Linux-Hardware Digest #415, Volume #14 Wed, 28 Feb 01 11:13:09 EST
Contents:
Re: New SCSI drive partition information different for fdisk and cfdisk???? (Rod
Smith)
Please help, can't use Backpack cdrw on Redhat machine ("Faldar")
Problem with Mustek 12000SE (Michael Felsberg)
RAID on Linux: What type of hardware to choose? (Toralf Lund)
Re: USR modem in Red Hat 7.0 (RDetenbeck)
Re: GLX driver for Riva TNT in RedHat (Harri Haataja)
Re: intel 810 chipset
Re: VIA IDE problems (was Re: Need LOTS of disks: Promise ATA RAID??) (Piercarlo
Grandi)
pc/104 modem problems (glen)
Linksys LNE100TX ver 4.0 SuSE 7.1 Probs (Mike McCann)
Re: Linux partitioning question ("Cjv")
Re: RAID on Linux: What type of hardware to choose? (Joshua Baker-LePain)
----------------------------------------------------------------------------
From: [EMAIL PROTECTED] (Rod Smith)
Subject: Re: New SCSI drive partition information different for fdisk and cfdisk????
Date: Wed, 28 Feb 2001 14:13:50 -0000
Reply-To: [EMAIL PROTECTED]
[Posted and mailed]
In article <Lq1n6.7147$[EMAIL PROTECTED]>,
"Pete Willemsen" <[EMAIL PROTECTED]> writes:
> Hello.
>
> I recently installed Linux (Debian woody) on a P4 machine that has two SCSI
> disks. Linux installed well on the second disk (/dev/sdb) which is an IBM
> Ultrastar 18GB SCSI disk. Everything seemed perfectly fine with the system
> until I attempted to look at the partitions with PartitionMagic.
>
> PartitionMagic reported that the Linux partition had an error. So, I booted
> Linux, ran fdisk and received the following report from "fdisk -l /dev/sdb":
>
> ----
> Disk /dev/sdb: 1 heads, 35565080 sectors, 1 cylinders
This is extremely unusual. Also, compare it to the cfdisk report....
> and the cfdisk table print produced:
>
> ---
> Partition Table for /dev/sdb
>
> ---Starting--- ----Ending---- Start Number of
> # Flags Head Sect Cyl ID Head Sect Cyl Sector Sectors
> -- ----- ---- ---- ---- ---- ---- ---- ---- -------- ---------
> 1 0x80 1 1 0 0x83 255 63 1023 63 35304129
You can't have 255 heads and 1023 cylinders on a disk with just 1 head
and 1 cylinder. Apparently fdisk and cfdisk are using different methods
to determine the CHS geometry for the disk. This probably interacts with
the SCSI BIOS and/or Linux's SCSI drivers.
Having inconsistent CHS geometry is an invitation to disaster. I
therefore recommend that you back up, fix the problem, and restore.
Unfortunately, "fix the problem" is easier said than done. Without
knowing the precise cause of the problem, it's hard to say how to fix
it. As a first stab, though, I'd do this:
1) In Linux, use "dd if=/dev/zero of=/dev/sdb bs=512 count=1". This will
completely wipe out the partition table, along with any flaky
information it might contain that could be the root of the problem.
2) Since you mentioned you've got Partition Magic, I'd recommend
booting it and using it to create new partitions. I believe that PM
has some way to report the CHS geometry, so check that it doesn't use
1 for the number of heads or cylinders. If it does, start over or try
another tool.
3) When you re-install or restore Linux, check with fdisk and cfdisk to
see that the CHS geometry is being reported consistently.
You might also investigate Linux driver options for your SCSI host
adapter. It's possible that this is caused, at least in part, by a SCSI
driver bug. Post information on what SCSI adapter you're using if you
continue to have problems and post again.
--
Rod Smith, [EMAIL PROTECTED]
http://www.rodsbooks.com
Author of books on Linux & multi-OS configuration
------------------------------
From: "Faldar" <[EMAIL PROTECTED]>
Subject: Please help, can't use Backpack cdrw on Redhat machine
Date: Wed, 28 Feb 2001 14:41:44 GMT
I am trying to get a Backpack cdrw drive (parallel port external, fyi)
installed and working on a Redhat 7.0 based computer running kernel 2.4.2.
