Linux-Hardware Digest #627, Volume #9 Thu, 11 Mar 99 00:13:35 EST
Contents:
Re: PCI Modems & Linux (Rob Clark)
Re: Strange fdisk problem (Brian)
Re: Linux on a Celeron? (BL)
Re: Can Linux use 36-bit Xeon addressing? (brian moore)
Re: why AMD386 = GenuineIntel? (Chris Dukes)
Re: Creative Labs Awe32 (egray7)
Re: Xeon Processor... (Eric Lee Green)
Re: Can Linux use 36-bit Xeon addressing? (Johan Kullstam)
Re: fdisk partitioning > 8GB drive (Stephen J Bovy)
Re: fdisk partitioning > 8GB drive (Stephen J Bovy)
Re: Linux and a 8088 (Lew Pitcher)
Re: bad superblock? (Lew Pitcher)
RH 5.2 and Hayes V.90 int modem ("Charley")
Re: ESS 1688 with 16-bit DMA (Wilmarcd)
----------------------------------------------------------------------------
Subject: Re: PCI Modems & Linux
From: [EMAIL PROTECTED] (Rob Clark)
Date: Thu, 11 Mar 1999 04:03:54 GMT
In article <36e72da3$2$ewyncunz$[EMAIL PROTECTED]>,
Jerry Lapham <[EMAIL PROTECTED]> wrote:
>In <7c6hdo$ont$[EMAIL PROTECTED]>, on 03/10/99
> at 07:33 PM, Richard Nelson <[EMAIL PROTECTED]> said:
>
>> My name is Richard Nelson, and I am a Technical support engineer for
>> Actiontec Electronics, INC
>
>> We do have a PCI modem that is NOT a "Win" modem, it is controller
>> based, and uses the Lucent Venus chipset.
>
>What's the model number? Do you also make controllerless PCI modems or
>can we assume that any Actiontec PCI modem is OK?
The model number is PCI56012 or PM560LKI. It's not on their web site yet.
IBM resells this modem as 33L4618. The original poster tested the modem for
OS/2 in his capacity at ActionTec.
This is ActionTec's only PCI hardware modem, AFAIK.
Rob Clark, [EMAIL PROTECTED]
http://www.o2.net/~gromitkc/winmodem.html
>
>Also, you might want to post this info in the OS/2 newsgroups, too. They
>have the same problem as Linux users.
>
> -Jerry
------------------------------
From: Brian <[EMAIL PROTECTED]>
Crossposted-To: comp.os.linux.setup
Subject: Re: Strange fdisk problem
Date: Wed, 10 Mar 1999 21:48:23 -0600
> Thirdly, there are cases where there is confusion about the units.
>
>
> You can investigate things for yourself.
> # tune2fs -l /dev/hda9
> ...
> Block count: 4095976
> ...
> (or so) will tell you the size of the filesystem, to be contrasted
> with the df output
> # df /dev/hda9
> Filesystem 1024-blocks Used Available Capacity Mounted on
> /dev/hda9 3958475 38849 3714828 1% /g3
> and the fdisk output
> # fdisk -s /dev/hda9
> 4095976
>
> If tune2fs shows half the fdisk output, you are wasting disk space.
tune2fs agrees with df and shows half of fdisk. What do you think is
going on why that would happen? How do you think it should be fixed?
> If tune2fs agrees with fdisk but df shows only half then either
> you use a ridiculous number of inodes (check "df -i") or, more likely,
> there is a confusion of units.
>
=====================================================
Brian Feeny (BF304) [EMAIL PROTECTED]
318-222-2638 x 109 http://www.shreve.net/~signal
Network Administrator ShreveNet Inc. (ASN 11881)
------------------------------
From: BL <[EMAIL PROTECTED]>
Subject: Re: Linux on a Celeron?
