Linux-Hardware Digest #315, Volume #13 Fri, 28 Jul 00 19:13:07 EDT
Contents:
Speedstar A55 Driver ("Gerald Jensen")
Re: scsi hdd optimization (Michael Meding)
Re: Ripping CD's with Linux??? (sandrews)
TV capture card without video card ? ([EMAIL PROTECTED])
Re: who is responsible for the cache? (Larry Ebbitt)
linuxcluster recommendation? ([EMAIL PROTECTED])
Linux and Adaptec 39160 (James Myers)
Re: Building a Linux Server from scratch: Experiences? (Bob Martin)
usb zip? (Wenming Ye)
Re: Building a Linux Server from scratch: Experiences? (Henrik Carlqvist)
Re: Entry-level server (Henrik Carlqvist)
Re: Internal modem stopped working (Henrik Carlqvist)
Re: dot clock frequency question (Steven Fosdick)
Re: harddisk recover (Steven Fosdick)
Re: Network/Ethernet, still need help, more information this time. (Steven Fosdick)
Re: Modem reset (Steven Fosdick)
Re: Logitech pilot combo: in search of 3 button mouse for serial port (Steven
Fosdick)
----------------------------------------------------------------------------
Reply-To: "Gerald Jensen" <[EMAIL PROTECTED]>
From: "Gerald Jensen" <[EMAIL PROTECTED]>
Subject: Speedstar A55 Driver
Date: Fri, 28 Jul 2000 15:12:06 -0500
Is there a Linux (RH 6.1) driver available for the Diamond Speedstar A55 AGP
card?
Any ideas on how I can configure my system to use this card?
------------------------------
From: Michael Meding <[EMAIL PROTECTED]>
Subject: Re: scsi hdd optimization
Date: Fri, 28 Jul 2000 21:22:43 +0200
Reply-To: [EMAIL PROTECTED]
Excuse me,
but to my understanding you most common tune the transfer speed between
the hostadaptor (in this case mostly the onboard ata interface) and the
hdd. This happens to be set by hdparm which sets the hostadaptor
settings. Nothing to do with the drives, they deliver according to their
capabilities and to the hostadaptors.
With SCSI hostadaptors you do set this a) in the linux kernel (i.e.
tagged command queing) b) in the hostadaptor within the
onhostadaptorbios.
Please correct me if I am wrong.
With best regards
Michael Meding
------------------------------
Date: Fri, 28 Jul 2000 16:39:09 -0400
From: sandrews <[EMAIL PROTECTED]>
Subject: Re: Ripping CD's with Linux???
dionysis wrote:
>
> I'm unable to iron out driver problems in Win 98 to be able to digitally rip
> audio CD's. Someone suggested I run Linux tandem with Win 98. I have a
> crappy Samsung SCR 3232, would be willing to install a CD-R or CD-RW to make
> this work.
>
> Are the Linux CD drivers better at digitally ripping audio CD's to MP3?
>
> I found Caldera Linux 2.2 with documention used for $20.
>
> I also found Linux 6.5 Mandrake used with documentation for $20. Should I
> go
> with that if I decide to try this?
>
> Which Linux is "better"?
Grip using Bladenc works well.
--
M$ Windows is aptly named, after all, it's easily broken, and
offers little
security, just like the glass ones...
------------------------------
From: [EMAIL PROTECTED]
Subject: TV capture card without video card ?
Date: Fri, 28 Jul 2000 20:59:18 GMT
Hi All,
I was rooting around for a webcam, but couldn't seem to find anything
available that would also work under linux. So I started thinking about
using a TV/video capture card and connecting it to a videocamera.
The computer I plan to run this in is quit old, and the connector pins
on the video card are broken. Question: can I install a capture card
(say, ATI TV-Wonder) in my machine *without* connecting it to a video
card ?
I expect I wouldn't be able to watch TV without connecting the two, but
I should still be able to capture images with it, right ?
thanks for any advice,
terry
Sent via Deja.com http://www.deja.com/
Before you buy.
------------------------------
Date: Fri, 28 Jul 2000 17:29:34 -0400
From: Larry Ebbitt <[EMAIL PROTECTED]>
Crossposted-To: comp.os.linux.misc
Subject: Re: who is responsible for the cache?
