Subject: Linux-Development Digest #751
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:     Mon, 23 May 94 14:13:08 EDT

Linux-Development Digest #751, Volume #1         Mon, 23 May 94 14:13:08 EDT

Contents:
  Linux and PCMCIA Ethernet (F. David Sinn)
  Re: SIGHUP: We tried it! (Theodore Ts'o)
  Re: SIGHUP - Deep Kernal Guts question! (Rocco Caputo)
  Re: SIGHUP - Deep Kernal Guts question! (Theodore Ts'o)
  Re: Anybody working on BSD dump porting? (Malcolm Beattie)
  Re: 32-bit Novell desktop OS combines Unix, DOS 7 (Brandon S. Allbery)
  Trantor T348 MiniScsi Plus (J. Daniel Kulp)
  AZTECH CDA 268-01A DS CDROM - configuration? (John M. Collinson)
  Re: net3 tcp window sizes (NOT SEQUENCE NUMBERS!), Please read (Mark Evans)
  Re: umsdos for 1.1.12? (GLAUDE DAVID)
  Re: 32-bit Novell desktop OS combines Unix, DOS 7 (Kai Petzke)
  verdict on the NCR 83c510 drivers? (John Wagner)
  Re: Emacs 19.22: error on compile for linux (Byron Thomas Faber)
  Re: Linux and Bernoulli? (Bob Perkins)
  Re: PCI or VL bus SCSI Host Adapters (James Soutter)
  Re: 32-bit Novell desktop OS combines Unix, DOS 7 (Mark A. Davis)
  Re: Anybody working on BSD dump porting? (Michael Beirne)

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

From: dsinn@coho.halcyon.com (F. David Sinn)
Subject: Linux and PCMCIA Ethernet
Date: 23 May 1994 15:25:38 GMT

Is anyone, or does anyone know of someone working on PCMCIA ethernet
hooks for LINUX?  I would like to install Linux on a portable, but my
only Ethernet options are through the PCMCIA slot (since I already have
the card!).

If no one is working on such a project, could anyone point me in the
direction to find out how to write a ethernet driver for Linux, and I
will atempt if myself.

David Sinn
dsinn@halcyon.com

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

From: tytso@ATHENA.MIT.EDU (Theodore Ts'o)
Subject: Re: SIGHUP: We tried it!
Date: 23 May 1994 12:14:39 -0400
Reply-To: tytso@ATHENA.MIT.EDU (Theodore Ts'o)

   From: troc@loreli.ftl.fl.us (Rocco Caputo)
   Date: Sat, 21 May 1994 16:41:13 GMT

   Theodore Ts'o penned:
   > The "ivory tower tribe", huh?  I'll excuse you, and assume you don't
   > know how much work I've done working on the Linux kernel.  Some of us
   > just have had enough experience doing Unix work that we *know* when some
   > idea is half-baked, without having to try it.  :-)

   You'll excuse him because you're an Alpha and he's not.  You're allowed
   to make assumptions because you're an Alpha and he's not.  Some of us
   have enough experience doing Unix work, and that makes them Alphas.

   And some of us don't, so we get treated like garbage when we ask an
   honest question.  And maybe if the Alphas would give some straight
   answers from time to time, people wouldn't have to go around patching
   their kernels to find out why SIGHUP shouldn't be propagated past a
   session leader.

I gave several good reasons why SIGHUP shouldn't be propagated past a
session leader.  He just choose not to believe me, and called me an
"ivory tower type".  He's perfectly allowed not to believe me, but he
should expect some ribbing when he turns out to be wrong.

Actually, though, you should be lucky that us "Alpha's" bother to
respond at all.  Most mathematicians simply throw proofs that you can
square a circle with a compass and straightedge into the trash, unread.
I know --- we get a couple of them sent to postmaster@mit.edu every so
often, with people flaming on and on about how there's this great
conspiracy of the mathmaticians, since no one will even review their
work.  Apparently the concept that they're just idiots hasn't crossed
their minds, and that mathmaticians have better things to do than to
answer crackpots --- especially when the crackpots aren't even being
respectful.

     "Why does SIGHUP only go to session leaders?"
         "Because POSIX says so."

   Is that what folks around here consider a decent answer?

