From owner-linux-activists@joker.cs.hut.fi Thu Dec 10 04:53:41 1992
Status: RO
X-VM-v5-Data: ([nil nil nil nil nil nil nil nil nil]
["1799" "Fri" "4" "December" "1992" "08:25:34" "+0200"
"Alex C. Liu" "aliu@chaph.usc.edu " nil "49" "Porting to amiga" "^From:" nil nil "12"])
Received: from joker.cs.hut.fi by hutcs.cs.hut.fi with SMTP id AA00775
(5.65c8/HUTCS-S 1.4 for < arl@cs.hut.fi>); Thu, 10 Dec 1992 04:53:38 +0200
Received: from joker.cs.hut.fi by niksula.hut.fi id <61699-2>;
Thu, 10 Dec 1992 04:53:08 +0200
Received: from santra.hut.fi ([130.233.224.1])
by niksula.hut.fi with SMTP id <64857-4>; Fri, 4 Dec 1992 11:25:16 +0200
Received: from chaph.usc.edu by santra.hut.fi
(5.65c/8.0/TeKoLa) id AA12777; Fri, 4 Dec 1992 08:27:24 +0200
Received: from aludra.usc.edu by chaph.usc.edu (4.1/SMI-4.1+ucs-3.0)
id AA29884; Thu, 3 Dec 92 22:26:05 PST
Received: by aludra.usc.edu (4.1/SMI-4.1+ucs-3.0)
id AA25304; Thu, 3 Dec 92 22:26:02 PST
Message-Id: <9212040626.AA25304@aludra.usc.edu>
Sender: owner-linux-activists@joker.cs.hut.fi
X-Note1: Remember to put 'X-Mn-Key: normal' to your mail body or header
From: aliu@chaph.usc.edu (Alex C. Liu)
To: linux-activists@joker.cs.hut.fi
Subject: Porting to amiga
Date: Fri, 4 Dec 1992 08:25:34 +0200
X-Mn-Key: 680X0
Q: Who do you request the test kernel boot code from?
Also, I think the reason why everybody has been so quiet here is
because the EASY things have been done and starting up just seems soo
hard. The current kernel, 0.98p6 is getting bigger and better and
there are so many things to worry about.
I think what we should do instead of start porting 0.98p6, we should
try our hand on 0.10 first. (The first released kernel) or maybe
0.11.
There are several reasons why to do this.
For one, the size of the source tree is much smaller:
This is from the tsx-11.mit.edu:/pub/linux/attic/sources/system directory
134568 Nov 5 1991 linux-0.10.tar.Z
52672 Dec 16 1991 include-0.11.tar.Z
172438 Dec 16 1991 library-0.11.tar.Z
143365 Dec 8 1991 linux-0.11.tar.Z
198747 Jan 14 1992 linux-0.12.tar.Z
58309 Jan 15 1992 include-0.12.tar.Z
216777 Mar 9 1992 linux-0.95.tar.Z
321243 May 12 1992 linux-0.96.tar.Z
444413 Aug 1 09:23 linux-0.97.tar.Z
704363 Nov 9 01:34 linux-0.98.4.tar.Z
As you can see, the kernel size has been increasing, and linux0.98p6
is about 5 times bigger than 0.10.
Also, since the older kernel has less features, there are less things
to port in order to get an useable system.
I propose using 0.10 or 0.11 (I dont know what are the differences
between them) mainly because those versions do not support swapping
yet. Trying to port a 0.98 linux is hard because you MUST have a
complete filesystem to swap and to do demand loading.
What do you guys think? Should we start from 0.12 instead. (That
adds swapping but doesn't have shared executables yet, I think...)
We should also find somebody willing to archive all the LInux versions
between our starting version to the current so we can upgrade
incrementally our port.
