Linux-Development-Sys Digest #906, Volume #7 Thu, 25 May 00 17:13:20 EDT
Contents:
Re: Writing driver for AMCC 5933 based PCI card (Bill Waddington)
Re: coredump files (John Gluck)
Re: Need ideas for university funded project for linux (John Hasler)
Re: unresolved symbol pmd_offset - huh? (Timur Tabi)
Re: Call BIOS from device diver ? (Alan Donovan)
Re: how-to develop? (John Gluck)
Re: exit system call troubles... (Alan Donovan)
Walking through Linux file system (Serge NOEL)
Re: Need ideas for university funded project for linux (Jim Richardson)
Re: Walking through Linux file system (H. Peter Anvin)
Re: 2.2 -> 2.3.99pre9 kernel API changes (Timothy J. Lee)
Source code for driver(s) of any of these NICs..... (Mark Darby)
Re: Need ideas for university funded project for linux (Nix)
Re: Need ideas for university funded project for linux (Nix)
Re: Need ideas for university funded project for linux (Nix)
Re: 2.2.15 and 2.3.99 USB-Backport and no mouse (Christian Winter)
Kernel module-user process comms (Dave Donaghy)
Driver level Ethernet High Availability ("Robichaud, Jean-Philippe [BAN:6S33:EXCH]")
Re: Need ideas for university funded project for linux (Nix)
virt_to_phys vs pgd/pmd/pte_offset calls (Timur Tabi)
----------------------------------------------------------------------------
From: Bill Waddington <[EMAIL PROTECTED]>
Subject: Re: Writing driver for AMCC 5933 based PCI card
Date: Thu, 25 May 2000 14:03:31 GMT
In article <[EMAIL PROTECTED]>,
[EMAIL PROTECTED] wrote:
> Thanks for reading my post Bill.
You're welcome. I hope I haven't given you any false hope - I'm just a
beginner myself. I have written equivalent DMA drivers for Solaris and
2.2.12-20 (RH 6.1). You are welcome to look at the code, but it
contains masses of stuff that doesn't relate to DMA.
http://home.att.net/~ikoncorp follow the FTP link.
Again, the code is wordy, and pretty ugly, but perhaps someting will
jump out at you. I would be happy to take a quick peek at your code if
it is available somewhere. No false hope, I'm not familiar with your
particular hardware, but you never know...
Bill
--
Bill Waddington
[EMAIL PROTECTED]
[EMAIL PROTECTED]
Sent via Deja.com http://www.deja.com/
Before you buy.
------------------------------
From: John Gluck <[EMAIL PROTECTED]>
Subject: Re: coredump files
Date: Thu, 25 May 2000 11:04:35 -0400
Gregory Taleck wrote:
>
> Is there a way on linux to place coredumps under different names and
> paths? For example, rather than dumping core in the current directory of
> the process under the filename 'core', can you specify a path and name
> format (e.g '/var/coredump/%s-$d.core')? I know there had been some
> interest in doing this on BSD but couldn't find anything on linux.
>
> Thanks,
> Greg
The unix way is to create a coredump where you're at. Part of the reason
is because there can be several users of the same app at the same time.
If user A crashes and a short time after user B crashes the same app,
the coredumps would trample each other if they were in the same place.
I have never really looked at the mechanism for producing coredumps. I
think it's in the kernel somewhere. I suppose it would be possible to
change it so that it does something like append the date and time to the
dump. Something along lines of "core_24052000.195400"
That way each core would have a different name. Still better would be to
add the name of the executable as well.
--
John Gluck (Passport Kernel Design Group)
(613) 765-8392 ESN 395-8392
Unless otherwise stated, any opinions expressed here are strictly my own
and do not reflect any official position of Nortel Networks.
------------------------------
From: John Hasler <[EMAIL PROTECTED]>
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: Thu, 25 May 2000 12:49:28 GMT
David Steuber writes:
> On my machines, I have code with quite a variety of licenses. Those
> include GPL, BSD, QPL, Perl's Artistic License, the TCL license, and a
> bunch of others. The base system is GNU/Linux. A bunch of libraries in
> use are either GPL or LGPL, including libc.
> Have I violated someone's license?
Not the GPL, certainly. The GPL is about distribution: you can do whatever
you want with GPL code in the privacy of you own machine.