Please note that I have never tried to use or install a cdr or cdrw on a
Linux computer before.
I compiled the bpck protocol as a module and also put directly in the kernel
the parport and the paride as a module. I also put in the scsi emulation
(not sure as a module or not) and I have the kernel set up to auto load
modules.
I have also checked the manufacturers web site with no luck. The only thing
available was a download of the backpack module for 2.4.0 kernel. As I
already have the bpck protocol set up in the kernel, it is not terribly
useful to download it. Besides, I tried it and can't install the download
since it was compiled for a different kernel version.
I have also checked any documentation I could find on parallel port drives
and installing cdr drives on www.linuxdoc.org and a few other sites that I
could find including www.linuxnewbie.org Naturally, I tried following the
'readme' that came with the downloaded module, but also with no luck.
If I try to mount the drive, I get an error saying 'incorrect major or minor
node number' when I try to mount the drive as pcd, and pg returns error 'pg
not block device'. Further, trying to load modules manually yields also no
luck. I can manually load paride just fine, but trying to load pg just
returns error:
pg version 1.02, major 97
pg0: Autoprobe failed
no atapi device detected
operation not permitted
trying to load pcd yields almost identical results with only the following
difference:
pcd version 1.07, major 46 nice 0
....
no cdrom detected.
Thanks for any help that I can get.
Mike Cooper
------------------------------
From: Michael Felsberg <[EMAIL PROTECTED]>
Subject: Problem with Mustek 12000SE
Date: Wed, 28 Feb 2001 15:38:14 +0100
Hi,
I had the Mustek 12000SE running with the original adapter in my
desktop. Now I use a notebook with the APA1460 PCMCIA-SCSI adapter. The
scanner gives totally destorted results: there is only a small stripe of
the image which is repeated until the bottom border of the image. Do you
have any idea why?
a. I use that scanner with the 50pin hd plug which is shipped with the
apa1460.
b. The scanner is the only device on the bus and I use no terminator
(generally this is not OK, but for SCSI-I with one device it should
work).
c. I disabled sync and reconnect for the apa1460, but I do not know how
to disable queuing.
d. Maybe there is something wrong in my mustek.conf (I use no special
options, if I enable buffersize and blocksize the scanner fails to
detect the end of its mechanics (brrrrr) - I immediately switched it off
and shot my notebook into DEEP sleep ...)
System: SuSE 7.1, SANE 1.04, Kernel 2.4
Michael
------------------------------
From: Toralf Lund <[EMAIL PROTECTED]>
Crossposted-To: redhat.linux.misc
Subject: RAID on Linux: What type of hardware to choose?
Date: Wed, 28 Feb 2001 15:50:44 +0100
We want to set up a RAID system consisting of 6 to 8 36 or 73 Gb disks
(initially; we may or may not want to add more disks later) for a Linux
host , which is Intel (dual-Pentium-III) based and runs Red Hat 7.0.
However, we are uncertain about what type of hardware to choose. As I
understand, there are 3 different types of configurations, namely:
1. An external, self-contained SCSI unit (SCSI-to-SCSI)
2. Internal (PCI) RAID controller and SCSI disks.
3. Internal RAID controller with IDE/ATA disks.
I was just wondering about what other peoples experience with the
different setups are, and what you would recommend. - At the moment, I'm
inclined to go for alternative 1. as it seems to be the most reliable
one, and also the one where the installation involves the least amount
of hassle. I'm assuming that there will be a certain amount of messing
around with drivers with 2. and 3., but maybe this isn't really a
problem? Also, 1. is probably the most expensive solution, which isn't a
major concern as such, but an added cost obviously has to be justified.
Finally - and this may be a silly question, but I really have no
experience with ATA disks - would selecting the IDE/ATA setup mean that
all disks need to fit into the main cabinet? (I seem to remember that
there was such a limitation with IDE back in the old days.)
