Subject: Linux-Development Digest #781
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:     Tue, 31 May 94 13:13:08 EDT

Linux-Development Digest #781, Volume #1         Tue, 31 May 94 13:13:08 EDT

Contents:
  Re: 1.1.16 NFS hangs on write when wsize >1k (Re: 8k NFS client in 1.1.13) (Holger Linne)
  Re: Why is my Emacs binary so big (2.2M) (Michael MNUK)
  Re: 8k nfs performance (Alan Cox)
  Re: 1.1.15 breaks SCSI (Alan Cox)
  Re: 1.1.16 NFS hangs on write when wsize >1k (Re: 8k NFS client in 1.1.13) (Alan Cox)
  Re: Latest and Greatest networking code (Alan Cox)
  Re: Serial drivers in 1.1.16 (autoconfig for IRQ's) (Rob Janssen)
  Re: SIGHUP - Where do we go from here? (Ken Pizzini)
  Toshiba 3101 CD-ROM audio commands wanted (ramon.via@sni.de)
  Re: 1.1.16 NFS hangs on write when wsize >1k (Re: 8k NFS client in 1.1.13) (Alan Cox)
  Re: bug on kernel 1.1.15 (Tim Cutts)
  developing X apps in Linux (Paul Kolonay)
  Re: 1.1.15 breaks SCSI (Tim Cutts)
  HELP! 20MB of mem and Linux 1.0 --swapper problems (Dorwin Shields)
  Telnet, FTP and NFS down under >1.1.13 (lcvanveen@et.tudelft.nl)
  Canon BJC-600 support? (Yves Arrouye)
  ARP broken with kernel 1.1.13 (Harry C. Johnson)
  Re: 1.1.15 breaks SCSI (NetDog)
  Re: 1.1.16 NFS hangs on write when wsize >1k (Re: 8k NFS client in 1.1.13) (Janne Sinkkonen)
  Re: Appletalk support? (James Hammett)
  Client/Server exampes? (Ron Watkins)

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

From: hll@lidarh.dkrz.de (Holger Linne)
Subject: Re: 1.1.16 NFS hangs on write when wsize >1k (Re: 8k NFS client in 1.1.13)
Date: Tue, 31 May 94 07:12:46 GMT

In article <2sdkde$l7g@hpsystem1.informatik.tu-muenchen.de>, Oliver Schoett <schoett@informatik.tu-muenchen.de> writes:
|> In article <2rr7sb$506@galaxy.ucr.edu> grif@tempest.ucr.edu (Michael
|> Griffith) writes:
|> 
|>  > The 8k NFS client in Linux 1.1.13 works so nicely, that I wanted
|>  > to publically thank Alan, Linus, and everyone else involved.  I am
|>  > seeing 3-4x speedup which is quite nice considering almost everything
|>  > is NFS mounted.  Great work!
|> 
|> Having waited eagerly for this, I immediately tried 1.1.16 on my 486
|> DX2-66 VL with 3c509 Ethernet card and the Slackware 1.2.0.3
|> distribution.  I mount various file systems
|> 
|>    rw,hard,bg,intr,posix,nosuid,rsize=8192,wsize=<read on>
|> 
|> from sun (pre-Solaris), HP, and SNI RM400.  Mounting and reading works
|> fine, but with wsize=2048, 4096 and 8192 I get hangs on write: For
|> example, 'cp' and 'iozone' both create an empty target file and then
|> hang such that only kill -9 will abort them.  This happens the same
|> way with all three manufacturer's servers.  Wsize=1024 works, but is
|> slow as ever.
|> 
|> Does anyone have an idea what might be wrong here?
|> 
|> Oliver Schoett  <schoett@informatik.tu-muenchen.de>
|> Schraudolphstr. 24, 80799 Muenchen, Germany.  Phone +49 89 272 51 58

... And I thought I am alone -- 
SAME WITH MY LINUX - BOX:

486 DX2-66 with SMC Elite card. The only difference difference is -
i can't kill this job with -9, but only a reboot 'solves' the problem.
Running 
'etherfind host my-linux-box'
on the SUN (SunOS 4.1.3) beeing the client or server (I have tried both)
gives a problem at that moment when an udp - packet of length 1498 will
be tarnsmitted.
Maybe its some problem with MTU-size (still 1500) and packet sequence ?????
BTW: I recompiled about every network program involved in this
(amd, rpc.nfsd, rpc.mountd, ifconfig, ....) -- no help 8-(

And one more to Alan Cox:
GATED works great ! Thanks !
But where to get the source or the diffs to gated-R3_0_3.tar.gz ??
I had a look at gated.cornell.edu but there is no linux-target even
in the new (ALPHA? / BETA?)-release and I don't like your paths
(gdc is looking for gated at /usr/local/etc ??)

