Subject: Linux-Development Digest #789
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:     Fri, 3 Jun 94 20:13:06 EDT

Linux-Development Digest #789, Volume #1          Fri, 3 Jun 94 20:13:06 EDT

Contents:
  ET4000/W32P chipset (Rolf Riesen)
  Re: Let's rename v1.0.9! [Was: Frustrated with new kernels] (Albert D. Cahalan)
  Re: Colorado Trakker 250 (Bruce@renzo.usra.edu (Bruce Pleat, ))
  Re: Let's rename v1.0.9! [Was: Frustrated with new kernels] (Alex Frausin)
  Re: Why is my Emacs binary so big (2.2M) (Rusty Trainham)
  Re: Logitech Scanner (Re: Twain Scanner) (Joel M. Hoffman)
  Re: 1.1.15 breaks SCSI (Rene COUGNENC)
  Re: Linux game development (Was Re: Why [DOS, W (Rene COUGNENC)
  Re: Can't compile latest ncurses for Linux (Stuart Herbert)
  Re: 1.1.15 breaks SCSI (Scott Snyder)
  Re: Frustrated with new k (Rick Emerson)
  Re: Frustrated with new k (Rick Emerson)
  Re: Why is Linux monolithic? (Art Hays)
  Re: Frustrated with new k (Rick Emerson)
  [Q] tracing I/O port activity ? (Daniel Hirschberg)

----------------------------------------------------------------------------

Crossposted-To: comp.os.linux.admin,comp.os.linux.help
From: riesen@chaco.cs.unm.edu (Rolf Riesen)
Subject: ET4000/W32P chipset
Date: Thu, 2 Jun 1994 20:18:35 GMT

I have a Gateway 2000 pentium system with a PCI graphics card that
uses the ET4000/W32P chip (note the P) and 2 MB of video memory. I'm
driving a NEC 6FGp monitor.

I'm using the SVGA driver. It can't see the second MB of memory. If
I specify 2048 in my Xconfig file, then things go wrong. The mouse
pointer disappears in certain regions of the screen. Parts of windows
get drawn in the wrong place or not at all, etc.

As long as I contend myself using only half the memory, things
work ok. Of course, I can't get 1280x1024 resolution. (What bugs
me, is that under MS Windows I can ;-(

A second problem is that the SVGA driver doesn't allow dot clock
frequencies beyond 90MHz. (I think I need more for the 1280x1024
mode.)

So, here are my specific questions:
    - What do I need to do to use all the memory on my ET4000/W32P
      card? If the answer is "modify the svga driver", can you give
      me some hints on what file in the X11 distribution needs to
      be updated, and what information I need to make the change;
      e.g. register settings for the ET4000/W32P.
    - What is the deal with the 90MHz limitation?
    - Where can I get the specifications for the ET4000/W32P?
      (Register maps, clock frequencies, etc.)

And one more general question:
    - How does X -probeonly determine the clock frequencies? I 
      understand that there are chip sets with fixed frequencies. I
      believe the ET4000 allows variable clock rates. If they are
      variable, what is the range, increment, limitation? How can
      X -probeonly say that only a certain set of 32 dot clocks
      work? (The largest clock reported is around 35MHz, but I
      drive the card succesfully at 72MHz.)

BTW, superprobe thinks I got an MGA or something. I will make this
setup work in 1280x1024 mode. However, I hate do waste time if
someone has already done it. The more help I get, the sooner I'll
reach my goal.

Currently I have an 800x600 and 1024x720 (or whatever) setting
that works with my setup. If you're interested in my Xconfig file
I'll mail it to you.

Thanks in advance.   Rolf

------------------------------

From: adc@zeta.coe.neu.edu (Albert D. Cahalan)
Subject: Re: Let's rename v1.0.9! [Was: Frustrated with new kernels]
Date: 03 Jun 1994 18:27:28 GMT

In article <2sna1n$cp8@winx03.informatik.uni-wuerzburg.de> tom@wpzd07.pzlc.uni-wuerzburg.de (Thomas Heiling) writes:
   Bjorn Ekwall (bj0rn@blox.se) wrote:
   : There are a lot of people that haven't been able to understand
   : the different aims of 1.0.X and 1.1.X, but beleive that the
   : kernel with the "highest" sequence number is meant for everyone.
   : I have a suggestion:
   : Let us rename 1.0.9 to 1.2.0!

