Linux-Development-Sys Digest #381, Volume #6 Sat, 6 Feb 99 18:14:32 EST
Contents:
Re: use theramin as input device (Ben Russo)
Re: Why I'm dumping Linux, going back to Windblows (Thomas Becker)
glibc or glibc? ("asdf")
Re: NFS - client cache (Chris Sorenson)
Re: dir structure in tarball ("asdf")
Re: New free widget library: Notif-0.1 (Christopher Browne)
Re: Why I'm dumping Linux, going back to Windblows (John Hasler)
Re: 2.0.36 compile error: Help! (Dirk Foersterling)
The "make config' (Kevin Miller)
Re: The "make config' (jamie)
Re: glibc2 & libstdc++ (Takeyasu Wakabayashi)
Re: What's the best way to do process communication ? (Warren Young)
Re: can kernel be locked on one processor in dual processor system? (Arun Sharma)
Re: K6-2 and Linux, Are there any Bug? (Hilaire Fernandes)
Adding Stats to inetd? ("Sean Watkins (home)")
Purely Technical ("Eugene K.")
C/C++ docs? ("Scott Nelson")
Re: 2.0.36 compile error: Help! ([EMAIL PROTECTED])
Debugging the Linux Kernel ([EMAIL PROTECTED])
Re: addition of system calls (Arun Sharma)
Re: Process shared semaphores (Pthreads) (Andi Kleen)
addition of system calls (Marx Rajangam)
----------------------------------------------------------------------------
From: Ben Russo <[EMAIL PROTECTED]>
Crossposted-To: comp.os.linux.development.apps,comp.os.linux.misc
Subject: Re: use theramin as input device
Date: Fri, 05 Feb 1999 11:31:52 -0500
Jehan Sappideen wrote:
> In comp.os.linux.development.system Rik van Riel <[EMAIL PROTECTED]> wrote:
>
> : What kind of interface will it use to the computer,
> : serial, joystick or something else?
>
> : Or, to put it another way: what kind of an interface
> : would it need to be useful to us?
>
> What exactly would be expected of the therEmin? I've building and
> experimenting with them for a few months now, and one thing I can say
> authoritatively is that ABSOLUTE positioning is quite hard to do, and
> in addition the
> effective range is generally limited to 5->30 cm away from the antenna (
> too close and the capacitance increases exponentially, too far and it
> becomes unresponsive). Changes in position are easier to detect,
> so perhaps a more basic plan might involve utilizing this aspect.
>
> Jehan
What about having a seperator of non-interfering material that keeps your
hand 5cm away from the plate? I don't know precicely how these things work,
or even what they are, (from the context of the conversation it sounds neat
though)
but seems that you could have a layer of plastic between
you and the plate?
Also, what if you then coated the top of the seperator with a soft spongy
shape memory foam like
those disposable earplugs?
-Ben.
------------------------------
From: Thomas Becker <[EMAIL PROTECTED]>
Crossposted-To: alt.os.linux,comp.os.linux.development.apps,comp.os.linux.setup
Subject: Re: Why I'm dumping Linux, going back to Windblows
Date: Sat, 06 Feb 1999 12:10:07 +0100
I am quite new in linux, but worked with hp-ux before.
So I am used to man-pages.
Although they are often hard to understand, they are usefull as they
are. Basical introduction are written in many books.
But what really makes me nervous, is the fact, that the documentation of
linux is so wide spread. There are man-pages, howtos, mini-howtos,
readmes, readme-firsts, last-minite-information and at last many, many
web-sites.
That may depend on the uncentralized structure of developing linux,
but IMHO it should be cenralized.
Nice weekend
Thomas Becker
------------------------------
Reply-To: "asdf" <[EMAIL PROTECTED]>
From: "asdf" <[EMAIL PROTECTED]>
Subject: glibc or glibc?
Date: Fri, 5 Feb 1999 14:33:21 -0800
What is the difference between glibc-2.0.112 and glibc-2.0.7pre6? Which one
is better, why is there a number skip?
Pardon my ignorance.
Thanks.
------------------------------
From: Chris Sorenson <[EMAIL PROTECTED]>
Subject: Re: NFS - client cache
Date: Fri, 05 Feb 1999 23:19:47 -0600
Leszek Gryz wrote:
> Hi,
> Do you know if linux has implemented NFS client cache?
