote:

> By reasoning I would assume that all the kernel has is a 'generic module
> hook' that you can load any module into. Therefore, your sequence above
> should work. If it did not, how would you be able to use binary only
> modules such as the lucent winmodem driver?
> 
> So in answer to your questions:
> 
> Kent West wrote:
> > When you do a "make menuconfig" (or one of the other methods), and you
> > specify to include support for, say, a 3c905 NIC, as a module, are you
> > doing anything to the kernel, or are you just making changes to a script
> > to tell it to compile the module?
> 
> You're making changes to a script. Of course, you have to compile module
> support into the kernel as well, otherwise it won't be able to load any
> modules.
> 
> > In still other words, can you do the following?
> >   * use "make menuconfig" today to specify kernel options for a minimal
> > kernel, and not mark module stuff like NICs and sound cards, etc,
> >   * then "make dep" and "make zImage" to compile the kernel,
> >   * then boot off that kernel and run for a day or two
> >   * then come back in a day or two and re-run "make menuconfig" and
> > specify some modules
> >   * then "make modules" and "make modules_install" without compiling the
> > kernel
> >   * resulting in a working kernel that can use the modules compiled a
> > day or two later
> 
> As long as you say yes to module support, yes.
> 
> > In still other words, can you use "make menuconfig" to compile a minimal
> > kernel and then add modules later from whatever source even though you
> > didn't tell the kernel to expect these modules when you did the "make
> > menuconfig".
> 
> Yes. That's how binary-only modules work.
> 
> Matthew
> 
> 
> -- 
> Unsubscribe?  mail -s unsubscribe [EMAIL PROTECTED] < /dev/null
> 

Re: [why is kernel recompilation necessary?]

[Brian May]

> "Kent" == Kent West <[EMAIL PROTECTED]> writes:

Kent> In still other words, can you use "make menuconfig" to
Kent> compile a minimal kernel and then add modules later from
Kent> whatever source even though you didn't tell the kernel to
Kent> expect these modules when you did the "make menuconfig".

I have done just that. That is, built modules for the kernel at a
latter date.

I think you need to be careful that the kernel version is the
same though...

I consider modules (don't take my word for this on accuracy) to be just
the same as shared libraries, only for the kernel instead of user
level processes.
-- 
Brian May <[EMAIL PROTECTED]>

Re: why is kernel recompilation necessary?

[Brian May]

> "Sean" == Sean 'Shaleh' Perry <[EMAIL PROTECTED]> writes:

Sean> if you compile a kernel once with all the odd devices as
Sean> modules you get the equivalent of how windows works.

... except you don't need to reboot just to load/unload a driver.
-- 
Brian May <[EMAIL PROTECTED]>

Re: [why is kernel recompilation necessary?]

[Matthew Dalton]

By reasoning I would assume that all the kernel has is a 'generic module
hook' that you can load any module into. Therefore, your sequence above
should work. If it did not, how would you be able to use binary only
modules such as the lucent winmodem driver?

So in answer to your questions:

Kent West wrote:
> When you do a "make menuconfig" (or one of the other methods), and you
> specify to include support for, say, a 3c905 NIC, as a module, are you
> doing anything to the kernel, or are you just making changes to a script
> to tell it to compile the module?

You're making changes to a script. Of course, you have to compile module
support into the kernel as well, otherwise it won't be able to load any
modules.

> In still other words, can you do the following?
>   * use "make menuconfig" today to specify kernel options for a minimal
> kernel, and not mark module stuff like NICs and sound cards, etc,
>   * then "make dep" and "make zImage" to compile the kernel,
>   * then boot off that kernel and run for a day or two
>   * then come back in a day or two and re-run "make menuconfig" and
> specify some modules
>   * then "make modules" and "make modules_install" without compiling the
> kernel
>   * resulting in a working kernel that can use the modules compiled a
> day or two later

As long as you say yes to module support, yes.

> In still other words, can you use "make menuconfig" to compile a minimal
> kernel and then add modules later from whatever source even though you
> didn't tell the kernel to expect these modules when you did the "make
> menuconfig".