     A very good Idea :-)

   : This way, those who haven't read the notices from Linus et al
   : will go for that version, and stay away from 1.1.X...
   : That would make everyone happy, wouldn't it?  :-)
Suggestion: version x.y.z is considered stable when y is even, and
considered hackers paridise when y is odd.  An unstable kernel is odd.  :)
This way every version x.0 would be considered stable.  (What a concept!
x.0 software that works)
--

Albert Cahalan
adc@meceng.coe.neu.edu

------------------------------

From: Bruce@renzo.usra.edu (Bruce Pleat, <Bruce@usra.edu>)
Subject: Re: Colorado Trakker 250
Date: 3 Jun 1994 17:01:56 -0400
Reply-To: Bruce@usra.edu

|From: emrys@cellar.org (The Fool On The Hill)
|Date: Fri, 03 Jun 94 07:24:24 EDT
|
|
|        Does a driver exist to allow me to access my Trakker?  If so, could
|I please get a pointer to it?  If not, why?  Is it _that_ difficult to
|access the tape on the parallel port?  Hell, if MS can do it, we oughta be
|able to!!

Try looking at the FTAPE HOWTO, and at the tar man page.
     Please feel free to contact me if you have any questions or 
need further assistance!

Virtually,
     Bruce

    //::      //::    Information Systems Coordinator
   //  //    //  //  Universities Space Research Association
  //::      //:::   Goddard Visiting Scientist Program
 //  //    //      7501 Forbes Blvd. #206; Seabrook, MD 20706-2253 USA
//:::  ee //      MC610.3 Bldg26 Rm215; NASA/Goddard; Greenbelt, MD 20771 USA

Two Roads Diverge in a Wood
And I took the One Less Travelled By
And That Has Made All the Difference
     --Robert Frost, "The Road Not Taken"


------------------------------

From: alex@myhost.subdomain.domain (Alex Frausin)
Subject: Re: Let's rename v1.0.9! [Was: Frustrated with new kernels]
Reply-To: alex@psyalex.psy.gu.se
Date: Fri, 3 Jun 1994 15:08:36 GMT

Thomas Heiling (tom@wpzd07.pzlc.uni-wuerzburg.de) wrote:
: : >Bjorn Ekwall == bj0rn@blox.se

: : I say let them mess around in their own ignorance.

:  O.K it's Friday afternoon,the sun is shining and I am in the mood.

:  Your statement is very harsh, all those who missed that *one* post
:  are ignorant. But you know the answer ? And you got it from the
:  newsgroup - I think.
:  So why haven't you answered "Look here and there,
:  and pick up this code/programs ?".
:  I think if somewhone has the knowledge, how to solve this
:  problem , he should share it with others.
:  For what else is this newsgroup ?

It's raining and I'm in a lousy mode :-(
This newsgroup is flooded with morons that doesn't even try. I know
that it can be hard sometimes to find the answer, but if they already know 
how to post on the net, I guess they must be able to read news aswell.
A very good for people that feel unsecure about how stuff works is to wait
and read. After a while there is always someone that "report" trouble
and the needed changes. And the difference between the production release
1.09 and 1.1.18 aren't _so_ big, that one has to bee on the bleeding edge 
as an ordinary user.