From owner-linux-activists@joker.cs.hut.fi Mon Dec 14 05:21:01 1992
Status: RO
X-VM-v5-Data: ([nil nil nil nil nil nil nil nil nil]
["14714" "Mon" "14" "December" "1992" "05:20:06"
"+0200" "\"Linux Activists\"" "linux-activists@joker.cs.hut.fi" nil "377"
"Linux-Activists - Channel 680X0 digest. 92-11-13" "^From:" nil nil "12"])
Received: from joker.cs.hut.fi by hutcs.cs.hut.fi with SMTP id AA07657
(5.65c8/HUTCS-S 1.4 for < arl@cs.hut.fi>); Mon, 14 Dec 1992 05:20:57 +0159
Received: from joker.cs.hut.fi by niksula.hut.fi id <61549-2>;
Mon, 14 Dec 1992 05:20:10 +0200
Reply-To: "Linux-Activists" < linux-activists@joker.cs.hut.fi>
X-Note1: Remember to put 'X-Mn-Key: 680X0' to your mail body or header
Sender: owner-linux-activists@joker.cs.hut.fi
Message-Id: <92Dec14.052010eet.61549-2@niksula.hut.fi>
From: "Linux Activists" < linux-activists@joker.cs.hut.fi>
To: "Linux-Activists" < linux-activists@joker.cs.hut.fi>
Subject: Linux-Activists - Channel 680X0 digest. 92-11-13
Date: Mon, 14 Dec 1992 05:20:06 +0200
X-Mn-Key: 680X0
Topics:
680X0-port progress report?
Porting to amiga
Re: Porting to amiga
Basing the first 680x0 version on early 80x86 versions
linux on Mac??
Re: Porting to amiga
Porting to amiga
----------------------------------------------------------------------
From: "Michael (M.W.) Norman" < mwnorman@bnr.ca>
Subject: 680X0-port progress report?
Date: Wed, 9 Dec 1992 13:32:46 +0200
Hi! I'm still new to this sort of stuff (mail servers, relays, etc)
I hope I don't goof up! Anyway, is there a summary/FAQ/progress
report on the 680X0-port of Linux? I've been following the main
Linux Usenet group for about a year now, but I'm probably going
to be getting a Macintosh computer in the new year. So, if
there is ANYTHING to report, could someone please mail it to
me at mwnorman@bnr.ca? Thanks in advance,
------------
Mike Norman,c/o Bell-Northern Research
P.O. Box 3511 Station 'C' Ottawa
Ontario Canada K1Y 4H7 M/S 115 Internet: mwnorman@bnr.ca
tel: (613) 763-7717 (FAX:613-763-7742)
"Hasta la vista, Baby!" A. Schwarzenegger
------------------------------
From: johnsonm@stolaf.edu (Michael K Johnson)
Subject: Porting to amiga
Date: Thu, 10 Dec 1992 05:41:36 +0200
Q: Who do you request the test kernel boot code from?
harp@netcom.com
Also, I think the reason why everybody has been so quiet here is
because the EASY things have been done and starting up just seems soo
hard. The current kernel, 0.98p6 is getting bigger and better and
there are so many things to worry about.
But much of it does not have to be included. It is not very hard to
exclude all but one filesystem, leave out the tcp/ip code, etc.
I think what we should do instead of start porting 0.98p6, we should
try our hand on 0.10 first. (The first released kernel) or maybe
0.11.
There are several reasons why to do this.
There are also several reasons why not: The best is: These releases
were buggy. Really. The buffer-cache code, for instance, had race
conditions, even in version .11, and I think it was .95 before the
mismatched ys and os links disappeared. Such things are /not/
features. Furthermore, several of the interfaces have changed since
then.
As you can see, the kernel size has been increasing, and linux0.98p6
is about 5 times bigger than 0.10.
True, but you don't need to deal with it all at first, and you want to
use the newer interfaces.
Also, since the older kernel has less features, there are less things
to port in order to get an useable system.
And many more bugs to fix.
I propose using 0.10 or 0.11 (I dont know what are the differences
between them) mainly because those versions do not support swapping
yet. Trying to port a 0.98 linux is hard because you MUST have a
complete filesystem to swap and to do demand loading.
Not really. You are writing your own memory management, so this is
not true. You only do what you can. However, I would suggest
starting out with demand loading, as it is not too hard (that's why it
was one of the first nifty features to get added to linux) and makes
it easier to move the rest of the kernel over. Swapping can
/definately/ wait, as it is perfectly fine even now to run linux
without swapping. Just put in hooks, or at least leave space.
That's not the difficult part.
What do you guys think? Should we start from 0.12 instead. (That
adds swapping but doesn't have shared executables yet, I think...)
Shared copy-on-write text was in .12, I am quite sure. Dynamic
loading came even earlier, if my memory serves me correctly.
We should also find somebody willing to archive all the LInux versions
between our starting version to the current so we can upgrade
incrementally our port.
