Linux-Development-Sys Digest #880, Volume #7 Fri, 19 May 00 14:13:16 EDT
Contents:
Re: Need ideas for university funded project for linux (Leslie Mikesell)
Interrupt handler not being called. (Ian Collinson)
Re: porting NT device drivers to linux (Paul D. Smith)
debugging ("Aurelie Fonteny")
Re: zip with password ([EMAIL PROTECTED])
R: embedded Linux system ("Fabrizio C.")
Re: polling device drivers? (Rick Ellis)
Re: Need ideas for university funded project for linux (Leslie Mikesell)
Re: Need ideas for university funded project for linux (brian moore)
Re: Need ideas for university funded project for linux (brian moore)
Re: linux kernel not C++ friendly? (Bryan Hackney)
Re: Need ideas for university funded project for linux (brian moore)
Re: ping souce code (Mario Klebsch)
simple wait semaphore in kernel module (Travis Hein)
----------------------------------------------------------------------------
From: [EMAIL PROTECTED] (Leslie Mikesell)
Crossposted-To:
comp.os.linux,comp.os.linux.development,comp.os.linux.development.apps,comp.os.linux.misc,comp.os.linux.setup,comp.os.linux.advocacy
Subject: Re: Need ideas for university funded project for linux
Date: 19 May 2000 10:05:39 -0500
In article <8g2oqh$4ij$[EMAIL PROTECTED]>,
Peter T. Breuer <[EMAIL PROTECTED]> wrote:
>: On an stock rpm-installed Redhat - and Mandrake:
>: /usr/bin/netscape
>: /usr/bin/netscape-communicator
>: /usr/bin/netscape-navigator
>
>:-). Well, that's wrong then. Netscape is not part of a distribution
>in any sense I can think of and its the single thing that's most likely
>not to have come from the original distro o my system.
RedHat seems to take the approach that if they are allowed to
include it on the CD, it is part of the distribution and belongs
in /usr/bin.
>Surely it should
>go in /opt! I.e. "large package put together by someone else". Or has
>someone finally understood a sufficient fraction of the source to
>actually be able to compile it meaningfully?
You can go to extremes either way - from any Linux packager's perspective
almost everything is ultimately done by someone else but the object
is to simplify the system as installed. I kind of like the /opt
configuration as a means to have a potential mount point there, but
I'd really rather have more thought put into what should be network
mounts, or a decent overlay filesystems to completely hide the
cruft. For the moment I'm willing to throw a 3 gig partition at
a combined / and /user and just not worry about it. It's hard to
buy a drive less than 8 gigs these days anyway.
Les Mikesell
[EMAIL PROTECTED]
------------------------------
From: Ian Collinson <[EMAIL PROTECTED]>
Subject: Interrupt handler not being called.
Date: Fri, 19 May 2000 16:35:25 +0100
Posting this for someone else. Please reply to him (adonovan) not me.
Hello everyone. I am developing a driver for a custom bit of PCI
hardware for an x86 PC. The device uses an interrupt, but my driver's
handler is not being called when the interrupt is active.
I can scope the lines on the card and on the motherboard, and it shows
that the PCI INTA# line is indeed being asserted (pulled low) when it
should be. I would expect that this should cause my interrupt handler
to be called, but it never is. My handler calls printk, and the
message is never found on the log (and yes, syslog.conf is all set up
fine). /proc/interrupts reports a figure of zero for the kirq_stat
field (number of times it has been called) for my interrupt, so it
seems the kernel does not believe my interrupt is being called either.
What I expect is that it should repeatedly call my interrupt handler
until the cause of the interrupt has been cleared. In fact, I have seen
it do this in the past so I am surprised that it does not now.
Other things of note:
- request_irq returns successfully; my interrupt is listed
in /proc/interrupts.
- enable_irq does nothing as the interrupt is already enabled (in fact
I get an "unbalanced" warning).
- I have tried temporarily removing all refs to cli/sti.
- I have tried using different physical PCI slots, as I remember
something about INTA#, INTB# etc lines being rotated from slot to slot
to balance the interrupt lines.
None of these has made any difference.
I am about to rebuild the kernel with some additional diagnostics in
the do_none "unhandled" interrupt handler and other places to see if I
can see my interrupt arriving.
Can anyone suggest what might be causing the asserted INTA# line not to
invoke my handler?
Many thanks in advance,
alan
PS. I would be very grateful if you could copy me by mail
([EMAIL PROTECTED]) on any replies. It's not because I'm too lazy
to read the group but because I am working at another company's site
and have no easy access to a news server.