Robert Schweikert wrote:
> > AFAIK, it's the motherboard chipset. Sounds like you've got hardware
> > problems to me. Compiling something large (liek gcc or the kernel) puts a
> > lot of strain on memory, and I've had machines with flakey hardware that
> > seemed fine until I tried to compile something.
> I've tested a different memory stick and had the same problem.
The cache is definitely a hardware responsibility. It's unlikely that a
memory stick would have any effect. It's difficult to imagine what software
could do to mess it up. Cache memory access patterns can be vastly
different between two compiles of different source. About all software can
do is invalidate cache under certain circumstances like context switch.
--
Larry Ebbitt - Linux + OS/2 - Atlanta
------------------------------
From: [EMAIL PROTECTED]
Subject: linuxcluster recommendation?
Date: Fri, 28 Jul 2000 21:18:52 GMT
I'm fairly new to the Linux world, and have been
given the responsibility of purchasing a
Linuxcluster for running compute intensive
software, some of which already runs on
Pentiums. VALinux has been recommended to me,
but I wonder if I should be looking at other
vendors and other chips, esp. either the athlon
or the alpha. Anyone out there have experience
with this?
Thanks,
Sarah
Sent via Deja.com http://www.deja.com/
Before you buy.
------------------------------
From: James Myers <[EMAIL PROTECTED]>
Subject: Linux and Adaptec 39160
Date: Fri, 28 Jul 2000 14:03:36 -0700
==============232C09E5B35A0DF754A3A599
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit
Hello,
I am running Linux 2.2.14 with an Adaptec 39160 and cannot get the
channel B to work properly. I am
forced to put all my SCSI devices on channel A. The problem with this
is I have a tape drive with a much
slower effective throughput and consequently it is slowing down my
10K rpm disk drives.
Is there a patch available for this problem? Am I doing something
wrong?
Any help would be greatly appreciated.
James
--
James Myers
[EMAIL PROTECTED]
==============232C09E5B35A0DF754A3A599
Content-Type: text/html; charset=us-ascii
Content-Transfer-Encoding: 7bit
<!doctype html public "-//w3c//dtd html 4.0 transitional//en">
<html>
Hello,
<p>I am running Linux 2.2.14 with an Adaptec 39160 and cannot get the channel
B to work properly. I am
<br>forced to put all my SCSI devices on channel A. The problem with
this is I have a tape drive with a much
<br>slower effective throughput and consequently it is slowing down my
10K rpm disk drives.
<p>Is there a patch available for this problem? Am I doing something
wrong?
<p>Any help would be greatly appreciated.
<p>James
<pre>--
James Myers
[EMAIL PROTECTED]</pre>
<pre></pre>
</html>
==============232C09E5B35A0DF754A3A599==
------------------------------
From: Bob Martin <[EMAIL PROTECTED]>
Crossposted-To: comp.os.linux.misc
Subject: Re: Building a Linux Server from scratch: Experiences?
Date: Fri, 28 Jul 2000 21:32:43 +0000
David Steuber wrote:
>
[snip]
> Has anyone else had similar experiences to mine? Does life get easier
> in the build your own PC game? I feel a lot like I should have just
> gone with VA Linux and be done with it.
>
> Now I have an e-mail to send.
>
The last computer I bought "pre-built" was an XT back in '85, since then
I have built all of mine from parts. Back then you could save a great
deal of money because you paid a premium to have an IBM label on your
box. These days PCs have become a commodity item and you can end up
paying more by building it yourself. It stills gives you the advantage
of getting exactly what you want and if something goes wrong your more
likely to know how to fix it. But places like emachines are basically
making disposable computers, keep it 2 years and buy a new one, dump the
old one.
As far as vendors go , sometimes your parts can come at different times
because they can be shipped from different warehouses in different
locations. One thing you have to watch is the shipping, they can
blindside you with that. I have generally good luck with most place I
have ordered from , a few of the ones I like now are:
allstarshop.com , axiontech.com, and onsale.com. Shipment was on time,
no missing parts and shipping was reasonable.
Check sysopt.com for some technical articles and they also have link to
a reseller ratings page that rates vendors.
--
Bob Martin
------------------------------
From: [EMAIL PROTECTED] (Wenming Ye)
Subject: usb zip?