>
> Leszek Gryz
I think the best client cache for Linux isn't in the kernel, it's the
CODA project at Carnegie-Mellon:
http://www.coda.cs.cmu.edu/
------------------------------
Reply-To: "asdf" <[EMAIL PROTECTED]>
From: "asdf" <[EMAIL PROTECTED]>
Subject: Re: dir structure in tarball
Date: Sat, 6 Feb 1999 04:59:35 -0800
I've done it plenty of times. Hey, I've reinstalled the system more than
remember too. :̃
------------------------------
From: [EMAIL PROTECTED] (Christopher Browne)
Crossposted-To: comp.os.linux.advocacy,comp.os.linux.development.apps
Subject: Re: New free widget library: Notif-0.1
Reply-To: [EMAIL PROTECTED]
Date: Sat, 06 Feb 1999 04:55:57 GMT
On 5 Feb 1999 22:42:47 GMT, [EMAIL PROTECTED] <[EMAIL PROTECTED]> wrote:
>On 5 Feb 1999 16:55:12 GMT, Stan <[EMAIL PROTECTED]> wrote:
>>Matthias Warkus ([EMAIL PROTECTED]) wrote:
>>: Berlin has a thread running in every widget (called a `reactor').
>>: Whenever there is an event, the reactor processes it and dispatches
>>: something via CORBA.
>>
>>BTW, does anybody know of any good human-readable COBRA documentation ?
>>
>>How-to introduce COBRA in your widget library ?
>>
>
>Pedant point...that's CORBA, not "cobra". :-) (This may explain
>your difficulties in finding it. :-) )
It is not appropriate to flame misspellings; in this case, it seems
*necessary* to provide a correction, as if you do a search for COBRA,
you're likely to find either things about poisonous snakes, or about
some peculiar US labour legislation/health care stuff.
It should be CORBA, as in "Common Object Request Broker Architecture."
>http://www.corba.org should have something, but I can't say precisely
>what.
I'm not sure what you'll find there; the more relevant place to go is:
<http://www.omg.org>; OMG is the Object Management Group that are
responsible for CORBA standards.
Also appropriate would be such places as:
<http://www.cs.wustl.edu/~schmidt/corba.html>
<http://www.acl.lanl.gov/CORBA/>
<http://linas.org/linux/corba.html>
<http://www.berlin-consortium.org/corba_intro.html>
<http://www.ntlug.org/~cbbrowne/corba.html>
--
OS/2: Why marketing matters more than technology...
[EMAIL PROTECTED] <http://www.ntlug.org/~cbbrowne/corba.html>
------------------------------
From: John Hasler <[EMAIL PROTECTED]>
Crossposted-To: alt.os.linux,comp.os.linux.development.apps,comp.os.linux.setup
Subject: Re: Why I'm dumping Linux, going back to Windblows
Date: Sat, 6 Feb 1999 13:28:04 GMT
Thomas Becker writes:
> That may depend on the uncentralized structure of developing linux,
> but IMHO it should be cenralized.
That is just another way of saying that it should not exist.
--
John Hasler
[EMAIL PROTECTED] (John Hasler)
Dancing Horse Hill
Elmwood, WI
------------------------------
From: [EMAIL PROTECTED] (Dirk Foersterling)
Crossposted-To: comp.os.linux.setup
Subject: Re: 2.0.36 compile error: Help!
Date: 6 Feb 1999 04:15:32 +0100
Reply-To: [EMAIL PROTECTED]
>>Thomas Simonson wrote in message <79ei1t$[EMAIL PROTECTED]>...
>>ld: cannot open crtbegin.o: No such file or directory
On Fri, 5 Feb 1999 07:15:45 -0500, Fred Forester <[EMAIL PROTECTED]> wrote:
>Might be the second make clean.
Nonsense. The file crtbegin.o is part of the C library. Usually, this
file gets installed to /usr/lib/. If it's not there, re-installation of
the C library (libc or glibc) will help. If the file exists, try to
re-install the binutils package.
-dirk
--
D i r k F "o r s t e r l i n g
[EMAIL PROTECTED] ******** http://www.DeathsDoor.com/milliByte/
-------------
Dieser Satz kein Verb
------------------------------
From: Kevin Miller <[EMAIL PROTECTED]>
Subject: The "make config'
Date: Sat, 06 Feb 1999 07:58:33 GMT
Hi everyone!!
I'm trying to update my kernel to the lastest version 2.2.1 and when I
go to use the "make config" command, it barks back at me and says
command not found. Can you please tell me what I need to do from there
to get that to work.
Thanks in advance!