Holger Linne <hll@lidar.dkrz.d400.de>

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

From: mmnuk@risc.uni-linz.ac.at (Michael MNUK)
Crossposted-To: comp.os.linux.help
Subject: Re: Why is my Emacs binary so big (2.2M)
Date: 31 May 1994 10:08:13 GMT

>>>>> In article <MMNUK.94May27152953@lion.risc.uni-linz.ac.at>, mmnuk@risc.uni-linz.ac.at (Michael MNUK) writes:
MM> NNTP-Posting-Host: lion.risc.uni-linz.ac.at

MM> I compiled the Emacs 19.24 using the standard distribution (no extra packages
MM> dumped). The resulting binary is about 2.2M big which is about twice as much
MM> as the binary I got with the Slackware distribution. I only changed the
MM> compilation flag from "-g" to "-O2".

I "solved" the problem by linking not with -lX11 but with libX11.sa, and
enabling the use of shared libraries when compiling the toolkit. This 
reduced the binary to about 1.6M. 
--
Michal Mnuk                                  Phone: +7236 3231 75             
Research Institute for Symbolic Computation  Fax  : +7236 3231 30             
A-4040 Linz, Austria                         E-mail: mmnuk@risc.uni-linz.ac.at


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

From: iiitac@uk.ac.swan.pyr (Alan Cox)
Subject: Re: 8k nfs performance
Date: Tue, 31 May 1994 09:15:59 GMT

In article <CqMH2r.K9o@cwi.nl> roelofs@cwi.nl (Marcel Roelofs) writes:
>Is this client caching in any way connected to the large quantities of
>nfsd & biod daemons floating around on almost every Unix system I have 
>ever worked on (that is, except for Linux)?

It's related. Basically NFS needs threads, BSD Unix didn't have threads so
they used user processes locked in kernel mode (ie a system call that never
returns) to give threads. The Linux NFS client could be done this way or
by using a cache and the timer services. The two big problems are the
Linux virtual fs layer isn't up to using the block cache for nfs caching
without some fixing (that ought to get done) and the fact it is currently
single threaded.

Alan




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

From: iiitac@uk.ac.swan.pyr (Alan Cox)
Subject: Re: 1.1.15 breaks SCSI
Date: Tue, 31 May 1994 09:17:59 GMT

In article <2sdae6$8q7@colin.muc.de> michael@muc.de (Michael Schmidt) writes:
>Hmmm. Perhaps slow machines? I have a 386/40. Rene has also a 386, if
>I remember correctly.
>
Running a 386DX40 here with a 330Mb conner on a 1542B - not a whisper of
complaint.

Alan



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

From: iiitac@uk.ac.swan.pyr (Alan Cox)
Subject: Re: 1.1.16 NFS hangs on write when wsize >1k (Re: 8k NFS client in 1.1.13)
Date: Tue, 31 May 1994 09:19:44 GMT

In article <2sdkde$l7g@hpsystem1.informatik.tu-muenchen.de> Oliver Schoett <schoett@informatik.tu-muenchen.de> writes:
>from sun (pre-Solaris), HP, and SNI RM400.  Mounting and reading works
>fine, but with wsize=2048, 4096 and 8192 I get hangs on write: For
>example, 'cp' and 'iozone' both create an empty target file and then

Cunningly the fragmentation got broken fixing some tcp bugs in 1.1.16 - I've
mailed Linus the fixes. Theres also a small memory tracking race that is now
hopefully buried (thanks Alex and co...)

Alan



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

From: iiitac@uk.ac.swan.pyr (Alan Cox)
Subject: Re: Latest and Greatest networking code
Date: Tue, 31 May 1994 09:23:05 GMT

In article <2se8dv$14e@styx.uwa.edu.au> michael@ecel.uwa.edu.au (Michael Simmons  - division) writes:
>The sunacm.swan.ac.uk directory appears empty.
>The mirror on sunsite.unc.edu has only old README files.

Then get a decent ftp client because it is not.

>Why can't the kernel developers include a Patchxx.README file that
>describes what has changed and what software is directly affected.

What the [******] do you think net/inet/README is

Sigh......