--
John Hasler
[EMAIL PROTECTED] (John Hasler)
Dancing Horse Hill
Elmwood, WI
------------------------------
From: Timur Tabi <[EMAIL PROTECTED]>
Subject: Re: unresolved symbol pmd_offset - huh?
Date: Thu, 25 May 2000 15:27:58 GMT
That was it - my makefile was screwed up and didn't include the -O2 on
the compiler's command line. Thanks!
--
Timur Tabi
Interactive Silicon - http://www.interactivesi.com
Remove "nospam_" from my email address when replying
Sent via Deja.com http://www.deja.com/
Before you buy.
------------------------------
From: Alan Donovan <[EMAIL PROTECTED]>
Subject: Re: Call BIOS from device diver ?
Date: Thu, 25 May 2000 16:45:11 +0100
Robert Kaiser wrote:
> is it possible to call a PC's BIOS from the kernel (e.g.
> a device driver) ? Specifically, I would like to call int #13
> for Disk I/O. I'm aware that this would be terribly inefficient
> compared to a native Linux device driver, but would it be possible
> at all ? If it is, can anyone point me to an example in the kernel
> code ?
Besides the other comments, you're immediately tying yourself to x86
Linux too.
alan
--
========================================================================
Alan Donovan [EMAIL PROTECTED] http://www.imerge.co.uk
Imerge Ltd. +44 1223 875265
------------------------------
From: John Gluck <[EMAIL PROTECTED]>
Crossposted-To: comp.os.linux.development.apps,comp.os.linux.misc
Subject: Re: how-to develop?
Date: Thu, 25 May 2000 11:34:05 -0400
Nicola Attico wrote:
>
> Hello,
>
> I'm serching for a word of advice.
> I'm searching for a project to work at
> (starting from zero [maybe better] or
> running).
> The point is that I've not strong C
> knowledge, so I would like to learn...
> Anyway I'm not a total unable
> and I've years of experience using
> Linux and some practice in administration...
> I think moreover that the better way
> to learn how to write software is to write
> it, so I would like to learn doing it!
> So, I'm searching for a *baby* project,
> where it is possible to learn...
>
> Does it exists?
>
> Thanks,
>
> Nicola
I think a good way to get started is to take a simple utility program
that exists. Get the source and try to add a feature or 2 to it.
After trying to do something like that a few times, you'll start getting
the hang of things.
The next step is to try and write a small program from scratch. What you
do doesn't need to be useful for any purpose other than to help you
learn.
An example of such a program is a program that simulates a lottery
drawing. It picks say 6 random numbers from a possible 76 and lists the
results in ascending order. It's a useless program but it gets you using
things like the random number generator and doing a sort.
--
John Gluck (Passport Kernel Design Group)
(613) 765-8392 ESN 395-8392
Unless otherwise stated, any opinions expressed here are strictly my own
and do not reflect any official position of Nortel Networks.
------------------------------
From: Alan Donovan <[EMAIL PROTECTED]>
Subject: Re: exit system call troubles...
Date: Thu, 25 May 2000 16:52:04 +0100
Tim Muir wrote:
> Does the exit system call occur EVERY time a process exits, or is there
> another one I should be looking at ?
Either:
1) a process calls the _exit() system call (note the underscore: exit(3)
is part of the C library; _exit(2) is a UNIX syscall). Remember, _exit
never returns as there is no process for it to return to as it causes
the OS to remove the process from the process table. Or:
2) it encounters a signal which causes the OS to kill it with extreme
prejudice.
See WIFEXITED/WIFSIGNALED in man waitpid(2)
alan
--
========================================================================
Alan Donovan [EMAIL PROTECTED] http://www.imerge.co.uk
Imerge Ltd. +44 1223 875265
------------------------------
From: Serge NOEL <[EMAIL PROTECTED]>
Subject: Walking through Linux file system
Date: 25 May 2000 16:26:00 GMT
Hi, I would walk in the tree of a linux filesystem, under �$Oft, i use t
he FindFile for walking. But with Linux, I do'nt know how to do that, I'
ve seen the getdents() but the dent entrie does'nt say if you see a dire
ctory or a file. Perhaps is-it the wrong way ? Could someone help me to
do that.