Yes. That's how binary-only modules work.

Matthew

Re: [why is kernel recompilation necessary?]

[Kent West]

Krzys Majewski wrote:
> 
> Hm, I guess my question was unclear. What I'm wondering about is
> how the linux kernel works and how the windows kernel works. I know
> that one of them is open source and the other isn't. At the same time
> I'm too lazy to dig deeply for this information. So I guess I was hoping
> someone would post an answer like, "The windows kernel maps a piece of
> memory to an adapter on the motherboard, and the drivers sit in that memory"
> or "The linux kernel has hooks for every possible module, and these
> hooks do/don't have to be explicitly compiled in every time a new
> module is installed", etc.
> -chris

This touches on a question that popped into my head the other day, so if
I may be permitted to add to Chris' question

When you do a "make menuconfig" (or one of the other methods), and you
specify to include support for, say, a 3c905 NIC, as a module, are you
doing anything to the kernel, or are you just making changes to a script
to tell it to compile the module?

In other words, does the "make menuconfig" do one thing (specify what
the kernel will look like including modules/hooks to modules), or does
it do two things (specify what the kernel will look like (without any
hooks to the modules) and what module code needs to be
compiled/installed)?

In still other words, can you do the following?
  * use "make menuconfig" today to specify kernel options for a minimal
kernel, and not mark module stuff like NICs and sound cards, etc,
  * then "make dep" and "make zImage" to compile the kernel,
  * then boot off that kernel and run for a day or two
  * then come back in a day or two and re-run "make menuconfig" and
specify some modules
  * then "make modules" and "make modules_install" without compiling the
kernel
  * resulting in a working kernel that can use the modules compiled a
day or two later

In still other words, can you use "make menuconfig" to compile a minimal
kernel and then add modules later from whatever source even though you
didn't tell the kernel to expect these modules when you did the "make
menuconfig".

Thanks!

Re: [why is kernel recompilation necessary?]

[John Hasler]

Krzys Majewski writes:
> The whole point of my post was to not have to read 17 web pages!

Why?  Do you find learning painful?
-- 
John Hasler
[EMAIL PROTECTED] (John Hasler)
Dancing Horse Hill
Elmwood, WI

Re: [why is kernel recompilation necessary?]

[John Hasler]

Krzys Majewski writes:
> For example, if I build a new device and write a driver for it, can I add
> support for this device to both windows and linux without having to
> modify either kernel?

You can for Linux, and probably for Windows as well.  In a sense you are
modifying the kernel when you load a module.

Even back in the days of Unix System III we could add drivers without
recompiling the kernel, but we did have to relink it.
-- 
John Hasler
[EMAIL PROTECTED]
Dancing Horse Hill
Elmwood, Wisconsin

Re: [why is kernel recompilation necessary?]

[Krzys Majewski]

Hm, I guess my question was unclear. What I'm wondering about is 
how the linux kernel works and how the windows kernel works. I know 
that one of them is open source and the other isn't. At the same time
I'm too lazy to dig deeply for this information. So I guess I was hoping
someone would post an answer like, "The windows kernel maps a piece of
memory to an adapter on the motherboard, and the drivers sit in that memory"
or "The linux kernel has hooks for every possible module, and these 
hooks do/don't have to be explicitly compiled in every time a new 
module is installed", etc. 
-chris

Re: [why is kernel recompilation necessary?]

[Krzys Majewski]

Yes but what I'm wondering is not why linux users recompile their kernels,
or why windows users can't, but how is it that windows users get away with
not having to? The closest answer I got is that windows kernels have 
a bunch of drivers already compiled in, and any additional drivers
compiled as modules. What I'm still not clear on is whether either 
windows or linux kernels (or both) need to have some sort of hooks 
enabling them to expect whatever modules at runtime, or not. 
For example, if I build a new device and write a driver for it,
can I add support for this device to both windows and linux without
having to modify either kernel? -chris



On Thu, 27 Jul 2000, Erik Mathisen wrote:

> custom kernels are smaller they only have what you want in the, no
> extra bloat, more secure (in theory) because it does not have excess
> stuff.  besides it customizes your system so it is your system.
> 
> erik
> 
> On Thu, Jul 27, 2000 at 06:04:57PM -0700, Krzys Majewski wrote:
> >The whole point of my post was to not have to read 17 web pages!
> >chris
> >
> >
> >On 27 Jul 2000, Sherab Puntsok wrote:
> >
> >> For general information on kernel visit
> >> http://linuxdoc.org/
> >> 
> >> For articles about kernel visit
> >> Taking the Plunge: Compiling the Kernel
> >> http://www.linux.com/newsitem.phtml?sid=60&aid-8841
> >> 
> >> Compiling the Kernel: Part 2
> >> http://www.linux.com/firststep/newsitem.phtml?sid=60&aid=8931
> >> 
> >> Kernel Basics
> >> http://www.linuxpapers.org/print_article.html?KERNEL_BASICS
> >> 
> >> Quest for a Leaner and Meaner Kernel
> >> http://www.linuxcare.com/viewpoints/tales/04-12-00.epl
> >> 
> >> Alan
> >> 
> >> Krzys Majewski <[EMAIL PROTECTED]> wrote:
> >> > Why is it that under Windows or whatever I don't have to recompile
> >> > the kernel just to add a new driver? Is it a protection thing? 
> >> > Or an optimization thing? Or something else? -chris
> >> > 
> >> > 
> >> > 
> >> > -- 
> >> > Unsubscribe?  mail -s unsubscribe [EMAIL PROTECTED] <
> >> /dev/null
> >> 
> >> 
> >> =
> >> (Optional)
> >> If you say " Om Ma Ne Pad Mei Hung ",
> >> The Avalokiteshvara Buddha blesses you.
> >> 
> >> For more info on Buddhism please visit
> >> http://victorian.fortunecity.com/holbein/272/
> >> =
> >> 
> >> 
> >> 
> >> Get your own FREE, personal Netscape WebMail account today at 
> >> http://webmail.netscape.com.
> >> 
> >> 
> >> --
> >> Unsubscribe?  mail -s unsubscribe [EMAIL PROTECTED] < /dev/null
> >> 
> >
> >
> >-- 
> >Unsubscribe?  mail -s unsubscribe [EMAIL PROTECTED] < /dev/null
> 

Re: [why is kernel recompilation necessary?]

[Krzys Majewski]

The whole point of my post was to not have to read 17 web pages!
chris


On 27 Jul 2000, Sherab Puntsok wrote:

> For general information on kernel visit
> http://linuxdoc.org/
> 
> For articles about kernel visit
> Taking the Plunge: Compiling the Kernel
> http://www.linux.com/newsitem.phtml?sid=60&aid-8841
> 
> Compiling the Kernel: Part 2
> http://www.linux.com/firststep/newsitem.phtml?sid=60&aid=8931
> 
> Kernel Basics
> http://www.linuxpapers.org/print_article.html?KERNEL_BASICS
> 
> Quest for a Leaner and Meaner Kernel
> http://www.linuxcare.com/viewpoints/tales/04-12-00.epl
> 
> Alan
> 
> Krzys Majewski <[EMAIL PROTECTED]> wrote:
> > Why is it that under Windows or whatever I don't have to recompile
> > the kernel just to add a new driver? Is it a protection thing? 
> > Or an optimization thing? Or something else? -chris
> > 
> > 
> > 
> > -- 
> > Unsubscribe?  mail -s unsubscribe [EMAIL PROTECTED] <
> /dev/null
> 
> 
> =
> (Optional)
> If you say " Om Ma Ne Pad Mei Hung ",
> The Avalokiteshvara Buddha blesses you.
> 
> For more info on Buddhism please visit
> http://victorian.fortunecity.com/holbein/272/
> =
> 
> 
> 
> Get your own FREE, personal Netscape WebMail account today at 
> http://webmail.netscape.com.
> 
> 
> --
> Unsubscribe?  mail -s unsubscribe [EMAIL PROTECTED] < /dev/null
> 

Re: [why is kernel recompilation necessary?]

