Files
oldlinux-files/ftp-archives/tsx-11.mit.edu/1996-10-07/mail-archive/linux-devel/Volume2/digest250
2024-02-19 00:24:15 -05:00

736 lines
26 KiB
Plaintext

From: Digestifier <Linux-Development-Request@senator-bedfellow.mit.edu>
To: Linux-Development@senator-bedfellow.mit.edu
Reply-To: Linux-Development@senator-bedfellow.mit.edu
Date: Sat, 1 Oct 94 11:13:21 EDT
Subject: Linux-Development Digest #250
Linux-Development Digest #250, Volume #2 Sat, 1 Oct 94 11:13:21 EDT
Contents:
Re: Could TCP/IP be implemented over SCSI? (jbarrett@onramp.net)
Re: how to install SCSI tape drive (Karsten Steffens)
NetWare client? (Dimitrios Valsamis)
Re: SMail security hole? (John Henders)
Adaptec 2940? (Lee J. Silverman)
Re: PROBLEM: Adaptec 1542 with SMC-Ultra (Rob Janssen)
Re: Try this IPX bridging code ... (Rob Janssen)
Re: Compiling progs using port I/O (Uwe Bonnes)
Re: Adaptec 1542/SCSI under Linux (Matthias Bruestle)
Re: Where's my corefile? (Daniel Quinlan)
Does linux implement semaphores? (Neal Patrick Howland)
Re: OpenGL Extensions to X server?
Re: i486 Word length, anyone? (Jim Finnis)
Re: Multiprocessing Pentium Systems (Cees de Groot)
3com509 corrupt packets (esp. w/ Fl.La Roche NFS-patch)-fix! (Wolfram Gloger)
Re: ParcPlace OI builder?? (Klaus Schniedergers)
Re: Linux on multiple processors? (Corey Sweeney)
fiber optic ethernet cards (Corey Sweeney)
Re: hda: read_intr: error (Bernd Eckenfels)
Re: 1+ Gig SCSI Drives (S. Lee)
Re: Memory in 1.1.50: What is data? (S. Lee)
Re: i486 Word length, anyone? (Doug Dejulio)
----------------------------------------------------------------------------
From: jbarrett@onramp.net
Subject: Re: Could TCP/IP be implemented over SCSI?
Date: Sat, 01 Oct 94 06:43:15 PDT
<rob@pe1chl.ampr.org> writes:
>
> In <1994Sep29.120947.16789@taylor.infi.net> mark@taylor.infi.net (Mark A.
Davis) writes:
>
> >dwm@shell.portal.com (David - Morris) writes:
>
> >>While ya'll are at it, shared scsi hard drives and/or CDROMs might be
> >>intersting as well ... even if restrictions to one host r/w others r/o.
> >>Source trees, netnets, etc. would be obvious candidates.
>
> >Actually, I think that is allowed within the SCSI specs..... Same SCSI
> >bus, with two controllers, one in each host. I don't know how it is
> >done, but I'm quite sure I heard that somebody had done it before to
> >share a tape drive....
>
> It sure isn't a problem at the block device level, but it needs to be
> supported at higher levels as well to yield meaningful results on devices
> you want to share on a 'permanent' basis (like disks).
> Having it r/o on one system and r/w on another is not going to cut it,
> given cached disk blocks on the r/o system and delayed writes on the
> r/w system...
> You could send a special message that causes "disk changed" handling
> whenever you write to the disk, but it would need to be implemented on
> both systems.
>
Sharing an RO device like a CDROM is no problem... simply make sure the two
host systems have different SCSI ID# (this is required anyway for two systems
to share a SCSI Bus)... The SCSI BUS will handle arbitration between the two
systems attempting to access the device simultaneously.
By sending LOCK (I think... or RESERVE) commands to the SCSI device before and
after disk access, you can temporarily limit access to the device to a single
host, so that the other system cannot seek the drive just before your read
command. This should be a relatively minor change to the existing SCSI block
device driver.
