Linux-Development-Sys Digest #19, Volume #7 Wed, 4 Aug 99 04:14:10 EDT
Contents:
Re: Video mode switching with asm (Jylam)
Re: Size limitations kernel (Frank v Waveren)
Re: struct byte alignment ??? (ellis)
Re: Web Based Linux Management (Norman Levin)
Re: APM_BAD_PARAM Error ("Shamsuddin, Amir")
The beginnings of a protected-mode boot loader (H. Peter Anvin)
Re: Redhat boot initrd (H. Peter Anvin)
More kind words from M$. ("Charles Sullivan")
Re: DMA from/to user space? ([EMAIL PROTECTED])
Re: The beginnings of a protected-mode boot loader (Frank v Waveren)
Re: self expanding system (H. Peter Anvin)
Re: The beginnings of a protected-mode boot loader (H. Peter Anvin)
Re: address? (H. Peter Anvin)
Re: Web Based Linux Management (Girish Kamath)
Re: Redhat boot initrd (Villy Kruse)
Re: Need kernel compile help (Gary Lawrence Murphy)
Re: /proc & lm_sensor kernel Oops... feature or bug? :-) (Paolo Pedaletti)
Size limitations kernel (Arnoud de Geus)
----------------------------------------------------------------------------
From: Jylam <[EMAIL PROTECTED]>
Subject: Re: Video mode switching with asm
Date: Tue, 03 Aug 1999 23:31:49 +0200
The problem is that I don't want to use the fb. You don't have to use it for
320x200x256, I'm sure. I think the svgalib sources are too hard for for, and
in C language. I'm sure you can do it until having a 3Mb executable ...
Tell me if I'm in the wrong way ...
Jylam
Tristan Wibberley wrote:
> Jylam wrote:
> >
> > Hello
> > I would like to know how switch to 320x200x256c for example without any
> > library, in assembly.
> > I can't use any lib because it's for any 4kb demo competition ...
>
> You'll need to use a kernel with framebuffer support. I suggest you
> subscribe to the linux framebuffer mailing list (though I don't remember
> the address right now).
>
> But then the problem is that the total code to do this is much greater
> than 4kb because you're using the kernel.
>
> --
> Tristan Wibberley
------------------------------
From: [EMAIL PROTECTED] (Frank v Waveren)
Subject: Re: Size limitations kernel
Date: Wed, 04 Aug 1999 01:51:41 GMT
In article <[EMAIL PROTECTED]>,
Arnoud de Geus <[EMAIL PROTECTED]> writes:
> Hello,
>
> Can somebody give me a pointer to information
> about the size limitations of the linux kernel on
> Intel computers.
>
> Have the size limitations something to do with the
> boot sequence and processor modes? What is
> the maximum size of the kernel or what is it
> dependent of?
IIRC, the kernel has to fit into the 1024k that can be easily accessed in real
mode. I believe using bzImage partially get's round this by using extended
memory.
--
Frank v Waveren
[EMAIL PROTECTED]
ICQ# 10074100
------------------------------
From: [EMAIL PROTECTED] (ellis)
Crossposted-To: comp.os.linux.development.apps
Subject: Re: struct byte alignment ???
Date: 3 Aug 1999 23:49:41 GMT
In article <[EMAIL PROTECTED]>,
Kaz Kylheku <[EMAIL PROTECTED]> wrote:
>> From what I understand, this is due to some alignment issuse. Is
>>there any compilation option that we can turn off the "auto-align"
>>feature ? (Means that the first printf will return 3 instead of 4).
>Why? Are you that hard up for memory? Sometimes it can be useful to have
>packed structures in order to conform to some external storage format. GCC has
>__attribute__ ((packed)) to support that.
>
>For example:
>
> struct s1 { char field1; unsigned short field2 } __attribute__ ((packed));
Have you ever used __attribute__ ((packed))? I've heard it doesn't
work (at least in some versions). On the other hand I've heard that
#pragma pack(1) *does* work. YMMV so it's best to do a test before
depending on either form.
--
http://www.fnet.net/~ellis/photo/linux.html
------------------------------
Date: Tue, 03 Aug 1999 21:27:39 -0400
From: Norman Levin <[EMAIL PROTECTED]>
Reply-To: [EMAIL PROTECTED]
Crossposted-To: alt.os.linux.caldera,comp.os.linux.networking,comp.os.linux.security
Subject: Re: Web Based Linux Management
The "linuxconf" tool can be web enabled - according to the RedHat install
manual.
--
Norman Levin
vm/dynAmIX inc.
