Linux-Development-Sys Digest #545, Volume #6 Mon, 29 Mar 99 00:14:32 EST
Contents:
Re: Anyone working with ORBit? (Elliot Lee)
Q: Terratec Xlerate PCI sound card with linux (Dominik Bodi)
Re: Idea: Make a seperate "i686" tree for Redhat Linux 6.0 (Chris Mauritz)
Re: Idea: Make a seperate "i686" tree for Redhat Linux 6.0 (David Fox)
Re: Security issues ("G. Sumner Hayes")
Re: After Week 1 With Linux -- licking wounds. (John Myers)
Re: Programming tools for Linux/Unix: Editor, IDE, Frontend to GCC. (Conrad
Sanderson)
bizzare problem with kernel 2.2.3 (Conrad Sanderson)
Re: no EOF from /dev/ram (Andy Isaacson)
Re: After Week 1 With Linux -- licking wounds. (Jeremy Taylor)
Re: Programming tools for Linux/Unix: Editor, IDE, Frontend to GCC. (David Pace)
IP Checsums bad (Todd)
Re: Programming tools for Linux/Unix: Editor, IDE, Frontend to GCC. (Amy Bellows)
Re: no EOF from /dev/ram ([EMAIL PROTECTED])
----------------------------------------------------------------------------
From: [EMAIL PROTECTED] (Elliot Lee)
Subject: Re: Anyone working with ORBit?
Date: 28 Mar 1999 23:19:10 GMT
On Sat, 27 Mar 1999 02:57:26 GMT, Christopher B. Browne
<[EMAIL PROTECTED]> wrote:
> /* POA_sample__init(&poa_sample_servant, &ev); */
This should be uncommented.
>Has anyone some relatively trivial examples of ORBit-based server code? Once
>I get such a sample working, I'd be glad to comment it contribute it to the
>documentation...
See ORBit/test/echo-{server,client}.c
-- Elliot
Lbh'er fhccbfrq gb gel bar bs gubfr fvyyl EFN-va-guerr-yvarf .fvtangherf
ba guvf, svefg
------------------------------
Date: Sun, 28 Mar 1999 21:56:39 +0200
From: Dominik Bodi <[EMAIL PROTECTED]>
Reply-To: [EMAIL PROTECTED]
Crossposted-To:
comp.os.linux.hardware,linux.dev.sound,redhat.hardware.arch.intel,comp.os.linux.help,alt.os.linux,alt.uu.comp.os.linux.questions,comp.os.linux.questions,linux.redhat,linux.redhat.misc,comp.os.linux.misc
Subject: Q: Terratec Xlerate PCI sound card with linux
Hi,
I am a linux newbie and have a Terratec Xlerate sound card in my PC. I am
searching for a possibility to get this run with linux.
The Terratec Xlerate is a PCI sound card with an Aureal Chipset and SigmaTel
AC-97 codec.
Equivalent (=using the same chipset) cards are the Aztech PCI338-A3D, Diamond
Sonic Impact S-90, Intresource TeraSound A3D PCI and the Turtle Beach Montego.
These are the names they are sold under in Germany, they may vary abroad.
The card fits nicely into Win98 and can be made 100% SB compatible with a
special DOS-Driver and setting the interrupts etc. to SB-like values. But this
does nothing for Linux.
The kernel (2.2.4) does not have support for it, neither does OSS.
Does someone know how to make this card work with linux?
Thanks in advance
[EMAIL PROTECTED]
------------------------------
Crossposted-To:
comp.os.linux.misc,linux.redhat.misc,alt.linux,alt.os.linux,comp.os.linux.hardware
From: Chris Mauritz <[EMAIL PROTECTED]>
Subject: Re: Idea: Make a seperate "i686" tree for Redhat Linux 6.0
Date: Sun, 28 Mar 1999 14:30:58 GMT
In comp.os.linux.misc Enkidu <[EMAIL PROTECTED]> wrote:
> wizard wrote:
>>
>> It would be foolish for redhat not to develop a 686 specific
>> version of Linux. The reality is if they don't someone else will.
