Linux-Hardware Digest #566, Volume #13 Tue, 12 Sep 00 16:13:08 EDT
Contents:
Re: Still not all memory detected even with 'mem = ' ("Peter T. Breuer")
Re: Poor hde performance ~ 1.84MB/sec (Harald van Pee)
Mammoth tape drive I/O errors ([EMAIL PROTECTED])
Re: SCSI Tape Drive Question ([EMAIL PROTECTED])
Re: modems (winmodems?) ("Green Goat")
Re: SCSI Tape Drive Question ([EMAIL PROTECTED])
Re: Im building a new computer to dual boot w2k and mandrake 7.1and (Stanislav
Kogan)
Re: help me identify my TV card (Stanislav Kogan)
Re: Linux won't boot with Thunderbird! ("Chris Ripp")
Re: promise controller ultra66 problems ("Greg James")
Re: Linux software RAID vs. hardware RAID ([EMAIL PROTECTED])
Re: Logitech Optical Mouse (Brian Langenberger)
Re: Baffling IDE HD performance difference (Andrey Vlasov)
Re: Im building a new computer to dual boot w2k and mandrake 7.1and (Stanislav
Kogan)
Re: SCSI Tape Drive Question ([EMAIL PROTECTED])
Re: Memory questions, Device questions; Remove LILO on RH6.2 ("Simple Simon")
----------------------------------------------------------------------------
From: "Peter T. Breuer" <[EMAIL PROTECTED]>
Subject: Re: Still not all memory detected even with 'mem = '
Date: 12 Sep 2000 18:02:38 GMT
kyle matteson <[EMAIL PROTECTED]> wrote:
: you must make sure that the M in 64M is capital
Whilst this is good practice, people got tired of telling other people
that and changed the recent kernel codes to ignore the case instead.
: append="mem=64M"
: and then you must remember to run lilo afterwards to make the change take
: affect.
Peter
------------------------------
From: Harald van Pee <[EMAIL PROTECTED]>
Subject: Re: Poor hde performance ~ 1.84MB/sec
Date: Tue, 12 Sep 2000 20:39:39 +0200
I would suggest:
Try the bonnie benchmark with 1GB test file and compare read/write
performance
(see my posting: read performance slows down). If write performance is
good (>10MB/sec) I think you have a hardware problem.
If not, try kernel 2.2.16/17 and apply ide patches.
All dma modes with disks from the last years should give a read
performance of at least 10MB/sec (I think).
The slowest I know is an older disk with 6MB/sec.
Look into your /var/log/boot.msg file to identify your hardisk (there
are more than one new maxtor 40 GB harddisks).
Regards
Harald
------------------------------
From: [EMAIL PROTECTED]
Subject: Mammoth tape drive I/O errors
Date: Tue, 12 Sep 2000 18:28:28 GMT
I recently acquired an Exabyte Mammoth tape drive for my Red Hat
6.2 Linux machine. It replaces an old Exabyte 8200 drive. I can
read the old 8200 tapes on the Mammoth, but I can't write anything
to the special Mammoth tapes. I have ExaTape 170mAME tapes.
After I read the 8200 tape, it told me I had to clean the drive, which
I did. When I try to write a tape using
dd of=/dev/nst0 if=<some big file>
I get an I/O error. If I do a "mt -f /dev/nst0 status" I get a
normal status message back except it doesn't like the density setting.
I sent email to Exabyte and they acknowledged my message, but haven't
gotten back to me for a week.
I searched this newsgroup and found a few people who were having
the same problem, but no solutions.
Any ideas? Any help would be greatly appreciated.
[EMAIL PROTECTED]
Sent via Deja.com http://www.deja.com/
Before you buy.
------------------------------
From: [EMAIL PROTECTED]
Subject: Re: SCSI Tape Drive Question
Date: Tue, 12 Sep 2000 18:37:22 GMT
In article <[EMAIL PROTECTED]>,
[EMAIL PROTECTED] wrote:
>
> >scsi : 0 hosts.
> >scsi : detected total.
> >
> >So far, so good.
> So far NO Good! Linux is not seeing your Scsi host adaptor!