Date: 28 Jul 2000 22:04:45 GMT
is this supported? If so, where can I find more info...
thanks,
W
------------------------------
From: Henrik Carlqvist <[EMAIL PROTECTED]>
Crossposted-To: comp.os.linux.misc
Subject: Re: Building a Linux Server from scratch: Experiences?
Date: Thu, 27 Jul 2000 20:53:49 +0200
David Steuber wrote:
> It's mission: to serve HTTP connections with Apache and run a
> smallish database with PostgreSQL. The web pages will contain
> dynamic content using mod_perl and PostgreSQL.
You might want to have a look at PHP. It's great for dynamic pages with
a connection to a database. PHP could be compiled as a modle for apache
or as a standalone CGI executable.
> Come to think of it, I can get them for credit card fraud now that
> they charged me for something I did not receive.
Sometimes it is better to buy from a local seller, however it is
probably not the cheapest solution. It is mostly easier to complain at
the local seller and you carry the stuff you bought out from the store
yourself.
> On top of all that, I recently found this on Tom's Hardware:
>
> http://www.tomshardware.com/cpu/00q2/000619/index.html
>
> It makes me think I might have gone with a Duron instead, or even an
> Athelon.
This is always a problem when buing computers. Only a few days after you
bought your machine you will find another one which is better and
cheaper.
> Has anyone else had similar experiences to mine? Does life get
> easier in the build your own PC game? I feel a lot like I should
> have just gone with VA Linux and be done with it.
There is one more advantage with building your own PC: You really get to
know your PC and will know exactly what to replace and how to replace it
the day you want to upgrade your PC.
regards Henrik
--
spammer strikeback:
root@localhost
[EMAIL PROTECTED]
------------------------------
From: Henrik Carlqvist <[EMAIL PROTECTED]>
Subject: Re: Entry-level server
Date: Thu, 27 Jul 2000 23:39:00 +0200
newly minted linux mofo wrote:
> > > (I know the reasons to prefer SCSI, our pocketbook prefers IDE.)
An interesting solution might be a box containing 6 cheap IDE drives
which connects to a SCSI controller. The box will cost a little extra,
but then you will get hardware RAID performance for RAID0 or RAID5. At
work we use such a box with 6 25 GB IDE drives running in a RAID5
configuration. It also has two power supplies.
> We were looking at an SMP box for scalability, not for redundancy.
I haven't tried it myself, but you might want to have a look at the
Linux Virtual Server project. By combining several machines it will give
you both scalability and redundancy. I don't remember the URL, but start
looking at freshmeat with keywords like "Linux Virtual Server",
"UltraMonkey" or "piranha".
> BTW, I thought RAID-0 was striped, RAID-1 was mirrored, and RAID-5
> was something of a combination of the two (striped / mirrored, with
> parity)?
You are right about this. There is also RAID-3 which is striped and has
an extra disk for parity. RAID-5 is like RAID-3, but the parity isn't on
any special disk, it is scattered on all disks.
regards Henrik
--
spammer strikeback:
root@localhost
[EMAIL PROTECTED]
------------------------------
From: Henrik Carlqvist <[EMAIL PROTECTED]>
Subject: Re: Internal modem stopped working
Date: Fri, 28 Jul 2000 00:15:16 +0200
David Yan wrote:
> I have changed the port settings for the internal modem from ttyS0 to
> ttyS3 without any success. wvdialconf will always find the external
> modem in ttyS0 or ttyS1 but not the internal one.
Are you able to see the internal modem with something like:
dmesg | grep tty
?
regards Henrik
--
spammer strikeback:
root@localhost
[EMAIL PROTECTED]
------------------------------
From: Steven Fosdick <[EMAIL PROTECTED]>
Subject: Re: dot clock frequency question
Date: Fri, 28 Jul 2000 00:33:34 +0000
In article <[EMAIL PROTECTED]>,
"Gene Heskett" <[EMAIL PROTECTED]> wrote:
> Most cards can operate at dotclocks well above what the monitor can
> reproduce unless you've spent a premium amount on the monitor.
True. Also for many cards the X server is able to discover the max.
dot clock from the card itself rather than it needing to be specified in
the config file.