>>
> Redhat do not develop anything, they "just" package it.
It would be nice if they would "package" a source-based distribution
similar to what you get with FreeBSD's source. Then, if I wanted
686 optimization, I could replace gcc with egcs, use the appropriate
cflags, type "make world", go to bed, and wake up to a system where
every single binary had been recompiled from scratch on the running
system. That is one of the things I miss most about FreeBSD. I
also miss the ability to track a development tree with cvsup, and
then do a "make world" every couple of months to completely catch
my system up to the latest/greatest.
Any chance of RedHat doing something similar?
Chris
--
Christopher Mauritz
[EMAIL PROTECTED]
------------------------------
From: d s f o x @ c o g s c i . u c s d . e d u (David Fox)
Crossposted-To:
comp.os.linux.misc,linux.redhat.misc,alt.linux,alt.os.linux,comp.os.linux.hardware
Subject: Re: Idea: Make a seperate "i686" tree for Redhat Linux 6.0
Date: 28 Mar 1999 15:45:52 -0800
Tomasz Korycki <[EMAIL PROTECTED]> writes:
> Johan Kullstam wrote:
> >
> > d s f o x @ c o g s c i . u c s d . e d u (David Fox) writes:
> >
> > > "Idea Man" <[EMAIL PROTECTED]> writes:
> > >
> > > > Does anyone else think this would be a good idea? Keep the i386 tree, and
> > > > add an i686 tree that is optimized for P-II/Celeron/P-III processors.
> > > >
> > > > This might be a pain in the butt for the mirrors (more hard drive space
> > > > used), but for some mirrors this would be just fine. This would also make
> > > > Linux higher performing for all the people with flashy new Pentium-III
> > > > machines...
> > >
> > > How much performance improvement would there be?
> >
> > based on my experience with egcs over the past year, not much.
> >
> > the pentium classic seems to be hypersensitive to scheduling, but the
> > i686 (i have a pentiumpro) seems have roughly the same performance
> > (using time on a few of my programs) for compiles with -march=i386,
> > i486 or pentiumpro. -march=pentium hurt speed by about 10%.
> >
> Actually, if You have FP-heavy code (graphics, raytracing, FEA, CFD,
> Stats, simulation), P6-optimised code can gain You as much as 15-20%,
> compared to i386.
I seem to recall that i686 code will still run on the i386, right? If
so the best thing to do is compile it for the 686 and call it an i386
binary. Binaries that need to go faster on a 686 mostly won't be
worth running on an i386 anyway.
--
David Fox http://hci.ucsd.edu/dsf xoF divaD
UCSD HCI Lab baL ICH DSCU
------------------------------
From: "G. Sumner Hayes" <[EMAIL PROTECTED]>
Subject: Re: Security issues
Date: Sun, 28 Mar 1999 18:59:43 -0500
Michael Schuerig wrote:
> I'm interested in the Unix/Linux environment in particular. I'm
> looking for advice similar to "If you really need to give root
> privileges to a program, then do it like this...".
Search engines are your friend. If you use them before posting,
you might get faster results and you'll help cut down on the
number of repeat questions.
My first search (entering "secure setuid programming" at
http://www.google.com) revealed:
http://c5.hakker.com/links/
http://seclab.cs.ucdavis.edu/~bishop/secprog.html
The second has a number of papers, some slightly dated but still
very useful. They're all worth a read.
The first one includes links to:
The UNIX secure programming FAQ: Just what it sounds like.
Secure programming: How to write secure setuid root programs
Designing Secure Software: Tips for writing secure software
Secure Checklist: Checklist for secure UNIX programming
Security holes: Howto find security holes
setuid7(): Checklist for setuid programs
Perl Security: Perl scripting security information
Robust programing: Lets make the code robust.
Security code review: Security Code Review Guidelines
Is this what you're looking for?
