Linux-Hardware Digest #510, Volume #9 Fri, 26 Feb 99 23:13:32 EST
Contents:
Re: Can Linux use 36-bit Xeon addressing? (Robert Krawitz)
Re: Overclocking (was: Re: K6-2 and Linux, Are there any Bug?) (BL)
Anyone using a PCMCIA 3Com 3CCFEM656BT? (Taylor Hutt)
Re: Midi with SB AWE64 ISA ? (Bas van Nunen)
Re: USRobotics Modem ("Bob Fahey [[EMAIL PROTECTED]]")
Re: Overclocking (was: Re: K6-2 and Linux, Are there any Bug?) (Julian Robert Yon)
Re: 3com 3c905b card loses Mac address ([EMAIL PROTECTED])
Any chance for Linux to use an MDP7800-U PCI Modem (John L. Spetz)
Re: RedHat 5.2 Workstation Install: SIS6326 (Jeff Javits)
Re: EPSON Stylus Color 740 (1440*720 now works!) (Grant Taylor)
Re: Raw writing to PCMCIA SRAM cards ("Scot E. Wilcoxon")
Raw writing to PCMCIA SRAM cards ("Mark Smith")
Serial port problem/P2/Linux ("S. John Ilagan")
Re: ISA Card (3C509 TPO) (Greg H.)
6x86 trashes X screen (Silvio Romagnoli)
Re: SCSI Zip Zoom card (Carlos Eduardo Rodrigues - LEI)
Re: Presario 1210 and Linux (Sparkzz)
Re: fsck problem on boot-up (Burkard B. Kreidler)
Re: Can Linux use 36-bit Xeon addressing? ("David A. Frantz")
----------------------------------------------------------------------------
From: Robert Krawitz <[EMAIL PROTECTED]>
Crossposted-To: comp.os.linux,comp.os.linux.misc
Subject: Re: Can Linux use 36-bit Xeon addressing?
Date: 25 Feb 1999 10:11:05 -0500
"David A. Frantz" <[EMAIL PROTECTED]> writes:
> Try this site http://humbolt.geo.uu.nl/Linux-MM/more_than_1GB.html to gets a
> little info on the current I386 capability. Nothing specific on XEON
> there, well at least I didn't find anything. Sounds like your trying to
> apply a low end (Yes I mean the XEON) PC chip to a project that requires a
> 64 bit CPU. You may want to consider an Alpha, or a POWERPC box from IBM.
I think this is a tad unfair. I'm disappointed that Linus doesn't
want to enable large memory addressing on the x86.
There's a lot more software available for the x86, even on Linux, than
on other platforms. Also, x86 boxes are much cheaper than Alphas and
PowerPC's. The cost of the extra memory somewhat softens the blow,
but 4 GB of RAM is about $6000 these days, so the economics are very
different from what they were a few years ago.
Job mixes that are more memory/IO than computation intensive (which is
the case for a lot of commercial data processing) would benefit
greatly from the availability of large memory on commodity hardware.
[Disclaimer: that I'm not a disinterested observer: I work for Torrent
Systems: http://www.torrent.com/. However, this posting is completely
my own opinion, and does not reflect any official company policy.]
> dave
> [EMAIL PROTECTED] wrote in message <7b0un2$i3e$[EMAIL PROTECTED]>...
> >Intel Xeon processor + NX chipset can support up to 8GB DRAM. Is there any
> >Linux support for this? If not, does anyone know if it's in the works?
> >
> >I'm looking for an OS platform which will handle these large memories.
> >NT addresses the >4GB range as a sort of "cache buffer" accessible only
> from
> >user more. Normal NT kernel code will be able to access the lower 4GB only.
> >This solution is a poor one for my application - I would like to be able to
> >access the entire address space from kernel mode as well, e.g. DMA, etc.
> >Will Linux do something better than this?
> >
> >Thanks,
> >-Mark
> >
> >-----------== Posted via Deja News, The Discussion Network ==----------
> >http://www.dejanews.com/ Search, Read, Discuss, or Start Your Own
--
Robert Krawitz <[EMAIL PROTECTED]> http://www.tiac.net/users/rlk/
Tall Clubs International -- http://www.tall.org/ or 1-888-IM-TALL-2
Member of the League for Programming Freedom -- mail [EMAIL PROTECTED]
"Linux doesn't dictate how I work, I dictate how Linux works."