This is too much boondoggleing. Just start by writing a memory
manager (no need for a filesystem yet) and get a few tasks going.
Then add hooks for a filesystem that are compatible with the linux
vfs. At the same time, someone should be porting the assembly in the
386 version to 680x0 assembly, and finishing off the basic device
drivers. You will want to make sure that the device drivers follow
the linux standard interface -- read any device driver to find the
tabular structure.
Then take the effort to merge it into the ported rest of the kernel,
and get it working there. Then start applying the patches by hand to
get to the latest release, but don't make unnecessary work for
yourselves by starting with an old buggy kernel.
(not that .12 wasn't great, but too many bugs have been fixed since to
bother playing with it now.)
I suppose my suggestions don't mean much, since I don't even have an
amiga, and so can't do much writing. However, I have major
modification plans of my own in mind, and have at least thought
whereof I speak...
michaelkjohnson
------------------------------
From: "Gregory O. Harp" < harp@netcom.com>
Subject: Re: Porting to amiga
Date: Thu, 10 Dec 1992 09:01:00 +0200
Alex writes:
>Q: Who do you request the test kernel boot code from?
Me. ;) Do you have your system running now? I recall that you had
some problems and therefore dropped out of the development team. Or
is that brain damage on my part? ;)
>Also, I think the reason why everybody has been so quiet here is
>because the EASY things have been done and starting up just seems soo
>hard. The current kernel, 0.98p6 is getting bigger and better and
>there are so many things to worry about.
Well, it's not necessarily that the east parts are done, but that the
next step is rather hard to determine. In developing one part it
appears that we'd almost have to develop everything (at least all the
low level parts) at the same time. That could get rather nasty, and
I'm not even sure that any of us has the kind of spare time it takes
to pull it off.
>I propose using 0.10 or 0.11 (I dont know what are the differences
>between them) mainly because those versions do not support swapping
>yet. Trying to port a 0.98 linux is hard because you MUST have a
>complete filesystem to swap and to do demand loading.
>
>What do you guys think? Should we start from 0.12 instead. (That
>adds swapping but doesn't have shared executables yet, I think...)
Perhaps we should stick to the basics. After all, in order to add
swapping we have to have the FS and the rest of the MM code done. In
other words, perhaps we should start with 0.11 or 0.10, then once that
is going we should work on the diffs between that and 0.12. Unless
something else has radically changed between the versions, this should
be a feasible way to go about it.
>We should also find somebody willing to archive all the LInux versions
>between our starting version to the current so we can upgrade
>incrementally our port.
That would be nice...
At the moment I _think_ I might have some time to get to it this
weekend. If so, I'll grab either 0.10 or 0.11 (or both) and have a
look. For that matter, I might grab several of the early versions.
I'm down to one monitor for my two machines (486 and Amiga) right now,
and getting that taken care of is sort of a priority right now...
FMS, what do you think about getting "everything else" such as it is
running before adding the VM code? It could help with your version of
the system, since that'd be one less part you'd have to take back out
to make a non-MMU version.
--Greg
------------------------------
From: "Gregory O. Harp" < harp@netcom.com>
Subject: Basing the first 680x0 version on early 80x86 versions
Date: Thu, 10 Dec 1992 14:54:05 +0200
I'm hoping this message gets out. There appear to be some problems at
the site where the messages are distributed. I'm not sure I totally
trust the digested message delivery system they've apparently started
using.
After having a bit of a look at the sources for the 386 kernel
versions 0.11 and 0.12 I think it would be a good idea, as Alex
suggested, to build the original port from these sources rather than
the current release.
There are a few inconsistencies in the source trees that we will have
to deal with, but it doesn't appear to be a real problem. I should
have a chance to spend more time with the sources this weekend (I
_think_).
So, unless anyone has any real objections, I'd recommend we start with
the 0.11 sources. If you want to look at them yourself, they're at
tsx-11.mit.edu in the /pub/linux/attic/sources/system directory.
You'll need all three files: linux-0.11.tar.Z, library-0.11.tar.Z, and
include-0.11.tar.Z. These filenames are typed from memory, so they
may be slightly off, but that should be close enough. Just 'mget'-ing
*.11.tar.Z should be sufficient.