Good luck,
Sumner
------------------------------
From: John Myers <[EMAIL PROTECTED]>
Subject: Re: After Week 1 With Linux -- licking wounds.
Date: 29 Mar 1999 00:31:56 GMT
Very well put "Rupert" !!
Rupert K. Snoopowitz wrote:
>
> This marks the end of my first week with Linux.
>
> First, let me state: This is not a troll. I am going to criticize
> Linux, but not bash it. I have a great deal of respect for it and the
> enormous amount of effort so may have put in to developing and
advocating
> it.
I submitted a question with some similar rantings attached, and was asked
if "this was a troll"...I did not know what "troll" meant, but the answer
in my case is also NO, it is not. I was just looking for some constructive
help...before scrapping the whole linux idea.
> I am a long time NT user and fan, and before that a big Mac guy -- ever
> since they first came out. I am a programmer, database developer, and
> web application developer. I am not a "newbie", but I'm not a hard-core
> geek either. My experience with Unix thus far has been using it lightly
> to serve Perl scripts through an ISP years ago.
Nor am I a "newbie" (except where linux is concerned). I have configured
and maintained DOS, Windows, and NetWare systems for many years, and have
at least a rudimentary understanding of C. After one very frustrating
month with debian linux, I can't even get this box to send garbage to the
printer... X-Windows is installed and working, but seems to be simply
a "brighter" version of the command line...very impressive!
I won't requote your entire message, but it did my heart good to read it.
I have read this forum extensively, and as a result have been hesitant to
submit my own basic questions. The standard response ...read the docs
(/usr/doc/whatever/examples/examples/examples/what.the.#%&#$%!!!!), is
somewhat insulting. I have read hundreds of pages of docs, FAQs, man pages
and have found precious few answers to seemingly simple questions. The
resultant conclusion from geekdom, of course, is that I must be a moron.
I've been thought of and called worse, but this does not answer my
questions, or help to get my system operating.
I was interested in a multi-tasking, multi-user operating system for use
in my office. To accomplish real work. I am not interested in endlessly
playing with a sophisticated "geek toy". I am willing to put in the time
and effort required, so long as that investment remains reasonable and
shows SOME return. To date debian. X-Windows, Samba have not met this
requirement.
I have not given up on linux, yet, but if I have learned anything from
this project it is that Mr. Gates needn't lose any sleep over the "linux
threat"
Still trying (for a while, anyway)
Thanks for listening
John Myers
================== Posted via SearchLinux ==================
http://www.searchlinux.com
------------------------------
From: [EMAIL PROTECTED] (Conrad Sanderson)
Crossposted-To:
comp.os.linux.advocacy,comp.os.linux.development.apps,comp.os.linux.help,comp.unix.programmer
Subject: Re: Programming tools for Linux/Unix: Editor, IDE, Frontend to GCC.
Date: 29 Mar 1999 01:43:44 GMT
On Sat, 27 Mar 1999 16:53:01 -0800, FM <[EMAIL PROTECTED]> wrote:
> What do most Linux programmers use as programming tools? I'm
> currently using Win98 despite its extreme instability and other
> fallacies because I do not at the moment have my own computer.
A decent editor is enough. People will probably suggest some
mammoth like emacs or Xemacs, which is fine if you want to
take some time to learn all the commands. Don't go near anything
known as "vi" or "vim" - you'd die from frustration coming
from a Windoze environment.
A good editor is Nedit - menu driven and still very powerful.
Syntax highlighting, its own macro language, etc.
You can find a binary of it on ftp://ftp.fnal.gov
or on the Red Hat PowerTools CD, but Red Hat's "version" uses
Lesstif (a free cloned version of Motif), which isn't working
quite well yet.