- Toralf
------------------------------
From: [EMAIL PROTECTED] (RDetenbeck)
Date: 28 Feb 2001 15:11:56 GMT
Subject: Re: USR modem in Red Hat 7.0
concerning difficult modem installations:
The best reference material I have found is the Modem--HowTo at www.redhat.com
Bob Detenbeck
------------------------------
From: [EMAIL PROTECTED] (Harri Haataja)
Subject: Re: GLX driver for Riva TNT in RedHat
Date: Wed, 28 Feb 2001 15:14:16 GMT
Marcus H�gglund wrote:
>I've also got an ASUS V3400... 0.9.6 seems to work for some programs,
>like some of the demos delivered with Mesa. It doesn't support
>dynamic-texturing or something like that (can't really remember what
>it's called).
>
>I finally got it to support AGP but although the driver /proc/nv/...
>says that sideband is supported it doesn't use it?
>
>Have you tried to use 0.9.6?
No, and that was 2.2.17 I think.
So I might give it a try once I get my hands on it again.
I suppose no-one has had luck with the video chip in it...
--
We are the dyslexic borg! Your ass will be laminated!!
------------------------------
From: <[EMAIL PROTECTED]>
Subject: Re: intel 810 chipset
Date: Wed, 28 Feb 2001 15:30:04 -0000
I'm having trouble getting X to work with my Optiplex GX110 system. If you
would, please send me a copy of your XF86Config file. Also, which server
are you using? I'm running Red Hat 6.2 with the standard kernel.
Thanks,
John Pfaff
[EMAIL PROTECTED]
Marc van Duivenvoorde wrote:
>
>
> I have a dell optiplex gx 100 system with a intel 810 chipset for video.
I'm
> running mandrake 7.1 and I have a weird problem.
>
> After editing the XFconfig file myself because mandrake didn't detect the
> card at all I have a fine running X system when using the standard kernel
> 2.2.15. I want to use the 2.4.1 kernel, but when boot that one X starts
> whining that it doesn't know the intel chipset while I do have the right
> drivers.
>
> Thanks a lot.
>
> Marc
>
>
--
Posted via CNET Help.com
http://www.help.com/
------------------------------
From: [EMAIL PROTECTED] (Piercarlo Grandi)
Crossposted-To: uk.comp.os.linux
Subject: Re: VIA IDE problems (was Re: Need LOTS of disks: Promise ATA RAID??)
Date: 28 Feb 2001 15:56:57 +0000
>>> On Tue, 27 Feb 2001 14:15:27 +0000, [EMAIL PROTECTED] (Tony Houghton)
>>> said:
tony> In <[EMAIL PROTECTED]>,
tony> Jason Clifford <[EMAIL PROTECTED]> wrote:
>> My understanding is that there are serious problems with the VIA IDE
>> chipset support for ATA-100 although not everyone appears to have the
>> problems. I don't use it here though so I will be selfish and say I
>> don't care too much.
tony> I keep hearing of problems with VIA IDE, in particular that
tony> enabling DMA can cause corruption (only with kernel 2.4 though I
tony> think). Is this only for ATA100 or can VIA motherboards safely be
tony> used with ATA33 with DMA?
I have been using lots of releases of 2.2.x and 2.3.x and 2.4.x with
both a VIA 586 chipset with ATA33 and a VIA 686 with ATA66/100, and
quite successfully. I suspect there is no real problem with the Linux
IDE driver, perhaps maybe only some tighter timing settings than is
prudent, but this is only speculation.
However, there have been reports that ATA100 is susceptible to
electrical noise; and I well believe that because I used to have
them with ATA33, until I shortened the cables in half.
So, even if I have ATA100 disks, but I have set them to actually run at
ATA66 using a manufacturer's utility.
Given that ATA100 gives essentially no advantage over ATA66 even with an
ATA100 capable disk, I'd rather be safe than sorry.
tony> I'm thinking of upgrading one of my systems to a Duron later in
tony> the year, and apart from that problem, the VIA motherboards seem
tony> to be the best. I don't mind having to stick to ATA33, because 66
tony> and 100 hardly make a difference in practice,
Actually, if you have a fast recent disk, ATA66 can make a (largish)
difference wrt ATA33.