Kevin Miller
------------------------------
From: [EMAIL PROTECTED] (jamie)
Subject: Re: The "make config'
Reply-To: [EMAIL PROTECTED]
Date: Sat, 06 Feb 1999 11:09:20 -0600
Kevin Miller <[EMAIL PROTECTED]> wrote:
>Hi everyone!!
>
>I'm trying to update my kernel to the lastest version 2.2.1 and when I
>go to use the "make config" command, it barks back at me and says
>command not found. Can you please tell me what I need to do from there
>to get that to work.
Most likely, you're in the wrong directory.
You need to do it from /usr/src/linux
I prefer make menuconfig to make config, myself, it's easier to
review what you entered.
(I hope you upgraded the other requirements, if you only retrieved the
new kernel sources you probably will not even be able to boot it
if it compiles.)
--
jamie ([EMAIL PROTECTED])
"There's a seeker born every minute."
------------------------------
From: Takeyasu Wakabayashi <[EMAIL PROTECTED]>
Subject: Re: glibc2 & libstdc++
Date: 06 Feb 1999 16:24:26 +0900
Daniel Zurcher <[EMAIL PROTECTED]> writes:
> programs. I got the latest package 2.8.1.1 and it compiled just fine,
> but when I run 'make check', libstdc++.so complains of an undefined
> reference to 'filebuf virtual table' and 'exception virtual table'.
> Anyway, I really need to get the c++ libraries working again. If anyone
> out there knows anything about how to remedy this problem, please help
> me out.
I also suffered the problem when I was using glibc-2.0.4 on my
Alpha/Linux box. Also, there is a bug in glibc before v.2.0.7 which
causes shutdown/halt to work improperly.
Try glibc-2.0.7pre6 at
ftp://alpha.gnu.org/gnu/
Remember to get crypt and pthread add-ons.
PS. Now egcs-1.1.1 uses libstdc++.2.9.0. And compiling
libstdc++ separately is not recommended by README.
####################################################
Takeyasu Wakabayashi
[EMAIL PROTECTED]
``When you ask him about the economic policy,
You get an answer obscure like mist falling down''
-- Li Po
------------------------------
From: Warren Young <[EMAIL PROTECTED]>
Subject: Re: What's the best way to do process communication ?
Date: Sat, 06 Feb 1999 01:39:59 -0700
Aurel Balmosan wrote:
>
> Markus Kohler <[EMAIL PROTECTED]> wrote:
>
> > I need a way to communicate between two or more processes on the
> > same machine and exchange data as fast as
[snip]
> I would choose shared memory with some kind of memory management to add
> a new segment to the already existing.
That's a fairly nasty solution, IMHO, as it implies a whole lot of
synchronization (e.g. a message queue with a protocol built on top just
to exchange control messages) just to make it work. Shared memory also
has a major flaw: if the program dies without cleaning up, the shared
memory segments stay allocated until someone explicitly cleans them up.
And, it opens doors for two programs to crash each other -- at least
with a FIFO-style communication method, only one program gets confused
if something's broken. Personally, I'd only use shared memory if I had
a smallish, fixed block of memory that two programs needed to share.
A better idea, if you need to send entire blocks of data from one box to
another, would be to use Unix domain sockets. They're optimized for
bulk data transfer across a single machine (i.e. no Nagle-algorithm-like
delays), and they're fairly foolproof.
The disadvantage, of course, is that both processes then keep a copy of
the data. If that's a problem for you, maybe you _should_ take a look
at some shared memory solution. Or, reconsider your architecture: maybe
only one process needs a complete copy of the data, and the other can
just ask the data owner questions about the data when it needs to know
things: again, I'd suggest Unix domain sockets for something like that.
One other architecture idea you might consider is merging both processes
into a single process so they can share the data naturally. Threads
might be helpful for this.
Good luck,
--
= Warren -- http://www.cyberport.com/~tangent/
= ICBM Address: 36.8274040 N, 108.0204086 W, alt. 1714m
= Don't judge a book by its movie.
------------------------------
Subject: Re: can kernel be locked on one processor in dual processor system?
From: Arun Sharma <[EMAIL PROTECTED]>
Date: Sat, 06 Feb 1999 08:08:06 GMT
"Ross Henderson" <[EMAIL PROTECTED]> writes:
> Hello,
>
>
> We are interested in running a realtime process
> on a dual-processor Linux machine. Can anyone tell
> me if the 2.2 Linux kernel can be made to run on
> one CPU while the other CPU locks the application
> process? We're doing data acquisition, and the
> less interrupts due to OS calls, the better.