Date: Thu, 11 Mar 1999 02:42:30 GMT
what is a 'celeron motherboard'? I am using a proper pentium-2 <sic> mobo
which is needed for dual operation. it SURELY supports ecc ram.
and if/when you want to upgrade to the next (overly expensive) slot-1 cpu(s)
you can.
I guess I don't see what you're talking about...
Andrew Comech <[EMAIL PROTECTED]> wrote:
: Hi, I noticed that if you build a Celeron machine, then it is
: too easy to end up with a system which does not support parity
: checking. Something like newer cheaper motherboards for
: celeron would not have ECC.
: Besides, people say newer intels could not be overclocked
: while my K6-2 300 and K6-2 333 both run stable at 350MHz without
: voltage increase.
: Also, later on you could swap K6-2 for K6-3, while I am not sure
: about the future of celeron motherboards. It looks that intel
: would be glad to force to upgrade them.
: Best,
: Andrew
: PS. Just looked up -- both K6-2 333 and celeron 333 cost around $65.
------------------------------
From: [EMAIL PROTECTED] (brian moore)
Crossposted-To: comp.os.linux,comp.os.linux.misc
Subject: Re: Can Linux use 36-bit Xeon addressing?
Date: 10 Mar 1999 17:21:43 GMT
On Wed, 10 Mar 1999 17:00:03 GMT,
Stuart R. Fuller <[EMAIL PROTECTED]> wrote:
> Tomasz Korycki ([EMAIL PROTECTED]) wrote:
> : "Stuart R. Fuller" wrote:
> : >
> : > brian moore ([EMAIL PROTECTED]) wrote:
> : > : >
> : > : > Yes. And Your point, as related to "the last holdout from basing their
> : > : > systems on Unix concepts is Microsoft" bit? Mind You, if You look deep
> : > : > enough into NT architecture, You'll see.... VMS!
> : > :
> : > : VMS is based on Unix?
> : > :
> : > : Very interesting news indeed.
> : >
> : > It might be interesting, but it is certainly wrong news. If you read the
> : > paragraph above, it implies that the NT architecture is based on VMS.
> : >
> : > Stu
> :
> : It is. Just look who the main architect was and who he brought with him
> : to create NT.
>
> I'm not disputing the fact that NT looks a lot like VMS. I was disputing the
> "VMS is based on Unix?" comment.
It's called sarcasm.
--
Brian Moore | "The Zen nature of a spammer resembles
Sysadmin, C/Perl Hacker | a cockroach, except that the cockroach
Usenet Vandal | is higher up on the evolutionary chain."
Netscum, Bane of Elves. Peter Olson, Delphi Postmaster
------------------------------
From: [EMAIL PROTECTED] (Chris Dukes)
Subject: Re: why AMD386 = GenuineIntel?
Date: 10 Mar 1999 17:46:51 GMT
On 10 Mar 1999 14:17:05 GMT in <7c5ut1$mi4$[EMAIL PROTECTED]> Georg Schwarz
<[EMAIL PROTECTED]> wrote:
>On an AMD 386 DX 33 Linux 2.0.36's /proc/cpuinfo reports:
>
>cpu : 386
>model : 386 SX/DX
>vendor_id : GenuineIntel
>
>Bug or feature?
Neither. As memory serves that was perfectly okay with the licensing
agreement AMD had with Intel.
--
Chris Dukes <[EMAIL PROTECTED]>
"Hacking's just another word for nothing left to kludge."
------------------------------
From: egray7 <[EMAIL PROTECTED]>
Crossposted-To: comp.os.linux.misc,comp.os.linux.setup,comp.os.linux.questions
Subject: Re: Creative Labs Awe32
Date: Thu, 11 Mar 1999 03:55:47 GMT
[EMAIL PROTECTED] wrote:
> You can put this command in your boot.local file to
> automatically turn sound on at boot time.
Don't have a boot.local, is it safe for me to put it in
/etc/rc.d/rc.local?
------------------------------
From: [EMAIL PROTECTED] (Eric Lee Green)
Subject: Re: Xeon Processor...