--
Conrad Sanderson - Microelectronic Signal Processing Laboratory
Griffith University, Queensland, Australia
http://hive.me.gu.edu.au/~cam/
------------------------------
From: [EMAIL PROTECTED] (Conrad Sanderson)
Subject: bizzare problem with kernel 2.2.3
Date: 29 Mar 1999 01:34:32 GMT
I've recently upgraded to kernel 2.2.3 and have been running
it for about a week non-stop, until suddenly it wouldn't run
apps anymore - ie. shell windows refused to open. I've logged out
of the X session and logged back in straight away, and the problem
seemed to have gone. But today trying to load netscape
resulted in this error: "too many files open in system".
A reboot was the only way to fix things.
--
Conrad Sanderson - Microelectronic Signal Processing Laboratory
Griffith University, Queensland, Australia
http://hive.me.gu.edu.au/~cam/
------------------------------
From: [EMAIL PROTECTED] (Andy Isaacson)
Subject: Re: no EOF from /dev/ram
Date: 29 Mar 1999 02:01:30 GMT
In article <_fWK2.116$[EMAIL PROTECTED]>, Phil Howard wrote:
> I realize that /dev/ram is an expandable device. But then again,
> regular files are expandable, too. When I read /dev/ram sequentially,
> I don't get EOF at the end. Instead, I get I/O error.
>
> Is this a mistake or a feature? If feature, what good is it?
>
> Actually, what I'd like to see is a ram-filesystem. This would be
> a filesystem that itself resides in ram (not formattable on a device)
> and can be expanded as the file contents need, until it reaches
> configured or physical limitations.
Sounds a lot like Solaris' tmpfs. tmpfs uses swap as backing store,
and is not saved across reboot. This is an even bigger advantage on
Solaris than it would probably be on Linux, because UFS does
synchronous metadata updates and so creating temporary files on real
filesystems takes at least one disk write (4ms probably). Creating a
file on tmpfs takes just a couple memory writes. On Linux's ext2fs,
in contrast, metadata updates are asynch and therefore creating a
temporary file does not take as long.
I would love to see tmpfs for linux, but apparently it's hampered by
the kernel's memory management architecture. I don't know enough
about the issues to really comment, though.
-andy
------------------------------
From: Jeremy Taylor <[EMAIL PROTECTED]>
Subject: Re: After Week 1 With Linux -- licking wounds.
Date: Mon, 29 Mar 1999 12:57:22 +1200
Reply-To: [EMAIL PROTECTED]
John Myers wrote:
> I was interested in a multi-tasking, multi-user operating system for use
> in my office. To accomplish real work. I am not interested in endlessly
> playing with a sophisticated "geek toy". I am willing to put in the time
> and effort required, so long as that investment remains reasonable and
> shows SOME return. To date debian. X-Windows, Samba have not met this
> requirement.
> I have not given up on linux, yet, but if I have learned anything from
> this project it is that Mr. Gates needn't lose any sleep over the "linux
> threat"
>
> Still trying (for a while, anyway)
>
> Thanks for listening
>
> John Myers
>
Perhaps you should consider purchasing a turnkey linux system from a
linux-friendly supplier? Then you get the value they add - ensuring that
when you turn the system on it is ready to go, re-suited to your needs?
You also then know you need not worry about configuring the hardware at
all, it has been done for you. You only need to know a few things about
administering the system, and you are up and running...
Many Linux users enjoy doing all this themselves. However, it sounds to
me as if you have a simple business requirement, are willing to put in a
little effort, but don't want to have to learn ALL about Linux before
being useful with it, earning some return on investment with it. Sounds
to me like you need to get someone else to do the stuff you don't care
about, so you can concentrate on the stuff you *do* care about.
Regards
Manaia
------------------------------
From: David Pace <[EMAIL PROTECTED]>
Crossposted-To:
comp.os.linux.advocacy,comp.os.linux.development.apps,comp.os.linux.help,comp.unix.programmer
Subject: Re: Programming tools for Linux/Unix: Editor, IDE, Frontend to GCC.