------------------------------
From: [EMAIL PROTECTED] (Paul D. Smith)
Subject: Re: porting NT device drivers to linux
Date: 19 May 2000 12:05:25 -0400
Reply-To: [EMAIL PROTECTED]
%% [EMAIL PROTECTED] (Mario Klebsch) writes:
mk> [EMAIL PROTECTED] (Paul D. Smith) writes:
>> There's one other reason to use shared libs: you can swap out the
>> shared library with a newer version (if the interface is
>> sufficiently similar) without recompiling or re-installing any apps
>> that use it.
mk> Yes, this is right, but this often is the cause of serious
mk> trouble, too.
It's certainly true that you must be careful when doing this. It's also
true that people have screwed this up, and probably will continue to do
so. Almost as important as carefully handling the new shared lib is
properly creating the original!! You need to be careful that you only
export symbols that are really public, etc. to avoid people
"accidentally" relying on unsupported features, etc.
Nevertheless, it _can_ be a great boon when handled correctly. For
example, often you can install bugfix versions of shared libraries
without much trouble. Security bugs are often good candidates, since
they're critical, and need to be fixed quickly.
--
===============================================================================
Paul D. Smith <[EMAIL PROTECTED]> Network Management Development
"Please remain calm...I may be mad, but I am a professional." --Mad Scientist
===============================================================================
These are my opinions---Nortel Networks takes no responsibility for them.
------------------------------
From: "Aurelie Fonteny" <[EMAIL PROTECTED]>
Subject: debugging
Date: Fri, 19 May 2000 11:09:28 -0500
Hi all!
I've already wrote a message not long ago about the problem I have but I
think I wasn't clear enough.
I'm writing a device driver that is going to do compression and
decompression. Right now, I have to debug it and the best way to do this is
to print the packet into a file at the sender side before compression and
print the packet into a file at the receiver side after decompression, and
do a diff() on those files to see if the data after decompression and the
data before compression match.
In user space, it is very easy to write data into files just by including
the stdio.h library and calling fopen(), fwrite() and fclose().
But this doesn't work in Linux Kernel, and I've tried to do it with
printk(KERN_INFO""), but it's not really efficient because I still get the
line "May 19 08:54:57 milou kernel:" that doesn't match at both ends so, I
can't a diff after that.
So does anyone know about a better way to do that? Is there really no simple
way to write into a file from the kernel? Is there another way to be able to
capture the packets at the IP level or PPP level (with IP + UDP/TCP headers
and all different port numbers.)
Another advice?
Aurelie
------------------------------
From: [EMAIL PROTECTED]
Subject: Re: zip with password
Date: Fri, 19 May 2000 16:31:10 GMT
Christian Winter <[EMAIL PROTECTED]> wrote:
>> Can I get a zip version that does encrypt?
> Just look round the web, there are a lot of servers providing
> binary packages, most of them homed in europe.
It's also worth noting that breaking the zip encryption is a fairly
trivial task which can be accomplished in a matter of minutes. If you
really need more security than that and it's feasible, you should
consider using a better cipher.
--
Matt Gauthier <[EMAIL PROTECTED]>
------------------------------
From: "Fabrizio C." <[EMAIL PROTECTED]>
Crossposted-To: comp.os.linux.development.apps
Subject: R: embedded Linux system
Date: Fri, 19 May 2000 18:37:02 +0200
sllai <[EMAIL PROTECTED]> wrote in message [EMAIL PROTECTED]
> If I would like to bulid an application software that is using embedded
> linux system.
> What kind of embedded Linux , what kind of development tool and do I need
to
> have a X windows based computer.
> Can somebody provide some info regarding this
Start from here:
http://www.rtlinux.com/
Fabrizio
------------------------------
From: [EMAIL PROTECTED] (Rick Ellis)
Subject: Re: polling device drivers?
Date: 19 May 2000 17:07:02 GMT
In article <8ftfvo$8gi$[EMAIL PROTECTED]>, <[EMAIL PROTECTED]> wrote:
>i'm working on a developmental network driver, and i've been reading up
>on rubini and examining 3c509.c, etc.
>
>while the ultimate goal is an interrupt based driver, for target
>specific reasons, i won't be able to use interrupts to start with, and
>so need to use polling. in chapter 14, "interrupt-driven operation"
>rubini actually alludes to the fact that some kernel drivers do use
>polling, but unfortunately doesn't go into any more detail.
>
>can anyone give me some pointers on how i would go about this, or which
>kernel sources to check out?
Take a look at the lp driver.