Just FYI, I didn't attempt to confirm the status of things like VM in
version 0.11 versus 0.12. I'm taking it on faith right now that
swapping VM is the main difference, other than obvious bug fixes.
--Greg
------------------------------
From: "Michael (M.W.) Norman" < mwnorman@bnr.ca>
Subject: linux on Mac??
Date: Thu, 10 Dec 1992 13:33:46 +0200
Has anyone thought about trying a Macintosh 'port' of
linux? Most Macs these days are 68030-based so there
is MMU-functionality available. I've never hacked a
kernel, what sort of difficulty am I in for if I try
this?
Thanks in advance,
------------
Mike Norman,c/o Bell-Northern Research
P.O. Box 3511 Station 'C' Ottawa
Ontario Canada K1Y 4H7 M/S 115 Internet: mwnorman@bnr.ca
tel: (613) 763-7717 (FAX:613-763-7742)
"Hasta la vista, Baby!" A. Schwarzenegger
------------------------------
From: dwilliam@jabba.ess.harris.com (Dave Williams)
Subject: Re: Porting to amiga
Date: Thu, 10 Dec 1992 21:35:14 +0200
> Q: Who do you request the test kernel boot code from?
Greg Harp is the maintainer - He's at harp@net.com
What do you intend to test? There's not much there.
> Also, I think the reason why everybody has been so quiet here is
> because the EASY things have been done and starting up just seems soo
> hard. The current kernel, 0.98p6 is getting bigger and better and
> there are so many things to worry about.
This is true, but we haven't even used *any* of the code out of
the real kernel sources yet. At this point, it dosen't matter how
complex the real kernel is - we don't have the underpinnings there
yet to support any of it. (Ex - task scheduler & switcher, console
driver, etc.)
Part of the reason that things have been so quiet around here is
the fact that bums like me aren't doing their jobs, and getting code
up and running. (I'm currently doing my *real* job, catching back
up... Not to mention my machine is doing wierd things, but GCC works
correctly now. (finally) :-) Since we don't have a nifty wowser
kernel to send out, people aren't excited about us. Oh well.
> I think what we should do instead of start porting 0.98p6, we should
> try our hand on 0.10 first. (The first released kernel) or maybe
> 0.11.
Hummmm. Now there's an idea.
> There are several reasons why to do this.
>
> For one, the size of the source tree is much smaller:
[sizes from tsx deleted to save space]
> As you can see, the kernel size has been increasing, and linux0.98p6
> is about 5 times bigger than 0.10.
>
> Also, since the older kernel has less features, there are less things
> to port in order to get an useable system.
>
> I propose using 0.10 or 0.11 (I dont know what are the differences
> between them) mainly because those versions do not support swapping
> yet. Trying to port a 0.98 linux is hard because you MUST have a
> complete filesystem to swap and to do demand loading.
>
> What do you guys think? Should we start from 0.12 instead. (That
> adds swapping but doesn't have shared executables yet, I think...)
On the surface, this looks like a great idea. However, I think
we shouldn't go this route - The reason is that once we have most of
the driver code completed, we can simply use the kernel source as-is.
The high-level features of the 0.98 kernel (such as file system, TCP
support, etc., etc., etc.) should be useable as-is. We need to
concentrate on doing the low-level stuff first, such as all the drivers.
Ex - Before we can even think about a file system, we need the hd.c
driver. But once that's up, we just copy the file system code from
whatever version i386 Linux is up to. If we ported our hd.c so it
has all the same calls as the i386 version, nothing should have to
change. Now, I know this is taking a somewhat rosy-eyed view of the
situation, but I don't think I'm too far off of the mark.
Your idea has merit, since porting to the current i386 version
is difficult since it's a moving target. If we nail down the old
version and port it, we'll avoid problems porting to a changing
standard. However, I don't think we should go back as far as 0.10.
So much has changed between then and now, we'd have to do a complete
re-write once we *DID* get a kernel up and running. No, better to
take the current kernel, and simply strip out the parts that we
don't currently need (or support). As we write more low-level
drivers, we can graft the stuff we stripped out back in, one piece
at a time.
Do you guys agree with me, disagree, what?