Alan




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

From: rob@pe1chl.ampr.org (Rob Janssen)
Subject: Re: Serial drivers in 1.1.16 (autoconfig for IRQ's)
Reply-To: pe1chl@rabo.nl
Date: Tue, 31 May 1994 07:19:58 GMT

In <2sdopd$bt9@nic.lth.se> d91mn@efd.lth.se (Mikael Nordqvist) writes:

>In article <2sdfgb$9tb@plootu.helsinki.fi> sinkkone@cc.Helsinki.FI (Janne Sinkkonen) writes:
>>
>>In article <CqMp9p.EzF@scifi.uucp>, Nick Simicich <njs@scifi.uucp> wrote:
>>>Personally, I found this feature quite useful and reliable, and was
>>>hoping that the symbol would someday be addeed to the autoconfig.  Any
>>>chance of getting it back in?  On Linux, the '4,3,4,3' irq pattern for
>>>com1-4 (as hard coded in serial.c) doesn't make much sense.
>>
>>As a remedy, use the setserial program in your /etc/rc*

>It seems that setserial doesn't do the job under some circumstances (like
>mine). When I use setserial to change irqs it reports them as changed, but
>the device (a mouse) doesn't work. If I enable the autoconfig in the kernel
>it works flawlessly without any need for setserial.

You must run the setserial before you start the mouse application.
(e.g. selection)

Rob
-- 
=========================================================================
| Rob Janssen                | AMPRnet:   rob@pe1chl.ampr.org           |
| e-mail: pe1chl@rabo.nl     | AX.25 BBS: PE1CHL@PI8UTR.#UTR.NLD.EU     |
=========================================================================

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

From: ken@chinook.halcyon.com (Ken Pizzini)
Subject: Re: SIGHUP - Where do we go from here?
Date: 31 May 1994 08:54:15 GMT

In article <2s9njq$6bs@nwfocus.wa.com>,
I <ken@halcyon.com> wrote:
>And yes, my interpretation too is that Linux is not POSIX
>compliant with regards to 3.2.2 item (6), based on the
>black-box experimentation I've conducted.  (Actually,
>the testing I've done can't distinguish between a 7.1.1.10
>noncompliance or a 3.2.2 noncompliance; I'm just assuming
>the latter.)  (I really should look at the kernel source
>for this, but I'm being lazy this evening.)

Having seen a bash code fragment posted recently by Matthias Urlichs
(urlichs@smurf.noris.de), I think my testing was flawed -- it was
probably bash doing the killing.  (Really gotta look at that
source code...)

                --Ken Pizzini

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

From: ramon.via@sni.de
Crossposted-To: alt.cd-rom,comp.periphs.scsi,comp.sys.ibm.pc.hardware.cd-rom
Subject: Toshiba 3101 CD-ROM audio commands wanted
Date: 31 May 94 11:00:44 GMT

I'm looking for the Toshiba 3101 CD-ROM commands for the audio part. It
uses non-SCSI commands for this and I already got some commands from the
'xmcd' player of Ti Kan. I already can play tracks with it and read TOCs,
but I don't know how to calculate offsets when I uses index within a track.
Can anyone send me these commands? Thanks in advance...

[For Linux users]
I'm working on a kernel enhancement for Linux, so it will automatically
support the audio functions of this drive. Is somebody interested in these
patches if they're finished?


--

Keep on hacking,                        \\\//
                                        (o o)
Ramon de Klein                  +----ooO-(_)-Ooo----+
Th. de Keyserstraat 298         |  _                |
7545 AJ  Enschede               | |_)        |/     |
The Netherlands                 | | \amon de |\lein |
                                |                   |
Email: ramon.via@sni.de         +-------------------+

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

From: iiitac@uk.ac.swan.pyr (Alan Cox)
Subject: Re: 1.1.16 NFS hangs on write when wsize >1k (Re: 8k NFS client in 1.1.13)
Date: Tue, 31 May 1994 11:34:56 GMT

In article <1994May31.071246.1536@news.dkrz.de> hll@lidarh.dkrz.de (Holger Linne) writes:
>... And I thought I am alone -- 
>SAME WITH MY LINUX - BOX:
Yep - 1.1.16 fixes some TCP bugs and in doing so broke fragmentation rather
subtly.. took me 4 hours to find. Fixes as usual on their way to Linus.

>And one more to Alan Cox:
>GATED works great ! Thanks !
Thanks for that go to Steven Davies not to me for that.

