Linux-Hardware Digest #596, Volume #14 Mon, 9 Apr 01 17:13:03 EDT
Contents:
Re: A Linux emulator for Linux, does this exist? ("Norm Dresner")
Re: Lucent AMR modem on suse (kernel 2.4.0-4GB) (M. Buchenrieder)
Re: SCSI issue with two cards (Michael Meissner)
Re: sys 6326 video driver (Raoul Markus)
Re: Linux on Intel Or Celeron? what is the best choice? ("Peter T. Breuer")
Re: A Linux emulator for Linux, does this exist? (Grant Edwards)
Re: Sound on a Dell Inspiron 3800 (ESS Maestro3) - Loading Modules fails ("Jason
Advani")
Re: Switchboxes for keyboard, mice, video? (Anthony Hill)
Re: A Linux emulator for Linux, does this exist? (paul beard)
Re: A Linux emulator for Linux, does this exist? (Uwe Bonnes)
Re: A Linux emulator for Linux, does this exist? (Timo =?ISO-8859-1?Q?St=F6ckigt?=)
Re: SCSI issue with two cards ("Peter T. Breuer")
Re: Linux on Intel Or Celeron? what is the best choice? ("BDS")
Serial port driver is making me crazy (Mike Lackey)
----------------------------------------------------------------------------
From: "Norm Dresner" <[EMAIL PROTECTED]>
Crossposted-To: comp.os.linux.development.system,comp.os.linux.misc
Subject: Re: A Linux emulator for Linux, does this exist?
Date: Mon, 09 Apr 2001 20:12:09 GMT
Hermann Samso <[EMAIL PROTECTED]> wrote in message
news:9at3ri$gc3$[EMAIL PROTECTED]...
> I am searching for a Linux emulator for Linux. Does such
> a beast exist? I think this could be useful for System
> developement, without needing to have more than 1 powerful
> computer for programming and testing.
>
> Saludos,
> SoLo2
Let's see. I can buy a cheap PC clone on which I can run Linux -- in a
networked environment so I don't even need anything more than the OS
itself -- for about $600US. At even a bargain basement (costing) rate of
$50US hour, that's only 12 hours of programmer time. Considering the
complexity of Linux, you'd have to expend thousands of hours to get
something that might -- just might -- provide a moderately faithful
emulation of exactly one version of Linux. Just one version! To get three
versions, you'd probably have to double that. And who's going to do the
program maintenance every time a new kernel version comes out? You?
Probably not. You wouldn't have enough time to understand all of the
differences from the last version in time to get it coded before you'd have
to start working on the next new kernel version.
OTOH, If I'm really paranoid, I can get removable shells for my HD's so I
can remove all but one OS drive every time I'm ready to test a new "feature"
of my hardware. Perhaps for an investment of $1000US I can create a totally
"safe" computer with two HDs, one of which is used only to restore the OS on
the other in the case of a total wipeout. And that's buying new hardware.
I can probably scrounge up enough hardware in my basement to create one of
these machines for under $100US.
While it might be a nice thing to have, it's totally economically
infeasible. Maybe it might make sense for a stable OS like Windows ;-) but
not for something that's changing as rapidly as Linux.
Norm
------------------------------
From: [EMAIL PROTECTED] (M. Buchenrieder)
Subject: Re: Lucent AMR modem on suse (kernel 2.4.0-4GB)
Date: Mon, 9 Apr 2001 15:18:45 GMT
Roberto <[EMAIL PROTECTED]> writes:
[...]
>NO misspoke, the winxxxx driver tells me really Lucent AMR modem, it
>should be the same/similar modem equipped with IBM iseries notebook.
Although AMR stands for AudioModemRisercard and basically means a
combo socket solution, this kind of software-controlled devices
has been implemented on quite a number of all-in-one design
motherboards.
>I visited already the link you give me, but i get errors when trying to
>load the module 'modprobe ltmodemxxxx'.
Hopeless. AFAIR, the PCTel driver does not work with that kind
of chipset. Get yourself a nice real PCMCIA modem instead.