> Check your DMA, IRQ and I/O addressing of your card. You may need to
> change them to the factory defaults. I have not used such a card as I
> use a Vesa Localbus AVA-2825VL. Check the Scsi-HOWTO to find out how
> to manually enter your Host Adaptor info into the Lilo command line.
> Unless your host adaptor gets set right. Forget the tape! Once
> your adaptor gets noticed. the tape devices you will use are very
> generic /dev/st0 and /dev/nst0.
>
>
Hi B'ichela,
Thanks for the suggestions. I read the SCSI-HOWTO and played around
more, but I am still having trouble. Here is where I am at. On bootup,
the SCSI card's BIOS does display a message on the screen (as Steve Wolfe
suggested I check), so I know that the card (1) is working and (2) is
correctly detecting the tape drive with the right SCSI id.
I then boot into Linux, but I still get the old:
scsi : 0 hosts.
scsi : detected total.
I have tried various LILO command line options (per the suggestions in
the HOWTO) such as:
aha274x=extended
(which is the one suggested; I have also played around with a bunch of
other ones)
but none of them seem to help. Linux never successfully sees the SCSI
card.
Does anyone have any other ideas as to what I might do? Does anyone else
have an Adaptec 2740 card that works correctly under linux?
Thanks,
Mike
Sent via Deja.com http://www.deja.com/
Before you buy.
------------------------------
From: "Green Goat" <[EMAIL PROTECTED]>
Subject: Re: modems (winmodems?)
Date: 12 Sep 2000 18:53:47 GMT
I suggest you to see the website,
http://www.o2.net/~gromitkc/winmodem.html
it can make you clearly.
Mike <[EMAIL PROTECTED]> wrote in message
news:8pkqus$r1i$[EMAIL PROTECTED]...
> Hi all, I'm looking for a new external modem, but I'm having trouble
> finding one that I can use with Linux, largely because all the modem
> manufacturers seem to put "Windows" under system requirements, leaving
> me uncertain as to whether or not they are winmodems. Are external
> modems generally not windmodems?
>
> I was looking at the USR Faxmodem (external), until I saw the line
> "Operating Systems Modem is compatible with Windows NT 4.0 (Windows
> 2000 support expected)"
>
> Bah. Can anyone recommend a good external modem?
>
> --
> mjo
>
>
> Sent via Deja.com http://www.deja.com/
> Before you buy.
------------------------------
From: [EMAIL PROTECTED]
Subject: Re: SCSI Tape Drive Question
Date: Tue, 12 Sep 2000 18:54:13 GMT
New development:
I just tried loading the aic7xxx.o module manually by doing a 'insmod /
lib/modules/2.2.15mdk/scsi/aic7xxx.o' and got the following output:
[root@qed /root]# insmod /lib/modules/2.2.15-4mdk/scsi/aic7xxx.o
scsi0 : Adaptec AHA274x/284x/294x (EISA/VLB/PCI-Fast SCSI) 5.1.29/3.2.4
<Adaptec AHA-294X Ultra2 SCSI host adapter>
scsi : 1 host.
(scsi0:0:6:0) invalid cur_addr:0x99feb8e during WIDE_RESIDUE
(scsi0:0:6:0) sg_address[-1]:0x0 sg_length[-1]:0
(scsi0:0:6:0) sg_address:0x0 sg_length:0
(scsi0:0:6:0) cur_address:0x99feb8e resid_dcnt:0x0000d1
(scsi0:0:6:0) invalid cur_addr:0x99feb8e during WIDE_RESIDUE
(scsi0:0:6:0) sg_address[-1]:0x0 sg_length[-1]:0
(scsi0:0:6:0) sg_address:0x0 sg_length:0
(scsi0:0:6:0) cur_address:0x99feb8e resid_dcnt:0x0000d1
(scsi0:0:6:0) invalid cur_addr:0x9c77dc2 during WIDE_RESIDUE
(scsi0:0:6:0) sg_address[-1]:0x0 sg_length[-1]:0
(scsi0:0:6:0) sg_address:0x0 sg_length:0
(scsi0:0:6:0) cur_address:0x9c77dc2 resid_dcnt:0x0000d2
Vendor: ARCHIVE Model: Python 06408-XXX Rev: 8071
Type: Sequential-Access ANSI SCSI revision: 03
Detected scsi tape st0 at scsi0, channel 0, id 6, lun 0
The tape drive works now! However, what do all these errors mean?