------------------------------
From: "Shamsuddin, Amir" <[EMAIL PROTECTED]>
Subject: Re: APM_BAD_PARAM Error
Date: Tue, 03 Aug 1999 14:37:43 +0100
Wes Strater wrote:
>
> I have a newly aquired Compaq 5133 and I am getting the following error:
>
> apm: busy: Parameter out of range
>
> These error occur when the screen is blank. When I press shift to
> redisplay the screen, most of the time the screen goes blank again.
>
> I am new to linux but traced the error back to the apm.c.
>
> Is this a known problem with my machine and can I set parameters to
> avoid this or do I have problems with my machine?
>
> If it is a parameter setting, could you please be explicit in your
> explenation or tell me where there is a good explenation since I am new.
>
> Thanks, wes.
Is this a laptop ?
Have you checked for the latest bios for this computer ?
------------------------------
From: [EMAIL PROTECTED] (H. Peter Anvin)
Subject: The beginnings of a protected-mode boot loader
Date: 4 Aug 1999 02:43:37 GMT
Reply-To: [EMAIL PROTECTED] (H. Peter Anvin)
Hi all,
Just for kicks, I uploaded the feeble beginnings of an ext2-reading
bootloader for Linux/i386 onto
ftp.zytor.com:/pub/linux/lbcon-0.01.tar.gz. This is far from
complete (in fact, it doesn't really do anything yet), but since there
was some talk about a boot loader using gcc to generate real-mode code
I thought I'd suggest an alternative.
Right now, all this code does is bootstrap itself from an ext2
filesystem (using either the regular BIOS or -- if available -- the
ABIOS calls which don't have the 1024-cylinder limit), prints "Hello,
World!" and sits in a loop accepting keyboard input. The interesting
part of this whole exercise, though, is that the loop is all in 32-bit
protected mode, while it still uses the real-mode BIOS for interrupt
and devices. In other words, this code is a "BIOS extender" which
allows 32-bit flat gcc-generated code to run while still leaving the
BIOS functional and available.
I just thought I'd post what I have so far in case someone wants to
play around with it. The code is GPL.
-=hpa
--
"The user's computer downloads the ActiveX code and simulates a 'Blue
Screen' crash, a generally benign event most users are familiar with
and that would not necessarily arouse suspicions."
-- Security exploit description on http://www.zks.net/p3/how.asp
------------------------------
From: [EMAIL PROTECTED] (H. Peter Anvin)
Subject: Re: Redhat boot initrd
Date: 4 Aug 1999 02:44:46 GMT
Reply-To: [EMAIL PROTECTED] (H. Peter Anvin)
Followup to: <7o7l53$6ue$[EMAIL PROTECTED]>
By author: Hugh Sparks <[EMAIL PROTECTED]>
In newsgroup: comp.os.linux.development.system
>
> Why does the Redhat 6.0 want to boot with an
> initrd image?
>
> Is there any reason to keep this around?
>
> I compiled my kernel and skipped
> building this with no ill effects...
>
Because the RedHat kernel doesn't contain each and every device under
the sun. Rather, they put the needed device drivers as modules on an
initrd and load them that way.
-hpa
--
"The user's computer downloads the ActiveX code and simulates a 'Blue
Screen' crash, a generally benign event most users are familiar with
and that would not necessarily arouse suspicions."
-- Security exploit description on http://www.zks.net/p3/how.asp
------------------------------
From: "Charles Sullivan" <[EMAIL PROTECTED]>
Crossposted-To: comp.os.linux.misc
Subject: More kind words from M$.
Date: Tue, 3 Aug 1999 23:25:11 -0400
While waiting to be selected for jury duty I found a copy
of the "Microsoft Internet Developer" magazine inexplicably
included in the stack of mostly ladies magazines in the
waiting room. This is the August 1999 issue.
On Page 6, "Editor's Notes", is found the following:
"What starts off hot, gets unpleasantly sticky after
awhile, starts to stink like a minty goat soon after
that, and ends up making you wish you never experienced
it? No, we're not talking about the experience of
getting Linux actually working on your machine. It's
summer in New York City! ...".
The note goes on to describe how easy they found it to
install Beta 3 rev of Windows 2000, and how slick it is.
(Would you expect anything different when a disclaimer
on page 8 says: "... All editorial content is created by
Microsoft Corp, which is solely responsible for the
contents, ...".)
------------------------------
From: [EMAIL PROTECTED]
Subject: Re: DMA from/to user space?
Date: 4 Aug 1999 03:09:05 GMT
In article <7mmt14$tpu$[EMAIL PROTECTED]>,
Robert Kaiser <[EMAIL PROTECTED]> wrote:
>IMHO, this kind of functionality should be in the standard kernel.