Sequential transfer speed on my IBM 46GB/7200RPM disk is something like
37MB/s, which is 7MB above ATA33.
BTW, this number somewhat surprised me, it's amazing. My other disk,
another really good IBM 25GB/5200RPM does top sequential transfer speed
at 15MB/s, which is less than half (the RPM difference does not
necessarily explain all or most of the difference).
Even more importantly the 46GB disks has a measured average access
time of 8.5 milliseconds, and the 25GB one of 14, which is also
somewhat suprising, and higher than the difference in RPM implies.
With a 37MB/s disk one wants ATA-66; with one or even two 15MB/s disks
ATA-33 is enough though.
Quite surprisingly with VIA chipsets the 686B (which supports up to
ATA-100) is a better choice for _ATA-66_ than the 686A (which supports
up to ATA-66). Check this:
http://www6.tomshardware.com/mainboard/00q4/001026/via-05.html
Here the 686A tops at 30MB/s, less than half of what ATA-66 is
theoretically capable, while the 686B in the same mode runs it much
faster.
tony> but IME, turning off DMA altogether can make a big difference.
Yes, not only to transfer rate, but also to CPU utilisation. With _any_
DMA on CPU utilisation during IO is very small compared with PIO modes.
[ ... ]
------------------------------
From: glen <[EMAIL PROTECTED]>
Subject: pc/104 modem problems
Date: Wed, 28 Feb 2001 15:22:48 +0000
Hi,
Trying to use a pc/104 modem under linux (redhat 6.2) - its called an
IPC2-5600 (??) and contains a rockwell chipset, its
supposed to be a pukka modem rather than a winmodem.
Its got two dip-switch banks which select com setting and port/irq - its
mapped to port 0x2e8 (on irq 3), however with both
com and pnp settings nothing appears when I do a pnp dump (I've tried
changing the port number but to no avail).
Now I was under the impression that this modem should be able to plug in
and be seen as a regular ISA device - does anyone know if this is true -
do I have to do anything with the kernel (I've had a good look on the
web and could not find
anything specific)?
Has anyone got any ideas or suggestions of how I can get this to work -
I'm a bit clueless as to where to start looking.
Cheers,
Glen
------------------------------
From: Mike McCann <[EMAIL PROTECTED]>
Crossposted-To: alt.os.linux.suse
Subject: Linksys LNE100TX ver 4.0 SuSE 7.1 Probs
Date: Wed, 28 Feb 2001 15:46:15 GMT
Hi
Can't get the known fixes for SuSE 6.3 or 7.0 to get my 7.1 working.
Has anyone got any solutions.
I don't think it is a IRQ problem because part of the problem when I
tried to compile the updates was there was an error compiling
pci-scan.o. So, not being an old hand at this upgrading process I do
have a couple of questions.
there are some commands like
gcc -DMODULE -D_KERNEL_
-l/usr/src/linux/net/inet
-Wall -Wstrict-prototypes -O6 -c tulip.c
'[ -f /usr/include/linux/modversion.h]
&& echo -DMODVERSIONS
Do I enter the command verbatum or are there something I should
substitute for -DMODULE or /usr/src/linux/net/inet ?
Thanks
Mike
------------------------------
From: "Cjv" <[EMAIL PROTECTED]>
Crossposted-To: comp.os.linux.setup,comp.os.linux.misc
Subject: Re: Linux partitioning question
Date: Wed, 28 Feb 2001 07:34:06 -0800
"Floyd Davidson" <[EMAIL PROTECTED]> wrote in message
news:[EMAIL PROTECTED]...
> [EMAIL PROTECTED] (Rod Smith) wrote:
> For instance, in most of the multi-partition examples shown in
> this thread there have been both a /tmp and a /var partition.
> That is an unnecessary waste of disk space. The /var partition
> is going to suffer high fragmentation, which is one reason it
> should be a separate partition, but that also makes it a great
> place for /tmp to physically be located. Both /tmp and /usr/tmp
> should be symlinks to /var/tmp.
>
How would I symlink /tmp and /usr/tmp to /var/tmp when they seem to be
created automatically when I do the install?? (Im a newbie)
> Other obvious candidates for locating on other partitions with a
> symlink are /usr/local, /usr/X11, /opt, and where ever it is
> that emacs/xemacs or tex are located.