Dave Williams | "What time is it?" "9:00AM"
dwilliam@jabba.ess.harris.com | "What day?" "Monday"
"Huh? What? Could you repeat the question?" | "Go away. Try me Tuesday"
< mumble mumble> opinions < mumble mumble> mine < mumble mumble mumble>
------------------------------
From: David Crooke < dcc@dcs.ed.ac.uk>
Subject: Porting to amiga
Date: Fri, 11 Dec 1992 15:36:52 +0200
There are two sides to this coin...it might be easier to port an old
beta version but we don't want to get in the situation of being miles
behind the PeeCee version. Eventually it would be nice to be able to
upgrade from PC Linux patches with the higher level stuff.
I'm sure plenty people would be prepared to archive, I certainly could
stash *some* stuff...
Dave
P.S. How's the kernel coming? Any chance of some stuff to start
hacking drivers on? Also, I take it that we will be supporting the 040
MMU and the IDE/AGA stuff (for 4000 (and 1200? - someone in the US has
apparently announced a 1200 upgrade with 65881 (unreliable source) -
imagine that, Unix on a 400 quid machine!))
------------------------------------------------------------------------------
David Crooke, Dept. of Computer Science, Uni. of Edinburgh. dcc@dcs.ed.ac.uk
Home: 22a Montgomery St., Edinburgh, SCOTLAND EH7 5JS. Tel: 031 556 0265
Work: JCMB Rm 3310, King's Bldgs, W Mains Rd., Edinburgh EH9 3JZ. 031 650 6013
------------------------------
End of 680X0 Digest
*******************
-------
From owner-linux-activists@joker.cs.hut.fi Sun Dec 20 20:31:34 1992
Status: RO
X-VM-v5-Data: ([nil nil nil nil nil nil nil nil nil]
["5196" "Sun" "20" "December" "1992" "20:30:41" "+0200"
"\"Linux Activists\"" "linux-activists@joker.cs.hut.fi" nil "138"
"Linux-Activists - Channel 680X0 digest. 92-11-20-10:5" "^From:" nil nil "12"])
Received: from joker.cs.hut.fi by hutcs.cs.hut.fi with SMTP id AA15715
(5.65c8/HUTCS-S 1.4 for < arl@cs.hut.fi>);
Sun, 20 Dec 1992 20:31:31 +0200
Received: from joker.cs.hut.fi by niksula.hut.fi id <61851-1>;
Sun, 20 Dec 1992 20:30:41 +0200
Reply-To: "Linux-Activists" < linux-activists@joker.cs.hut.fi>
X-Note1: Remember to put 'X-Mn-Key: 680X0' to your mail body or header
Sender: owner-linux-activists@joker.cs.hut.fi
Message-Id: <92Dec20.203041eet.61851-1@niksula.hut.fi>
From: "Linux Activists" < linux-activists@joker.cs.hut.fi>
To: "Linux-Activists" < linux-activists@joker.cs.hut.fi>
Subject: Linux-Activists - Channel 680X0 digest. 92-11-20-10:5
Date: Sun, 20 Dec 1992 20:30:41 +0200
X-Mn-Key: 680X0
Topics:
Re: ATARI port
re:ATARI port...
----------------------------------------------------------------------
From: < fms@ccgr.technion.ac.il>
Subject: < none>
Date: Sun, 20 Dec 1992 11:14:22 +0200
[ Hi all, sorry for the dealy in answering mails. I've been in transition,
so to speak. I'm answering a rather big packet of mail now (including
the digest)]
On Nov 25, 9:20pm, David Crooke wrote:
: In view of the increased interest in non-Amiga platforms, does
: everyone consider it worth trying for binary compatibility (of
: applications, not the kernel :-) across all 680X0 platforms, even if
: we must differ from all existing standards to do so? Given that we are
: assuming VM I can't see this posing any problems. Perhaps it is worth
: looking at software and support and choosing one of Amix/AUX/etc. to
: go with.
Going for binary compataility seems like a great idea (And easy to implement).
Most binaries are based on system calls, and since these are compatible (You
get my drift.). A good example for this "feature" is the 68000-Minix, which
allows it (Mac binaries on Amiga/ST and vice-verse, as long as you have the
right CPU version).
Trying to get binary compatability with Amix/AUX is not practical, as we are
dealing with different *nix variants here.
: Kernel guys, how goes? I am keen to get stuck in with some driver
: stuff etc. If no-one else is into it, I have a bit of experience with
: X and am happy to put my head on the block for porting the MIT X
: server - then again this should be trivial if CBM are willing to part
: with their driver sources.