Michael
--
Michael Buchenrieder * [EMAIL PROTECTED] * http://www.muc.de/~mibu
Lumber Cartel Unit #456 (TINLC) & Official Netscum
Note: If you want me to send you email, don't munge your address.
------------------------------
Subject: Re: SCSI issue with two cards
From: Michael Meissner <[EMAIL PROTECTED]>
Date: 09 Apr 2001 16:24:37 -0400
"Peter T. Breuer" <[EMAIL PROTECTED]> writes:
> Don Gingrich <[EMAIL PROTECTED]> wrote:
> > "Peter T. Breuer" wrote:
> >> Don Gingrich <"gingrich"@melbpc(dot)org(dot).au> wrote:
> >> But you can force the detection order .. pci
> >> cards are detected in ascending slot order. Slot 1 (furthest from cpu)
> >> is detected first. You can reverse the pci scan order too if you
> >> prefer. It's a boot param.
> >>
> >> In general you can force the detection order between two differnet
> >> drivers by just choosing which driver you load first!
>
> > OK, but in the case of already having the drivers in the
> > kernel this is not an option. BTW where is setting the load
> > order documented?
>
> If both drivers are in the kernel, then you will always get one
> installed before the other CONSISTENTLY, since you have different
> dirvers, and the kernel can only load one at a time, and it cannot do
> two things at once.
>
> So you have no problem.
Until you change your kernel. Linux 2.4 uses a different order to initialize
scsi adapters than 2.2 did.
> If one driver is in the kernel, and the other is a module, then the one
> in the kernel will be loaded first by definition (as the kernel starts
> before anything takes place in userspace, such as loading modules),
> CONSISTENTLY.
>
> So you have no problem.
Which is the solution I current ly.
> If both devices are modules, then they load when you choose to load
> them, and so the "order is determined by YOU.
>
> So you have no problem.
>
--
Michael Meissner, Red Hat, Inc. (GCC group)
PMB 198, 174 Littleton Road #3, Westford, Massachusetts 01886, USA
Work: [EMAIL PROTECTED] phone: +1 978-486-9304
Non-work: [EMAIL PROTECTED] fax: +1 978-692-4482
------------------------------
From: Raoul Markus <[EMAIL PROTECTED]>
Subject: Re: sys 6326 video driver
Date: Mon, 9 Apr 2001 22:16:00 +0200
dario mendez wrote:
> wehere i can find drivers to install a agp sys 6326 video on
> linux(genome), please help me
>
>
>
> [EMAIL PROTECTED]
>
>
>
the XFree 4.0.2 manages this card quite o.k. (but to my experience only
without acceleration)
Hope I could help you... ;-)
Raoul
------------------------------
From: "Peter T. Breuer" <[EMAIL PROTECTED]>
Subject: Re: Linux on Intel Or Celeron? what is the best choice?
Date: Mon, 09 Apr 2001 20:29:26 GMT
Eric P. McCoy <[EMAIL PROTECTED]> wrote:
> "Peter T. Breuer" <[EMAIL PROTECTED]> writes:
> ["overclock"]
>> That word is your invention.
> I don't think I'm being pedantic in saying that no, it's not.
Well, it's not your invention then, but honestly, it has no semantics.
There is no god-given rate at which a cpu should go.
> [snip Intel "underclocking" CPUs to fill a market niche]
>> Undo intels attempted sabotage and you have the chip the way back it
>> was supposed to be.
> So you're saying that any P3 CPU manufactured after the introduction
> of the P3-933 will operate reliably at that speed?
No, I am not. I'm saying that when the yield of 933's from the P3
fabrication process passes 90%, it's practically a dead cert that a
P3 800 that you buy will do 933 too. The case of the celeron 300a's
was even more marked .. they were P2s made with a BETTER manufacturing
process (smaller technology) and without most of the heat-producing
cache, and with what cache there was going at twice the P2's speed
(i.e. at full speed, instead of half speed). All you had to do
was mask the pin that told them they were celerons and hey presto, rock
solid P450's at a time when P450's cost the earth. Intel simply
marketed them as 300s.