--
http://www.fnet.net/~ellis/photo/linux.html
------------------------------
From: [EMAIL PROTECTED] (Leslie Mikesell)
Crossposted-To:
comp.os.linux,comp.os.linux.development,comp.os.linux.development.apps,comp.os.linux.misc,comp.os.linux.setup,comp.os.linux.advocacy
Subject: Re: Need ideas for university funded project for linux
Date: 19 May 2000 12:04:29 -0500
In article <[EMAIL PROTECTED]>,
Bob Hauck <hauck[at]codem{dot}com> wrote:
>>>Oh yes you can! Try FTE. It does exactly all what you requested :-)
>>>http://fte.sourceforge.net/
>
>0.49.13 comes with a slang version called "sfte". It isn't compiled by
>default, you have to edit the TARGETS line in ~/src/fte-unix.mak. Seems
>to work ok on the console, with the PuTTY telnet client for Windows, and
>in a kvt (with the exception of c-pgup and c-pgdn not working) but the
>keybindings get funky in an xterm (hint: alt -> esc).
This is pretty nice - it seems to be mostly functional in a
kconsole window except for selecting with the shift-motion
keys. And it doesn't replace selected text when you type
on top of it.
>I'm still playing around with it (downloaded when I saw the previous post)
>so don't yet have a good feel for how well it will hold up over the long
>term. But it is in fact pretty easy to get started with for DOS/Win
>converts due to the CUA-style default key bindings and menus, so it'll
>probably end up staying on my system for that reason alone.
I think it would be nice to have as close as possible to a dos
EDIT clone for people's first encounter with Linux and for when
you have to talk someone through fixing a config file. A tiny
version for use on rescue floppies would be especially good.
Sfte looks pretty close for the user interface, but it is
pretty big with all the syntax parsing stuff.
But, back to the question of a university project: how about a
real overlay filesystem where you could mount something else
transparently over an existing directory. The main use would
be to make a 'master' tree on a CD or network that you could
mount read-only and overlay your custom files for the specific
machine. Or overlay a ramdisk on top of a source tree for
fast compile/link/install times.
An interesting variation would be to do something similar at the
block device level. Build a log of block-writes where the contents
are different than the matching device block (a gdbm file might
work) and satisfy reads from the log if the block exists.
You would also need some tools to start and stop the logging
with an option to merge the changes back.
Les Mikesell
[EMAIL PROTECTED]
------------------------------
From: [EMAIL PROTECTED] (brian moore)
Crossposted-To:
comp.os.linux.development.apps,comp.os.linux.misc,comp.os.linux.setup,comp.os.linux.advocacy
Subject: Re: Need ideas for university funded project for linux
Date: 19 May 2000 17:49:31 GMT
On 19 May 2000 09:56:27 -0500,
Leslie Mikesell <[EMAIL PROTECTED]> wrote:
> In article <8g31si$6ri$[EMAIL PROTECTED]>,
> Peter T. Breuer <[EMAIL PROTECTED]> wrote:
>
> >/opt is for independent packages. But there is a new fsstd coming up.
>
> Oh good - we didn't have quite enough standard layouts already...
No, we don't have enough detail and vendors that are anal about
following the rules.
> More like 20 - and they still don't correctly address issues of
> having local copies AND (perhaps multiple) network-mounted directories
> of the same thing, or things where the config files should be
> network-shared but not the binaries or vice-versa.
Why would you local -and- shared copies of the same thing?
You may want to look at /usr/share, which is precisely for things that
can be shared across platforms. (Man pages, docs, pixmaps,
soundfiles, dictionaries, all sorts of things should be living in
/usr/share -- I've got a couple hundred meg worth of stuff there)
See http://www.pathname.com/fhs/
--
Brian Moore | Of course vi is God's editor.
Sysadmin, C/Perl Hacker | If He used Emacs, He'd still be waiting
Usenet Vandal | for it to load on the seventh day.
Netscum, Bane of Elves.
------------------------------
From: [EMAIL PROTECTED] (brian moore)
Crossposted-To:
comp.os.linux,comp.os.linux.development,comp.os.linux.development.apps,comp.os.linux.misc,comp.os.linux.setup,comp.os.linux.advocacy
Subject: Re: Need ideas for university funded project for linux
Date: 19 May 2000 17:53:10 GMT
On 19 May 2000 07:00:33 GMT,
Peter T. Breuer <[EMAIL PROTECTED]> wrote:
> In comp.os.linux.development Leslie Mikesell <[EMAIL PROTECTED]> wrote:
>
> :>I use /usr/local for things that weren't in my original system and
> :>aren't likely to be in it for the foreseeable future. Netscape would
> :>be an example, though I can't think of any good ones.
