Linux-Hardware Digest #399, Volume #12 Sat, 4 Mar 00 02:13:10 EST
Contents:
Re: CPU Over clocking ("LordCompaq")
Re: Wireless Network (Bryan)
Re: Drivers for Real Magic 64/GX display card for Linux ("A. Teutonico")
Re: 96Mb = 64Mb RAM??? (Mark)
Re: HPT366 and Linux (Brian Thomson)
Re: Binary compatibility (Dave Acker)
Re: Linux smp kernel UNSTABLE? (Robert Redelmeier)
Re: Binary compatibility (Christopher Browne)
Re: Advice on PartitionMagic on all-Linux system? (Rod Smith)
Re: Jaton (Trident 9880) videocard problem (Larry Morley)
Re: Ultra66 RH6.1 installs, but LILO hangs (wayne rattz)
Re: ABIT BE6 HighPoint ATA66 controller (wayne rattz)
----------------------------------------------------------------------------
From: "LordCompaq" <[EMAIL PROTECTED]>
Subject: Re: CPU Over clocking
Date: Fri, 3 Mar 2000 20:07:05 -0600
I forgot to mention that I will be installing RedHat 6.1 on this little
beastie....
"LordCompaq" <[EMAIL PROTECTED]> wrote in message
news:38bed58b$0$[EMAIL PROTECTED]...
> Does anyone know if this configuration would ever work?
>
> I have a couple processor cards for a compaq proliant server. They run
> 486DX/50 chips. The chips can be removed. What would happen if I put in a
> dx4/100 chip or the like. Do you think it might work?
>
> I have taken intel overdrive processors and put them in other compaq
> processor boards (P5/66 board with a 133MHz upgrade chip, and it works
> fine).
>
> I know that the DX/? is a factor of a multiplier of the processor bus, in
> this case 50MHz. Is anyone aware of a DX2/100 or something like that? I
have
> been tempted to put a DX2/66 chip in the thing to see what happens, but I
am
> not sure as this chip has a bus speed of 33MHz. I dont know what would
> happen if I tried to run it at 50MHz. I guess I might end up with a 100MHz
> board?
>
> If you know the answer, I would like to hear about it.
>
> Thanks,
> Todd
>
>
------------------------------
From: Bryan <Bryan@[EMAIL PROTECTED]>
Subject: Re: Wireless Network
Crossposted-To: comp.os.linux.networking
Date: Sat, 04 Mar 2000 02:57:46 GMT
In comp.os.linux.hardware Mary K. Conner <[EMAIL PROTECTED]> wrote:
: Okay, now I'm pumped. Here I was looking at having to pop $800 for an
: access point, when instead I can stick an $80 card in my Linux server.
: Only problem is it already has two NIC's in it (one connects to the DSL
: modem, the other to the hub that serves the rest of the LAN). Anybody
: ever stick three NIC's in one machine?
yes - 3 tulip pci cards. no problem.
--
Bryan, http://Grateful.Net (ANTISPAM: email is my name at my web's domain)
(c) 2000. Publishing and/or relaying of this material on all forums other than
USENET implies agreeing to a consultancy fee of US$150 per posting. You must
obtain a written permit before you publish. Violators are subject to civil
prosecution for Copyright Infringement as applicable. Publication by C|NET
and Microsoft Networks expressly prohibited.
------------------------------
From: "A. Teutonico" <[EMAIL PROTECTED]>
Subject: Re: Drivers for Real Magic 64/GX display card for Linux
Date: Fri, 3 Mar 2000 22:19:58 -0500
On Thu, 27 Jan 2000, Kshitij S. Velhal wrote:
> I checked Sigma design's site for the driver but could not find any driver
> for Linux. Can anybody help me finding the Real Magic Drivers?
The Real Magic 64/GX Card is not supported in Linux. In fact, it's a
dead, discontinued product (I used to have it). Gotta get a new
card.
Good Luck!
------------------------------
Date: Sat, 04 Mar 2000 03:36:17 +0000
From: Mark <[EMAIL PROTECTED]>
Subject: Re: 96Mb = 64Mb RAM???