> I was under the impression that the variance in clock speeds was due
> to differing quality in the materials used. And that some ludicrous
No, that is not the case. All materials are "top" quality. The
fabrication process is the same for all on the line. Yes, the wafer
varies slightly across it, but you don't need to worry about that.
The problem is the manufacturing process, which has to be "tuned"
and perfected as the run goes on. Minor changes in layout are made in
order to correct defect-prone areas, and to add or eliminate
redundancies. This results in ever increasing yields at the target
clockspeed. Then the target becomes silly, and they raise the target,
etc. etc. But if they are already smashing the opposition and the
market into the ground, they can't STOP the improvement process. They
just have to market fast chips as slow ones, to stop their market share
being taken.
You can spot when this is happening.
> figure like 75% of all chips designed for, say, 1GHz fail reliability
> tests at that speed, but work just fine at lower speeds.
>> > fond of warranty and stuff. Offcourse you're right about the performance,
>> Warranty is not affected by clockspeed. No harm is done by clocking a
>> device at a different clockspeed .. it simply works or does not work.
>> If it's iffy, clock it down. If it goes well, clock it up until you
>> find where it stops working. There is no guarantee that chips will work
>> at their RATED clockspeed, let alone at any other. This is all
>> statistics, not certainties.
> Yes, sure. But a CPU is far more likely to operate correctly at its
> rated speed than at a faster one. And the tendency of CPUs to
No .. it is (not "far") more likely that a randomly chosen cpu will
operate at any given speed than at any given higher one. That
says nothing at all about any individual cpu, which may be a good one
or a lemon. You don't know until you look. All you know is that intel
sold it to you. No .. there is no guarantee that it will work at its
"rated" speed. The shop will take it back if it doesn't. If you SPOT
that it doesn't, that is. I've taken back enough that I know how hard
it can be to show that it doesn't work.
I still remember my boxed intel P120, which turned out on close
inspection to be a P100 run over-voltage and with a huge heatsink.
I eventually learned how to run it at 133, which allwed the bus to go
at 66 instead of 60, but by then I had several P100's going at 120
by just turning up the voltage the way intel had done.
These people are marketing their products. Don't take what they say as
gospel. It isn't.
> demonstrate erratic glitches when hot makes thorough testing of an
> overclocked box difficult to manage.
Absolutely. But not "over"clocking it doesn't make it any easier. I
have plenty of machines that are showing glitches at their rated speed.
Who knows what it is .. cpu? memory? motherboard? I can tell you
that from my experience, running a oem'ed machine at its rated speed leaves
you with approximately a 20% chance of failure. No, the compoments are
NOT tested by the factories, by the shops or by anyone. I was once sold
a memory stick only half of which worked .. on locking at it you could
see that one of the chips had been stuck on askew, so that the legs
didn't fit in the holes on the PCB. Ask anyone who works in a computer
assembly shop!
> I don't really object to any of what you've said, but I think you're
> giving the impression that overclocking is easy and largely
It is absolutely risk free: if it doesn't run at the speed you want,
drop the speed until it does work. There is no penalty for trying it.
That's "risk free". In contrast, the penalty for NOT trying to find the
correct maximum speed for the processor in question is the certainty
of a 20-25% failure rate in installed OEM equipment at their installed
rates. Try running burn tests on all your new machines. Do it again
each year.
> risk-free. Which may be true in some cases, but is certainly not in
> others.
Peter
------------------------------
From: [EMAIL PROTECTED] (Grant Edwards)
Crossposted-To: comp.os.linux.development.system,comp.os.linux.misc
Subject: Re: A Linux emulator for Linux, does this exist?
Date: Mon, 09 Apr 2001 20:30:45 GMT
In article <9at3ri$gc3$[EMAIL PROTECTED]>, Hermann Samso wrote:
> I am searching for a Linux emulator for Linux. Does such
> a beast exist?