--Eric Crampton
------------------------------
From: BL <[EMAIL PROTECTED]>
Crossposted-To:
alt.os.linux.slackware,comp.os.linux.development.system,comp.os.linux.misc,comp.os.linux.setup
Subject: Re: Overclocking (was: Re: K6-2 and Linux, Are there any Bug?)
Date: 27 Feb 1999 02:18:36 GMT
In comp.os.linux.development.system GBP <[EMAIL PROTECTED]> wrote:
: Yeah point well taken. But is it true that overclocking can result in
: system instability? When netscape crashes how am i going to know it was
: a bug and not my CPU doing and instruction wrong or something? When
: people say instability what exactly do they mean? do these machines
: freeze?
easy to tell - run your system to-spec for a while to find out which bugs are
os- or app-related. then overclock and test again. the delta in bugs are
your local system.
------------------------------
From: Taylor Hutt <[EMAIL PROTECTED]>
Crossposted-To: comp.os.linux.portable
Subject: Anyone using a PCMCIA 3Com 3CCFEM656BT?
Date: 25 Feb 1999 15:43:30 GMT
I've got a Dell Inspiron 7000 running RedHat 5.2 (kernel 2.0).
I purchased the Dell-offered 3Com 10-100 LAN/56K Modem card to go with
the machine, but I can't get it to work under Linux -- neither the
LAN nor the modem section.
I've called 3Com and they claim "the card is supported by Dell,
and oh, by the way, even though we manufacture it, we have no idea
if it is compatible with anything else we make".
Dell was a bit more help and personable on their tech support line,
but still they had no idea if it was compatible with anything else
in the world.
I have found that 3Com makes a few other cards with similar part
numbers:
3Com 3CCFEM556B (10/100 & 56K Modem)
3Com 3CXEM556BT (10/100 & 56K Modem)
3Com 3CCEM556BT (10/100 & 56K Modem)
(The last 3 digits of the Dell part are 656; the 3Com: 556 -- that's
not a typo)
I have found a reference that at least the 3CCFEM556B using the
3c574_cs driver has worked for at least one machine at the following
URL:
http://www.wht.mmtr.or.jp/~semoto/PLP/memo/kernel/support-card.html
My Linux setup will give a high then a low beep, which AFAIK, indicates
the card is recognized by not properly initialized.
Does anyone have any clue what I need to do to get this card recognized
and running?
I've got no eth0 device at the end of the boot sequence.
/var/run/stab indicates the 3Com card is installed.
cardctl ident reports the card as being installed.
What else do I need to do to get an eth0?
Taylor Hutt
------------------------------
From: Bas van Nunen <[EMAIL PROTECTED]>
Subject: Re: Midi with SB AWE64 ISA ?
Date: 25 Feb 1999 14:08:33 GMT
Daniele Bernardini <[EMAIL PROTECTED]> wrote:
> /usr/src/linux/Documentation/sound
> -
> and the file is
>
> AWE32
Yep, I know,
but what I meant was:
the 'sound' directory is not included....... ;-)
--
BvN
---
B.M.J. van Nunen, 013-5801805 / 06-22458289.
Student of Information Management and Technology.
[EMAIL PROTECTED] / [EMAIL PROTECTED]
http://stuwww.kub.nl/people/b.vannunen
------------------------------
From: "Bob Fahey [[EMAIL PROTECTED]]" <[EMAIL PROTECTED]>
Subject: Re: USRobotics Modem
Date: Thu, 25 Feb 1999 23:09:23 -0500
Reply-To: <[EMAIL PROTECTED]>
=====BEGIN PGP SIGNED MESSAGE=====
Hash: SHA1
The command sets are different for each chipset. Anyone with a
'rockwell', please advise how to ensure the compression/error control
is 'turned on'. Sometimes just doing a 'AT&F' from a comm program
will load factory modem presets (as long as you didn't mess those up
by saving a 'bad' config), those presets 'usually' set the modem for
rts/cts with error control/compression. Please take a look at the
faq's and manuals on how to set up comm links for linux.
feliz wrote in message <[EMAIL PROTECTED]>...