I would :
Begin in a tree. Scan the entries and determine if the answer if a fil
e, a link, a directory, ...
Thank you in advance.
Serge NOEL
=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-
Article poste via Voila News - http://www.news.voila.fr
Le : Thu May 25 18:25:59 2000 depuis l'IP : 193.56.211.252 [VIP 3015391]
------------------------------
From: [EMAIL PROTECTED] (Jim Richardson)
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: Wed, 24 May 2000 22:52:38 -0700
Reply-To: [EMAIL PROTECTED]
On 25 May 2000 04:54:09 GMT,
Peter T. Breuer, in the persona of <[EMAIL PROTECTED]>,
brought forth the following words...:
>In comp.os.linux.misc Jim Richardson <[EMAIL PROTECTED]> wrote:
>: Sure, you can replicate the functionality in RPM or Deb easily enough, but
>: not with just the ./configure;make;make install mentioned. (At least not
>: without the connivence of the writer of the ./configure script.) RPM allready
>
>setenv INSTALL "pkginstall install -c"
>
>(and thus log what goes where)
>
>Peter
Having logged it, what tool do you use to check before removing/upgrading
something?
(this is slackware you are discussing, right? )
--
Jim Richardson
Anarchist, pagan and proud of it
WWW.eskimo.com/~warlock
Linux, because life's too short for a buggy OS.
------------------------------
From: [EMAIL PROTECTED] (H. Peter Anvin)
Subject: Re: Walking through Linux file system
Reply-To: [EMAIL PROTECTED] (H. Peter Anvin)
Date: 25 May 2000 09:58:47 -0700
Followup to: <8gjk6o$n0i$[EMAIL PROTECTED]>
By author: Serge NOEL <[EMAIL PROTECTED]>
In newsgroup: comp.os.linux.development.system
>
> Hi, I would walk in the tree of a linux filesystem, under �$Oft, i use t
> he FindFile for walking. But with Linux, I do'nt know how to do that, I'
> ve seen the getdents() but the dent entrie does'nt say if you see a dire
> ctory or a file. Perhaps is-it the wrong way ? Could someone help me to
> do that.
>
> I would :
> Begin in a tree. Scan the entries and determine if the answer if a fil
> e, a link, a directory, ...
>
RTFM(opendir, readdir, closedir, stat)
-hpa
--
<[EMAIL PROTECTED]> at work, <[EMAIL PROTECTED]> in private!
"Unix gives you enough rope to shoot yourself in the foot."
------------------------------
From: [EMAIL PROTECTED] (Timothy J. Lee)
Subject: Re: 2.2 -> 2.3.99pre9 kernel API changes
Date: 25 May 2000 17:50:33 GMT
Reply-To: see-signature-for-email-address---junk-not-welcome
[EMAIL PROTECTED] (Aki M Laukkanen) writes:
|Unfortunately there is no comprehensive list. Softnet specific changes
|however are detailed in a separate document distributed on the netdev
|and linux-kernel lists:
|
|http://www.uwsg.indiana.edu/hypermail/linux/kernel/0001.0/0229.html
Thanks for the pointer.
At what kernel version did these changes to the network driver API
occur? (needed for #if in the drivers)
--
========================================================================
Timothy J. Lee timlee@
Unsolicited bulk or commercial email is not welcome. netcom.com
No warranty of any kind is provided with this message.
------------------------------
From: Mark Darby <[EMAIL PROTECTED]>
Subject: Source code for driver(s) of any of these NICs.....
Date: Thu, 25 May 2000 14:31:11 -0400
All:
I am searching for linux driver module source code for
any of the following NIC cards:
Netgear PA301
3Com 3c410
Linksys HPN200
Intel AnyPoint Phoneline 10M PCI
All of these cards support the HPNA 2.0 standard (10Mbps over
traditional home phone wiring). Unless I am mistaken,
all of these cards are made with the Broadcom iLine10
controller chip.
Any help or pointers would be greatly appreciated.
Regards,
Mark Darby
[EMAIL PROTECTED] (home)
[EMAIL PROTECTED] (work)
------------------------------
From: Nix <$}xinix{[email protected]>
Crossposted-To:
comp.os.linux.development,comp.os.linux.development.apps,comp.os.linux.misc
Subject: Re: Need ideas for university funded project for linux
Date: 25 May 2000 20:35:56 +0100
[EMAIL PROTECTED] (Christopher Browne) writes:
[config tools]
> I _seriously wish_ that these tools took the approach of not merely
> modifying files, but rather generating some form of "closures."