Yes. You can run multiple instances of Linux on an IBM 390.
One guy at IBM reportedly had 4000+ instances of Linux running
on a signle box. It's a very cool concept: instead of racks
and racks of servers that all need to be maintained, you buy a
single 390 and run umpteen copies of Linux: one as a DNS
server, one as an IMAP server, one as an NNTP server, a few
dozen as HTTP servers, a few as database servers, etc. There
are virutual network channels between the instances, and they
can share disk space.
390's have auto-failover, hot-swappable _everything_ and they
quite commonly run 99.99+ percent uptime. That's something
that's simply not possible with a room full of x86 machines.
Once the number of x86 machines gets large enough, the cost of
ownership of a 390 is actually less than the "many racks of
cheap shit" approach.
> I think this could be useful for System
> developement,
Definitely. That's way machines like the 390 have always had
fully virtualizable designs. That way you can run multiple
OSes on one box. Can't affort to give every system programmer
his own physical machine? Then give each one a virtual machine
or two.
> without needing to have more than 1 powerful
> computer for programming and testing.
Unfortunately, due to the mind-numbing backwardness of the
Intel CPU design, it is not easily virtualizable. So there's
no nice to set up multiple virtual machines on one physical
machine if you're talking about the x86.
If Intel had done a better job on the 386, we probably would be
able to run multiple virtual machines on nice cheap hardware
(vmware does it, but they put a _lot_ of work into it, and
they want some payment for that work).
--
Grant Edwards grante Yow! Yow! I like my new
at DENTIST...
visi.com
------------------------------
From: "Jason Advani" <[EMAIL PROTECTED]>
Crossposted-To: comp.os.linux.portable,osu.sys.linux
Subject: Re: Sound on a Dell Inspiron 3800 (ESS Maestro3) - Loading Modules fails
Date: Mon, 9 Apr 2001 16:37:49 -0400
"Iain Lea" <[EMAIL PROTECTED]> wrote in message
news:Mw0A6.1$[EMAIL PROTECTED]...
> Install kernel 2.4.3 as it supports the Maestro3 audio soundcard :)
>
> I have it running here... neat... the only thing that does not work
> is the Fn-Vol+/- keys have no effect... anyone have a fix for that ?
I have the same problem...no volume change w/ those keys. I don't think it
would be too difficult to write a kernel module to fix that. Just capture
the keys and poke around in the appropriate device, no? I was reading the
Module Programing Howto and was surprised at how easy, but by no means a
piece of cake, it was to write modules.
Jason
------------------------------
From: Anthony Hill <[EMAIL PROTECTED]>
Crossposted-To: comp.sys.ibm.pc.hardware.chips
Subject: Re: Switchboxes for keyboard, mice, video?
Date: Mon, 09 Apr 2001 20:33:13 GMT
On 08 Apr 2001 21:03:06 -0400, [EMAIL PROTECTED] (Eric P. McCoy)
wrote:
>Anthony Hill <[EMAIL PROTECTED]> writes:
>
>> >Yes, I've just purchased an older system on ebay and plan to use
>> >Linux to turn it into a server/router with NAT to distribute my
>> >PPP connection to two PCs.
>
>> Well, in case you haven't read it yet, I'd throw out a VERY strong
>> recommendation to read the IP Masquerading HOWTO (available at
>> http://ipmasq.cjb.net/).
>
>Has it been updated to apply to iptables in 2.4.x kernels?
Unfortunately no. There is a small section about 2.4.x kernels in the
HOWTO, but it mostly just outlines a few differences between ipchains
and iptables and explains some pro's and con's of the two. It has
been about 3 months since the document has been updated, and since
it's usually updated fairly regularly, my guess is that the author is
working (at least a little bit) at the iptables section.
>In any case, there's no howto necessary for a simple iptables NAT
>setup.