>But where to get the source or the diffs to gated-R3_0_3.tar.gz ??
Dunno.. I'd assumed he would have released them by now. Maybe they have
been submitted to cornell ??

Alan



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

From: tjrc1@cus.cam.ac.uk (Tim Cutts)
Subject: Re: bug on kernel 1.1.15
Date: 31 May 1994 12:25:07 GMT

supat@nuntana.animal.uiuc.edu (Supat Faarungsang) writes:

>I got this bug in 1.1.12 1.1.13 1.1.14 and 1.1.15
>it work fine up to 1.1.11

>here the bug

>stty: TCGETS: Operation not supported on socket

Are you using term?  I get this from SunOS, not Linux, when running a shell
script on the Sun with 'trsh -s scriptname'.  I don't think trsh allocates a
pty if you use the -s option.

Tim.

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

From: paulk@cs.pitt.edu (Paul Kolonay)
Subject: developing X apps in Linux
Date: 31 May 1994 12:28:02 GMT

What do I need in general to develop X apps under Linux... hardware
requirements, software packages, etc... and specifically, what do I
need to use PHIGS stuff...

thanx 
paul

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

From: tjrc1@cus.cam.ac.uk (Tim Cutts)
Subject: Re: 1.1.15 breaks SCSI
Date: 31 May 1994 12:22:01 GMT

michael@muc.de (Michael Schmidt) writes:

>Hmmm. Perhaps slow machines? I have a 386/40. Rene has also a 386, if
>I remember correctly.

Nope.  I have a 486/33 and have this problem.

Tim.

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

From: parprods@ecn.uoknor.edu (Dorwin Shields)
Subject: HELP! 20MB of mem and Linux 1.0 --swapper problems
Date: 31 May 1994 14:10:22 GMT

   HELP!!--I can hardly get Linux to boot now--I get swapper problems
during boot up--the messages say that a nonexistant swap page was 
trying to be duplicated--then later I get Oops:0002 and the message
says that it is unable to fulfill a kernel paging request.  I cant
rebuild my kernel--I tried to reinstall from scratch from the trans-
ameritech CD but I started getting segmentation faults during installation.
     All of this started when I took out 4 1MB simms and put 4 4MB simms in.
I now have 20MB--4 1MB simms(3 chip) and 4 4MB(9 chip)--Can anyone help?
Thanks,
Dorwin

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

From: lcvanveen@et.tudelft.nl
Subject: Telnet, FTP and NFS down under >1.1.13
Date: 31 May 94 15:06:25 +0200

When compiling 1.1.16 we noticed we couldn't telnet, nfs or ftp
from one machine to another anymore. Ping does work and so does
the windows ftp daemon. Alas we weren't able to do any of the above
stuff anymore. Backtracing through versions we discovered that 
after patch 1.1.13 it broke down. 1.1.12 worked like a charm 
though.
I suspect that there's something wrong with our config file.
Does anybody know whats wrong? We run on (among others) a 486-40
with 16 Mb, 240 Mb and 370 Mb IDE drive and a panasonic CD-ROM
drive.
We followed the default config stuff except for ISO9660 support.
Any help?
Thanks in advance.
Martijn.

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

Crossposted-To: comp.os.linux.help
From: arrouye@petole.imag.fr (Yves Arrouye)
Subject: Canon BJC-600 support?
Date: Tue, 31 May 1994 12:57:17 GMT

Hi,

This is rather a ghostscript-oriented question, but I thought one of
you may help me: I would like to know if there is some support for
color-printing on a CAnon BJC-600?

Thanks in advance,
Yves.
--
Advocates for the C++ school claim that a well designed           Yves Arrouye
    program does not need the extra flexibility (a lie),  Yves.Arrouye@imag.fr
while advocates for the Objective-C school claim that         (33) 76 57 48 64
    the errors are no problem in practice (another lie).             NeXT Mail

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

From: hcj@gull.uncc.edu (Harry C. Johnson)
Subject: ARP broken with kernel 1.1.13
Date: 31 May 1994 04:42:31 GMT

I just loaded version 1.1.13 of the kernel on the system I am using as
a slip server, the machine that is called in to.  It seems however
that arp is broken with this version of the kernel.  Execution of arp
gives the error:

/bin/arp: /dev/arp device not present 

or something similar, but it is there and it is a character special
file (16/1).  This never failed before, though publishing permanent
proxies didn't work.  Do I need to change the major and minor numbers
of the /dev/arp file, or do I need a new version of arp and if so,
where do I get it, or is arp just broken with this kernel and I need
to get a new/old one?  Thanks for your time.