> The single, most important specification for 99% of the monitors is the
> horizontal sync frequency. It is difficult to hurt a monitor from
> going too high in most cases because the circuitry becomes both more
> efficient, and less, depending on what parameter you are looking at.
> They are size regulated by regulating the voltage applied to the h-scan
> transistor, with the voltage riseing with the applied frequency in order
> to maintain both the size and the high voltage developed. When the
> upper limit of the available voltage has been reached, and the rate
> further increased, the usual effect is a dimming of the image while it
> remains at approximately the same size, the lower high voltage being
> developed compensates for the loss of scan power.
Another effect of driving monitors beyond the speed they were really designed
for seems to be a loss of focus. Of course if the EHT supply voltage is reduced
then a loss of focus would be expected.
> Going down below the monitors rated lower frequency however can and
> often will destroy the scan parts in wholesale fashion. This occurs
> because the lengthened time for the currents to build up can allow the
> ferrite in the scan transformer core to go into magnetic saturation. At
> this point, the inductance rather effectively disappears and there is no
> other ready mechanism to control the riseing rush of current thru the
> scan transistor and and other connected parts. This current can rise to
> the many tens of amperes region in a fraction of a microsecond. It does
> little good to shut it all off 4 microseconds later, and the heat in the
> transistor as it comes out of saturation, trying to stop 20 amps when it
> was expecting maybe half an amp will cause it to fail shorted in just a
> few milliseconds. The short then destroys as much of the rest of the
> circuit as it can long before any fuses or circuit breakers can act to
> protect things.
It's worth noting here that the transformer in the h-scan circuit is a
particularly expensive component, second only in cost to the tube
itself. On a cheap monitor burning out a winding on this transformer
normally puts the monitor beyond econnomic repair.
> So basicly, a monitor rated for 31 to 72 kilohertz horizontal can
> probably be run at 80 kilohertz without any long term damage. But it
> may well fail instantly if fed a 27 kilohertz h-sync and it manages to
> lock to it. Some monitors are self protecting in that event, either
> doing a total shutdown, or defaulting to some safe rate showing you
> either garbage or nothing during such times. This NEC 5FG is one such,
> it gets fed some 16khz while this machine is booting and all it shows is
> a dim, blank raster at such times.
Many newer, higher spec monitors with on screen display (OSD) detect an
out of range input and put up an OSD message saying scan freq. too high
(or too low) but it's best not to put this to the test.
Interestingly having told X about your monitor it may default to a very high
resolution at a relatively low vertical refresh rate which may cause the display
to flicker. If this happens a useful trick is to lie to X and tell it that the
monitor's
slowest Vertical refresh rate is the minimum rate you are prepared to accept.
It is quite common for a monitor to be capable of 50Hz V Refresh but flicker
intolerable anything below about 75 to 80 Hz (depending on the persistance
of the phosphor and the person viewing).
------------------------------
From: Steven Fosdick <[EMAIL PROTECTED]>
Subject: Re: harddisk recover
Date: Fri, 28 Jul 2000 00:45:10 +0000
In article <8lhi70$61d$[EMAIL PROTECTED]>, "Diana Block" <[EMAIL PROTECTED]> wrote:
> after a system crash i can�t boot my server from it�s ibm scsi disk. i tried
> to repartition the disk in another linux system with fdisk but there is the
> following error message:
> SCSI disk error : host 0 channel 0 id 0 lun 0 return code = 8000002
> [valid=0] Info fld=0x0, Current sd08:00: sense key Not Ready
> scsidisk I/0 error : dev 08:00, sector 0
> anybody knows if there is a way to recover the data on the damaged disk,
> cause there is an very important database on it with no backup...
Make sure the drive is being supplied with power on both the 5V and
12V rails and that the start-spindle-on-power-up jumper hasn't fallen
off.
If neither of the above turns out to be the fault then unfortunately it sounds
like a hardware fault inside the HDA. Maybe the spindle motor has jammed,
maybe the head actuator is stuck, or maybe on of the heads has been
damaged so it can't read any data from the disk.
The only way you are likely to be able to recover the data is to employ a
firm of data recovery experts to do it for you. It requires specialised
equipment and clean room conditions to be sucessful. You'll want to tell
them what partitions the disk should have on it and what filesystem type
is to be found on each (prob. ext2fs for most Linuxes).