>Anything else (patch or module) is a kludge.
I think I'd have to agree 100% there. We also needed this sort of
functionality, and the only viable alternative is to add the code to the driver
even though it duplicates a slab of kernel stuff. It's just not practical to
expect every user to patch their kernel just so they can use your device. It
sure would be nice to have this facility available in the kernel...
Regards,
Graham
------------------------------
From: [EMAIL PROTECTED] (Frank v Waveren)
Subject: Re: The beginnings of a protected-mode boot loader
Date: Wed, 04 Aug 1999 04:27:21 GMT
I can just imagine it: One day, a guy at his works says: I need a program
that'll print "Hello world!". His co-workers say: "Oh dear, how are we
going to handle that with only 1M of memory address space? We're going to
need something a lot better." And so, protected mode was invented. :-)
But seriously: Nice work, sounds interesting. I'm certainly going to have a
look at it (Or at least poke it and see if it moves).
In article <7o89cp$sut$[EMAIL PROTECTED]>,
[EMAIL PROTECTED] (H. Peter Anvin) writes:
> Hi all,
>
> Just for kicks, I uploaded the feeble beginnings of an ext2-reading
> bootloader for Linux/i386 onto
> ftp.zytor.com:/pub/linux/lbcon-0.01.tar.gz. This is far from
> complete (in fact, it doesn't really do anything yet), but since there
> was some talk about a boot loader using gcc to generate real-mode code
> I thought I'd suggest an alternative.
>
> Right now, all this code does is bootstrap itself from an ext2
> filesystem (using either the regular BIOS or -- if available -- the
> ABIOS calls which don't have the 1024-cylinder limit), prints "Hello,
> World!" and sits in a loop accepting keyboard input. The interesting
> part of this whole exercise, though, is that the loop is all in 32-bit
> protected mode, while it still uses the real-mode BIOS for interrupt
> and devices. In other words, this code is a "BIOS extender" which
> allows 32-bit flat gcc-generated code to run while still leaving the
> BIOS functional and available.
>
> I just thought I'd post what I have so far in case someone wants to
> play around with it. The code is GPL.
>
> -=hpa
--
Frank v Waveren
[EMAIL PROTECTED]
ICQ# 10074100
------------------------------
From: [EMAIL PROTECTED] (H. Peter Anvin)
Subject: Re: self expanding system
Date: 4 Aug 1999 05:02:56 GMT
Reply-To: [EMAIL PROTECTED] (H. Peter Anvin)
Followup to: <01beddc3$2a5c3600$160a0180@toto>
By author: "Olivier Nibart" <[EMAIL PROTECTED]>
In newsgroup: comp.os.linux.development.system
>
> Once I have a complete Linux system installed a running well, I'd like to
> create a compressed image to put on a FlashDisk, with something to expand
> it in RamDisk at boot time.
>
> How can I do that ?
>
See /usr/src/linux/Documentation/initrd.txt
-hpa
--
"The user's computer downloads the ActiveX code and simulates a 'Blue
Screen' crash, a generally benign event most users are familiar with
and that would not necessarily arouse suspicions."
-- Security exploit description on http://www.zks.net/p3/how.asp
------------------------------
From: [EMAIL PROTECTED] (H. Peter Anvin)
Subject: Re: The beginnings of a protected-mode boot loader
Date: 4 Aug 1999 04:43:31 GMT
Reply-To: [EMAIL PROTECTED] (H. Peter Anvin)
Followup to: <JiPp3.14639$[EMAIL PROTECTED]>
By author: [EMAIL PROTECTED] (Frank v Waveren)
In newsgroup: comp.os.linux.development.system
>
> I can just imagine it: One day, a guy at his works says: I need a program
> that'll print "Hello world!". His co-workers say: "Oh dear, how are we
> going to handle that with only 1M of memory address space? We're going to
> need something a lot better." And so, protected mode was invented. :-)
>
Heh :)
To answer the implicit question (i.e. "WTF were you thinking?"): I
figured it would be easier to use libext2 than to write and maintain a
real-mode ext2 library. The idea is to have a boot loader than can
read and write ext2 and read iso9660, and do cool stuff like generate
initrd on the fly.
But for right now, it's just a toy. Rather like when Linux first was
released :)
-hpa
--
"The user's computer downloads the ActiveX code and simulates a 'Blue
Screen' crash, a generally benign event most users are familiar with
and that would not necessarily arouse suspicions."