"Victor A. Grinberg" wrote:
> Marcelo Muzilli wrote:
> >
> > Try this,
> >
> > alter your /etc/lilo.conf for:
> > append="mem=96M"
> >
> > Regards,
> >
> > Muzilli
> >
> > trevorjf <[EMAIL PROTECTED]> wrote in message
> > news:[EMAIL PROTECTED]...
> > > I have 96 Mb ram, but Linux only recognises 64Mb. Win 95 uses all 96Mb,
> > > so its installed OK, how can I get Linux to play with the rest?
> > >
>
> And why is it that I have to do this?? Kernel docs say it's because the
> BIOS doesn't report the size >64M. Well, windoze found it, so how come
> linux can't?
> -vg
Depends on your kernel version. At some point they added autodetection for
>64Mb.
I'm a 2.2.14 kernel with 128Mb RAM without having to add 'mem=128Mb' to my
boot switches.
Mark
-----------== Posted via Newsfeeds.Com, Uncensored Usenet News ==----------
http://www.newsfeeds.com The Largest Usenet Servers in the World!
======== Over 73,000 Newsgroups = Including Dedicated Binaries Servers =======
------------------------------
From: Brian Thomson <[EMAIL PROTECTED]>
Subject: Re: HPT366 and Linux
Date: Sat, 04 Mar 2000 03:53:15 GMT
I have encountered the same problem with a Plextor CDRW.
With crafty insertion of printk() calls I have discovered this happens
when the drive is being enabled for DMA.
In file ide.c function ide_config_drive_speed when the drive control
register is set by
if (IDE_CONTROL_REG)
OUT_BYTE(drive->ctl | 2, IDE_CONTROL_REG);
it causes a continuous interrupt condition from the interface.
I don't know what to do about this yet, so right now I am running
with a "hdx=none" boot parameter to disable the drive.
Ryan Maloney wrote:
> Here's a good one I hope someone can help me with: I have a BE6 w/
> HPT366 controller. I have been unable to mount either my CDRW (Ricoh
> 7060A) or my DVD drive (Pioneer 103A) if they are cabled on my HPT366
> controller. The drives will mount fine if attached to the ATA33
> controller, but not the HPT366. I have tried both the 80 pin cable
> and
> the standard 40 pin cable. I have tried the 2.2.14 kernel and the
> 2.3.39 kernel. Both kernels will recognize the drives attached to the
> cables, however upon mounting, the computer seizes up solid and
> nothing short of a hardware reset will bring the computer back.
> My non-working IDE setup looks like this:
> #Onboard DMA33 controller:
> /dev/hda Maxtor HD #1
> /dev/hdb Iomega Zip
> /dev/hdc Maxtor HD #2
> /dev/hdd Pioneer DVD
> #Onboard HPT366 controller:
> /dev/hde Western Digital HD
> /dev/hdg Ricoh 7060A CDRW
>
> Any ideas?
>
> Thanks,
> Ryan
------------------------------
From: Dave Acker <[EMAIL PROTECTED]>
Subject: Re: Binary compatibility
Date: Fri, 03 Mar 2000 23:50:43 -0500
Here goes...
>
> 1. Linux and Windows both run on Intel x86 processors. What stops code
> compiled on one platform from not running on the other (ie why does the OS
> make a difference)?
>
As one reply pointed out, system calls. This is the biggest headache.
Without system calls, you really can not anything.
> 2. Even on the same platform (e.g win95) the object code generated from one
> source (C say) doesn't work with object code from another source (pascal for
> instance). Why not?
>
Actually they can...with some work. Different languages use different
calling conventions when compiled. Some may use registers, some may use
the stack, some may use reverse order, etc. Most compilers allow
modifiers to control this somewhat. I use to write C/C++ DLLs that were
called from all sorts of languages like Pascal, Delphi, Visual Basic,
etc.
By the way...C++ in paticular is nasty because there is no standard
about how "name mangling" is done. This is the process of decorating a
function name to indicate more information, like type and number of
parameters, the class it is a member, etc. This means object code from
different C++ compilers often does not work together. The inventor of
C++, Bjarne Stroustrup, recently gave a Q and Q at http://slashdot.org/
and mentioned this issue. Check out
http://slashdot.org/article.pl?sid=00/02/21/0915255&mode=thread for the
interview.