I also said

        * because it's more flexible; session leaders can decide whether
                or not to propagate signals down to their children.
        * because users occasionally want to run "make >& MAKEFILE" and
                then logout for the night, and expect to see results the
                next morning, as opposed to having the SIGHUP take out
                the make.
        * because the vast majority of the programs written for Unix
                assume the POSIX standard these days, and if you break
                the POSIX model, programs will break.

This last one turned out in fact to be true very quicky (mercifully) ---
as opposed to six months from now, in which case people wouldn't be
tearing out their hair trying to debug it --- although people wouldn't
believe me until they tried it out for themselves.  Oh, well...

                                                        - Ted

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

From: troc@loreli.ftl.fl.us (Rocco Caputo)
Subject: Re: SIGHUP - Deep Kernal Guts question!
Date: Mon, 23 May 1994 13:49:54 GMT

Leon Garde penned:
> yeah. a daemon could do it. but the kernel almost does the right job.
> what 'incompatibility' is there if it did ?

The most notable incompatibility with the POSIX specification (and remeber,
we tried this) is that some daemon authors are careless enough not to trap
and ignore terminating signals like SIGHUP.  They rely on POSIX-compliant
operating systems to leave them alone when the owner hangs up.

Previously "working" daemons suddenly became "broken" when the kernel
sent them notification of the user hanging up.  We've unpatched the
kernel to avoid having to continually change or work around weak
daemon programs.  Like a lot of other compliant administrators, we'll
have to spackle over a hole in the POSIX specifications with yet
another program.

On a multiuser system, this POSIX-compliant process leakage can, if
left alone, eventually take over a system.  The performance hit
will be even more noticeable if these processes go into tight input
loops.  This is not a problem?

--
-><- Signature.

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

From: tytso@ATHENA.MIT.EDU (Theodore Ts'o)
Subject: Re: SIGHUP - Deep Kernal Guts question!
Date: 23 May 1994 12:39:02 -0400
Reply-To: tytso@ATHENA.MIT.EDU (Theodore Ts'o)

   From: jmorriso@bogomips.ee.ubc.ca (John Paul Morrison)
   Date: Sun, 22 May 1994 06:09:40 GMT

   so 'more' got HUP'ed even though -hupcl was set (maybe that's because
   I'm trying with a pty. anyway, that's not the point. SIGHUP does get
   sent when the tty is closed (by the shell, so I'm told) and the
   process in this case is getting killed like it's supposed to).

hupcl has nothing to do with whether or not a process gets a SIGHUP ---
hupcl stands for "hangup on close", and controls whether or not a tty
should have its modem control lines dropped when the last process closes
a tty device.  hupcl has no effect on pty's, of course....

                                                - Ted

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

Crossposted-To: comp.os.linux.help
From: mbeattie@black.ox.ac.uk (Malcolm Beattie)
Subject: Re: Anybody working on BSD dump porting?
Date: Mon, 23 May 1994 16:31:49 GMT

In article <2rh32n$7fu@apollo.west.oic.com> dillon@apollo.west.oic.com (Matthew Dillon) writes:
>In article <2reh76$hfv@bosnia.pop.psu.edu> barr@pop.psu.edu (David Barr) writes:
>:In article <1994May18.210039.6822@belvedere.sbay.org>,
>:David E. Fox <root@belvedere.sbay.org> wrote:
>:>But what advantages does 'dump' have that 'tar' or 'cpio' do not?
>:
>:4. Does not affect atimes of backed up files
>
>    True.  This is not important for most people.
[I can't quite figure out who followed up to what here, so watch attributions.]

Affecting atime causes problems with people backing up /var/spool/mail
since biff and friends think new mail has arrived. However, GNU tar has
an option --atime-preserve which does exactly what you'd hope: leaves
access times alone while dumping files.

--Malcolm

-- 
Malcolm Beattie <mbeattie@black.ox.ac.uk>
Oxford University Computing Services
"Widget. It's got a widget. A lovely widget. A widget it has got." --Jack Dee

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

From: bsa@kf8nh.wariat.org (Brandon S. Allbery)
Subject: Re: 32-bit Novell desktop OS combines Unix, DOS 7
Date: Mon, 23 May 1994 16:09:27 GMT

In article <1994May23.105418.22438@cato.robots.ox.ac.uk>, jon@gtex02.us.es says:
+---------------
| The only way to prevent this would to put in the linux copywrite that
| the act of loading a module puts it under the GPL, but I'm not sure that
| we would want that.
+------------->8

According to RMS, the GPL *already* does this:  it covers, among other things,
interfaces specific to the GPLed program.  Or are you claiming the details of
the modules interface aren't specific to the Linux kernel?