Sharing RW devices gets a little more difficult. DEC implemented a ClusterDisk
scheme where access arbitration and cache invalidates were communicated between
systems via DECNET.... If we were to use TCP/IP over SCSINET all data transfer
and arbitration would be handled on the SCSI Bus. That would leave the task of
deriving a new file system type that would handle the arbitration.
Hmmm... rethink that a second....
The preliminary SCSINET spec has an underlying protocol, TMMP (Target Mode
Message Passing), which allows multiple protocols to share the SCSI bus by
assigning each protocol to a different SCSI LUN. Using TMMP instead of TCPIP to
handle arbitration between systems sharing a device would eliminate the TCPIP
overhead from the arbitration process... mucho faster....
Are we on the verge of a new shared file system standard here ??? Hmmm....
It appears that there is going to be a dedicated Mailing List for SCSINET and
TMMP... details and net addresses as soon as the server is up and running!!!
John Barrett <jbarrett@onramp.net>
------------------------------
From: karsten@kshome.ruhr.de (Karsten Steffens)
Crossposted-To: comp.os.linux.admin
Subject: Re: how to install SCSI tape drive
Date: 30 Sep 1994 06:47:38 GMT
Woody Weaver (woody@hermes.stmarys-ca.edu) wrote:
: I can talk to the card fine: $ mount -t iso9660 /dev/sr0 /mnt
: mounts the CD ROM normally, and gives me access. However, I don't have any
: st0 devices. I tried
: $ mknod /dev/st0 c 46 0
: (and a couple of other minor numbers, 8 and 128) but each time I try
: $ mt -f /dev/st0 fsf
: or whatever, I get "/dev/st0: No such device". Am I being particularly
: dense? What is the fix?
Who the hack told you that 46 was the correct major device number?
Try 9 instead, and it should work.
BTW: the official lists of device-numbers can be found in:
/usr/src/linux/include/linux/major.h
This is official because its the one that the kernel incorporates during
compilation...
Karsten
--
==================> Karsten Steffens <=====================
karsten@kshome.ruhr.de | steffens@ikp.uni-muenster.de
Marl - close to Recklinghausen | Institut fuer Kernphysik
North of the Ruhrgebiet | Westf.Wilhelms-Universitaet Muenster
------------------------------
From: dvalsami@ernie.eecs.uic.edu (Dimitrios Valsamis)
Subject: NetWare client?
Date: 29 Sep 1994 14:01:00 GMT
Hello everybody,
Is there any NetWare client available for Linux. If
yes, where can I get it?
Thanks
Dimitrios Valsamis
dvalsami@eecs.uic.edu
u45561@uicvm.uic.edu
------------------------------
Crossposted-To: comp.os.linux.help
From: jhenders@jonh.wimsey.com (John Henders)
Subject: Re: SMail security hole?
Date: Thu, 29 Sep 1994 00:27:13 GMT
In <36bppk$ml@midget.wg.saar.de> bof@wg.saar.de (Patrick Schaaf) writes:
>Your conclusion (smail must be misconfigured) is correct, your proof
>is not; the hole mentioned allows unwanted _creation_ of files in
>inaccessible directories, with the file being owned by the user
>(when append_as_user is set). Checking the source of transports/appendfile.c
>you'll find that the attribute to set in the transport is called
>'check_path'. The bug is gone now. I have no idea why that isn't the
>default setting - does anybody know?
I tested the creation of a new file at the same time as I tested
trying to overwrite /etc/passwd and creation didn't work on my system
either. I would suspect another possibility is that it is not a smail
problem but a permission problem with /. I don't have check_path set
either, but I'll look into what it does. Perhaps the version of smail I
have has this as a default.
--
John Henders - Wimsey Information Services
http://www.wimsey.com/ (teletimes, gnn and more)
GAT/MU/AE d- -p+(--) c++++ l++ u++ t- m---
e* s-/+ n-(?) h++ f+ g+ w+++ y*
------------------------------
From: lee@netspace.students.brown.edu (Lee J. Silverman)
Subject: Adaptec 2940?