>
> iptables -t nat -A POSTROUTING -o eth1 -j MASQUERADE
> echo 1 > /proc/sys/net/ipv4/ip_forward
True, though a setup for ipchains would only need one or two more
lines to get the thing working, however to get it working with good
security is a slightly more complicated issue. Also, probably the
biggest problem right now with IP masquerading with iptables is that
the IP masquerading modules for things like ICQ, Quake, RealPlayer,
etc. would need to be re-written to be fully functional, and to the
best of my knowledge no one has done that yet.
>> It's very well written and describes how to set all these things up
>> to work well with good security. If you've already read it, then
>> never mind :>
>
>I think the iptables howtos are still works in progress. Bah, I can't
>remember the URL now, and it was nontrivial (for me) to find; it might
>be mentioned at the start of the Adv-Routing-HOWTO.
Iptables itself is a work in progress, let alone it's documentation!
:> The main website for info on it seems to be here:
http://netfilter.filewatcher.org/
They do have an iptables HOWTO on that site, though it's a bit sparse
still.
And now in a vain attempt to bring this thread back on-topic...
Anyone have any idea how the 2.4.x kernel and iptables would compare
performance-wise to the 2.2.x kernel and ipchains for a
not-too-powerful system (Pentium 100 with 32MB of memory)?
=======================
Tony Hill
[EMAIL PROTECTED]
------------------------------
Subject: Re: A Linux emulator for Linux, does this exist?
From: paul beard <[EMAIL PROTECTED]>
Crossposted-To: comp.os.linux.development.system,comp.os.linux.misc
Date: Mon, 09 Apr 2001 20:38:48 GMT
Ever tried VMware?
in article 9at3ri$gc3$[EMAIL PROTECTED], Hermann Samso at
[EMAIL PROTECTED] wrote on 4/9/01 12:51 PM:
> I am searching for a Linux emulator for Linux. Does such
> a beast exist? I think this could be useful for System
> developement, without needing to have more than 1 powerful
> computer for programming and testing.
>
> Saludos,
> SoLo2
------------------------------
From: Uwe Bonnes <[EMAIL PROTECTED]>
Crossposted-To: comp.os.linux.development.system,comp.os.linux.misc
Subject: Re: A Linux emulator for Linux, does this exist?
Date: 9 Apr 2001 20:27:57 GMT
In comp.os.linux.misc Hermann Samso <[EMAIL PROTECTED]> wrote:
: I am searching for a Linux emulator for Linux. Does such
: a beast exist? I think this could be useful for System
: developement, without needing to have more than 1 powerful
: computer for programming and testing.
You can boot Linux in Linux with vmware and there is s user space
kernel.
Bye
--
Uwe Bonnes [EMAIL PROTECTED]
Institut fuer Kernphysik Schlossgartenstrasse 9 64289 Darmstadt
========= Tel. 06151 162516 ======== Fax. 06151 164321 ==========
------------------------------
From: Timo =?ISO-8859-1?Q?St=F6ckigt?= <[EMAIL PROTECTED]>
Crossposted-To: comp.os.linux.development.system,comp.os.linux.misc
Subject: Re: A Linux emulator for Linux, does this exist?
Date: Mon, 9 Apr 2001 22:36:03 +0200
Reply-To: [EMAIL PROTECTED]
Hermann Samso wrote:
> I am searching for a Linux emulator for Linux. Does such
> a beast exist? I think this could be useful for System
> developement, without needing to have more than 1 powerful
> computer for programming and testing.
>
> Saludos,
> SoLo2
>
If I didn't understand everything wrong VMWare provides you a virtual
computer, right? And you have a powerful computer. So what's the problem?
Ciao
Timo
------------------------------
From: "Peter T. Breuer" <[EMAIL PROTECTED]>
Subject: Re: SCSI issue with two cards
Date: Mon, 9 Apr 2001 22:38:22 +0200
Michael Meissner <[EMAIL PROTECTED]> wrote:
> "Peter T. Breuer" <[EMAIL PROTECTED]> writes:
>> Don Gingrich <[EMAIL PROTECTED]> wrote:
>> > "Peter T. Breuer" wrote:
>> >> Don Gingrich <"gingrich"@melbpc(dot)org(dot).au> wrote:
>> If both drivers are in the kernel, then you will always get one
>> installed before the other CONSISTENTLY, since you have different
>> dirvers, and the kernel can only load one at a time, and it cannot do
>> two things at once.