>I have the same problem with my 56K modem...its a rockwell chipset
with some
>cheapie company board...
>
>how does one turn the compression on and check if setserial is
working
>right???
>
>Anyone who has answers to the problem plz e-mail me at:
>[EMAIL PROTECTED]
>
>
>Bob Fahey [[EMAIL PROTECTED]] wrote in message
><7aoed3$57p$[EMAIL PROTECTED]>...
>>Your compression is turned off, also try making sure your setserial
is
>>working correctly. What does your console tell you about 'connnect
.....'?
>>
>>John E. Hagensieker wrote in message
<7a53tf$[EMAIL PROTECTED]>...
>>>Just installed Red Hat 5.2. Have an internal 56K USRobotics modem
and
>have
>>>it installed to the "Com3" port. When I query the modem with an AT
>command
>>>in minicom it takes about 15 seconds to respond with the modem
string. It
>>>comes back at me with just a couple charachters and it takes a
while.
>>>
>>>I can dial out and connect to my ISP but the login prompt comes in
just a
>>>few charachters at a time also. Normal login can take minutes
instead of
>>>seconds.
>
>
>
=====BEGIN PGP SIGNATURE=====
Version: PGP Personal Privacy 6.0.2
iQA/AwUBNtYecoZlCXlo/b1yEQL2KQCeLDEzuPAcVg71gReynNXZzP0POMkAn3X1
UQW2jdTvDFaAdZylBwMiUeww
=NxoI
=====END PGP SIGNATURE=====
------------------------------
From: Julian Robert Yon <[EMAIL PROTECTED]>
Crossposted-To:
alt.os.linux.slackware,comp.os.linux.development.system,comp.os.linux.misc,comp.os.linux.setup
Subject: Re: Overclocking (was: Re: K6-2 and Linux, Are there any Bug?)
Date: Thu, 25 Feb 1999 17:18:09 +0000
BL wrote:
> you're partially correct.
>
> if you overclock by choosing nonstandard bus speeds (75, 83, 102, etc) then
> you are running pci at nonstandard data xfer speeds. but the intrinsic
> function of the board (unless they cut MAJOR corners in the design) derives
> timing locally, from a local clock source. ie, if you overclock, you should
> NOT be getting a "12megabit ethernet system", etc ;-) ;-)
I tried to overclock my 300A using a 83MHz bus and my i740 gfx card
didn't like it. It was, however, happy at 66 and 75MHz, and is labelled
as running at 100. This looks like a deliberate act on the part of Intel
(since some competitors' cpus run at 83), but maybe I'm just being
paranoid.
Of course, I'm having a few problems getting the card to work with linux
anyway, but that's another story...
Julian
~~~~~~~~~~~~~~~~~~~~
[EMAIL PROTECTED]
[EMAIL PROTECTED]
------------------------------
From: [EMAIL PROTECTED]
Subject: Re: 3com 3c905b card loses Mac address
Date: 26 Feb 1999 19:35:40 -0500
I have exactly the same problem. Any help would be very much
appreciated.
Craig Martell
[EMAIL PROTECTED]
[EMAIL PROTECTED] writes:
> Hey,
>
> I've been reading for hours now and I haven't been able to solve a problem I
> thought you might understand. I have rh5.2 and I compiled kernel 2.2.2.
> When I load the new kernel, the card loads exactly the same as in 2.0.36
> except DHCP fails and ifconfig reports no eth0 interface. I compiled support
> for the card and bootp. Any idea why i'm having this problem?