Date: 30 Sep 1994 06:58:22 GMT
I've seen people mention the Adaptec 2740, but a friend of
mine has a new Gateway 2000 with what he says is an Adaptec 2940 SCSI
card. I trust him to be competent enough to read the model number off
the card, but I haven't heard anything about it, and it isn't
mentioned in the SCSI Howto. Do the newest kernels support this card?
--
Lee Silverman, Brown class of '94, Brown GeoPhysics ScM '95
Email to: Lee_Silverman@brown.edu
Phish-Net Archivist: phish-archives@phish.net
"Nonsense - you only say it's impossible because nobody's ever done it."
------------------------------
From: rob@pe1chl.ampr.org (Rob Janssen)
Subject: Re: PROBLEM: Adaptec 1542 with SMC-Ultra
Reply-To: pe1chl@rabo.nl
Date: Sat, 1 Oct 1994 11:04:03 GMT
In <C4289.94Sep30125136@rphc2.physik.uni-regensburg.de> c4289@rphc2.physik.uni-regensburg.de (Olaf Jaeger) writes:
>problem:
> I am using an ISA-Adaptec-1542c and a SCSI-2-HD with an
>ext2-filesystem V. 0.5a on it. From the time that i put a
>SMC-Ultra into the machine, the filesystem on the HD begins to vanish.
I think there is a hardware problem in the ethernet card that causes
it to be incompatible with busmastering controllers.
This has been mentioned some times before, but I don't exactly remember
which version of the card fixed it.
Rob
--
=========================================================================
| Rob Janssen | AMPRnet: rob@pe1chl.ampr.org |
| e-mail: pe1chl@rabo.nl | AX.25 BBS: PE1CHL@PI8UTR.#UTR.NLD.EU |
=========================================================================
------------------------------
From: rob@pe1chl.ampr.org (Rob Janssen)
Subject: Re: Try this IPX bridging code ...
Reply-To: pe1chl@rabo.nl
Date: Sat, 1 Oct 1994 11:10:25 GMT
In <1994Sep30.175942.284@acad.ursinus.edu> STEVO@acad.ursinus.edu (Steve Kneizys) writes:
>Alan Cox (iialan@iifeak.swan.ac.uk) wrote:
>: In article <1994Sep25.223539.260@acad.ursinus.edu> STEVO@acad.ursinus.edu (Steve Kneizys) writes:
>: >If somebody wanted to isolate an IPX net/server from the main net
>: >in terms of packet density but did not want to change the net
>: >numbers, well, bridging would be an option! I may decide to add
>: >it to my above bridge, as bridging is faster than routing.
>: No bridging is normally slower than routing as you process more packets
>: at the software level.
>Nah...depends on the speed of your algorithms! Has nothing to do with
>packet density, unless your algorithms are so slow that packets come
>in faster than you can filter them. Think about it for a second...the
>CPU overhead may be different, but network throughput depends on how
>much time you spend processing the packet. If I can lookup an
>Ethernet address and determine what interface, if any, to send out
>faster than looking up in a table, changing the hop count and
>checksum, net number, then sending out then bridging is faster.
The point probably is that bridging is practical to be implemented on
dedicated hardware (possibly with some microcode), while routing usually
isn't. So, a dedicated bridge could be faster than a router.
Of course, this no longer holds once you build your bridge using two
normal ethernet cards and a PC. In that case, the disadvantage of having
to look at all packets becomes the deciding factor.
Rob
--
=========================================================================
| Rob Janssen | AMPRnet: rob@pe1chl.ampr.org |
| e-mail: pe1chl@rabo.nl | AX.25 BBS: PE1CHL@PI8UTR.#UTR.NLD.EU |
=========================================================================
------------------------------
From: bon@lte.e-technik.uni-erlangen.de (Uwe Bonnes)
Subject: Re: Compiling progs using port I/O
Date: Fri, 30 Sep 1994 18:41:41 GMT
Rob Janssen (rob@pe1chl.ampr.org) wrote:
> In <36bmo0$fmg@clarknet.clark.net> nardone@clark.net (Joe Nardone) writes:
> >Hey net-folks--
> >I'm trying to compile a program that uses the inb and outb
> >functions (macros, actually) but when it comes to link time
> >all my inb/outb calls are represented as unresolved references
> >to ___outb (or ___outcb) and ___inb...