>>
>> So you have no problem.
> Until you change your kernel. Linux 2.4 uses a different order to initialize
> scsi adapters than 2.2 did.
True. But changing kernels wasn't in the parameters of this problem
:-(. Nor is changing OS's :-).
A real masochist will grep /proc/pci and rename the special device
files !
>> If one driver is in the kernel, and the other is a module, then the one
>> in the kernel will be loaded first by definition (as the kernel starts
>> before anything takes place in userspace, such as loading modules),
>> CONSISTENTLY.
>>
>> So you have no problem.
> Which is the solution I current ly.
That clearly works across kernel releases.
>> If both devices are modules, then they load when you choose to load
>> them, and so the "order is determined by YOU.
So does that. I just hope he's not playing with initrd.
Peter
------------------------------
From: "BDS" <[EMAIL PROTECTED]>
Subject: Re: Linux on Intel Or Celeron? what is the best choice?
Date: Mon, 9 Apr 2001 15:56:47 -0500
Peter T. Breuer <[EMAIL PROTECTED]> wrote in message
news:[EMAIL PROTECTED]...
<snip>This results in ever increasing yields at the target
> clockspeed. Then the target becomes silly, and they raise the target,
> etc. etc. But if they are already smashing the opposition and the
> market into the ground, they can't STOP the improvement process. They
> just have to market fast chips as slow ones, to stop their market share
> being taken.
>
Sometimes it's hard to get people to believe this, but it happens in many
industries. Chances are that X-Out Titleist golf balls are perfectly good
first-rate golf balls. But, they have to package a certain amount to meet
the demands of the K-Mart market. Sure, they do some basic testing and the
failures get X-ed out and sold for less. But, the tests don't kick out
nearly as many as they need to meet that market demand. They X-Out a much
larger proportion of good balls than bad.
-BDS
------------------------------
From: Mike Lackey <[EMAIL PROTECTED]>
Subject: Serial port driver is making me crazy
Date: Mon, 09 Apr 2001 16:05:03 -0500
Not for sure I'm in the right forum, so bear with me. If I need to go
somewhere else, please point me there.
I have a very strange problem that I cannot seem to get a handle on. It
appears that under certain conditions my interrupt routine is not being
called.
I have written a serial port driver (loadable module) to interface with
our equipment. Standard 16550 kind of code. This code works OK for
long, long periods of time until...
With an HP logic analyzer, I can see that the UART is generating
interrupts to the 8259 chip. I can see that the 8259 chip is generating
interrupts to the CPU. However, for whatever reason, I can also see
that our interrupt code is NOT being called for an extended period of
time, on the order of 4-5 milliseconds (yes, that's right,
milliseconds). Next thing I know, I've got a receiver overrun error.
I know for a fact I'm not getting into the ISR, I've used some spare
bits in a parallel port to show me when I enter/leave. I do know for a
fact that during this 4-5 millisec window somebody else is generating
interrupts to the CPU, just not quite sure who yet. My guess is that a
higher priority ISR is "taking over" and not relinquishing control.
This kernel (Redhat 6.2) has network support enabled, but is not
actually hooked up to a network of any kind. The only things connected
are our equipment, a keyboard, and a monitor, text mode, no mouse or
graphics.
So... I just picked up a book on the Linux kernel and am about to start
delving into the particulars of interrupt handling. Before I make the
leap, does anybody out there have any suggestions? Where to look, docs
to read, etc., etc.
TIA,
Mike Lackey
Sr. Software Engineer
PESA Switching Systems
330-A Wynn Dr.
Huntsville, AL 35805
256.726.9200
------------------------------
** 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 by posting to comp.os.linux.hardware.
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-Hardware Digest
******************************