>
There is no such capability in the release kernel today. But someone
came up with a set of patches to do this. Look for "processor binding"
in the kernel archives.
This will be similar to pbind/processor_bind on Solaris.
-Arun
------------------------------
From: Hilaire Fernandes <[EMAIL PROTECTED]>
Crossposted-To:
alt.os.linux.slackware,comp.os.linux.hardware,comp.os.linux.misc,comp.os.linux.setup
Subject: Re: K6-2 and Linux, Are there any Bug?
Date: Sat, 06 Feb 1999 10:04:01 +0800
Javier Pulido wrote:
>
> ¡¡SOS!!
>
> Problems with the booting of linux in a computer AMD K6-2 (300 MHz) with
> i430TX (no AGP) and 128 MB, two hard drives and 1 CDROM. Can you help me?
>
> After many attempts of booting with several kernels (2.0.29, 2.0.30, 2.1.48,
> 2.0.35, 2.0.36) through a boot-disk, LILO or LOADLIN (Symbol of System,
> option F8 of Windows95), I realize only procedure that was operating:
>
> 1. I Start Windows95/98 (graphic environment)
> 2. I restart in MSDOS-Mode
> 3. C:> loadlin zImage2_2 root=/dev/hdc3 mem=128M no-hlt
>
> If I don´t use option no-hlt, the system starts but is hung in little
> minutes, and it shown me all the processor registers, the stack and the
> message "idle task may not sleep".
>
> In the attempts rest of starting were remained hung when was ending the
> load kernel in report: Loading linux ....... (!!stop!!)
>
> My system crashes while it´s booting. I
>
I've the same problem with K6 200 Mhz, 32Mb and a motherboard with
i430TX
When I install linux, the kernel load, it starts to make its uusal
check-in and after the PCI probe the system just reset.
I've almost 20 Pc where I can't install linux :(
--
Hilaire Fernandes
Dr Geo project http://members.xoom.com/FeYiLai/dr_geo/doctor_geo.html
------------------------------
From: "Sean Watkins (home)" <[EMAIL PROTECTED]>
Subject: Adding Stats to inetd?
Date: Sat, 06 Feb 1999 14:44:54 -0300
Has anyone done any work at providing inetd with some statistics?
I'd like to be able to graph the number of people hitting pop-3 with
mrtg?
Looking through the code, it appears trivial to add... why build a
better mouse trap though.
Remove the NOSPAM from my email to reply...
--
Sean Watkins
------------------------------
From: "Eugene K." <[EMAIL PROTECTED]>
Reply-To: [EMAIL PROTECTED]
Subject: Purely Technical
Date: Sat, 06 Feb 1999 07:30:58 -0600
Hi!
I have a modem hooked up to /dev/cua2. Once I accidently turned off my
computer without shutting down first, and since then I have not been
able to access my modem. I am pretty sure this has something to do with
lock files, because on some applications it says "Sorry, the modem is
busy". Has anyone had this problem before?
Thanx in advance,
-Bob
------------------------------
From: "Scott Nelson" <[EMAIL PROTECTED]>
Subject: C/C++ docs?
Date: Sat, 6 Feb 1999 13:49:50 -0500
I just installed Red Hat 5.2 and would like to do some development. Where
can I get some docs on the Linux API? There is a small HTML document that
is on my disk that lists some API functions but doesn't describe their
prototypes, return codes, etc. Very incomplete.
Any help would be greatly appreciated.
Scott Nelson
[EMAIL PROTECTED]
------------------------------
Subject: Re: 2.0.36 compile error: Help!
From: [EMAIL PROTECTED]
Crossposted-To: comp.os.linux.setup
Date: 6 Feb 1999 10:54:17 -0500
Fred Forester ([EMAIL PROTECTED]) wrote:
: Might be the second make clean.
:
: Thomas Simonson wrote in message <79ei1t$[EMAIL PROTECTED]>...