I have had my share of problems with different kerneluppgrades but by waiting
and reading, most of the problems has been solved. My machine is filled
with stuff (SCSI, SB16, 3Com509...) and it's sitting on the Internet.
I have problems with SLIP at home with my machine but you don't see me whine
about that here :-)

/Alex
--
+-----------------------------+----------------------------------------------+
| Alex Frausin                | Haven't been able to steal any good .sig     |
| Gothenburg University       | file lately.                                 |
| Department of Psychology    |                                              |
| Gothenburg, Sweden          | Opinions all mine.......                     |
|                             |                                              |
| alex@psyalex.psy.gu.se      |                                              |
+-----------------------------+----------------------------------------------+

------------------------------

Crossposted-To: comp.os.linux.help
From: trainham@penduick.saclay.cea.fr (Rusty Trainham)
Subject: Re: Why is my Emacs binary so big (2.2M)
Date: Thu, 2 Jun 1994 21:44:09 GMT

Robert Sanders (gt8134b@prism.gatech.edu) wrote:
: gt8134b@prism.gatech.edu (Robert Sanders) writes:

: >kevinl@NewsServer.cc.monash.edu.au (Kevin Lentin) writes:

: >>-g implies -static. And don't forget to check the CFLAGS and LFLAGS or
: >>LDFLAGS.

: >Not necessarily; it causes ld to link against libg.a instead of libc.sa.
: >If, like me, you made libg.a a symlink to libc.sa, it doesn't mean anything
: >at linktime.

: Er, of course, -g *does* imply -static.  What I meant to say was that -static
: wasn't necessarily the only problem.


Hey, -g implies linking in the debugging library, which of course is static,
but the essential is that a debugging library is huge.
 
Rusty Trainham    trainham@penduick.saclay.cea.fr


------------------------------

From: joel@wam.umd.edu (Joel M. Hoffman)
Subject: Re: Logitech Scanner (Re: Twain Scanner)
Date: Thu, 2 Jun 1994 13:26:03 GMT

[who wrote what deleted]
>>
>>module `logi32' (3 pages @ 0x01818000) created
>>initializing module `logi32', 8928 (0x22e0) bytes
>>  init entry @ 0x0181a0d4, cleanup entry @ 0x0181a1c4
>>LOGITECH handheld scanner driver  version 0.1a
>>Scanner-interface is incompatible.
>>logi32.c: scanner interface not found
>
> This would indicate to me that the interface for your ScanMan 256
> is not compatible with the logi32 software.   Neat how that's
> worded to avoid pointing fingers at software, huh? :)

Hmm.  I'm tyring to use the logi32 driver with my scanman256.  The
driver can talk to the scanner, but something's going slightly wrong.
Each scanned line is shifted to the right, so that a straight vertical
line ends up diagonal.  I suspect the scanman256 is sending protocol
bytes that the driver is interpreting as data bytes, but I can't be
sure.  Also, the scanner only works at all in 400dpi mode.

If you're not getting any results, it may be because your scanner is
using a different h/w address that the driver is expecting.

-Joel
(joel@wam.umd.edu)
-- 
=============================================================================
|_|~~ Germany, Europe. 1943.    "The diameter of the bomb was 30 centimeters,
__|~| 16 Million DEAD.           and the diameter of its destruction, about 7
                                meters, and in it four killed and 11 wounded. 
 cnc  Bosnia, Europe. 1993.     And around these, in a larger circle of  pain
 cnc  HOW MANY MORE?          and time,  are scattered two  hospitals and one
                          cemetery.   But the young woman who was  buried  in
                    the place from where she came, at a distance of more than
             than 100 kilometers, enlarges the circle considerably.   And the 
      lonely man who is mourning her death in a distant  country incorporates
into the circle the whole world.  And I won't speak of the cry of the orphans
that reaches God's chair and from there makes the circle endless and godless."
=============================================================================
     Tell Clinton to stop the genocide:  president@whitehouse.gov

------------------------------

From: rene@renux.frmug.fr.net (Rene COUGNENC)
Subject: Re: 1.1.15 breaks SCSI
Date: 2 Jun 1994 11:43:49 GMT
Reply-To: cougnenc@hsc.fr.net (Rene COUGNENC)

Ce brave Eric Youngdale ecrit:

>       OK, I think I understand what is going on.  I have come up with 
> some patches which should correct this and a few other minor nits that 
> have recently been introduced, and I have uploaded this to tsx-11 in 
> pub/linux/ALPHA/scsi/scsifix-1.1.17.diff.gz.   
>       I have not sent this to Linus yet - I would like some direct 
> confirmation that it actually helps first :-).