Date: 10 Mar 1999 04:50:39 GMT
Reply-To: [EMAIL PROTECTED]
On Thu, 04 Mar 1999 21:26:18 GMT, John Burton <[EMAIL PROTECTED]> wrote:
>I'm planning on buying a low-end server machine to primarily do heavy
>duty number crunching (*BIG* atmospheric models) and generate large data
>files. I've been doing some research and it seems that the Pentium II
>Xeon processor running on an Intel SC450NX or AD450NX motherboard should
>be a screamer. But unfortunately my checkbook can't handle a full blown
>8 processor AD450NX system... My base requirements are:
>
> 450 Mhz Pentium II ??? processor (Initially single, but planning on SMP
>later)
> >256 MB memory
> 2x9 Gb SCSI LVD U2W disks (good swap & file I/O performance)
> 100 Base-T NIC
>
> I've found several options and am evaluating them, but I need some
>information...okay, relative to a Linux RedHat 5.2 base O/S with
>appropriate upgrades for the 2.2.2 (or greater) kernel:
>
> 1) is the ADAC A-466 Ultra2 SCSI Raid card supported under Linux?
No.
> 2) is the Adaptec 7980 (and 7985) SCSI chipset *well* supported under
>Linux?
Not well.
> 2) Can Linux make use of the 4-way interleaved memory access provided
>by the 450NX chipset?
Yes. The BIOS sets it up. The OS has nothing to do with it.
> 3) Under Linux is there much performance difference between a Dual P-II
>on a 440BX motherboard and a Dual P-II Xeon on a 440GX motherboard
>(there is a *significant* price difference).
I have not noticed much difference, unlike with the 450NX where there
is a very big and noticable difference (Red Hat Linux did a full
install on a 450NX system in 4 minutes!). BTW, unlike some vendors who
will use the on-board RAID on the SC450NX, we use a real RAID
controller (ICP-Vortex, see http://www.icp-vortex.com). Intel just
released a new PII/PIII server board, the L440GX+, with dual PCI bus
and lots of performance goodies which to me makes a dual Xeon on the
440GX chipset pretty much obsolete (a Xeon on the NX will still go
faster because of memory bandwidth). A Pentium III 500 sells for $250
less than a Xeon 450 and is just as fast on scientific codes (just
can't put four of them together, it's limited to 2-way SMP).
For scientific codes the biggest issue is memory bandwidth, not cache.
Cache helps, but not much, because you're looping over a data set
bigger than the cache (i.e., the cache is caching your program but not
your data). That is why classic Cray supercomputers did not have a
cache, but, rather, had a super-fast (and wide) memory architecture.
That is one reason why the Alpha is better for scientific codes, because
it has double the bandwidth of the typical Pentium board. But the
SC450NX solves that with the 4-way interleave.
The current downfall of the Alpha is that I don't know of any good
rock-solid RAID solutions for Alpha Linux. They're super-fast
scientific workstations, but not currently suited for mid-to-high end
file serving.
Some more tidbits:
For RAID controllers I recommend the ICP-Vortex line. Rock-solid,
plenty fast, goes all the way from single-channel widgets to
multi-channel Fibre Channel cards that'll handle up to 255 drives.
They write their own drivers for Linux, and have supported Linux since
the 1.3 kernel. They have even released their RAID configuration
programs as Linux binaries, both for glibc and libc5, unlike other
vendors where you must boot to DOS to configure the RAID array.
--
Eric Lee Green [EMAIL PROTECTED] http://www.linux-hw.com/~eric
"Linux represents a best-of-breed UNIX, that is trusted in mission
critical applications..." -- internal Microsoft memo
------------------------------
Crossposted-To: comp.os.linux.misc
Subject: Re: Can Linux use 36-bit Xeon addressing?