>
> : On an stock rpm-installed Redhat - and Mandrake:
> : /usr/bin/netscape
> : /usr/bin/netscape-communicator
> : /usr/bin/netscape-navigator
>
> :-). Well, that's wrong then. Netscape is not part of a distribution
> in any sense I can think of and its the single thing that's most likely
> not to have come from the original distro o my system. Surely it should
> go in /opt! I.e. "large package put together by someone else". Or has
> someone finally understood a sufficient fraction of the source to
> actually be able to compile it meaningfully?
Hrrm? Netscape is in the 'non-free' portion of debian and comes with
most (all?) other (non-tiny) distributions as well.
--
Brian Moore | Of course vi is God's editor.
Sysadmin, C/Perl Hacker | If He used Emacs, He'd still be waiting
Usenet Vandal | for it to load on the seventh day.
Netscum, Bane of Elves.
------------------------------
From: Bryan Hackney <[EMAIL PROTECTED]>
Subject: Re: linux kernel not C++ friendly?
Date: Fri, 19 May 2000 12:55:00 -0500
Kaz Kylheku wrote:
>
[...]
> >libc. You don't want to do this.
>
> I just wrote a module in C++ and it loaded just nicely into 2.2.12.
>
> Here is the module:
>
[...]
Bravo. Nothing like proving a point. Whether it's worth it is a judgement
call or a personal preference or something else, not a technical argument.
It's _amazing_ how many supposedly competent coders think C++ is mostly
about iostreams and operator overloading. It's also amazing how many
think C++ is an automatic performance hit.
Oh well. I think most people who hate C++ got introduced to it through
MFC, a big, nasty botch.
--
Bryan Hackney / BHC / [EMAIL PROTECTED]
http://www.FreeClassAds.com/
http://bhconsult.com/
http://bhconsult.com/bh/pgp.key
------------------------------
From: [EMAIL PROTECTED] (brian moore)
Crossposted-To:
comp.os.linux,comp.os.linux.development,comp.os.linux.development.apps,comp.os.linux.misc,comp.os.linux.setup,comp.os.linux.advocacy
Subject: Re: Need ideas for university funded project for linux
Date: 19 May 2000 17:56:28 GMT
On 19 May 2000 01:07:29 -0500,
Leslie Mikesell <[EMAIL PROTECTED]> wrote:
> In article <[EMAIL PROTECTED]>, <[EMAIL PROTECTED]> wrote:
>
> >> Never, never, never let user who doesn't understand things tweak the
> >> config files. For such users remote sysadmin service via SSH should be
> >> provided.
> >
> >Huh?
> >
> >Are you suggesting we start up a Centralized Linux Administration
> >Bureau or something? And remember that not all computers are on a
> >network, and very few of them are on one all the time.
>
> I've suggested something similar on the freebsd newsgroup before
> because they need it even worse, but they seem to think everyone
> should learn to be an expert.
>
> I think what we really need is some number of well-maintained
> 'master' system images (somewhere between 10 and 100 would
> suffice, but the number doesn't matter) and some tools
> to sync up your system to the master without breaking things
> due to hardware differences. Good system administrators
> would each maintain their own 'ideal' system as the master
> copy, tuned for whatever purpose they want. They would
> document the philosophy of the configuration (i.e. the
> purpose, not the details) and keep everything up to date,
> adding new things as they become available. This is work
> every system administrator does anyway - we just need the
> tools to share it easily. Then, instead of everyone having
> to configure and tune their own system, they would just pick
> a setup already built that matches their needs and periodically
> sync up any updates. There would still be a small amount of
> local setup to do, but the bulk of the work could be done.
That sounds like Debian.
Debian makes the sysadmin's job trivial.
--
Brian Moore | Of course vi is God's editor.
Sysadmin, C/Perl Hacker | If He used Emacs, He'd still be waiting
Usenet Vandal | for it to load on the seventh day.
Netscum, Bane of Elves.
------------------------------
From: [EMAIL PROTECTED] (Mario Klebsch)
Subject: Re: ping souce code
Date: Fri, 19 May 2000 18:22:00 +0200
[EMAIL PROTECTED] (Diego Betancor) writes:
> But does not tell you that the packet did not arrived. Just try to
>ping with verbose and then take out the ethernet cord and see if ping
>tells you anything.
$ ping -c 1 tabaluga
PING tabaluga (192.1.1.10): 56 data bytes
--- tabaluga ping statistics ---
1 packets transmitted, 0 packets received, 100% packet loss
$
It tells me 100% packet loss.
73, Mario
--
Mario Klebsch [EMAIL PROTECTED]
------------------------------
From: Travis Hein <[EMAIL PROTECTED]>
Subject: simple wait semaphore in kernel module
Date: Fri, 19 May 2000 13:49:29 -0400
looking for a way to implment a kernel thread to wait on another kernel thread
(in the same module) to complete something.
------------------------------
** 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
******************************