> 3. When you run your program, what is the extent of the interaction with the
> OS? (eg the OS stops user programs from accessing memory allocated to other
> processes. Does it have to vet each single machine code instruction in order
> to do this?)
>
It is usually some part of the OS that starts a program. You start from
some shell (DOS, Explorer, bash, etc). The previous post discussed the
whole kernel mode vs. user mode thing pretty well. The OS gets involved
with all memory allocation. For example, an OS will present a process
with a flat virtual memory space. When a process allocates memory, it
believes that it has been granted a contiguous chuck of real memory. In
reality the _physical_ memory may be scattered, or it may not exist.
Memory may be swapped to disk. Of course, as soon as that process tries
to access that memory it is swapped back in. This is all done behind
your back. Memory managers that keep track of this stuff can watch for
bad memory accesses as well.
I hope I was able to add a bit more help.
-Ack
------------------------------
From: Robert Redelmeier <[EMAIL PROTECTED]>
Subject: Re: Linux smp kernel UNSTABLE?
Date: Fri, 03 Mar 2000 22:53:58 -0800
You might want to give my hardware test routines a try.
SMP machines can be expecially demanding.
-- Robert author `cpuburn` http://users.ev1.net/~`redelm
Kyo-Bang Chung wrote:
>
> Dear Edward,
>
> Thanks for your valuable message.
>
> Taking a clue from your comment, I restored my CPU clock
> to the original, which was overclocked from 550MHz to 616MHz.
> Then my program ran well on dual CPU and SMP kernel.
>
> What I use is a Iwill DBD100 (dual) motherboard, two Pentium III (550MHz),
> IBM 15GB Hard Disk, Matrox G-200 video, generic LAN card,
> CD-ROM, floppy, and nothing else.
>
> The strange thing is that everything except for my Fortran program is OK
> on SMP kernel, even a couple of test parallel programs from a compiler
> vendor (PGI).
> My sources were tested in Win98/DVF5.0 and I tried three different
> compilers and different versions
> of Linux SMP kernels.
>
> Of course, I do not have resources to try more than one dual-motherboard.
>
> Would you advise more about this Linux/SMP stuff, because you seem to have
> much more experience
> than I do?
>
> Thanks again.
>
> Kyo-Bang Chung
>
> Edward Lee <[EMAIL PROTECTED]>��(��) �Ʒ� �޽�����
> news:[EMAIL PROTECTED]�� �Խ��Ͽ����ϴ�.
> > I use SMP kernels 2.2.12,13 and 14 on M2, K6, P5 and Dual P-II. My only
> > unstable system is due to marginal motherboard. Can you be more specific
> on
> > what you are exercising the kernel on? I/O or file system? Have you
> tried
> > more than one motherboard?
> >
> > Kyo-Bang Chung wrote:
> >
> > > Hi,
> > >
> > > I report that kernel-2.2.12smp and kernel-2.2.15smp is unstable
> > > for some programs, very unfortunately, such as mine.
> > >
> > > My program runs well on kernel-2.2.12 and kernel-2.2.15, and
> > > crashes on the smp versions of both.
> > >
> > > I tested with three compilers (PGI, Fujitsu, Absoft)
> > > on both single- and dual-CPU (Intel PIII). The result are the same.
> > >
> > > The reason I use Linux is the support of muliti-processor.
> > >
> > > Does anyone have any idea of what is going on the smp kernel
> development?
> > >
> > > Thanks in advance.
> > >
> > > Kyo-Bang Chung
> >
------------------------------
From: [EMAIL PROTECTED] (Christopher Browne)
Subject: Re: Binary compatibility
Reply-To: [EMAIL PROTECTED]
Date: Sat, 04 Mar 2000 04:58:03 GMT
Centuries ago, Nostradamus foresaw a time when Barry Kiernan would say:
>Hope somebody can teach me something here.