From: Johan Kullstam <[EMAIL PROTECTED]>
Date: 10 Mar 1999 21:54:11 -0500
[EMAIL PROTECTED] (Philip Brown) writes:
> On 10 Mar 1999 16:38:51 -0500, [EMAIL PROTECTED] wrote:
> >...
> >MS doesn't want to do business on unix. it could interfere with its
> >windows sales. if i could run ms-office on solaris, then i could make
> >a solid case to my boss to get rid of my pc running microsoft windows
> >and get me a sparcstation on my desktop.
>
> but why can't you run staroffice, or corel WP suite, or applixware, and
> still make that solid case to get rid of the PC right now?
> It's compatible with office?
it's not all *that* compatible. my documents tend to be chock full of
equations and figures. straight text is done ok. tables survive
usually. figures maybe. equations, not a chance. i'd love to use
AMS-LaTeX but the US DoD requires all documentation be submitted in
`MS-Word 6.0 for Windows' format.
--
J o h a n K u l l s t a m
[[EMAIL PROTECTED]]
Don't Fear the Penguin!
------------------------------
From: Stephen J Bovy <[EMAIL PROTECTED]>
Subject: Re: fdisk partitioning > 8GB drive
Date: 11 Mar 1999 02:31:39 GMT
Andries Brouwer wrote:
>
> Bryan McKinley <[EMAIL PROTECTED]> writes:
>
> : win98 is loaded and supports my very large disk (~11GB); my bios has
LBA
> : support. I have been reading the docs & howtos and they indicate that
I
> : should pass 'linear' to the kernel on boot to make it use logical
sector-like
> : addressing rather than CHS. I don't see this listed among options
> : available to pass to the kernel at boot time. Am I barking up the
wrong
> : tree? The docs indicate that fdisk can handle disks this large so long
> : as the kernel supports the linear disk addressing mode.
> : How do I set options at LILO-time to make this happen?
> : Is this already happening unbeknownst to me? Kernel logs CHS-style
disk
> : info on boot, not linear info--at a cursory glance.
>
> You do not mention any problem.
> There is no reason to expect any problems - a recent kernel will handle
> your disk just fine - there is no need to start doing obscure things.
>
> (i) The kernel always gives a C/H/S at boot time (for IDE disks) - it
just
> tells you what geometry it will give fdisk and LILO if hey ask for one.
> It does not mean that the kernel itself uses CHS addressing.
> (It will use LBA whenever that is supported by your disk, i.e., always,
> on recent hardware.)
>
> (ii) This `linear' thing is not something for the kernel, but for LILO.
> Edit /etc/lilo.conf adding a line
> linear
> (NOT append="linear" like some people are saying; append is for
> giving options to the kernel, this is an option for LILO itself.
> Read the LILO documentation, or lilo.conf(5)).
> Usually everything works fine both with and without this LILO option.
>
> (iii) The more you worry about these things the less it will work.
> For the great majority of people, doing absolutely nothing is the
> right way to get things working.
> Sometimes people have problems mounting a disk and start fiddling with
> geometries, but a mount problem is never a geometry problem.
> Look at the kernel boot messages. First there is a section about
> what disks are recognized, like
>
> hda: Maxtor 91728D8, 16479MB w/512kB Cache, CHS=1020/200/62, DMA
> hdb: QUANTUM Bigfoot TX12.0AT, 11497MB w/69kB Cache, CHS=23361/16/63, DMA
> ...
> SCSI device sda: hdwr sector= 512 bytes. Sectors= 12657717 [6180 MB]
[6.2 GB]
> SCSI device sdb: hdwr sector= 512 bytes. Sectors= 17755792 [8669 MB]
[8.7 GB]
> ...