-- Security exploit description on http://www.zks.net/p3/how.asp
------------------------------
From: [EMAIL PROTECTED] (H. Peter Anvin)
Subject: Re: address?
Date: 4 Aug 1999 05:08:59 GMT
Reply-To: [EMAIL PROTECTED] (H. Peter Anvin)
Followup to: <[EMAIL PROTECTED]>
By author: Konrad Mierendorff <[EMAIL PROTECTED]>
In newsgroup: comp.os.linux.development.system
>
> Ann Chen wrote:
> >
> > We are trying to find out where does the Linux operating system reside??
> > Does it live on the lower half of the memory, or the upper half? And how
> > big is the Linux kernel (how much memory does it occupy?)
>
> As to the last question, the answer is easy. The amount of memory the
> kernel is printed out when you boot your system. Of course it heavily
> depends on the drivers you compiled into the kernel.
>
> I don't know were the kernel resides but anyway why is it of interest?
>
> Bye
>
> - Konrad
The kernel lives near the bottom of memory, starting at the 1 MB
mark. This applies, of course, to the i386 architecture only; on
Alpha, for example, it lives at address 0x310000 (3 MB+64 K).
-hpa
--
"The user's computer downloads the ActiveX code and simulates a 'Blue
Screen' crash, a generally benign event most users are familiar with
and that would not necessarily arouse suspicions."
-- Security exploit description on http://www.zks.net/p3/how.asp
------------------------------
From: Girish Kamath <[EMAIL PROTECTED]>
Crossposted-To: comp.os.linux.networking
Subject: Re: Web Based Linux Management
Date: Wed, 04 Aug 1999 05:30:37 GMT
Hi
I use Webmin.
It's quite helpful.
For more details visit
www.webmin.com
Even Perl is very strong. But I prefer Webmin.
-Regards,
Girish Kamath.
Karl McMurdo wrote:
>
> Is there such an animal ?
>
> Unixware has their 'Webtop' interface that allows system management with
a
> web interface, and Samba has it as well, I'm looking for the ability to
> manage a linux server from a remote browser (or even a local browser
> although that makes less sense as you are in X and presumably X
management
> tools are available) I have a number of application servers that do not
> have X installed, and a couple that do, my 'desktop' machine is actually
a
> laptop running WinNT, and I run a windows X-Server to connect to those
> machines with X running, and telnet into the others to do sysadmin tasks.
I
> have been experimenting with perl and have created a script to do the
most
> common task (add/mod/del users) with a web interface, and am thinking of
> expanding it to cover more tasks, but don't want to reinvent the wheel.
> Any feedback would be appreciated.
>
> Karl McMurdo
>
>
>
================== Posted via SearchLinux ==================
http://www.searchlinux.com
------------------------------
From: [EMAIL PROTECTED] (Villy Kruse)
Subject: Re: Redhat boot initrd
Date: 4 Aug 1999 08:43:12 +0200
In article <7o7l53$6ue$[EMAIL PROTECTED]>,
Hugh Sparks <[EMAIL PROTECTED]> wrote:
>Why does the Redhat 6.0 want to boot with an
>initrd image?
It only does this if your disk is SCSI. IDE disk drivers are built
into the RH kernel and doesn't need initrd.
>Is there any reason to keep this around?
If you link your SCSI driver into the kernel initrd is no longer needed.
>I compiled my kernel and skipped
>building this with no ill effects...
If you upgrade to a newer kernel without linking in SCSI support and
you keep the old initrd you might get problem loading the SCSI module
frpm the initrd image because of version conflict. That would make your
system un-bootable.
Villy
------------------------------
From: Gary Lawrence Murphy <[EMAIL PROTECTED]>
Subject: Re: Need kernel compile help
Date: 04 Aug 1999 02:39:49 -0400
Reply-To: [EMAIL PROTECTED]
>>>>> "a" == aewell <[EMAIL PROTECTED]> writes:
a> If I compile all my critical stuff as non-modules, the
a> non-critical stuff won't work (like my sound card). Changing
a> these modules to non-modules and re-compiling fixes the
a> problem.
Ah ... the dangers of the UI strike again! Remember, What You See Is
ALL You Get :) That doesn't mean that what you see is all there is.
This is just a guess, but what may be happening is the depmod line in
/etc/rc.sysinit is encountering modules left over from a prior
install, and these modules have version mis-matches or depend on
symbols no longer in your kernel.
With your new modular kernel running, run "depmod -a" to rebuild the
dependencies. If you can then reboot and run the modules, the problem
is solved.