It works fine for me, no problem !
--
 linux linux linux linux -[ cougnenc@renux.frmug.fr.net ]- linux linux linux 

------------------------------

From: rene@renux.frmug.fr.net (Rene COUGNENC)
Subject: Re: Linux game development (Was Re: Why [DOS, W
Date: 2 Jun 1994 11:59:25 GMT
Reply-To: cougnenc@hsc.fr.net (Rene COUGNENC)

Ce brave John J. Sullivan ecrit:

> Considerer that Linux could actually make a pretty good game development
> platform:

> The main objection that I see to using Linux is that games developed 
> under it cannot be run directly from DOS, and generally require Linux
> to already be installed on a user's machine.  Here's what I suggest:

With Umsdos and loadlin you have the basic tools for your project.

> This eliminates the need for the user to install Linux on their machine
> or to reboot just to run the game.  On the other hand, it requires that
> a Linux kernel be distributed with the game (about 1M) 

Just a suggestion, based on MS/DOS users psychology :-)
If you put together a package based on a small Linux kernel to run games
(or other applications) from an MS/DOS prompt, call it "Game Extender"
or some kind of pompous name :-)

I know many DOS users around me that laugh, go away or delete the archive
when they see the word "Unix"... :-)

Anyway, it is a good idea. Linux can  be installed as a full-featured
Un*x system, but the kernel can be used in many other ways...

--
 linux linux linux linux -[ cougnenc@renux.frmug.fr.net ]- linux linux linux 

------------------------------

From: ac3slh@sunc.sheffield.ac.uk (Stuart Herbert)
Subject: Re: Can't compile latest ncurses for Linux
Date: 3 Jun 1994 16:03:32 GMT

Sam A Gentile (sgentile@world.std.com) wrote:
: I can't compile the latest ncurses package for Linux. I run Configure, select

I downloaded ncurses 1.8.5 from ftp.netcom.com, in /pub/zmbenhal/ncurses
and this compiled no problems at all (I'm running slackware 1.2).

The docs do say that if you're using SLS, you should remove any previous
version of ncurses (not that this should affect actual compilation ...)

Stuart
--
Stuart Herbert -- S.Herbert@shef.ac.uk

------------------------------

From: snyder@ee.ee.duke.edu (Scott Snyder)
Subject: Re: 1.1.15 breaks SCSI
Date: 3 Jun 1994 04:20:11 GMT

I have a Pro Audio Spectrum - 16 SCSI -2 interface and consistantly get
the scsi0: reseting for second half of retries error message with a hang.

I'm runing Slackware 1.2 and wnt to know what driver update is safe for me
to try so that I can boot Linux and have it see my Sanyo drive. 
(It does boot with the scsi cable disconnected from the CDROM drive.)

Thanks for your help,

Since this may not be of help to everyone perhaps response by email would
be preferable.

Scott snyder@ee.duke.edu


Tim Cutts (tjrc1@cus.cam.ac.uk) wrote:
: ericy@cais.com (Eric Youngdale) writes:

: >>> >The kernel 1.1.15 does not work at all with my Adaptec 1540B.
: >>> >I get "Unable to reset SCSI host 0, probably a SCSI bus hang."

: >     OK, I think I understand what is going on.  I have come up with 
: >some patches which should correct this and a few other minor nits that 
: >have recently been introduced, and I have uploaded this to tsx-11 in 
: >pub/linux/ALPHA/scsi/scsifix-1.1.17.diff.gz.   If you experienced any 
: >trouble with 1.1.16/1.1.17, please try this and see if it helps.

: >     I have not sent this to Linus yet - I would like some direct 
: >confirmation that it actually helps first :-).

: This patch works, people (at least, it did for me!)  Eric for Linux Hero of
: the Day!

: Tim.