-Harry
--
=============================================================================
                             Harry C. Johnson IV

                             Phone: (704)547-4339

                         E-Mail: hcj@gull.uncc.edu

                            Department of Chemistry
                   University of North Carolina at Charlotte
                        Charlotte, North Carolina 28223

"Seems to me, its all just chemistry..."
                                        -Neil Peart
=============================================================================



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

From: cdent@yod.honors.indiana.edu (NetDog)
Subject: Re: 1.1.15 breaks SCSI
Date: Tue, 31 May 1994 14:27:08 GMT

>>>>> "AC" == Alan Cox <iiitac@uk.ac.swan.pyr> writes:

    AC> In article <2sdae6$8q7@colin.muc.de> michael@muc.de (Michael
    AC> Schmidt) writes:
    >> Hmmm. Perhaps slow machines? I have a 386/40. Rene has also a
    >> 386, if I remember correctly.
    >> 
    AC> Running a 386DX40 here with a 330Mb conner on a 1542B - not a
    AC> whisper of complaint.

From the multiple postings there is obviously something wrong that was
introduced above 1.1.12. It is neither slow nor fast machines since
we've seen people with 386 and 486 machines complaining. It does
appear like it might be people with more than one device but that's
not clear.

So far, I've seen two solutions, both of which have worked for me:

back down to 1.1.12

or 

#define NEEDS_JUMPSTART 0x20 /* taken from old hosts.h */

in drivers/scsi/scsi.c and run 1.1.16.

I got both of these ideas from rene.

Chris



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

From: sinkkone@cc.Helsinki.FI (Janne Sinkkonen)
Subject: Re: 1.1.16 NFS hangs on write when wsize >1k (Re: 8k NFS client in 1.1.13)
Date: 31 May 1994 17:13:40 +0300

In article <1994May31.091944.4397@uk.ac.swan.pyr>,
Alan Cox <iiitac@uk.ac.swan.pyr> wrote:
>In article <2sdkde$l7g@hpsystem1.informatik.tu-muenchen.de> Oliver Schoett <schoett@informatik.tu-muenchen.de> writes:

>Cunningly the fragmentation got broken fixing some tcp bugs in 1.1.16 - I've
>mailed Linus the fixes. Theres also a small memory tracking race that is now
>hopefully buried (thanks Alex and co...)

I just installed 1.1.17 and 8k nfs write problems disappeared (so it
seems at least).

--
Janne

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

From: James Hammett <jamesh@apple.com>
Subject: Re: Appletalk support?
Date: Tue, 31 May 1994 15:19:15 GMT

In article <Cpx0Do.K2v@stortek.com> Jeff Andre, andre@hoth.stortek.com
writes:
> One of the real questions is how much of an interest is there in such
> a capability.  Beyond sharing printers, AppleTalk isn't a steller
performer.
> 
> I believe there is a way to attach Ethernet to old Macs; I believe there
> are SCSI devices.  If you're interested, I'll find the product.  I think
> we may have one or two here at work.
> 
> The other solution would be to get an IP <-> AppleTalk gateway.
> 
> Enjoy,
> 
> Jeff
> 

In terms of getting your Mac to mount the Linux drive, you could contact
Interconn, they make a product that allows the Mac to mount NFS shared
volumes.

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

From: ron@argus.lpl.Arizona.EDU (Ron Watkins)
Crossposted-To: comp.os.linux.help
Subject: Client/Server exampes?
Date: 31 May 1994 14:49:15 GMT

Im looking for some examples of writing simple client/server applications
on Linux. I have a quasi real time application which needs near continueous
readout of an event timer. The event timer must be running any time the
machine is up because several processes need this event timer to schedule
activities. I had thought that I could initialize the event timer then
set up a client/server situation so that each time the event time was
requested, the event timer could quickly compute it and return that value
to the calling program. I can't really do this as a subroutine as there
is quite a bit of overhead in computing the base event and hold times.
Once they are computed, the offset to current time is easy. So I thought
that a seperate server program could do this.
Any advice? Examples of client/server (I have never written one before)?
Alternatives to client/server model?
                        Ron Watkins
--
Ron Watkins    [ron@argus.lpl.arizona.edu]    /            /~~~~)     /
931 Gould-Simpson                            /            /____/     /
University of Arizona                       /            /          /
Tucson AZ. 85721 -- (602) 621-8606         (____ unar & / lanetary (____ ab.

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


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