> >Am I missing a library, or a path to one? gcc -v looks like it's
> >looking in all the right places for library files...
> >I'm running GCC 2.5.8 on Linux Kernel 1.1.50 w/ a 486dx2/66.
> >Any help would be much appreciated-
> You have to compile with optimization (-O2)
Is there some explanation for that behaviour?
--
Uwe Bonnes bon@lte.e-technik.uni-erlangen.de
------------------------------
From: m@mbsks.franken.de (Matthias Bruestle)
Subject: Re: Adaptec 1542/SCSI under Linux
Date: Wed, 28 Sep 1994 13:11:42 GMT
Mahlzeit
> I would like to know how reliable SCSI generally is under Linux. I have
> had some problems witj my Fujitsu floptical but I am quite prepared
> to accept that lies with the way the drive behaves, however I would
> be very interested to find out how people have been getting with
> large SCSI drives (>1 gig or so) as I am thinking of buying one!
I have here a 1.2Gig Toshiba and a 1542C. No Problems since about 10 months
(ie. since kernel 0.99pl14+scsi-diff).
Mahlzeit
--
A leap ahead... through insanity.
------------------------------
From: quinlan@freya.yggdrasil.com (Daniel Quinlan)
Subject: Re: Where's my corefile?
Date: 28 Sep 1994 19:01:45 GMT
Reply-To: quinlan@yggdrasil.com
Todd Klaus <klaus@indirect.com> writes:
> I'm getting segmentation faults, but no (core dumped)! I looked in
> /etc/profile and ~/.profile thinking this was a shell thing, but
> found nothing. How do I enable corefiles? I'm using the Yggdrasil
> summer '94 CD.
Please read the Linux FAQ. Also, please post questions to
comp.os.linux.help, where they belong.
Question 6.14. How can I produce core files ?
Since 0.99pl14 Linux has had corefiles turned off by default for all
processes. You can turn them on by using the ulimit command in bash, or
the limit command in tcsh. See the manpages or Q6.15 `How do I stop
producing core files ?' for more details.
After executing that command all programs run from that shell (directly or
indirectly) will be able to dump core.
If you wish to enable coredumping for all processes by default you can
change the default setting in <linux/sched.h> - see the definition of
INIT_TASK.
--
Daniel Quinlan <quinlan@yggdrasil.com>
------------------------------
From: nhowland@ksu.ksu.edu (Neal Patrick Howland)
Subject: Does linux implement semaphores?
Date: 28 Sep 1994 16:16:40 -0500
I was wondering in the standard linux develpment packages implemented
a semaphore synchronization call. If not, how do you synchronize two
processes to keep them from entering their critical sections at the same
time?
I haven't installed or used linux yet, but I am going to try soon and would
like to know what kind of system calls are available. I have several
unix programs that use the semaphore method and I would like to just be
able to transfer them to linux and compile them.
Neal Howland
nhowland@matt.ksu.ksu.edu
------------------------------
From: jsmith@red-branch.MIT.EDU ()
Subject: Re: OpenGL Extensions to X server?
Date: 1 Oct 1994 13:44:00 GMT
Jim Callahan (jim@jimcal.dialup.access.net) wrote:
: Anyone working on extending the XFree86 server to handle OpenGL calls?
: With several fast 3D hardware PC cards comming out soon which support OpenGL in
: their instruction set, it would be very nice to have a GL/X server under Linux.
: With this done, Linux users could get the benifit of the OpenInventor 3D OOP
: toolkit and tons of GL apps...
: I've seen OpenGL running on all of the more traditional UNIX platforms and even
: the dreaded NT and Mac will be supprting it soon. Wouldn't want uncle Bill to
: beat the Linux community on this one... ;^]
: Jim Callahan Computer Animation Cygnus Feedback, Inc.