: >
: >
: >I get an error when trying to recompile
: >the kernel (2.0.36) (for SMP) on a dual
: >PII 400 MHz machine:
: >
: >make config
: >make clean
: >make dep
: >make clean
: >make zImage ..... compiles happily for 3-4 minutes, then:
: >
: >gcc -I/usr/src/linux/include hexify.c -o hexify
: >hexify.c: In function `main':
: >hexify.c:5: warning: return type of `main' is not `int'
: >ld: cannot open crtbegin.o: No such file or directory
: >make[1]: *** [hexify] Error 1
: >make[1]: Leaving directory `/usr/src/linux-2.0.36/arch/i386/kernel'
: >make: *** [linuxsubdirs] Error 2
Your version of GCC is incorrectly installed or is incomplete. You
should re-install GCC, either from your distribution's CD (which
might be incorrect itself), or from the package:
ftp://tsx-11.mit.edu/pub/linux/packages/GCC/egcs-1.0.3-glibc.x86.tar.bz2
When unpacked from / as root, it will overwrite your version of the GCC
executable and it'll create a directory under /usr/lib/gcc-lib, which
should contain the crtbegin.o file.
------------------------------
Subject: Debugging the Linux Kernel
From: [EMAIL PROTECTED]
Date: 6 Feb 1999 10:59:15 -0500
How does one go about debugging the kernel? On a user app, I would
compile it with certain debug options, and run the program through
a debugger. I can't do this to the kernel, for obvious reasons. How
can a bug in the kernel be tracked down?
------------------------------
Subject: Re: addition of system calls
From: Arun Sharma <[EMAIL PROTECTED]>
Date: Sat, 06 Feb 1999 22:18:44 GMT
Marx Rajangam <[EMAIL PROTECTED]> writes:
>
>
> then I booted the kernel and wrote a simple c program.
>
> 1) _syscall0(int, assoc)
> This gave no syntax error but at the same time didn't return 1 either.
>
> 2) i = syscall(164,assoc);
> and I am getting an error like
>
> undefined reference to assoc..
>
> If any one can tell me where I am going wrong, it would help me a lot.
>
Use "strace" to trace exactly what's going on.
-Arun
------------------------------
From: Andi Kleen <[EMAIL PROTECTED]>
Subject: Re: Process shared semaphores (Pthreads)
Date: 05 Feb 1999 18:51:22 +0100
In article <[EMAIL PROTECTED]>,
"D. Emilio Grimaldo Tunon" <[EMAIL PROTECTED]> writes:
> Hi *,
> Anybody knows when on earth Linux's PThreads (LinuxThreads/glib)
> is going to support process-shared semaphores and p-s
> condition variables? I am looking for 'drop-in' replacement for
> the pthread_mutexattr_setpshared() and pthread_condattr_setpshared()
> functions.
> Also, these functions are Optional in the POSIX standard, but
> they are required in the Unix98 standard, since kernel 2.2 is
> supposed to have some (most? all?) Unix98 conformance, I wonder
> if 2.2 does support these two problem-makers?
Who claimed that 2.2 is Unix98 compliant? It has some useful Unix98 features,
but anyone claiming conformance or near-conformance is a fool (and AFAIK
there are no plans for full Unix98 compliance)
Process-shared semaphores are a 2.3 project and they are unlikely
to be ever supported in the 2.2.x stable kernel series. Current workaround
is to use SYSV style semaphores instead. You have to implement condition
variables on top of them.
-Andi
--
This is like TV. I don't like TV.
------------------------------
From: Marx Rajangam <[EMAIL PROTECTED]>
Subject: addition of system calls
Date: 6 Feb 1999 20:38:36 GMT
hi,
I am a newbie to Linux and I am trying to add a system call to the kernel.
I did the following :
1) I wrote a function
asmlinkage int assoc(void)
{
return 1;
}
and added that to fs/open.c
2) Then I added an entry for the system call in asm-i386/unistd.h file
that looks like,
#define __NR_assoc 164
3) Then I added space a pointer to the function thats handled by system
call in arch/i386/kernel/entry.S file that looks like,
.long SYMBOL_NAME(sys_assoc)
.space (NR_syscalls-167)*4
then I booted the kernel and wrote a simple c program.
1) _syscall0(int, assoc)
This gave no syntax error but at the same time didn't return 1 either.
2) i = syscall(164,assoc);
and I am getting an error like
undefined reference to assoc..
If any one can tell me where I am going wrong, it would help me a lot.
Thanks a lot in advance.
--
Marx Rajangam
Fisher 227, CS Dept,
Michigan Tech. Univ.
------------------------------
** 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: [EMAIL PROTECTED]
You can send mail to the entire list (and comp.os.linux.development.system) via:
Internet: [EMAIL PROTECTED]
Linux may be obtained via one of these FTP sites:
ftp.funet.fi pub/Linux
tsx-11.mit.edu pub/linux
sunsite.unc.edu pub/Linux
End of Linux-Development-System Digest
******************************