When you rebuild a new kernel with the same version number, you should
always move the old /lib/modules/<version number/ directory out of
the way before you "make modules_install"
--
Gary Lawrence Murphy <[EMAIL PROTECTED]> TeleDynamics Communications Inc
Business Telecom Services : Internet Consulting : http://www.teledyn.com
Linux/GNU Education Group: http://www.egroups.com/group/linux-education/
"Computers are useless. They can only give you answers."(Pablo Picasso)
------------------------------
From: [EMAIL PROTECTED] (Paolo Pedaletti)
Subject: Re: /proc & lm_sensor kernel Oops... feature or bug? :-)
Date: 4 Aug 1999 01:32:47 GMT
Reply-To: [EMAIL PROTECTED]
Ciao Alexander Viro,
> Bug, indeed, but you've omitted very interesting piece of data: version of the
> kerenl ;-)
2.2.10 .... sorry :-)
> Results of ksymoops on your kern.log would be also useful
Jul 30 14:50:09 niels kernel: Unable to handle kernel NULL pointer dereference at
virtual address 00000030
Jul 30 14:50:09 niels kernel: current->tss.cr3 = 07095000, %cr3 = 07095000
Jul 30 14:50:09 niels kernel: *pde = 00000000
Jul 30 14:50:09 niels kernel: Oops: 0000
Jul 30 14:50:09 niels kernel: CPU: 0
Jul 30 15:11:30 niels kernel: klogd 1.3-3#32, log source = /proc/kmsg started.
Jul 30 15:11:30 niels kernel: Cannot find map file.
Jul 30 15:11:30 niels kernel: Loaded 304 symbols from 39 modules.
Jul 30 15:11:30 niels kernel: Unable to handle kernel NULL pointer dereference at
virtual address 00000030
Jul 30 15:11:30 niels kernel: current->tss.cr3 = 028fb000, %cr3 = 028fb000
Jul 30 15:11:30 niels kernel: *pde = 00000000
Jul 30 15:11:30 niels kernel: Oops: 0000
Jul 30 15:11:30 niels kernel: CPU: 0
Jul 30 15:16:15 niels kernel: klogd 1.3-3#32, log source = /proc/kmsg started.
Jul 30 15:16:15 niels kernel: Cannot find map file.
Jul 30 15:16:15 niels kernel: Loaded 300 symbols from 24 modules.
Jul 30 15:16:15 niels kernel: Linux version 2.2.10 (root@niels) (gcc version
egcs-2.91.66 Debian GNU/Linux (egcs-1.1.2 release)) #2
Jul 30 15:16:15 niels kernel: Detected 350810353 Hz processor.
Jul 30 15:16:15 niels kernel: Console: colour VGA+ 132x44
Jul 30 15:16:15 niels kernel: Calibrating delay loop... 699.60 BogoMIPS
Jul 30 15:16:15 niels kernel: Memory: 128228k/131008k available (820k kernel code,
408k reserved, 1512k data, 40k init)
Jul 30 15:16:15 niels kernel: CPU: AMD AMD-K6(tm) 3D processor stepping 0c
--
Paolo Pedaletti, Como, ITALYa < paolo . pedaletti @ flashnet . it >
------------------------------
From: Arnoud de Geus <[EMAIL PROTECTED]>
Subject: Size limitations kernel
Date: Tue, 03 Aug 1999 23:38:33 +0200
Hello,
Can somebody give me a pointer to information
about the size limitations of the linux kernel on
Intel computers.
Have the size limitations something to do with the
boot sequence and processor modes? What is
the maximum size of the kernel or what is it
dependent of?
I have problems compiling a 2.2.10 kernel under
RedHat 6.0 (which uses egcs 1.1.2) in comparison
with RedHat 5.2 (which used gcc 2.7). The egcs
compiler produces a bigger image than the old
gcc compiler (using the same configuration file).
To be able to compile a bootable kernel, I had to
move a number of things to modules.
Does the egcs 1.1.2 compiler produce stable
kernels? Just recently, there were problems
using egcs for kernel build.
Greetings,
[EMAIL PROTECTED]
P.S. Is there a good book about the linux 2.2.x
kernel internals (besides /usr/src/linux)?
------------------------------
** FOR YOUR REFERENCE **
The service address, to which questions about the list itself and requests
to be added to or deleted from it should be directed, is:
Internet: [EMAIL PROTECTED]
You can send mail to the entire list (and comp.os.linux.development.system) via:
Internet: [EMAIL PROTECTED]
Linux may be obtained via one of these FTP sites:
ftp.funet.fi pub/Linux
tsx-11.mit.edu pub/linux
sunsite.unc.edu pub/Linux
End of Linux-Development-System Digest
******************************