>
>In order to appreciate the value of COM (component object model), I need to
>understand why it's hard to get compiled code to talk to other code. It
>would help if somebody could explain
>
>1. Linux and Windows both run on Intel x86 processors. What stops code
>compiled on one platform from not running on the other (ie why does the OS
>make a difference)?
As soon as you want to access anything more than the registers on the
CPU, you've got a problem, because you need to interact with the OS in
order to:
a) Request memory, or
b) Request access to a device.
You can write assembly code that demands these resources directly, but
unless you have some sort of protocol for doing so, that code won't
"play well" with any other programs that one might run on the same
machine.
As a result, you need to play by the OS's rules, and with
substantially different OS designs (e.g. - a UNIX-like design versus a
DOS-like design) underneath, Linux and W9x just can't play well
together without some substantial "scaffolding" (ala VMWare or WINE)
to help.
>2. Even on the same platform (e.g win95) the object code generated from one
>source (C say) doesn't work with object code from another source (pascal for
>instance). Why not?
Notably because they have different conventions for passing
parameters.
The only way you get binary interoperability is if you have a
well-specified ABI (Application Binary Interface).
COM is one; CORBA is another, that adds network-interoperability to
the mix. On UNIX systems, there is typically some convention for C
object code that is *nearly* well-specified, which allows some sort of
linking of things together by treating a C-style ABI as the lowest
common denominator.
>3. When you run your program, what is the extent of the interaction with the
>OS? (eg the OS stops user programs from accessing memory allocated to other
>processes. Does it have to vet each single machine code instruction in order
>to do this?)
As soon as you access memory, you "run into" OS control.
>If it helps in pitching your explanation, I'm familiar with assembly
>language (eg calling subroutines by pushing parameters/local vars onto the
>stack) though I don't know what the assembly language equivalent of dynamic
>memory allocation is.
>
>PS saying the OS performs this task doesn't help; after all, the OS boils
>down to some assembly program, no?
No. The OS most often boils down to a C program with some bits of
assembler. Sometimes there are bits of C++. Sometimes it's mostly
FORTH, with some bits of assembler. Sometimes it's basically Lisp,
with some rendering into assembler.
With Linux, the bulk of the kernel is written in C, with some crucial
bits written in assembler. Those bits will include some of the lowest
layer of memory management, and vary from platform to platform.
(Remember that Linux isn't an x86 OS - it has to support Alpha, SPARC,
MIPS, and PPC as well, just to mention the major platforms for which
there are hardware vendors...)
--
"Windows95, Word97, Office98: With all the criticisms of MICROS~1, at
least they include ``best-before'' dating on many of their
products..." -- [EMAIL PROTECTED]
[EMAIL PROTECTED] <http://www.ntlug.org/~cbbrowne/oses.html>
------------------------------
Reply-To: [EMAIL PROTECTED]
From: [EMAIL PROTECTED] (Rod Smith)
Subject: Re: Advice on PartitionMagic on all-Linux system?
Crossposted-To: comp.os.linux.misc
Date: Sat, 04 Mar 2000 05:18:33 GMT
[Posted and mailed]
In article <[EMAIL PROTECTED]>,
[EMAIL PROTECTED] (David C.) writes:
> Michael Meissner <[EMAIL PROTECTED]> writes:
>>
>> You don't need PM to create the initial system, but it is extremely
>> helpful if you want to resize disk partitions without doing a
>> dump/load. My PM has more than paid for itself for that one feature.
>
> Can the current version (5.0) resize an ext2fs partition?
Yes.
> The last time I looked, PM could only move ext2fs partitions, but
> couldn't resize them. If this feature has been added, I'm quite
> interested.
Your recollection is faulty. There was never a version of Partition Magic
that could move but not resize ext2fs partitions. (Except maybe internally
to PowerQuest.) Version 3.0 could recognize ext2fs partitions, but that's
it. Version 4.0 added the ability to move, resize, copy, and create ext2fs
partitions. Version 5.0 hasn't really added anything to that set of
operations, AFAIK.