I wholeheartedly agree.
> The idea is that rather than modifying the file, the tools should
> generate _little programs_ (e.g. - closures) to modify the files.
> That way, the tool may store the _program_, and provide the ability to
> rerun as needed.
One program does this already; autoconf. `configure' does not generate
any files at all; an AC_OUTPUT generates code in `configure' which
causes a config.status to be generated, and then *that* is run to
generate all the substituted output files. And, of course, you can run
them again to regenerate them, and automake-generated Makefiles do that
automatically as needed.
> (Theoretical CS Geek Aside: When you compile a C
> program using GCC, what GCC does is to transform C code into a whole
> bunch of machine language closures.)
Just because RTL looks like Lisp doesn't mean it shares many Lispy
properties. Among those it doesn't share is closure semantics.
The Lisp-likeness is only a facade; the precise form of RTL is very
machine-specific, and not anywhere near Lisp in expressivity. It is
notable that GCC is moving more and more towards doing things as trees,
because they retain far more context than does RTL, so optimizations can
be better.
> d) Lisp Machines
Emacs ;}
(Well, no, not enough :( )
--
`Q: Why did they deprecate a.out support in linux?
A: Because a nasty coff is bad for your elf.' --- James Simmons
------------------------------
From: Nix <$}xinix{[email protected]>
Crossposted-To:
comp.os.linux,comp.os.linux.development,comp.os.linux.development.apps,comp.os.linux.misc,comp.os.linux.setup,comp.os.linux.advocacy
Subject: Re: Need ideas for university funded project for linux
Date: 25 May 2000 20:49:48 +0100
[EMAIL PROTECTED] writes:
> [1] Having two separate "enlightenment" and "enlightenment-nosound"
> packages that differ only in their dependencies is, in my opinion,
> broken packaging.
You *do* comprehend the concept of virtual packages, don't you?
(That is, `this is not a bug. This is a feature.')
--
`Q: Why did they deprecate a.out support in linux?
A: Because a nasty coff is bad for your elf.' --- James Simmons
------------------------------
From: Nix <$}xinix{[email protected]>
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: 25 May 2000 20:51:49 +0100
Johan Kullstam <[EMAIL PROTECTED]> writes:
> the problem isn't with the install, it's when you go to remove or
> upgrade. sometimes it's hard (or at least tedious) to figure out what
> all things went where. then when you find a random file, you wonder
> where it came from.
GNU stow fixes this pretty well. STORE fixes it far too well (the same
way as stow) and is so complicated that nobody can work out how to use
it ;)
If you need external databases to work out where things come from, you
are in trouble if they get corrupted. If the database is the filesystem
itself, then that adds no additional points of vulnerability (as you are
using the fs anyway).
--
`Q: Why did they deprecate a.out support in linux?
A: Because a nasty coff is bad for your elf.' --- James Simmons
------------------------------
From: Christian Winter <[EMAIL PROTECTED]>
Subject: Re: 2.2.15 and 2.3.99 USB-Backport and no mouse
Date: Thu, 25 May 2000 14:39:19 +0200
Konstantinos Agouros <[EMAIL PROTECTED]> schrob:
> I tried the above combination and I can't use the mousedev anymore. I took
> the kernel-config from 2.2.14/2.3.50-Backport that worked. Anybody has an
> idea what might have changed?
Well, it seems like you formerly did use the "old" usb driver that
put its mouse devices into /dev/usb. The new driver wants to
access the mouse device in /dev/input/mouseXX (or all mice together
in /dev/input/mice).
>From devices.txt:
============================================================================
13 char Input drivers
32 = /dev/input/mouse0 # first mouse
33 = /dev/input/mouse1 # second mouse
...
64 = /dev/input/mice # Mixed input from all mice
The current input system is used for USB devices only, however,
it isn't limited just to those, and hopefully will be the way
to access all input devices in the future.
============================================================================
As USB for Linux is still beta, I'd suggest you join the linux-usb mailing
list, where all new patches, tricks and compatibility issues are discussed.