: jimcal@panix.com | Interactive 3D Apps | 200 East 11th Street 2E
: 212 477-6239 GUI Design New York, NY 10003
It is being worked on yes.
Jonathan Smith
------------------------------
From: white@elf.dircon.co.uk (Jim Finnis)
Subject: Re: i486 Word length, anyone?
Date: Sat, 1 Oct 1994 10:26:17 GMT
In article <36hoga$1f1@usenet.srv.cis.pitt.edu>,
Doug Dejulio <ddj+@pitt.edu> wrote:
>
>Well, "word" doesn't have an exact quantitative meaning -- it just
>means some number of bytes. It's often used to mean the number of
>bytes that it's most "natural" to work with on a particular computer
>(ie. 4 bytes on a 32 bit computer, 8 bytes on a 64 bit computer, etc).
>It *can* mean *any* number of bytes. My bet is that the number the
>original poster wants is 4.
>
Quite true, and the same used to apply to bytes. Nowadays, a byte is
accepted by most people to mean "8 bits", but I've worked on a 9-bit byte
machine with 3 of these bytes per word. Too weird.
Jim
------------------------------
From: cg@tricbbs.fn.sub.org (Cees de Groot)
Subject: Re: Multiprocessing Pentium Systems
Date: 30 Sep 1994 23:42:53 +0100
In article <Cwtt18.54u@syd.dms.csiro.au>,
David Monro <davidm@syd.dms.CSIRO.AU> wrote:
>
>As an aside, are there any really good MP OSs out there?
Yes. VMS does quite well on multiprocessor machines.
>Whatever, if Linux is going to run on MP hardware (whether it be 486s,
>Pentiums, DEC Alphas, Mips R4x00s, PowerPCs, 680x0s or anything else you
>care to dream up), we should probably better do our homework pretty
>thoroughly first.
>
I keep having these wonderful dreams of Linux running on a 4-Processor
300MHz AXP... Just haven't figured out yet what I would do with it :-)
--
Cees de Groot, Lake of Konstanz, Germany
PGP23a: 73 5D BA 7C F8 EF DD 65 56 68 AF BB 2B 58 2C 8B
Running Windows on a Pentium is like having a brand new Porsche but only
be able to drive backwards with the handbrake on
------------------------------
From: u7y22ab@sun2.lrz-muenchen.de (Wolfram Gloger)
Subject: 3com509 corrupt packets (esp. w/ Fl.La Roche NFS-patch)-fix!
Date: 1 Oct 1994 14:06:36 GMT
The 3com509 driver in Linux 1.1.51 fails to drop corrupt or overrun
packets. I only noticed this with Florian La Roche's NFS-performance
patch which for the first time caused a fair amount of overruns on my
network. Here is a fix that has even been accepted by Donald Becker:
--- 3c509.c.orig Wed Aug 10 19:00:12 1994
+++ 3c509.c Sat Oct 1 14:44:03 1994
@@ -550,9 +550,7 @@
case 0x2000: lp->stats.rx_frame_errors++; break;
case 0x2800: lp->stats.rx_crc_errors++; break;
}
- }
- if ( (! (rx_status & 0x4000))
- || ! (rx_status & 0x1000)) { /* Dribble bits are OK. */
+ } else { /* Good packet. */
short pkt_len = rx_status & 0x7ff;
struct sk_buff *skb;
Have fun and may your NFS be fast.
Wolfram.
------------------------------
From: eedksc@teamos50.ericsson.se (Klaus Schniedergers)
Crossposted-To: comp.os.linux.misc
Subject: Re: ParcPlace OI builder??
Date: 29 Sep 1994 09:20:34 GMT
Reply-To: eedksc@aachen.ericsson.se
>>>>> "David" == David A Vohwinkel <vohwi-d@acsu.buffalo.edu> writes:
David> and can you buy any manuals for
David> it?
Yes, a manual for the library (not about the builder) is available as
an ordinary book. I saw it in several bookstores. It seemed to be
identical to the OI-lib-manual that you get when you buy an OI license.