Should I be concerned about all this "invalid cur_addr"? Also, is this a
kosher way of loading the driver? Is it ok just to put the insmod
command in one of the startup scripts?
-mike
Sent via Deja.com http://www.deja.com/
Before you buy.
------------------------------
From: Stanislav Kogan <[EMAIL PROTECTED]>
Subject: Re: Im building a new computer to dual boot w2k and mandrake 7.1and
Date: Tue, 12 Sep 2000 15:04:28 -0400
Hi!
Michael D wrote:
>
> H i,
>
> I am considering this configuration and I am just curious befor I commit to
> my purchase what professionals think of it. If you could recomend another
> config please post it. The system will be for a tech student so dual os
> compatibility and value is more important than performance.
Well, I beleive that stability is always an issue. So:
>
> Mainboard Asus cuv4x
Don't touch this board. It sucks. In fact, ANY VIA based board is a bad
choice.
What you need is an Intel 440BX or 815 chipset based board. I'd go with
ABit BF6, BE6 or Asus CUBX.
> Processor Celleron 566
Very good choice. This CPU, BTW is VERY overclockable. If you are lucky,
you will be able to clock it to 850Mhz in no time. (The luck here
depends on weather you'll get a good chip. Your chance is about 50/50)
> Ram Micron 128mb
Good choice. Good RAM is always better than bad RAM. :-)
> Sound Card Creative Labs SB Live Value
Leave this card alone. SB Live's performance is greatly exaggerated by
Creative itself. If you are feeling adventurous, grab an Aureal Vortex2
(AU8830 based). The drivers for this card are still under heavy
development, but recently became stable on uniprocessor systems. SMP
systems, however is another story.
> Graphics Card 3dfs Voodoo3 3000
I'd take a TNT2 or maybe a Geforce MX based card. You see, 3dfx is on a
steep decline right now, and their cards tend to be extremely
overpriced.
If you are not planning to do much 3d (and under Linux you can't anyway,
unless you are a quake addict), then Matrox g400 may be a good choice.
This card has a much better graphics quality than those gamers' cards.
> Modem
Take a Hardware modem of course. I personally had a very good luck with
AOpen ITU90. Also, make sure it's ISA and not PCI. (PCI is harder to
configure under Linux). Also, Non-plug and play will help too.
> Hard Drive Maxtor DM 15gb 7200
Good choice.
> Floppy Drive 1.44mb
Here it doesn't matter much
> Cd Rom
Make sure you grab a good one. Don't be fooled by the advertised speeds,
what matters is the ability to read different kinds of CD-Rs. I'd take a
Sony or a Mitsumi. Make sure you won't take a piece of s%$t like
Creative or BTC or some unnamed brand.
> Ethernet Card Dlink10/100
Looks like a decent choice.
> Case Overcool
> Monitor 17"
Try to find AOC 7Glr. They are reletavily cheap, but offer great
performance at the same time. Of course it's not a perfect flat, but
it's .25, and 100Hz at 1024x768. (and the maximum is 1600x1200)
I hope this helped.
Best regards.
------------------------------
From: Stanislav Kogan <[EMAIL PROTECTED]>
Subject: Re: help me identify my TV card
Date: Tue, 12 Sep 2000 15:12:02 -0400
Hi!
TomZ wrote:
>
> Hi,
> thanks. I have got it working now.
> Still, i can not hear sound.
> But installing worked quite easy.
> ciao
> Sebastian
Don't worry, this is the usual problem. First check if you connected it
to the sound card properly, then look on the card again, and check if
you have any chips that start with TDA TEA or MSP.
D/L the LATEST BTTV package, and install it. Then load ALL the modules
that correspond to the chips that you found on your card. After that, it
should work.+
------------------------------
From: "Chris Ripp" <[EMAIL PROTECTED]>
Subject: Re: Linux won't boot with Thunderbird!
Date: Tue, 12 Sep 2000 13:23:20 -0500
That's funny,
usually *anyone* will boot with Thunderbird....
given enough of it....
oh, wait...
never mind.