To do that, simply send a mail to [EMAIL PROTECTED] with a body of
"subscribe linux-usb".
HTH
Christian
------------------------------
From: Dave Donaghy <[EMAIL PROTECTED]>
Subject: Kernel module-user process comms
Date: Thu, 25 May 2000 22:39:45 +0100
This is a multi-part message in MIME format.
==============B3A8267B98345A1AAF0F1B76
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit
I hope this isn't too stupid a question, but I've read Rubini, and I've
examined the source code, and I've played around (a lot!) and I'm still
stuck.
I'm trying to do some comms between a kernel module and a user process.
Sync would be nice (!), async will do.
I stumbled across sys_msgget etc, and got my hopes up, even though these
things aren't exported to modules. (Re-compiled the kernel and exported
them before I realised that the reason they're not exported is that
they're pretty much useless to modules.)
Is there anything obvious I'm missing, or do I have to all the work
myself?
Dave
==============B3A8267B98345A1AAF0F1B76
Content-Type: text/x-vcard; charset=us-ascii;
name="dave.donaghy.vcf"
Content-Transfer-Encoding: 7bit
Content-Description: Card for Dave Donaghy
Content-Disposition: attachment;
filename="dave.donaghy.vcf"
begin:vcard
n:Donaghy;Dave
x-mozilla-html:FALSE
adr:;;;;;;
version:2.1
email;internet:[EMAIL PROTECTED]
x-mozilla-cpt:;-4160
fn:Dave Donaghy
end:vcard
==============B3A8267B98345A1AAF0F1B76==
------------------------------
From: "Robichaud, Jean-Philippe [BAN:6S33:EXCH]" <[EMAIL PROTECTED]>
Subject: Driver level Ethernet High Availability
Date: Fri, 26 May 2000 04:26:18 -0400
Hi, I must implement an high availability thing that must be really
rapid. That's why I want to implement it at the driver level. I have 2
intel etherexpress100 card and I have the source code of the driver for
linux. (I'm on RedHat 6.1, kernel : 2.2.12-20.) Does anyone can tell me
how must I modify the driver so that he will ping eth0 and if it fails,
raise eth1 up with the same IP address as eth0 had ?
Thanks for your help, this is my first Linux program !!!
Jean-Philippe Robichaud
[EMAIL PROTECTED]
514-818-7750
------------------------------
From: Nix <$}xinix{[email protected]>
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: 25 May 2000 21:46:51 +0100
[EMAIL PROTECTED] (JEDIDIAH) writes:
> Given the non-package specific dependency checking tools already in
> Linux, one would think that you could build a dependency database
> quite independent of rpm or deb.
You can. It's called `the filesystem'.
(GNU stow, opt-depot, STORE and similar programs automate this process.)
--
`Q: Why did they deprecate a.out support in linux?
A: Because a nasty coff is bad for your elf.' --- James Simmons
------------------------------
From: Timur Tabi <[EMAIL PROTECTED]>
Subject: virt_to_phys vs pgd/pmd/pte_offset calls
Date: Thu, 25 May 2000 20:54:31 GMT
I'm trying to understand the Linux memory manager, so I'm hoping someone
could answer this question for me.
I'm a little confused about the difference between the virt_to_phys
function and the combination of the pgd_offset,pmd_offset, and
pte_offset functions.
The pgd/pmd/pte stuff is used, AFAIK, to obtain a physical address of a
USER SPACE virtual address, like this:
pgd = pgd_offset(current->mm, address);
pmd = pmd_offset(pgd, address);
pte = pte_offset(pmd, address);
printk("page: %08lx\n", pte_page(*pte));
(please stop me if I say something wrong)
If I try this approach on a virtual address in a driver, this won't
work, because the driver's "virtual address" are mapped directly to
physical memory, and if I want to obtain the physical address, I need to
use the virt_to_phys macro.
In short, to obtain the physical address of a user-space virtual
address, use pgd/pmd/pte_offset. To obtain the physical address of a
kernel-space virtual address, use virt_to_phys.
Is that correct?
--
Timur "too sexy for my code" Tabi
Remove "nospam_" from my email address when replying
Sent via Deja.com http://www.deja.com/
Before you buy.
------------------------------
** 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
******************************