Title is 'OI Programmer's Guide' or similar.
Klaus
--
Klaus Schniedergers, OSS Design, eedksc@aachen.ericsson.se
<A HREF="http://www-eed.ericsson.se:8001/org/b/eedksc.html"> Me </A>
------------------------------
From: corey@bbs.xnet.com (Corey Sweeney)
Subject: Re: Linux on multiple processors?
Date: 28 Sep 94 16:49:17 CST
iialan@iifeak.swan.ac.uk (Alan Cox) writes:
>In article <1994Sep20.231705.539@golem.greenie.muc.de>
andi@golem.greenie.muc.de
> (Andi Kleen) writes:
>>: I know that MP (and specifically SMP) is sort of "trendy" these days
>>: (vis. NT and OS/2 SMP), but the particular application for which I am
>>: considering using Linux as a platform (dialin Internet host, web server,
>>: fairly high volume) seems to me to be one for which SMP might give
>>: good results--or at least make it easier to stave off the purchase of
>>: a second machine.
>>: So is this being considered, or at least batted around as a possibility?
>>The HURD (the GNU OS) will (or is planing to) support multiprocessing
>>(through the Mach-kernel)
>
>For Linux doing none to smart (initially) SMP is being played with at the
>concept level here (pending possible hardware donations). Some other people
>'viper' are working on making the Linux kernel truely threaded but that is
>a much bigger project.
>
>Alan
>
>
>--
> ..-----------,,----------------------------,,----------------------------,,
> // Alan Cox // iialan@www.linux.org.uk // GW4PTS@GB7SWN.#45.GBR.EU //
> ``----------'`----------------------------'`----------------------------''
While this is in the concept stage I might as well propose a time-saving step.
How about adding multi-processor support to hurd. By the time the
multi-processor support is ready, one could reasonably assume that hurd would
have the linux server ready.
Actually i see a assumption with my own plan, that might not be true. If we
run linux as a server under mach in hurd, and we add multi-processor support to
hurd, does that mean that linux would be multi-processor? i'm assuming so.
Corey Sweeney
corey@bbs.xnet.com
------------------------------
From: corey@bbs.xnet.com (Corey Sweeney)
Subject: fiber optic ethernet cards
Date: 28 Sep 94 16:58:27 CST
has anyone seen a fiber optic ethernet driver, or does anyone have a intention
of creating one?
I havn't been able to find one.
Corey Sweeney
corey@bbs.xnet.com
------------------------------
From: ukd1@rzstud1.rz.uni-karlsruhe.de (Bernd Eckenfels)
Subject: Re: hda: read_intr: error
Date: 29 Sep 1994 02:52:01 GMT
Pierre Belanger (belanger@info.polymtl.ca) wrote:
: Hello everybody,
: I just received my P90/Intel Premiere Motherboard, before I had a 486 with
: no problem at all. I first switch the motherboards and Linux did not
: want to boot (1.1.51). It says: hda: read_intr: status = 0x59 and the
: error = 0x10. It was saying this when it was ready to mount the partitions.
I get those Errors on my HD with bad Blocks. I had an Headcrash, and
on one zylinder there are some tracks destroyed. Trying to read those
blocks may result in heavyly recalibration of the hd and the messages
above.
Conclusion->your HD is defect.
Is there a testing (nor formating) tool under linux for IDE HDs?
Greetings
--
(OO) -- Bernd_Eckenfels@Wittumstrasse13.76646Bruchsal.de --
( .. ) +4972573817 ecki@lina.ka.sub.org ukd1@rz.uni-karlsruhe.de
o--o *QUAK* Jetzt auch mit Plueschtier in der .Sig!