[Sherab Puntsok]

For general information on kernel visit
http://linuxdoc.org/

For articles about kernel visit
Taking the Plunge: Compiling the Kernel
http://www.linux.com/newsitem.phtml?sid=60&aid-8841

Compiling the Kernel: Part 2
http://www.linux.com/firststep/newsitem.phtml?sid=60&aid=8931

Kernel Basics
http://www.linuxpapers.org/print_article.html?KERNEL_BASICS

Quest for a Leaner and Meaner Kernel
http://www.linuxcare.com/viewpoints/tales/04-12-00.epl

Alan

Krzys Majewski <[EMAIL PROTECTED]> wrote:
> Why is it that under Windows or whatever I don't have to recompile
> the kernel just to add a new driver? Is it a protection thing? 
> Or an optimization thing? Or something else? -chris
> 
> 
> 
> -- 
> Unsubscribe?  mail -s unsubscribe [EMAIL PROTECTED] <
/dev/null


=
(Optional)
If you say " Om Ma Ne Pad Mei Hung ",
The Avalokiteshvara Buddha blesses you.

For more info on Buddhism please visit
http://victorian.fortunecity.com/holbein/272/
=



Get your own FREE, personal Netscape WebMail account today at 
http://webmail.netscape.com.

Re: why is kernel recompilation necessary?

[Krzys Majewski]

Really? So the kernel doesn't compile any hooks for itself to enable
loading latesthardwaredevice.o as a module? -chris

On Fri, 28 Jul 2000, Frodo Baggins wrote:

> Krzys Majewski scripsit:
> >Why is it that under Windows or whatever I don't have to recompile
> >the kernel just to add a new driver? Is it a protection thing? 
> >Or an optimization thing? Or something else? -chris
> 
> Well, stricly speaking kernel recompilation isn't mandatory. You cold
> just compile a modulke and load it with modconf. This way is exactly
> the same thing as under nullsoft windog, except that you have the
> source of the driver:). 
> 
> -- 
> Leo TheHobbit 
> IRCnet #leiene
> ICQ 56656060
> 
> -BEGIN GEEK CODE BLOCK-
> Version: 3.1
> GED/CS d? s-:+>-: a C+++ U+++ L++(+++)> P+++>+ E+(++) 
> W++ N+ K? o? !w O? M V--- PS+++ PE-- Y+ GPG+ t++ 5? X- R+ tv+ 
> b D? DI? G e h(+) r--(---) y+(--)>+++*
> --END GEEK CODE BLOCK--
> 
> 
> -- 
> Unsubscribe?  mail -s unsubscribe [EMAIL PROTECTED] < /dev/null
> 

Re: why is kernel recompilation necessary?

[Frodo Baggins]

Krzys Majewski scripsit:
>Why is it that under Windows or whatever I don't have to recompile
>the kernel just to add a new driver? Is it a protection thing? 
>Or an optimization thing? Or something else? -chris

Well, stricly speaking kernel recompilation isn't mandatory. You cold
just compile a modulke and load it with modconf. This way is exactly
the same thing as under nullsoft windog, except that you have the
source of the driver:). 

-- 
Leo TheHobbit 
IRCnet #leiene
ICQ 56656060

-BEGIN GEEK CODE BLOCK-
Version: 3.1
GED/CS d? s-:+>-: a C+++ U+++ L++(+++)> P+++>+ E+(++) 
W++ N+ K? o? !w O? M V--- PS+++ PE-- Y+ GPG+ t++ 5? X- R+ tv+ 
b D? DI? G e h(+) r--(---) y+(--)>+++*
--END GEEK CODE BLOCK--

RE: why is kernel recompilation necessary?

[Sean 'Shaleh' Perry]

On 27-Jul-2000 Krzys Majewski wrote:
> Why is it that under Windows or whatever I don't have to recompile
> the kernel just to add a new driver? Is it a protection thing? 
> Or an optimization thing? Or something else? -chris
> 

if you compile a kernel once with all the odd devices as modules you get the
equivalent of how windows works.