I think you'll have to wait with X. At least until you have something to test
your code with. As for CBM's X11 server. It is a standard X11R4 server, along
with some display drivers for the Amiga. I doubt CBM would be willing to give
us the sources for the drivers without some hefty fee.
--
-- Fishman Shmuel Technion -- Israel Institue of Technology
-- fms@ccgr.technion.ac.il Taub Computer center.
-- fms@technion.BITNET
-- FMS@irc "Complete and balanced -- back from the dead"
------------------------------
From: < fms@ccgr.technion.ac.il>
Subject: Re: ATARI port
Date: Sun, 20 Dec 1992 11:32:53 +0200
On Nov 26, 10:53am, "Gregory O. Harp" wrote:
[ General discussion of autovectoring removed ]
: >Or perhaps it would be better to have the low-level ISR stuff be
: >completely machine dependent, and have the low level ISRs call higher
: >level routines based on the actual device interrupting.
:
: I think it's better to separate these for each machine. Using tons of
: #ifdef's would make the code quite unreadable. It's not a horribly
: large amount of code, especially compared to the drivers, which will
: all have to be machine-dependant anyway. In the case of the general
: OS routines that call the ISR code, conditional compilation based on
: the system is the way to go, IMO.
I don't see any need to #ifdef's. The low level stuff (ie. Hardware accessing
stuff) would be located in different files, under different directories (Does
Alex' dir structure proposal come to mind ? Yes, it does ;-) ). All the low-
level stuff needs to do is preper the info for the higher levels, and setup
a signal to let the higher-levels the info is ready.
For exaple: SCSI devices: Only one routine to parse SCSI commands, and the low
levels just interface the drive itself. (Amiga people: Check SCSI-Direct to see
such an implementation).
Of couse, some stuff would still need to be left at the low levels, but I don't
think there's a lot that cannot be done at higher levels.
--
-- Fishman Shmuel Technion -- Israel Institue of Technology
-- fms@ccgr.technion.ac.il Taub Computer center.
-- fms@technion.BITNET
-- FMS@irc "Complete and balanced -- back from the dead"
------------------------------
From: < fms@ccgr.technion.ac.il>
Subject: re:ATARI port...
Date: Sun, 20 Dec 1992 11:55:33 +0200
On Nov 25, 4:29pm, "Hamish (H.I.) Macdonald" wrote:
: How much resolution do we need?
:
: I suspect that we want at least as low as 10ms for scheduling purposes
: and system time (i.e. the "big" timer would use 10ms), and that
: anything lower is probably unnecessary.
Well, 386-Linux runs it's timers at roughly 100HZ, (The scheduller is slower, I
think, though I'm not sure, I don't have the info before me now (Does anyone
remember of hand what frequency you get when you load the timer with the
value of 11931 ?). I think Mode 3 is continues with interrupt at reload).
: So do we really need two timers?
The general idea behind the two timers is not to cause too many interrupts when
givving higher percission timing. However, since 100 HZ should be enough for
most purposes, one timer might be enough.
--
-- Fishman Shmuel Technion -- Israel Institue of Technology
-- fms@ccgr.technion.ac.il Taub Computer center.
-- fms@technion.BITNET
-- FMS@irc "Complete and balanced -- back from the dead"
------------------------------
End of 680X0 Digest
*******************
-------
From owner-linux-activists@joker.cs.hut.fi Mon Dec 21 08:09:34 1992
Status: RO
X-VM-v5-Data: ([nil nil nil nil nil nil nil nil nil]
["1832" "Mon" "21" "December" "1992" "08:08:47" "+0200"
"\"Linux Activists\"" "linux-activists@joker.cs.hut.fi" nil "51"
"Linux-Activists - Channel 680X0 digest. 92-11-20-19:24" "^From:" nil nil "12"])
Received: from joker.cs.hut.fi by hutcs.cs.hut.fi with SMTP id AA18717
(5.65c8/HUTCS-S 1.4 for < arl@cs.hut.fi>); Mon, 21 Dec 1992 08:09:32 +0200
Received: from joker.cs.hut.fi by niksula.hut.fi id <61692-8>;
Mon, 21 Dec 1992 08:08:54 +0200
Reply-To: "Linux-Activists" < linux-activists@joker.cs.hut.fi>
X-Note1: Remember to put 'X-Mn-Key: 680X0' to your mail body or header
Sender: owner-linux-activists@joker.cs.hut.fi
Message-Id: <92Dec21.080854eet.61692-8@niksula.hut.fi>
From: "Linux Activists" < linux-activists@joker.cs.hut.fi>
To: "Linux-Activists" < linux-activists@joker.cs.hut.fi>
Subject: Linux-Activists - Channel 680X0 digest. 92-11-20-19:24
Date: Mon, 21 Dec 1992 08:08:47 +0200
X-Mn-Key: 680X0
Topics:
re:ATARI port...