++Brandon
-- 
Brandon S. Allbery         kf8nh@kf8nh.ampr.org          bsa@kf8nh.wariat.org
The FUDs at Microsoft are shouting "Kill The Wabi!"

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

From: dkulp@ccs.neu.edu (J. Daniel Kulp)
Subject: Trantor T348 MiniScsi Plus
Date: 23 May 1994 16:46:06 GMT

Does anyone know if a driver for the Trantor MiniScsi Plus 348 is being
developed.  I have a very small internal hard drive that I would like to put
Linux on but I woiuld like it to have access ot the large external hard drive
and cd-rom.  

Actually, If not, if anyone has specs on the adapter, in a few weeks my
classes would be over and I could devote some time to working on one.

Thanks.

-- 
J. Daniel Kulp                                       dkulp@ccs.neu.edu
Chemical Engineering Major at Northeastern University in Boston, Mass.


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

Crossposted-To: comp.os.linux.admin,comp.os.linux.help
From: csjmc@blaze.trentu.ca (John M. Collinson)
Subject: AZTECH CDA 268-01A DS CDROM - configuration?
Date: Mon, 23 May 1994 16:49:01 GMT

I have an AZTECH CDA 268-01A Double Speed CDROM drive.  I have been
trying to configure for a couple of days now.  Based on the manual and
the settings on the controller card this drive should be compatible with
the Mitsumi CDROM drive.  So I configured my system for a Mitsumi drive
including changing the memory address and irq settings in mcd.h and doing
a make config to include the mitsumi drivers in the kernal as well as
the iso9660 file system.  This all seemed to work properly.  However
when I reboot the kernel I get a message similar to:

mcd: mitsumi get version failed at 0x800

Has anyone had similar problems with this CDROM or am I going to have to
leave the drive dead until I can find someone who knows how to write a
driver for it?

Please reply to csjmc@blaze.trentu.ca

Thanks
John Collinson


-- 
============================================================================
|  John M. Collinson                      Internet: csjmc@blaze.trentu.ca  | 
|  Computer Studies and                              jcollinson@trentu.ca  |
|  Environmental Resource Studies              john.collinson@fleming.edu  |
|  Trent University, Peterborough, Ontario, CANADA                         |
============================================================================

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

From: evansmp@mb5194.aston.ac.uk (Mark Evans)
Subject: Re: net3 tcp window sizes (NOT SEQUENCE NUMBERS!), Please read
Date: Mon, 23 May 1994 16:48:42 GMT

Alan Cox (iialan@myhost.subdomain.domain) wrote:
: Tom Briggs (tbriggs@cutter.ship.edu) wrote:
: : When Linux recieves a short packet, it performs what is called a "keep
: : alive" check, it basically sends a ping (I_COMP) packet with an out of
: Nothing of the sort - sorry.
: : Software that I know has big problems with tis:
: :    NCSA Ftp (telnet works fine)
: Don't know about why this one should occur.

In my experience NCSA ftp will work, but it ends up operating in a sort of
'burst' mode, which is rather silly, especially for to machines on the same
piece of ethernet.

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

From: dglaude@vub.ac.be (GLAUDE DAVID)
Subject: Re: umsdos for 1.1.12?
Date: 23 May 1994 16:32:32 GMT

Mark Swanson (mswanson@hookup.net) wrote:
: Has anyone patched the umsdos 0.2 patches to Linux 1.0 to work
: with Linux 1.1.12?  I get the error on line 505 in the patch -
: vm_task->min_flt stuff.  
: Help!
: (and thank you)
Maybe you should try Umsdos 0.3 on linux 1.1.12 ;-)


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

From: wpp@marie.physik.tu-berlin.de (Kai Petzke)
Subject: Re: 32-bit Novell desktop OS combines Unix, DOS 7
Date: 23 May 94 14:22:49 GMT

jon@robots.ox.ac.uk (Jon Tombs) writes:

>As far as I see it , NOVEL can add better vm86 emulation for dosemu,
>better IPX support, several new filesystems all without breaking the
>GPL and without revaeling source code. It all depends on how you view
>loadable modules. If they put all there changes in a .o as a loadable module
>then I believe they can get away with only providing the changes needed
>to the kernel to support there modules (code like if (!ipx_loaded_hook) ...).