>
> and then a section about what partitions are seen on the disk, like
>
> Partition check:
> sda: sda1 sda2 sda3 < sda5 sda6 sda7 sda8 >
> sdb: sdb1 sdb2 sdb3 sdb4 < sdb5 sdb6 >
> sdc: sdc1 sdc2 < sdc5 sdc6 sdc7 > sdc3
> sdd: sdd1 < sdd5 sdd6 > sdd2 sdd3 sdd4
> hda: hda1
> hdb: hdb1 hdb2 < hdb5 hdb6 hdb7 > hdb3 hdb4
> hdc: hdc1 hdc2 hdc3 hdc4
>
> Once the kernel mentions a disk, you know that it has seen this disk.
> Once the kernel mentions the partitions, you know that it can read the
disk.
>
> Geometry-related problems only occur in relation to other systems
> like DOS and Windows (and perhaps the BIOS), and only for LILO and
> fdisk, never for the Linux kernel or mount or so.
>
>
> [And for the gurus a question: What is peculiar in the above kernel
messages?
> What causes it? Could the kernel have avoided it?]
I think you have missed the point !!!
We are dealing with fdisk within the context of the "installation process"
for a "distribution" Because the kernel is not getting the disk geometry
correct, neither fdisk, nor any other installation related partitioning
software can correctly install the software on the drive..
================== Posted via SearchLinux ==================
http://www.searchlinux.com
------------------------------
From: Stephen J Bovy <[EMAIL PROTECTED]>
Subject: Re: fdisk partitioning > 8GB drive
Date: 11 Mar 1999 02:31:44 GMT
Andries Brouwer wrote:
>
> Bryan McKinley <[EMAIL PROTECTED]> writes:
>
> : Ok, under RedHat lilo I tried:
>
> : boot: hda=10022,8,[200,256,330] <--From drive specs 200-330 s/t
>
> : The square brackets indicate 3 values I have tried with fdisk without
> : committing
> : the partition changes. Question is, since the sectors/track varies
from
> : 200-300,
> : what should I use? 200 produced no warnings but 256 and 330 produced
> : cylinder ending warnings. (Your e-mail is bounced by the way).
>
> Read the Large Disk HOWTO. A quote:
> "Only fools talk about the `real' geometry of the drive".
>
> You see - nobody is interested in the `real' geometry.
> It does not exist. It will not be used, and it cannot be used.
> Linux wants the geometry only for interaction with DOS and BIOS.
> But DOS and BIOS use a geometry concept where you have 6 bits
> for the number of sectors per track. So each of the values
> 200, 256, 330 is nonsensical, 63 is the maximum.
> If you say nothing at all, and avoid giving "hda=..." options,
> chances are good that all will be fine, especially if you use
> a recent kernel, like 2.0.34 or 2.0.36.
>
> Read the Large Disk HOWTO.
>
No this is not true , because if the kernel passes the wrong geometry
values to fdisk, it will not read the part table correctly, and if
you actually go ahead, and modify the part table with fdisk, and the
wrong geometry, it will screw up the part table for all other os's that
may have used that geometry when they originally created the partition
table.
================== Posted via SearchLinux ==================
http://www.searchlinux.com
------------------------------
From: [EMAIL PROTECTED] (Lew Pitcher)
Subject: Re: Linux and a 8088
Date: Thu, 11 Mar 1999 04:24:27 GMT
Reply-To: [EMAIL PROTECTED]
On Tue, 09 Mar 1999 20:07:41 -0800, Paul Hovnanian <[EMAIL PROTECTED]>
wrote:
>> Blake Thompson wrote:
>> >
>> > Hey guys, I thought I'd run this one through the group and see if anyone
>> > had any ideas. I've got this old Compaq portable which I believe to be an
>> > 8088 (but it could be a 286). Anyway, my question is what micro
>> > distribution to put on it. I've checked out LOAF, Tom's, muLinux,
>> > DOSLinux, etc, etc. They all seem to require a 386 w/ 4MB RAM.
>> > I've got a 20 MB Hard Drive to work with, so it doesn't _necessarily_ have
>> > to fit on a floppy, but the micro-distributions seem to have the least
>> > stringent system requirements.