Date: Sun, 28 Mar 1999 10:25:36 -0500
FM wrote:
> What do most Linux programmers use as programming tools? I'm
> currently using Win98 despite its extreme instability and other
> fallacies because I do not at the moment have my own computer.
> I'll soon get my own computer, hopefully this summer or fall,
> and do not wish to tolerate any more MS nonsense on it,
> although I might reserve a small partition for compatibility
> reasons. While I've used Linux before, my experience has been
> somewhat limited and my programming experience has been entirely
> confined to the Win/Dos environment. This also means that I'm
> quite accustomed to Borland Win/Dos interface, including RHIDE,
> a front-end to DJGPP, a DOS port of GCC. I would just like to
> know what tools the actual linux/unix programmers use and how
> they compare to typical Windows/Dos IDE's. Are there GUI/Menu-
> driven programming environments available? If so, are they widely
> used? Why/why not? In general, I'd like a comparison of various
> Unix programming tools/environments/sets of tools. Thanks in
> advance.
I just started searching for one of these, for C/C++ (gcc) and I
am trying gIDE. It is new and has some annoying bugs in the editor
like cut/paste failing sometimes. But, it looks very promising.
There are a bunch of choices in http://www.freshmeat.net
Look in appindex, development - Environment
and development tools, et cetera.
Also, check out:
http://www.xnet.com/~blatura/linapp3.html
http://www.linuxapps.com/
--
Free commodity/stock graphing software
and Linux links at http://www.daveware.com
------------------------------
From: Todd <[EMAIL PROTECTED]>
Subject: IP Checsums bad
Date: Mon, 29 Mar 1999 01:47:02 GMT
I have been trying to debug a problem that involves bad IP checksums.
First a description of my configuration:
+-----------------------+
10.5.1.14 ----eth---|-10.5.0.21----10.9.1.1-|--p-t-p--10.9.1.2
+-----------------------+
10.5.0.12 and 10.9.1.1 are the ethernet and point-to-point addresses,
respectively, for a single system. The 10.9.1.2 ---> 10.9.1.1
connection is between two modems over a PRI line terminated by an ISDN
board in the 10.5.0.21/10.9.1.1 system. The ISDN board has is running a
driver written by the company that I work for. This system is
configured to use synchronous PPP (isdn4linux), with van Jacobson
compression DISABLED (i.e. not even compiled into the kernel).
The purpose of this setup is to test the ISDN board contained in the
10.5.0.21 system. Here is what I have seen happen:
1. If I attempt a TCP connection from 10.9.1.2 to 10.5.1.14 using
'netperf', the test never even starts. Although, I can verify that the
socket connection has been established by using 'netstat.'
2. An FTP connection from 10.9.1.2 to 10.5.1.14 can be successfully
opened, however, if I attempt to pass even minimal data (e.g. run an
'ls' command), the connection hangs.
3. If I initiate the FTP connection from the 10.5.1.14 end and do a
'get' from the the 10.9.1.2 machine, it works, albeit a little slow (but
this could be the fact that the driver for the ISDN board is beta).
4. If I attempt a 'netperf' TCP test from 10.9.1.2 to 10.5.0.21, it
works, however, when I monitor the 10.5.0.21 end of the point-to-point
connection, I notice a very high rate of IP packet checksum errors
(approximately 90%-95% of the IP packets). This could also account for
the poor performance since TCP must keep retransmitting.
5. If I run the 'netperf' test from 10.9.1.2 to 10.5.0.21 using a UDP
stream, performance is slightly better and the number of IP checksum
errors drops to approximately 5% or less. In addition, the checksum
error count only increases during the very beginning and end of the
test, when, apparently, 'netperf' must do some housekeeping stuff that
requires bi-directional communication.
Based upon the above, it appears to me that the problem could be in one
of two places:
1. The driver (most likely culprit);
2. The synchronous PPP code.
In order to isolate the problem, I need to find out where the problem
lies, I need to find out where the the IP checksums are being done and
reported as bad. To the best of my knowledge, the driver code does not
do any checksum calculations (I didn't write it, I just started looking
at the code in an attempt to figure out what's going on here. The
person that wrote it is no longer with our company.), it just passes the
socket buffers on to the board and the firmware does some checksums, but
this is for the HDLC framing.