------------------------------

Subject: Re: Frustrated with new k
From: rick.emerson@dscmail.com (Rick Emerson)
Date: Fri,  3 Jun 94 10:28:00 -0640

 @SUBJECT:Re: Frustrated with new kernels                             N
LJ>  Someone posts complaining about the fact that Linux Kernels
LJ> and versions of Libc change very often, often breaking existing
LJ> programs or not working at all.
LJ> 
LJ>  My first response is to tell you to look at the rest of the
LJ> industry: If you want a stable, backwards compatible configuration,
LJ> spend several million dollars and buy an IBM Mainframe running VM/ESA;
LJ> they're backwards compatible to 1965 (maybe earlier).  
LJ>
LJ>  If you want Unix, spend several thousand dollars and get a
LJ> Sparcstation.  Keep in mind that the upgrade from SunOS 4.x to Solaris
LJ> broke almost every program in existence.  Or you could spend several
LJ> tens of thousands of dollars and get an RS/6000.  Keep in mind,
LJ> though, that the upgrade from AIX 3.1.x to 3.2 broke a lot of things
LJ> also.  Maybe you'd find Ultrix more stable (last I knew, Ultrix
LJ> upgrades didn't break lots of things) but keep in mind that many
LJ> analysts think that DEC is going down the tubes, so you may be stuck
LJ> with no upgrades at all.
LJ> 
LJ>  In short, you'd be hard pressed to find a stable version of
LJ> Unix from a commercial vendor.  HPUX strikes me as rather stable, as
LJ> does Irix, but you pay through the nose for SGI's or HP's.  Then
LJ> there's SCO...
LJ> 
LJ>  Second, you get what you pay for.  Linux Kernels and software
LJ> are developed by very skilled people who have donated their spare time
LJ> to the concept of free software.  They are doing their best, with
LJ> limited resources and limited time and zero budget, to create a really
LJ> good package.  If you want a stable version of Linux, find a kernel
LJ> version and a distribution you like, install them, make sure all your
LJ> binaries work, and then don't upgrade.
LJ> 
LJ>  Lastly, Linux is still very young and very much in the
LJ> development stage.  When you install Linux on your machine, the
LJ> documentation alone should be indication enough that there is still
LJ> much development going on.  Not all of this development will be
LJ> backwards compatible.  Not all of it will be filtered down to a level
LJ> where the non-developers can understand it.  I certainly don't
LJ> understand the point of a number of the Kernel mods, or why Libc 4.5
LJ> couldn't be backwards compatible.  I have faith in the people
LJ> developing Linux that they know what they're doing.  If you don't have
LJ> that kind of faith, buy a time-tested commercial package and then take
LJ> your chances with the vendor.
LJ> --
LJ> Lee Silverman, Brown class of '94, Brown GeoPhysics ScM '95
LJ> Email to: Lee_Silverman@brown.edu
LJ> Phish-Net Archivist: phish-archives@phish.net
LJ> "Nonsense - you only say it's impossible because nobody's ever done it."
LJ> 

I quote above verbatim because this is the sort of reply that doesn't
advance the discussion and misses the point of the question.  To be
quite clear, I have nothing against the author quoted here and this is
not meant to flame him; I'm using this article as an example of
several replies seen here and in e-mail.

No one with any sense is expecting *any* Linux kernel to be perfect.
For that matter, looking for bug-free code in anything much beyond
"Hello, world" is a futile effort.  Linux' track record is, in fact,
as good if not better than commercial products with a far longer
development history.  Even the 0.99 versions of the kernel were (and
are) surprisingly robust.  There was a thread in c.o.l.x on
applications using Linux in Antarctica, in shipboard environments, and
for hospital support; all of these are applications where reliability
is paramount.  Linux clearly delivered the goods.

We all know there are inefficiencies, minor buglets, and probably a
couple of real Bugs waiting in the code.  We all want to be sure of
getting the best performance possible, so we look for later versions
of the kernel (and other packages for that matter).

The point of the initial question is "how can I recognize a later
version of the kernel that's reliable?"  This acknowledges that life
on the leading edge is risky and asks for a version that's newer but
not too new to reasonably trust.

The current answer seems to be "trust 1.0.x and use 1.1.x at your
risk."  Several commentators, and the above is an example, take the
question to be some sort of assault on Linux' reputation or complaint
that Linix somehow should be much better than it is.  There are always
going to be whiners; my recommendation is to let them whine on their
own.  Just answer the question, please, and move on.