"Frank T. Allotta" <[EMAIL PROTECTED]> wrote in message
news:VKev5.1061$[EMAIL PROTECTED]...
> Hi,
>
> I am running an Abit KT7-RAID motherboard with an AMD 750 Thunderbird.
> I just installed them the other day. I installed Linux 6.2 and everything
> went great during the install. Now when Linux tries to boot up, it gives
an
> error. It gets to a point where it says "Disabling Athlon Serial Number"
> and then it gives a "General Protection Fault!" Then it it just locks up
at
> another error about "Not Syncing." What the heck is this? I had Linux
> running on the same hard drive with a different mb and processor earlier
in
> the week. That is the only thing that has changed. Is this some sort of
> BIOS setting problem? Give me a hand here!
>
> Frank
>
> You can email me directly if you like. Thanks!
>
>
>
>
------------------------------
From: "Greg James" <[EMAIL PROTECTED]>
Subject: Re: promise controller ultra66 problems
Date: Tue, 12 Sep 2000 14:30:28 -0500
When you install, pass "ide2=0xeff0,0xefe6" to the kernel. You do this by
add this to the prompt when you install. This assumes the hard drives are on
the primary interface of the promise card. If you have drives on the second
interface, also add "ide3=0xefa8,0xfe2" to the prompt. When you boot, you
also have to pass the same parameters to the kernel. Also, you need to pass
what drive you are booting from to the kernel. You do this with the "boot="
parameter. Let's say you have a system where you are booting from the first
drive on the primary interface on the promise controller. When lilo comes
up, you would type "linux ide2=0xeff0,0xefe6 ide3=0xefa8,0xfe2
boot=/dev/hde". Then once the system comes up, modify /etc/lilo.conf to to
do these. Mine looks like this:
/etc/lilo.conf:
boot=/dev/hde
map=/boot/map
install=/boot/boot.b
prompt
timeout=25
linear
default=linux
image=/boot/vmlinuz-2.2.16-3
label=linux
read-only
root=/dev/hde1
append="ide2=0xec98,0xecaa ide3=0xeca0,0xecae"
initrd=/boot/initrd-2.2.16-3.img
After you update lilo.conf, don't forget to update it by running "lilo -C
/etc/lilo.conf". I forgot once and it took me about two hours to recover.
What a pain.
I'm running RH 6.2 and I updated the Kernel to 2.2.16, so my /etc/lilo.conf
might look a little different.
I had to to a major work-around before I figured out this would work. I had
no luck with the promise boot disk they have on their website and they never
returned my calls or email.
This will get the drives working under RH, however they only run in slow PIO
mode. I think you have to upgrade to 2.2.17 or higher in order to get the
Ultra-DMA 4 support.
Hope this works,
Greg James
[EMAIL PROTECTED]
"Ken Siersma" <[EMAIL PROTECTED]> wrote in message
news:[EMAIL PROTECTED]...
> Thank you for your reply.
>
> I tried your suggestion, and I went into the RedHat 6.1 Installation,
> and installed every package. I formatted the /, /usr, and /tmp
> partitions (not /home, /usr2, etc). I tried to boot with my boot disk,
> and I got a kernel panic, can't mount root again. It never found my
> drive. Any suggestions?
> Thanks
> Ken
>
>
>
------------------------------
From: [EMAIL PROTECTED]
Subject: Re: Linux software RAID vs. hardware RAID
Date: Tue, 12 Sep 2000 19:28:51 GMT
In article <[EMAIL PROTECTED]>,
James Knowles <[EMAIL PROTECTED]> wrote:
> > That's a toy, not a server.
>
> If by definition a server must cost $10k, yes, it's a toy. But it does
> the job admirably.
>
> --
> How do I love thee? My accumulator overflows.
>
I just wanted to add my two cents to this thread even though its a
little old now...
I happen to have the Abit KT7 RAID Motherboard in question with two
Quantum LM drives in a striped config.
According to the manufacturer of the RAID controller on the board you
cannot run Linux using a RAID config but you can install linux on a
drive using the RAID controller... although you need to patch the
kernal.
I didn't realize this until after I had striped my drives and install
windoze but I plan on adding a third drive on my empty IDE channel and
putting linux there until I can put linux on my RAID config.