One of the thoughts that I had was to track the number of socket buffers
passed though the driver for the ISDN board and compare that with the
packets recorded by 'iptraf' at the point-to-point interface. This
would tell me whether or not all of the packets were making it to the
driver. If they are, I presume that that would confirm the suspiscion
of a driver bug.
So finally, here are my questions:
1. At what points are IP checksums calculated and how is the information
regarding a bad checksum passed back to whoever needs it? I assume
'errno', but have not yet looked at the kernel code.
2. Does a socket buffer correspond to a single IP packet? Or is it
possible that the IP packet has been fragmented by the time that I see
it in the driver as a socket buffer?
3. I have been using 'iptraf' in order to gather my statistics regarding
bad packets. Is there anything that I should know about this program
(e.g. where it derives its data from) in order to keep from heading down
the wrong path?
4. Does anybody have any suggestions?
Please keep in mind the following:
1. I'm not asking anyone to solve my problem for me (after all, that's
part of the fun), just point me in the right direction and I'll be
happy.
2. I am relatively inexpreienced in this area, so if it looks like I
don't know what I'm talking about, that's because, in all likelihood, I
don't. However, if you do notice something that doesn't look right,
PLEASE point it out so that I can learn from my mistakes.
I very much appreciate any help that I get.
--
Todd Urie
[EMAIL PROTECTED]
------------------------------
From: Amy Bellows <[EMAIL PROTECTED]>
Crossposted-To:
comp.os.linux.advocacy,comp.os.linux.development.apps,comp.os.linux.help,comp.unix.programmer
Subject: Re: Programming tools for Linux/Unix: Editor, IDE, Frontend to GCC.
Date: Sun, 28 Mar 1999 20:06:31 -0600
Does anyone (but me) use Pico?
David M. Cook wrote:
> On Sat, 27 Mar 1999 16:53:01 -0800, FM <[EMAIL PROTECTED]> wrote:
>
> >What do most Linux programmers use as programming tools?
>
> I've never seen any kind of survey about what editors productive Linux
> programmers use. vim is popular, as is emacs, but you should not feel
> obligated to use what other people use. For C programming most everyone
> will use the same tools: gcc, make, rcs, gdb.
>
> >quite accustomed to Borland Win/Dos interface, including RHIDE,
>
> You can use RHIDE under Linux, but AFAIK, it only works in the console. You
> can easily switch between console and X, but it's still inconvenient if you
> are writing GUI apps.
>
> >they compare to typical Windows/Dos IDE's. Are there GUI/Menu-
> >driven programming environments available? If so, are they widely
>
> See http://members.home.com/davecook/devel/
>
> Dave Cook
> --
> No Linux for you!
--
reply to enigma(ninetynine)atearthlinkdotnet
================================================================================
Have you FreeBSD today?
Your MUD or mine?
I have a UNIX perspective on life.
------------------------------
From: [EMAIL PROTECTED]
Subject: Re: no EOF from /dev/ram
Date: 29 Mar 1999 02:12:35 GMT
In article <_fWK2.116$[EMAIL PROTECTED]>,
Phil Howard <[EMAIL PROTECTED]> wrote:
>Actually, what I'd like to see is a ram-filesystem. This would be
>a filesystem that itself resides in ram (not formattable on a device)
>and can be expanded as the file contents need, until it reaches
>configured or physical limitations.
I'm not sure what this buys you over the existing ram disk implementation,
where blocks for the file system are taken from the buffer cache as needed.
The size you specify when you make the filesystem is a maximum size, but blocks
of RAM are only allocated to it as it fills up so it acts almost exactly as
you describe regardless what type of filesystem you make on it.
Regards,
Graham
------------------------------
** 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
******************************