>
> Likewise the /home directory can actually be on one or more
> other partitions. /home itself can be a symlink, but so can
> each user's directory if that is useful (as might be for the
> /home/ftp directory, as an example).
>
> Hence, while it is possible to get / or /usr partitions too
> small, they will be too small to even install the first time if
> that is true. If those partitions are large enough to actually
> manage a functional install to begin with, they *never* require
> resizing.
>
> --
Floyd, If I were to do a fresh install then, how many original partitions do
you suggest I create? Then how would I use symlinks (as you describe)?
Would this approach be harder to restore if one of the partitions became
unstable?
====== Posted via Newsfeeds.Com, Uncensored Usenet News ======
http://www.newsfeeds.com - The #1 Newsgroup Service in the World!
======= Over 80,000 Newsgroups = 16 Different Servers! ======
------------------------------
From: Joshua Baker-LePain <[EMAIL PROTECTED]>
Crossposted-To: redhat.linux.misc
Subject: Re: RAID on Linux: What type of hardware to choose?
Date: 28 Feb 2001 16:07:18 GMT
Toralf Lund <[EMAIL PROTECTED]> wrote:
> We want to set up a RAID system consisting of 6 to 8 36 or 73 Gb disks
> (initially; we may or may not want to add more disks later) for a Linux
> host , which is Intel (dual-Pentium-III) based and runs Red Hat 7.0.
> However, we are uncertain about what type of hardware to choose. As I
> understand, there are 3 different types of configurations, namely:
> 1. An external, self-contained SCSI unit (SCSI-to-SCSI)
> 2. Internal (PCI) RAID controller and SCSI disks.
> 3. Internal RAID controller with IDE/ATA disks.
4. External, self-contained unit with IDE disks which connects to host
via SCSI (IDE-SCSI RAID).
> I was just wondering about what other peoples experience with the
> different setups are, and what you would recommend. - At the moment, I'm
> inclined to go for alternative 1. as it seems to be the most reliable
That's been the "standard" for a while now, at least in the high end sector.
> one, and also the one where the installation involves the least amount
> of hassle. I'm assuming that there will be a certain amount of messing
> around with drivers with 2. and 3., but maybe this isn't really a
The 3ware cards (option 3) seem to be very well supported. The driver
is already in your distro. It *should* be as easy as a modprobe.
> problem? Also, 1. is probably the most expensive solution, which isn't a
> major concern as such, but an added cost obviously has to be justified.
You haven't given us the most important bit of information -- what is
this going to be used for? Are you going to have tons of users pounding
on this thing day and night? How much of a problem is downtime?
> Finally - and this may be a silly question, but I really have no
> experience with ATA disks - would selecting the IDE/ATA setup mean that
> all disks need to fit into the main cabinet? (I seem to remember that
> there was such a limitation with IDE back in the old days.)
3ware sells 4 and 8 channel cards for IDE RAID, one drive per channel.
That gets over IDE's limitation.
There's been a number of threads on this recently. The gist is, you need
to determine what your needs are and what you are willing to pay to fulfill
them. Myself, I need a fair bit of space (.5TB), but I don't need humongous
speed/multitasking abilities. So I'm looking very hard at option 4. In
fact, I just got a quote for a pretty nice system -- Syneraid 800T (UDMA/66
disks, U2W LVD host connection, 8 hot swap drive bays, dual power supplies)
with 8 80GB 5400RPM Maxtor disks for ~$5900US. The vendor claimed that
the 5400RPM disks only minimally affect throughput vs. 7200RPM models, and
that they have "timing issues" with the 75 and 80GB 7200RPM disks.
If performance is critical, there's just no beating SCSI. If you've
got the bucks and the need, go with option 1. I got quotes on some
*very* nice systems from Zzyzx (www.zzyzx.com). Check out their
RocketStor and RocketRaid lines. I highly recommend them as a vendor.
--
Joshua Baker-LePain
Department of Biomedical Engineering
Duke University
------------------------------
** 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
******************************