Sent via Deja.com http://www.deja.com/
Before you buy.
------------------------------
From: Brian Langenberger <[EMAIL PROTECTED]>
Subject: Re: Logitech Optical Mouse
Date: 12 Sep 2000 19:40:48 GMT
Trelin <[EMAIL PROTECTED]> wrote:
: Is the Logitech Optical Mouse supported in Linux?
Yes. They're just standard PS2/USB mice since all the
optical work is done in the mouse itself. Look here:
http://www-sop.inria.fr/koala/colas/mouse-wheel-scroll/
for info on getting the wheel working. I own one myself
and have had no trouble whatsoever.
------------------------------
From: Andrey Vlasov <[EMAIL PROTECTED]>
Subject: Re: Baffling IDE HD performance difference
Date: Tue, 12 Sep 2000 12:42:11 -0700
Hi John,
I am not expert in this area but have a big experience in computers.
So, my understanding is:
1. Most new disks have internal buffer from 0 KB to 8 MB
2. Most new disks support a command to read several sectors at ones
3. Some disks support SMART technology
4. RPM and etc.
So, when you test with hdparm -T you make a test how fast you can
write/read
to/from internal buffer of the disk. With hdparm -t you make physical
write/read to/from
disk. With option hdparm -m {number} you request to read n sectors at
ones. It can
improve performance if you try to get access to big file as a part of file
as big as 8 MB can
be in cache memory of the disk and you do not need make positioning of the
head -
which is very slow operation. You just get this data from buffer without
delay.
Now lets talk about RPM - it is how fast you disk spins. Suppose you read
fragmented file and it mean that disk head will move back and forth
looking
for data. Now a question "What disk will find it faster 5400 RPM or 7200
RPM?".
I guess that you already find answer on this question. So, I'd like to
remark that some
disks configured to read several sectors at once and it can be optimal -
if you will
try change this parameter you will get lower performance. But there is
another point
in disk performance as CACHE provided on OS level and OS sector size. In
case if you
have some extra money it will good to have more memory in computer and
some part
of memory can be used as cache. I remember from DOS times when cache in 1
MB
increase disk performance in 10-15 times. Just you write file into memory
cache and
it will write down onto disk later and for you it seems as write operation
already complete.
Sector size plays very important role as well. Just let say you have 10 KB
file and your
block size is 512 bytes - default DOS block size. To read it you need make
20 positioning
of the head to read in this block size. But if you will change block size
to 4 KB you need
make only 3 positioning. In case of blocksize 8 KB you need only 2
positioning. So, far
so good. But what will happen if you have 8 KB block size and on your disk
has million very
small files - between 1-2KB. I guess that you already guess that most of
this blocks will
empty or you just lost diskspace. This is why exists FAT/FAT16/FAT32 in
DOS/Windows.
In UNIX systems it is more flexible - you can change block size on running
system. For
that read "man tunefs" and "man mkfs".
I guess that your computers can look very close on hardware level but they
different.
Check with hdparm -v /dev/hd?? information about multiread and cache size.
Check information on manufacture web site for RPM and another parameters.
I'd like remark that when I switched from PII-350 to PIII-750 I got some
extra
performance - may be about 2MB/s more. So, speed of CPU also can play some
role but performance will not change 2 times for switching between 3500MHz
and 700MHz.
Andrey
John Harlow wrote:
> I recently switched notebooks from a Dell Lat CPx-450 to a Tecra 8100
> PIII-600. I was in a hurry when I switched, so I just swapped the HDs
> (Try doing that with NT!) They both have 12 gig drives with the same
> geometry (Dell is IBM, Tecra is Toshiba drive)
>
> In an effort to keep warranties straight, today I ghosted my install
> back to the Toshiba drive. This gave me an opportunity to run the same
> install in both. I stayed in a char session with sound off, so they
> were basically identical. I ran hdparm -t. Both were pretty close
> (13ish). Then I ran hdparm -T. The 450 mHz Dell score a 133 while the
> 600 mHz Toshiba score a 55.
>
> Both systems were running hdparm -c1d1u1k1. I started experimenting
> with settings and got the Toshiba up to 67 on hdparm -T, but no better.