Rick
  
...
 * ATP/Linux 1.42 * To excel at what you do, you must love doing it.

------------------------------

Subject: Re: Frustrated with new k
From: rick.emerson@dscmail.com (Rick Emerson)
Date: Fri,  3 Jun 94 10:28:00 -0640

JA> In article <1.8643.2382.0N27CD41@dscmail.com>,
JA> Rick Emerson <rick.emerson@dscmail.com> wrote:
JA> 
JA> >Anyway, the original post points out an ongoing problem: Where is the
JA> >latest *reliable* kernel?  Aside from the 1.0 announcement, I can't
JA> >recall another notice saying, in effect, "OK, gang, this kernel works
JA> >pretty well and won't surprise you."  
JA> 
JA> I have seen an announcement by L Torvalds which says that v1.1.x are
JA> leading edge kernels, while v1.0.y are stable. From that, one could
JA> conclude that the most stable kernel is the v1.0 version with the
JA> highest patch level (9, is it?).

That's the sense I got.  But notice that trying to recall which was
the last 1.0.x version is a chore.  An occasional note in the
announcements newsgroup would be helpful here.  Trying to find out just
what's been changed in the new kernel patches has always been tough.

JA> >So it *is* frustrating to think there's a later version of the kernel
JA> >that might be a little quicker or cleaner but not [know] which one it is.
                                                       ^^^^^^
                            Sigh... gotta proofread my stuff more carefully!

JA> What is frustrating here is that you _know_ that there exists a kernel
JA> with more features than yours, not necessarily cleaner
JA> (whatever that means) or more stable. If linux were commercial, you
JA> would not have that temptation to get the latest alpha/beta version,
JA> because it would simply not be available.
JA>
JA> A good strategy is maybe to stick with one distribution, maybe with a
JA> commercial one (so that the distributor has something to lose), and
JA> update the system when new versions of that distribution becomes
JA> available. I don't know what is the state of present distributions
JA> considering updating the system without destroying everything.  I have
JA> used this strategy since linux 0.2 (believe me things are easy
JA> nowadays... :-)) with a lot of all kind of piecewise updating as long
JA> as I'm able to. Now we have 1.1.17 in all three machines because it
JA> has better SLIP support & better buffer caching. It is stable enough.

Well, yes and no...  I think most people think or suspect there's a
super version of [your favorite program here] in the wings.  If you
could only get it life would be perfect... until the next version.
But the changes you mention for 1.1.17 address two very specific areas
I'm having problems with; I need a better SLIP engine and my IDE's
need all the buffering help they can get.  Now, do I bet the farm on
this version of the kernel (my Linux box earns its keep as a mail
system) or do I hope someday someone says Vx.x.x is now stable and has
better SLIP and buffer caching?  And who's going to say that because
this sort of news isn't being posted all that clearly or regularly and
*that* is the real frustration in trying to keep up with kernel changes.

JA> This still takes a lot of hacking time, and sometimes I think it is
JA> not worth it (maybe addiction is the most real thing here :)). But you
JA> surely learn something, too.
JA> 

Where Linux is being used as a hacker's toy, this makes sense but (and
this is a tribute to Linux) where it's being used in "real world"
applications, there's a need for a clearer sense of how the kernel is
progressing.  I share the initial poster's frustration that this isn't
happening.  

Rick  
...
 * ATP/Linux 1.42 * Why look here, the message is up there...