--
Rod Smith, [EMAIL PROTECTED]
http://www.rodsbooks.com
Author of books on Linux networking & WordPerfect for Linux
------------------------------
From: Larry Morley <[EMAIL PROTECTED]>
Subject: Re: Jaton (Trident 9880) videocard problem
Date: Sat, 04 Mar 2000 01:18:10 -0500
Don't know what RH version you have, but if it's 5.2, 5.0 etc.
you might as well forget AGP period, unless you've upgraded your
X software in the meantime.
Basically, if a card isn't on the list of supported cards for
XFree86 you're awfully lucky if you get the thing to work -
sometimes (very) similar chipsets & controllers will work, but
that's about it.
The only place I've seen the 9880 is on AGP cards. I've got
lots of 9680s (PCI), many of which are Jatons, that work fine,
but they're PCI cards...
If you get desperate, check out the ATI @Play or @Work (I think;
that's close anyway) - they pretty much always work.
Good Luck!
- Larry
steve wrote:
>
> Burak Serdar wrote:
>
> > I have a Jaton Video 107 Blad 3D PCI/AGP (Trident 9880). Redhat
> > installer attempts to use the SVGA server, but it crashes. I installed
> > using the text mode installer, and tried a few trident chipsets, with no
> > luck (again, crash). Should I just buy a new card, or is there anybody
> > out there who made it work?
> >
> > Thanks.
>
> It is quite possible you can use your card.
>
> Rod Smith's article might help you.
>
> http://www.rodsbooks.com/presario/
>
> Steve Kang
------------------------------
From: wayne rattz <[EMAIL PROTECTED]>
Subject: Re: Ultra66 RH6.1 installs, but LILO hangs
Date: Sat, 04 Mar 2000 06:30:12 GMT
John Noble wrote:
>
>
> So I've made some progress with my Promise Ultra66 and RedHat 6.1.
> I'm able to get the install done with the ide2=addr,adrr ide3=addr,addr
> trick, but
> then I run into trouble with LILO. When I try to boot off the first
> Ultra66 drive, LILO hangs at "LI".
>
> I've got 2 Ultra66 drives on this system. The first with Win98 and the
> other specifically for Linux. I put all my partitions on the second
> drive (/boot, /, all of it) and then install LILO on the MBR of the
> Win98 drive. I've added the ide2=addr,addr ide3=addr,addr to the append
> of the lilo.conf, but to no avail. I have not installed and new
> kernels/patches for Ultra66 support. I figure I shouldn't have to just
> to get LILO to boot the system right?
> I'm pretty sure /boot is under the 1024 cylinder limit, but it's on a
> different drive than where LILO is installed (LILO is on the MBR of the
> win98 drive, /boot is on the other drive) should this matter?
> I took the lilo.conf that the RH 6.1 installer created and tired adding
> "linear" and "bios=0x80" to the global section, but that did not help.
> If this is a disk geometry problem, how can I tell if it is and any
> ideas how I would solve it?
>
> Should I use another loader (maybe ChooseOs)?
>
> Any clues here to any of this?
>
> Thanks
> pulling out my hair
> -John
>
> Hello john:Take out the boot partition it dosent need it and more than
likely its confusing lilo as to where to boot from..Just put lilo in the
master boot record.GOOD LUCK WAYNE!
--
Posted via CNET Help.com
http://www.help.com/
------------------------------
From: wayne rattz <[EMAIL PROTECTED]>
Subject: Re: ABIT BE6 HighPoint ATA66 controller
Date: Sat, 04 Mar 2000 06:30:09 GMT
Preiz wrote:
>
>
> Hi, thks for any kindly help in advance. When I tried to install linux
on
> my pc with ABIT BE6 motherboard. Linux seems not able to recognise the
> HighPoint ATA66 controller. Since I only have one ATA66 disk on the
ATA66
> controller and nothing on the oridiance ata33 controller.
>
> Does anyone know if there is any driver or workaround for the HP366
> controller. Once again, thanks very much
>
HELLO:Try using my howto at http://www.geocities.com/wrattz/linux6.html
For my main page change the 6 to a 1 Good luck wayne!
>
--
Posted via CNET Help.com
http://www.help.com/
------------------------------
** 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.hardware) 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-Hardware Digest
******************************