(O____O) <A href=http://rzstud1.rz.uni-karlsruhe.de/~ukd1/>Eckes@IRC</A>
------------------------------
From: sl14@crux3.cit.cornell.edu (S. Lee)
Subject: Re: 1+ Gig SCSI Drives
Date: 29 Sep 1994 03:00:48 GMT
In article <36d9t7$68o@epaus.island.net>,
Daniel Rogers <rogersd@epaus.island.net> wrote:
>In article <369gba$b0@news.ED.RAY.COM>,
>Bill Heiser <heiser@spc280.ed.ray.com> wrote:
>>
>>This is something I've been wondering about. It appears to depend
>>on which SCSI HA you are using. In looking for a solution to my
>>instability problems with Buslogic BT445S/DEC DSP3107LS, I noticed
>>a comment in the buslogic driver that seemed to indicate it expects
>>the extended Translation to be switched ON.
>
>Well, I have a 1.7gig Maxtor on a Buslogic 445S, and I had problems when I
>went from using the Adaptec 1542 driver (which didn't want translation) to
>the Buslogic (which did).
>
>What I ended up with was two partitions which overlapped, and when my 1 gig
>news partition filled up to my swap space, all hell broke loose. So, I
>turned on translation, re-partitioned, and re-formatted the drive.
If what I've experienced is the same to yours, you can just patch the
Buslogic driver to make it return 'translation off' instead of on.
Actually, OFF was the default when the driver was introduced into the
kernel, but somewhere around 1.1.46 it changed, and I remember it
breaking LILO on my system (I have a Micropolis 1.7G but the boot
partition is on < 1G).
If you're interested I can dig out the patch for you. But you seem to
have already found it, no? It's just a one-line change.
Stephen
------------------------------
From: sl14@crux3.cit.cornell.edu (S. Lee)
Subject: Re: Memory in 1.1.50: What is data?
Date: 29 Sep 1994 03:04:44 GMT
In article <369gvj$b0@news.ed.ray.com>,
Bill Heiser <heiser@spc280.ed.ray.com> wrote:
>: Peter Suetterlin (pit@myhost.subdomain.domain) wrote:
>:: Memory: 12956k/16384k available (624k kernel code, 384k reserved, 2420k data)
>
>My SCSI (buslogic bt445s) system looks like that.
>However my IDE system doesn't!
>
>Memory: 23476k/24576k available (484k kernel code, 384k reserved, 232k data)
>Linux version 1.1.51 (root@xxxxxx) (gcc version 2.5.8) #1 Fri Sep 23 16:53:55 ED
>
Huh? My BT445S system looks like this:
Memory: 14900k/16384k available (652k kernel code, 384k reserved, 448k data)
How's that?
sl14@cornell.edu
Witty .sig under construction.
------------------------------
From: ddj+@pitt.edu (Doug Dejulio)
Subject: Re: i486 Word length, anyone?
Date: 30 Sep 1994 19:21:46 GMT
>> I tried setting it to 32 (since the 486 is a 32-bit processor <shrug>),
>
>Are you sure ? 32 bytes in a word ? ;) I would take 2 bytes for a
>word, but maybe they mean 4 bytes for 32bit.
Well, "word" doesn't have an exact quantitative meaning -- it just
means some number of bytes. It's often used to mean the number of
bytes that it's most "natural" to work with on a particular computer
(ie. 4 bytes on a 32 bit computer, 8 bytes on a 64 bit computer, etc).
It *can* mean *any* number of bytes. My bet is that the number the
original poster wants is 4.
Since the original 8086 was 16-bit, Intel decided to call 2 bytes a
word. A lot of people took this convention as a definition, and now
think all words are 2 bytes.
So:
In the most general case, a word is a collection of bytes.
One convention widely used in general is that a word is the most
natural number of bytes for a particular CPU to use.
One convention used among people who use Intel chips exclusively is
that a word is 2 bytes (regardless of processor).
--
Doug DeJulio
ddj+@pitt.edu
------------------------------
** 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: Linux-Development-Request@NEWS-DIGESTS.MIT.EDU
You can send mail to the entire list (and comp.os.linux.development) via:
Internet: Linux-Development@NEWS-DIGESTS.MIT.EDU
Linux may be obtained via one of these FTP sites:
nic.funet.fi pub/OS/Linux
tsx-11.mit.edu pub/linux
sunsite.unc.edu pub/Linux
End of Linux-Development Digest
******************************