>The can also put what ever restrictions they like on there .o module to
>prevent copying of it, and/or backward engineering.

>The only way to prevent this would to put in the linux copywrite that
>the act of loading a module puts it under the GPL, but I'm not sure that
>we would want that.

Well, the GPL makes it hard to impossible to write a module for the
Linux kernel, without being forced to release the module under the
GPL.  Loading a module is not too different from linking an object
file into a program.  The according paragraph of the GPL reads:

    These requirements apply to the modified work as a whole.  If
    identifiable sections of that work are not derived from the Program,
    and can be reasonably considered independent and separate works in
                          ^^^^^^^^^^^^^^^^^^^^^^
    themselves, then this License, and its terms, do not apply to those
    sections when you distribute them as separate works.  But when you
    distribute the same sections as part of a whole which is a work based
    ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
    on the Program, the distribution of the whole must be on the terms of
                        ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
    this License, whose permissions for other licensees extend to the
    ^^^^^^^^^^^^
    entire whole, and thus to each and every part regardless of who wrote it.


So, there are two points in here:

a) The new module must be considered independent of the kernel.  I
   have my problems considering a file system independent of the
   rest of the kernel.

   A normal user programme is (mostly) independent of the Linux
   kernel.  You can recompile it on another U*ix.  This is not
   true for a Linux file system code.

b) If you distribute the joint package (kernel + new fs), you have
   to put it under the GPL.
-- 
Kai Petzke                      | How fast can computers get?
Technical University of Berlin  |
Berlin, Germany                 | Sol 9, of course, on Star Trek.
wpp@marie.physik.tu-berlin.de   |

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

From: jwagner@mental.mitre.org (John Wagner)
Subject: verdict on the NCR 83c510 drivers?
Date: 23 May 1994 14:27:09 GMT
Reply-To: jwagner@mental.mitre.org (John Wagner)


Well is there any? I've heard anything from weeks to months, and
that was several weeks ago, anyone care to state a progress report?

John
waiting with baited breath to run Linux on my pentium 500, or that might
be what it is by the time I get those drivers. :) Ok so I'm alittle
push-e :) Then again I could just run out to the local toy store and
buy a cheap isa card to have a temp fix :)
-- 
+++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
+         Heck even I don't know what I do, so the company can't.         +
+                 empire isn't a game, it a life style! :)                +
+  Sure wish I had learned "C" code now that Linux is installed  <OPPS>   +
+            jwagner@mitre.org | John Wagner | PH# (703)883-3740          +
+++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++

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

From: bf11620@ehsn9.cen.uiuc.edu (Byron Thomas Faber)
Subject: Re: Emacs 19.22: error on compile for linux
Date: 23 May 1994 14:44:38 GMT

The problem was solved with some help.

Thanks guys.

Byron Faber

-- 
`Playing this disk at loud volume may permanently damage your speakers or
other sound components.'                                -LFO
                                b-faber@uiuc.edu

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

Crossposted-To: comp.os.linux,comp.os.linux.misc,comp.os.linux.help
From: news@edus.oau.org (Bob Perkins)
Subject: Re: Linux and Bernoulli?
Date: Mon, 23 May 1994 13:13:36 GMT

adap@andrews.edu (Edsel Adap) writes:


>Hi!
>Is there any way I can get linux to recognize my bernoulli 90?  Is
>anyone working on a device driver for it?
Depends. If the Bernoulli is on the SCSI bus, treat it just like a
second hard drive, or third or fourth, and address it as such; i.e.,
/dev/sd[bcd]x.

If you have one of the IDE interface units, I cannot help.