>
> My understanding of hdparm -T is that I am testing os buffers, not disk
> i/o. OS is RH 6.2. Kernel is 2.2.17 w/Toshiba flag set on.
>
> What is going on here? Any ideas?
>
> john
>
> --
> John R. Harlow
> United Systems Inc.
> email: [EMAIL PROTECTED]
>
> Sent via Deja.com http://www.deja.com/
> Before you buy.
------------------------------
From: Stanislav Kogan <[EMAIL PROTECTED]>
Subject: Re: Im building a new computer to dual boot w2k and mandrake 7.1and
Date: Tue, 12 Sep 2000 15:47:12 -0400
Hi!
Stanislav Kogan wrote:
> > Modem
> Take a Hardware modem of course. I personally had a very good luck with
> AOpen ITU90. Also, make sure it's ISA and not PCI. (PCI is harder to
> configure under Linux). Also, Non-plug and play will help too.
I made a mistake here. The modem is Aopen FM56-ITU. A real workhorse.
(cheap too)
------------------------------
From: [EMAIL PROTECTED]
Subject: Re: SCSI Tape Drive Question
Date: Tue, 12 Sep 2000 19:37:12 GMT
New development:
I just tried loading the aic7xxx.o module manually by doing a 'insmod /
lib/modules/2.2.15mdk/scsi/aic7xxx.o' and got the following output:
[root@qed /root]# insmod /lib/modules/2.2.15-4mdk/scsi/aic7xxx.o
scsi0 : Adaptec AHA274x/284x/294x (EISA/VLB/PCI-Fast SCSI) 5.1.29/3.2.4
<Adaptec AHA-294X Ultra2 SCSI host adapter>
scsi : 1 host.
(scsi0:0:6:0) invalid cur_addr:0x99feb8e during WIDE_RESIDUE
(scsi0:0:6:0) sg_address[-1]:0x0 sg_length[-1]:0
(scsi0:0:6:0) sg_address:0x0 sg_length:0
(scsi0:0:6:0) cur_address:0x99feb8e resid_dcnt:0x0000d1
(scsi0:0:6:0) invalid cur_addr:0x99feb8e during WIDE_RESIDUE
(scsi0:0:6:0) sg_address[-1]:0x0 sg_length[-1]:0
(scsi0:0:6:0) sg_address:0x0 sg_length:0
(scsi0:0:6:0) cur_address:0x99feb8e resid_dcnt:0x0000d1
(scsi0:0:6:0) invalid cur_addr:0x9c77dc2 during WIDE_RESIDUE
(scsi0:0:6:0) sg_address[-1]:0x0 sg_length[-1]:0
(scsi0:0:6:0) sg_address:0x0 sg_length:0
(scsi0:0:6:0) cur_address:0x9c77dc2 resid_dcnt:0x0000d2
Vendor: ARCHIVE Model: Python 06408-XXX Rev: 8071
Type: Sequential-Access ANSI SCSI revision: 03
Detected scsi tape st0 at scsi0, channel 0, id 6, lun 0
The tape drive works now! However, what do all these errors mean?
Should I be concerned about all this "invalid cur_addr"? Also, is this a
kosher way of loading the driver? Is it ok just to put the insmod
command in one of the startup scripts?
-mike
Sent via Deja.com http://www.deja.com/
Before you buy.
------------------------------
From: "Simple Simon" <[EMAIL PROTECTED]>
Crossposted-To:
comp.os.linux.help,comp.os.linux.questions,comp.os.linux.redhat,linux.redhat.misc
Subject: Re: Memory questions, Device questions; Remove LILO on RH6.2
Date: Tue, 12 Sep 2000 20:06:14 GMT
Ack!
"Bartek Kostrzewa" <[EMAIL PROTECTED]> wrote in message
news:39be6f7e$[EMAIL PROTECTED]...
>
> > > add append = "mem=192MB" to your /etc/lilo.conf > run /sbin/lilo to
> enable
> > > changes
Which is it?
append="mem=192"
append="mem=192m"
append="mem=192mb"
...and yes, my CMOS had the 15-16m memory hole enabled - even after I
disabled it once. ALWAYS CHECK YOUR SETTINGS AFTER YOU MAKE A CHANGE!
SS
------------------------------
** 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
******************************