>
> Thanks so much,
> Dave
>
>
------------------------------
From: [EMAIL PROTECTED] (John L. Spetz)
Subject: Any chance for Linux to use an MDP7800-U PCI Modem
Date: 27 Feb 1999 02:37:04 GMT
Reply-To: [EMAIL PROTECTED] (John L. Spetz)
The statistics gleaned from Windoze 98 are as follows:
MDP7800-U PCI Modem
IRQ 10
Mem Range E5821000-E58210FF
I/O Range E000-E007
I/O Range E400-E4FF
Max Speed 115200
UART NS 16550AN
Identifier No hardware ID for this modem
ATI1 E851
ATI2 OK
ATI3 LT V.90 Data+Fax Modem Version 5.17
ATI4 72
ATI5 5.17, 1, 19 122D, 4100
ATI6 OK
ATI7 OK
AT+FCLA 0,1
The communications driver is comm.drv
5872 bytes 5/11/1998 08:01 PM
The modem was manufactured by, I think, Aztech. I have not had any
luck trying to use setserial to access it. I have not taken off the
cover to look for jumpers but I don't have any docs on how they would
be set if there are any.
As I understand it there are three warning signs for modem compatibility:
1) Winmodems are almost certainly useless.
2) Modems that depended on Rockwell drivers are awkward to get working but
sometimes useable with error correction, etc, turned off.
3) Modems that use "shared high memory" to communicate with Windoze are
hard or impossible to get using.
I have a feeling my PCI modem falls under category 3 and perhaps 2. I
would love to hear different if anyone has gotten a MDP7800-u modem to
work.
Failing that, I was wondering if there is a list of widely available
modems currently on the market that are known to work properly with
Linux. Recent trips to CompUSA and MicroCenter here in Ohio, came
up with a wide selection of win-modems, various PCI internal modems
some of which claim to work under Dos, various external modems some
of which prominently mention Windoze on the packaging. I really
hate the idea of paying $50-150 on a modem and then experimenting. Are
there any specific recommendations for new modems out there? Also, is
there any chance that "shared high memory" modems might be supported some
time soon or are they to proprietary/undocumented ala winmodems?
------------------------------
From: [EMAIL PROTECTED] (Jeff Javits)
Subject: Re: RedHat 5.2 Workstation Install: SIS6326
Date: Thu, 25 Feb 1999 03:34:20 GMT
On Wed, 24 Feb 1999 10:29:59 -0800, "Frank Britton"
<[EMAIL PROTECTED]> wrote:
>Being new to Linux I am first trying the simplest install but seem not to be
>able to succeed. Is there a Linux drive for SIS6326? I have just tried SVGA
>and boom.
>
>Thanks
>
>
Frank:
Although I am having other troubles getting X Windows going, that
chipset is only supported by the newest release of XFree86, which
can be retrieved from the Red Hat updates ftp site at
updates.redhat.com.
Download the package (about 7MB) and use RPM to install.
Good luck!
Jeff Javits
San Francisco, CA
[EMAIL PROTECTED]
------------------------------
From: Grant Taylor <[EMAIL PROTECTED]>
Subject: Re: EPSON Stylus Color 740 (1440*720 now works!)
Date: 25 Feb 1999 12:45:13 -0500
[EMAIL PROTECTED] writes:
> All three major resolutions are now working. On to color tuning.
Woo-hoo! I've upgraded this printer's status from "Partially" to
"Mostly" in the HOWTO's compatibility listing.
--
Grant Taylor - gtaylor@picante<dot>com - http://www.picante.com/~gtaylor/
Cellphone information: http://www.picante.com/~gtaylor/cell/
Libretto information: http://www.picante.com/~gtaylor/portable/
Linux Printing HOWTO: http://www.picante.com/~gtaylor/pht/
------------------------------
Date: Fri, 26 Feb 1999 20:48:45 -0600
From: "Scot E. Wilcoxon" <[EMAIL PROTECTED]>
Crossposted-To:
comp.os.linux.development.system,comp.os.linux.misc,comp.os.linux.portable
Subject: Re: Raw writing to PCMCIA SRAM cards
> What, "dd" isn't nice enough??
Maybe "cat" or ">" will do.
(There's usually six ways to do things -- what are some more...?)
------------------------------
From: "Mark Smith" <[EMAIL PROTECTED]>
Crossposted-To:
comp.os.linux.development.system,comp.os.linux.misc,comp.os.linux.portable
Subject: Raw writing to PCMCIA SRAM cards
Date: Thu, 25 Feb 1999 17:53:05 -0000
Hello,
With the PCMCIA drivers under Linux is it possible to write a binary to an
SRAM card ?