>thanks.
>-- 
> Edsel Adap                         Computer Science / Mathematics Major
> adap@andrews.edu                   Andrews University, Berrien Springs, MI

>The Briggs/Chase Law of Program Development:
--
rcp@edus.oau.org

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

Crossposted-To: comp.os.linux.misc
From: cgjks1@lut.ac.uk (James Soutter)
Subject: Re: PCI or VL bus SCSI Host Adapters
Date: Mon, 23 May 1994 12:49:19 GMT

Doug McIntyre (merlyn@icicle.winternet.com) wrote:

: I don't know why anybody automaticly thinks of VL as soo much faster. I did
: some time comparisions with various VL bus and ISA bus controllers with 
: a fast SCSI-II hard drive (Fujitsu 1.2G), and the one that one had the
: top speed was still the Adaptec 1542CF. Granted, a couple of them came
: real close, like only 20-30k/sec slower :)
: These did not include the Adaptec 2842 though, so I don't know if that
: would have gained anything. I haven't gotten to play with one yet :(

SCSI adapters on a PCI bus are supposed to be much better than SCSI
adapters on an ISA bus because PCI supports 16 word burst most
transfer.  This will also improve CPU throughput because less CPU
cycles will be lost waiting for the SCSI controller to finish with the
bus.

I don't know much about VLB so I can only suggest some questions:

1) Does VLB support burst mode transfer (or is PCI better than VLB in
        this respect) ?
4) Is the gain of CPU cycles that significant ?
2) Were the VLB cards used in Adaptec 1542 compatibility mode rather
        than native mode (did the test use burst transfer mode) ?
3) Were the VLB cards bus mastering cards (did the test use burst
        transfer mode) ?
        
-- James

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

From: mark@taylor.wyvern.com (Mark A. Davis)
Subject: Re: 32-bit Novell desktop OS combines Unix, DOS 7
Date: Mon, 23 May 1994 14:59:24 GMT

jon@robots.ox.ac.uk (Jon Tombs) writes:

>As far as I see it , NOVEL can add better vm86 emulation for dosemu,
>better IPX support, several new filesystems all without breaking the
>GPL and without revaeling source code. It all depends on how you view
>loadable modules. If they put all there changes in a .o as a loadable module
>then I believe they can get away with only providing the changes needed
>to the kernel to support there modules (code like if (!ipx_loaded_hook) ...).

>The can also put what ever restrictions they like on there .o module to
>prevent copying of it, and/or backward engineering.

>The only way to prevent this would to put in the linux copywrite that
>the act of loading a module puts it under the GPL, but I'm not sure that
>we would want that.

I agree, that would be a bad idea, and severely limit the potential of
what could be done with Linux in the commercial sector.

-- 
  /--------------------------------------------------------------------------\
  | Mark A. Davis    | Lake Taylor Hospital | Norfolk, VA (804)-461-5001x431 |
  | Sys.Administrator|  Computer Services   | mark@taylor.wyvern.com   .uucp |
  \--------------------------------------------------------------------------/

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

From: beirne@MCS.COM (Michael Beirne)
Crossposted-To: comp.os.linux.help
Subject: Re: Anybody working on BSD dump porting?
Date: 23 May 1994 10:44:30 -0500

In article <2reh76$hfv@bosnia.pop.psu.edu>,
David Barr <barr@pop.psu.edu> wrote:
>In article <1994May18.210039.6822@belvedere.sbay.org>,
>David E. Fox <root@belvedere.sbay.org> wrote:
>>But what advantages does 'dump' have that 'tar' or 'cpio' do not?
>
>1. Does multiple dump levels
>2. Handles sparse files (and handles them correctly)
>3. Handles hard links (and handles them completely)
>4. Does not affect atimes of backed up files
>5. Supports multiple tape volumes (ok, so does GNU tar)
>6. No pathname length/filename restrictions
>7. Backs up devices/named pipes, etc.
>
>--Dave

And don't forget the "i" option of restore. Very nice if you need
to restore a file for someone and they have forgotten the exact name.
With tar and cpio backups, I always end up doing a "t" option to find
the correct name of the file and then an extraction pass. This is
two passes through the tape(s) instead of one. If you have multiple tapes
restore knows which tape the file is stored on so it skips reading the
whole tape and prompts you for the next tape after it has looked at the header
and found that the file you want is on different tape. Restore also does tape
seeks. The above can change the time required to restore one
file accidentally blown away from hours to minutes.

-- 
Michael G. Beirne |845 W. Washington #2D, Oak Park,IL 60302-3821|(708)386-2415
beirne@limerick.chi.il.us

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


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