I only know of one of these data recovery companies which is S&S in
London but there will be others probably closer to home.
------------------------------
From: Steven Fosdick <[EMAIL PROTECTED]>
Crossposted-To: alt.os.linux,alt.os.linux.mandrake,comp.os.linux.misc
Subject: Re: Network/Ethernet, still need help, more information this time.
Date: Fri, 28 Jul 2000 00:56:57 +0000
In article <964282067.143623@sj-nntpcache-5>,
"Brian Davis" <[EMAIL PROTECTED]> wrote:
> One thing that I did notice was that the TX packets (as reported by
> ifconfig) are increasing when I do pings, but the eth1 value in
> /proc/interrupts never increases higher than 1. The eth0 vaule increases
> with every packet I send/receive on that interface. I took eth1 out of the
> system and checked /proc/interrupts to make sure there was no other device
> was trying to use IRQ 9, and that checked out OK.
By far the most common reason for not receiving packets from a NIC is having
the driver use one interrupt line whilst the card is using a different one. In Linux
the /proc/interrupts and ifconfig will report which interuupt the driver has registered
with the kernel but that doesn't mean the card is configured to generate that
interrupt.
It is also not unkown for cards to forget their settings so the next thing to do would
probably to check the card settings with the 3com setup utility (you would probbably
have to boot DOS from a floppy to do this unless you have DOS or windows on
another partition.
------------------------------
From: Steven Fosdick <[EMAIL PROTECTED]>
Subject: Re: Modem reset
Date: Fri, 28 Jul 2000 01:04:37 +0000
In article <[EMAIL PROTECTED]>,
Andrew Higgs <[EMAIL PROTECTED]> wrote:
> What can be done with a modem once DTR has been dropped during transfer.
> All that seems to help is a physical reset of the modem. I have tried to
> use chat to reset the modem ( send it the init string again ). This does
> not seem to help. Have also removed the lock file (/var/lock/LCK..ttyS1
> on my slakware box).
DTR is a signal from the comnputer to the modem to tell the modem that
the computer is ready to communicate. If the computer drops DTR when
there is a call in progress the modem will hang up.
The most interesting signal going in the other direction is the CD (or DCD)
line. The modem raises this when the modem has trained up with the modem
at the other end and this tells the computer that it can start to send real data
(previously anything send would be interpeted as a command to the modem).
If the modem looses the link to the other end it drops CD to tell the comnputer
that the link has gone down. If you were running ppp over the link then pppd
will repond to losing CD by running the ip-down script and then exiting.
If there is a lock file being left behind then maybe your pppd is crashing.
Looking at the syslog to see what was happenning at the time the link goes
down may be useful. If it isn't obvious to you what the log means then post a
few lines for the relavent time here and see if we can see what is happenning.
------------------------------
From: Steven Fosdick <[EMAIL PROTECTED]>
Subject: Re: Logitech pilot combo: in search of 3 button mouse for serial port
Date: Fri, 28 Jul 2000 01:28:49 +0000
In article <[EMAIL PROTECTED]>, Guy Maskall <[EMAIL PROTECTED]> wrote:
> I'm after a mouse for linux (redhat 6.1 with kernel 2.2.16) with 3
> buttons for connection to the serial port (no ps/2 socket, y'see).
>
> I'm considering a logitech pilot mouse combo. Now ... I understand that
> this is available as a serial device, can anyone offer confirmation that
> this is a 'proper' serial mouse, rather than a ps/2 mouse with a
> ps/2-serial adaptor?
I have a 3 button Logitech Pilot Mouse (model M-C43) on my spare PC.
It is supplied with a PS/2 connector and one of those adapter things and
works perfectly under Linux in both serial and PS/2 mode.
It is worth noting that the adaptor doesn't really do any conversion - the
PS/2 anbd serial interfaces are completely different as PS/2 is syncronous
and the standard serial I/F is async. What happens is that some pins on the
PS/2 connector that the PS/2 interface doesn't use are connected to the part
of the mouse that outputs data in normal async format and the "adaptor"
connects these to the serial port lines.
So a dual standard mouse (one that will do serial with one of the converters)
is a specifc thing - the converter doesn't make just any PS/2 mouse work on
a serial port.
------------------------------
** 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
******************************