I want to do some development for my Amstrad PDA600 and dump the programs
onto SRAM so as they'll load. The PDA requires that the binaries are
written in a "raw" format to the card, the cards musn't be treated by the
development system as though it was a disk.
Anyone know the answers ?
Regards
Mark Smith
If replying via e-mail remove my spam blocker
------------------------------
From: "S. John Ilagan" <[EMAIL PROTECTED]>
Crossposted-To: comp.os.linux.networking
Subject: Serial port problem/P2/Linux
Date: Fri, 26 Feb 1999 21:25:18 -0600
I've recently moved to a new Linux box (Dell XPS 400, boots into RH 5.2,
Win98 and WinNT Server) and I'm having a problem with the serial ports.
One port is off the ATX case (9 pin 16550a) and the other is an ISA
Card, 16C650, high-speed serial portt (25 pin) from Byterunner. I hook
up my ISDN TA (Adtran Express XRT) to the 16C650 card.
The 16C650 card works fine in Win95 and WinNT Server, throughput is in
the 13-15Kbps range. However, in Linux, it pretty much stinks. I get
dropped packets constantly and stalled transmissions. Throughput is
only in the 100bps to 2Kbps range. I've compiled the kernel with
default serial code and also used patches I've found that modify
serial.c with code that addresses the special FIFO buffers on the 16C650
(location: http://www.devilsthumb.com/~rob) I've used them before with
great results on my old Cyrix P200+ box. Also I use irqtune to
reprioritize the irq so that /dev/cua2 (i.e. the 16C650 card which is
DOS COM3) is given the highest priority and of course setserial. This
exact setup worked great on my old Cyrix box but craps out totally on my
new P2. Just a default kernel with no mods yields the same results.
Same results also if just using my vanilla 16550A 9 pin serial port
native to the motherboard again with no kernel mods, just setserial.
I'm currently using an internal USR 56K modem w/linux with great speed
and no problems.
Anybody have any suggestions to try?
------------------------------
From: [EMAIL PROTECTED] (Greg H.)
Subject: Re: ISA Card (3C509 TPO)
Date: Thu, 25 Feb 1999 18:24:28 GMT
peteri ([EMAIL PROTECTED]) wrote:
> Does anybody have experience configuring this card under Linux? Is it
> compatible?
I have the 3c509b-TPO (Etherlink III) ISA NIC and it works great.
Greg H.
------------------------------
From: [EMAIL PROTECTED] (Silvio Romagnoli)
Subject: 6x86 trashes X screen
Date: Thu, 25 Feb 1999 18:31:08 GMT
I've put on my Linux box an used IBM 6x86MX 233 cpu,
and it works flawlessly in character mode, but under
X it trashes all windows when moving or scrolling
them, forcing me to do screen refreshes.
This didn't happen with other cpus (Intel P75
and IDT 200)
I was thinking of a defective cpu since I can't
clock it higher than 75x2, instead of the 75x2.5
rate marked on the package, but setting it to
lower clocks (50x2 and so) or disabling 1st level
cache doesn't solve the problem under X.
RH5.1 upgraded to 2.2.1 kernel (rebuilt for 6x86)
and related packages, XFree86 3.3.2 (SVGA), Cirrus
Logic 5446 PCI card and QDI Explorer2 m/b with
latest bios upgrade.
TIA
--
sr - [EMAIL PROTECTED] <strip NOS & PAM to email>
------------------------------
From: [EMAIL PROTECTED] (Carlos Eduardo Rodrigues - LEI)
Subject: Re: SCSI Zip Zoom card
Date: 25 Feb 1999 18:28:18 GMT
Jose Carlos Machado ([EMAIL PROTECTED]) wrote:
: Hi
: I'have RedHat Linux 5.2 with Kernel 2.2.1
: Ican't have my Zip Zoom scsi card recognized by the system.