------------------------------

From: art@lsr.nei.nih.gov (Art Hays)
Subject: Re: Why is Linux monolithic?
Date: Fri, 3 Jun 1994 16:13:44 GMT

Elizabeth Atkins (eatkins@alpha.wright.edu) wrote:
: In article <2rt19o$3mil@ns2.cc.lehigh.edu>,
: DAVID L. JOHNSON <dlj0@ns2.CC.Lehigh.EDU> wrote:
: >Perhaps you should re-think posting it at all, since people are developing
: >loadable drivers, in particular, SLS 1.05's kernel is modular.
: >
: >OTOH, the claims of microkernel arrangements are a little far-fetched in 
: >general.  I forget the source, but some afficionato was discussing a
: >microkernel system (could it be NT?) that had this incredibly small kernel.
: >Only thing was, one needed to add modules to drive the disk, the keyboard,
: >and the monitor.  It would seem that there is little advantage in making
: >such device drivers loadable (maybe the monitor to lighten the load on
: >a headless server).  
: >
: the Dr. Dobbs that had the interview with Linus had several articles on 
: up and coming OS's.  QNX was one object oriented microkernel where everything
: was loadable.  i believe they were also talking about either taligent or the
: new OS/2 that would be using a version of the mach kernel with loadalbe
: modules to emulate other OS's.  but i bought it just for the interview with
: linus.

Perhaps one advantage of a microkernel is for realtime?  My understanding of the
80X86 (please correct me if I'm wrong, I'm very weak on the 80X86 architecture)
is that it has only one external interrupt
pin, and one interrupt enable flag bit in the EFLAGS register.  This makes
realtime implementation difficult.  Every time the kernel must protect against
interrupts it masks them all (the CLI and STI instructions).  For comparison,
on the pdpd11 the driver for the serial line would only block interrupts
at level 4 or lower (the serial interface interrupted at priority level 4).  
Interrupts from higher priority level devices (levels 5-7) could still preempt.

With a microkernel I imagine it would be easier to reduce the amount
of time all interrupts would have to be blocked, for a lot of code
that would be running in the kernel with a monolithic kernel is now 
out in user space.


--
Art Hays
National Eye Institute, National Institutes of Health
Bethesda, MD  20892
art@lsr.nei.nih.gov     (301) 496-7143

------------------------------

Subject: Re: Frustrated with new k
From: rick.emerson@dscmail.com (Rick Emerson)
Date: Thu,  2 Jun 94 12:18:00 -0640

 @SUBJECT:Re: Frustrated with new kernels                             N
PE> In <2sje4b$ntu@charm.magnus.acs.ohio-state.edu>
PE> ewalton@magnus.acs.ohio-state.edu (Elaine Walton) writes:
PE> 
PE> >I have tried to compile the new kernels from off of tsx-11--no luck.
PE> >I am really confused how others are able to compile, yet I am not.
PE> >Maybe it's because I'm trying to go from 1.0.0 directly to 1.1.13-6?
PE> >Either way, I have pretty much given up..
PE> 
PE> Many people also have pretty much given up replying to vague complaints
PE> like that.
PE> 
PE> Rob

Well, *you* replied... Hmmm???  

Anyway, the original post points out an ongoing problem: Where is the
latest *reliable* kernel?  Aside from the 1.0 announcement, I can't
recall another notice saying, in effect, "OK, gang, this kernel works
pretty well and won't surprise you."  

So it *is* frustrating to think there's a later version of the kernel
that might be a little quicker or cleaner but not which one it is.

Rick
  
...
 * ATP/Linux 1.42 * Where law ends, there tyranny begins.



------------------------------

From: daniel@ing.puc.cl (Daniel Hirschberg)
Subject: [Q] tracing I/O port activity ?
Date: Fri, 27 May 1994 06:14:56 GMT

I'd like to write a device driver for an internal UPS card I own, but there is
no info about it specs so I would like to trace it's activity through
the I/O ports it reports in use... in order to figure out the way in
works!


Is there any way for doing this ?

any help is really appreciated... the card works great under DOS... but
DOS... you know!

by the way... does anybody know a way to reach a brand "The Complete PC"
to try to get the specs for their Complete Answering Machine voice mail
solution... It's really nice... that woud make possible a voice mail/email
gateway... real nice!!

best regards,
--
       _______________________
      /\                      \
      \_| Daniel Hirschberg J. |
        | daniel@ing.puc.cl    |
        | Santiago, CHILE.     |
        |   ___________________|_
         \_/____________________/

PS: if possible send any answer via email... the local news server is a
week late... but their working to bring it up to date :-) 

------------------------------


** 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
******************************