----------------------------------------------------------------------
From: johnsonm@stolaf.edu (Michael K Johnson)
Subject: re:ATARI port...
Date: Sun, 20 Dec 1992 20:49:12 +0200
: I suspect that we want at least as low as 10ms for scheduling purposes
: and system time (i.e. the "big" timer would use 10ms), and that
: anything lower is probably unnecessary.
Well, 386-Linux runs it's timers at roughly 100HZ, (The scheduller is slower, I
think, though I'm not sure, I don't have the info before me now (Does anyone
remember of hand what frequency you get when you load the timer with the
value of 11931 ?). I think Mode 3 is continues with interrupt at reload).
386 linux runs its system timer at 100 HZ. The scheduler is not run at n
HZ, but rather whenever required. do_timer() (sched.c, I beleive)
gets called at 100 HZ, and when necessary, it calls schedule() (sched.c,
I know... ;-). schedule() also is called by all sorts of pieces. For
instance, my non-interrupt-driven parallel prot driver calls schedule
whenever the need_resched flag is set from interrupt-driven code, like
the serial port code (as I recall).
: So do we really need two timers?
The general idea behind the two timers is not to cause too many interrupts when
givving higher percission timing. However, since 100 HZ should be enough for
most purposes, one timer might be enough.
If you are doing real-time work, then multiple high precision timers
are a bonus. I don't think we need it here unless there are
brain-damaged pieces of hardware that require high-speed
high-precision clocks as interrupts from the system, and I have heard
that amiga hardware is pretty decent...
michaelkjohnson
------------------------------
End of 680X0 Digest
*******************
-------
From owner-linux-activists@joker.cs.hut.fi Tue Dec 22 20:30:01 1992
Status: RO
X-VM-v5-Data: ([nil nil nil nil nil nil nil nil nil]
["581" "Tue" "22" "December" "1992" "20:29:09" "+0200"
"\"Linux Activists\"" "linux-activists@joker.cs.hut.fi" nil "31"
"Linux-Activists - Channel 680X0 digest. 92-11-22-16:39" "^From:" nil nil "12"])
Received: from joker.cs.hut.fi by hutcs.cs.hut.fi with SMTP id AA00603
(5.65c8/HUTCS-S 1.4 for < arl@cs.hut.fi>); Tue, 22 Dec 1992 20:29:59 +0159
Received: from joker.cs.hut.fi by niksula.hut.fi id <61709-5>;
Tue, 22 Dec 1992 20:29:11 +0200
Reply-To: "Linux-Activists" < linux-activists@joker.cs.hut.fi>
X-Note1: Remember to put 'X-Mn-Key: 680X0' to your mail body or header
Sender: owner-linux-activists@joker.cs.hut.fi
Message-Id: <92Dec22.202911eet.61709-5@niksula.hut.fi>
From: "Linux Activists" < linux-activists@joker.cs.hut.fi>
To: "Linux-Activists" < linux-activists@joker.cs.hut.fi>
Subject: Linux-Activists - Channel 680X0 digest. 92-11-22-16:39
Date: Tue, 22 Dec 1992 20:29:09 +0200
X-Mn-Key: 680X0
Topics:
How far is AmiLinux?
----------------------------------------------------------------------
From: a73ae009@IS.TWI.TUDelft.NL (P.R. Vervoorn- beg Pompstra)
Subject: How far is AmiLinux?
Date: Tue, 22 Dec 1992 12:59:43 +0200
Hi, I just recently joined this channel, because I was wondering how
far the port of Linux to the Amiga is...
So, what has been ported and what Amigas does it work on?
Well, that's all, so have a merry christmas and a happy new year!!
Regards,
Patrick.
------------------------------
End of 680X0 Digest
*******************
-------