: scsi support and scsi disk support are built into the kernel but I get
: the following message during boot:
: SCSI: 0 HOSTS
: SCSI: Not Detected
: The driver aha152x with proper parameters is loaded as a module.
: If I use insmod aha152x.o I get a message saying:
: init_module: device or resource busy
: Hepl will be really appreciated. Thanks a lot!
: ZC
Use modprobe aha152x aha152x=IO_ADRESS_OF_THE_CARD
------------------------------
From: [EMAIL PROTECTED] (Sparkzz)
Subject: Re: Presario 1210 and Linux
Date: 27 Feb 1999 03:34:01 GMT
Hello,
I finally got Linux loaded on a 1235, but keep in mind the buildin modem is a
winmodem.
I took me about a week to find all the parts for Xwindows .
.
.
....Ken
------------------------------
From: [EMAIL PROTECTED] (Burkard B. Kreidler)
Crossposted-To: comp.os.linux.setup
Subject: Re: fsck problem on boot-up
Date: Thu, 25 Feb 1999 18:13:41 GMT
>Well, I tried several times to run fsck-- without luck!!
>
>First fsck/dev/hda3 ... no results, just a new prompt.
>
Dont't forget the blank!
fsck /dev/hda3
------------------------------
From: "David A. Frantz" <[EMAIL PROTECTED]>
Crossposted-To: comp.os.linux,comp.os.linux.misc
Subject: Re: Can Linux use 36-bit Xeon addressing?
Date: Fri, 26 Feb 1999 20:47:15 -0500
Robert Krawitz wrote in message ...
>"David A. Frantz" <[EMAIL PROTECTED]> writes:
>
>> Robert Krawitz wrote in message ...
>
>> >I think this is a tad unfair. I'm disappointed that Linus doesn't
>> >want to enable large memory addressing on the x86.
>>
>> As with any general purpose operateing system there are trade offs, one
>> outstanding feature of Linux is the freedom to transform it into
something
>> that suits your purposes. The reallity is that there is nothing to be
>> gained by trying to use a special capability of the XEON just to fillfull
>> the special needs of a few users. This is especially the case when the
>> Chip and Chip SETs are not suited for the application. I firmly
believe
>> that if you really need 64 bit addressing to main memory then you need to
>> look at a 64 bit system.
>
>Well, Xeon boxes seem to be awfully popular these days. And again:
>there's a lot of software (even for Linux) that only runs on x86.
>Folks who want to use Oracle don't have the option of getting an Alpha.
They may be popular some where in the world, but the cost is a consideration
especial considering there limit capability increase over the rest ot the
Intel family. I still of the opinion if you need more than 32 bit
addressing then you should jump to a 64 bit system. Why screw around with
a solution that at best is only temporary.
>
>> >Job mixes that are more memory/IO than computation intensive (which is
>> >the case for a lot of commercial data processing) would benefit
>> >greatly from the availability of large memory on commodity hardware.
>>
>> Why would anyone do commercial data processing in large pools of main
>> memory? Seems awfully risky. Actually large memory systems and
heavy
>> computation base apps go hand in hand.
>
>Example: something that's trying to join a stream of transactions to
>accounts. Database (and non-database) joins can always use all the
>memory they can get their grubby little paws on.
>
>Actually, on further thought Linus's last message on the topic
>suggested using the extra RAM as a ramdisk. If the machine then
>swapped to the ramdisk, things would work reasonably well.
This is something that is totally different than addressable application
space. RAM disks can be very useful but they really aren't part of the
applications address space or shouldn't be.
>
>
>[Disclaimer: that I'm not a disinterested observer: I work for Torrent
>Systems: http://www.torrent.com/. However, this posting is completely
>my own opinion, and does not reflect any official company policy.]
>
>--
>Robert Krawitz <[EMAIL PROTECTED]> http://www.tiac.net/users/rlk/
>
>Tall Clubs International -- http://www.tall.org/ or 1-888-IM-TALL-2
>Member of the League for Programming Freedom -- mail [EMAIL PROTECTED]
>
>"Linux doesn't dictate how I work, I dictate how Linux works."
>--Eric Crampton
------------------------------
** 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
******************************