>
>Well, if you have too much time on your hands, anything is possible:
>
IIRC, Linux can't be run natively on anything less than an 80386.
If you really want a unix clone on an 8088 or 80286, and you can't hack
Linux down to that level, try Minix. Ask on news:comp.os.minix about
support.
Lew Pitcher
Joat-in-training
------------------------------
From: [EMAIL PROTECTED] (Lew Pitcher)
Subject: Re: bad superblock?
Date: Thu, 11 Mar 1999 04:33:56 GMT
Reply-To: [EMAIL PROTECTED]
On Wed, 10 Mar 1999 16:27:47 -0500, Rick Luna <[EMAIL PROTECTED]> wrote:
>Hi,
>
>I am trying to mount a partition on my disk which has win98 on it. Linux
>(RH 5.0) is on the same disk in a seperate partition.
>
>This is the error I receive when I try to mount it using fstool:
>
>mount /dev/hda1
try mount -t vfat /dev/hda /mnt
>returned the following error
>mount: wrong fs type, bad option, bad superblock on /dev/hda1, or too
>many file systems
>
>The file system must be vfat, since win98 is there. fstool shows the
>partition type as being "unknown".
Is it, by chance, a FAT32 partition? AFAICR, Linux doesn't yet have FAT32
support, and cant read such filesystems.
>When I run cabaret, sometimes i can see all the mountable drives,
>sometimes not, sometimes i see multiple instances of /dev/hda1.
>
>Any ideas on how to proceed? Can't get WINE to run if I can't get to
>windows...
>
>What is a "bad superblock?. Win98 has been crashing a lot lately, so I'm
>seriously considering reformatting that partition and starting over.
>
>Thanks,
>Rick
>--
>________________________________________________________________
>Rick Luna Phillips Design Group [EMAIL PROTECTED]
>Graphic Designer 930 N Meridian Street www.pdgroup.com/rick/
> Indianapolis, In 46204
> http://www.pdgroup.com
> Ofc:317.955.8435
> FAX:317.955.8551
Lew Pitcher
Joat-in-training
------------------------------
From: "Charley" <[EMAIL PROTECTED]>
Subject: RH 5.2 and Hayes V.90 int modem
Date: Thu, 11 Mar 1999 00:05:40 -0500
Hey!
I can't seem to get the modem to communicate. I was able to get
"pnpdump --c > isapnp.conf" to set the correct irq and address for COM2, but
Minicom nor the ppp module can get a response out of the modem. Minicom is
able to initialize the modem, (or so it reports) but it will not dial nor
can I force it to respond to AT commands. PPP doesn't do a thing either.
/dev/modem is directed to /dev/cua1, which I think is correct. The setup
string being sent to the modem is ATZ. This is very frustrating - my
coworkers are snickering behind my back! Anybody have any suggestions?
Thanx
Maus
------------------------------
From: [EMAIL PROTECTED] (Wilmarcd)
Subject: Re: ESS 1688 with 16-bit DMA
Date: 11 Mar 1999 01:57:11 GMT
In article <7c2f6j$la4$[EMAIL PROTECTED]>, [EMAIL PROTECTED] writes:
>I am unable to get linux to recognize the 16-bit DMA on my
>ESS 1688 sound card. Kernel 2.1.97 uses the 8-bit DMA but
>2.2.2 won't play any audio at all. I am using all default
>settings, and have verified that the settings are correct.
>
Need to know exactly what sound card you have. Usually some of the ESS 1688
and 18xx chipset cards are just Soundblaster Pro compatibile, no 16-bit DMA.
'course that info is a bit old.
Marc
---
Marc D. Williams
[EMAIL PROTECTED] / [EMAIL PROTECTED]
http://www.geocities.com/SiliconValley/Platform/8269/ -- Windows 3.x Makeover
http://www.agate.net/~tvdog/internet.html -- DOS Internet
------------------------------
** 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
******************************
