Linux-Development-Sys Digest #777, Volume #6 Fri, 4 Jun 99 12:14:14 EDT
Contents:
Modules aren't loaded after 2.2.5 or 2.2.9 kernel recompile ("Ausias")
Framebuffer device? (Craig Graham)
Re: Redhat 6 & NFS (Bill Anderson)
Re: NFS Lock Deamon for Linux. Is there one? (Miquel van Smoorenburg)
Re: Kernel Changes 2.0.x->2.2.x???? (ellis)
Re: kernel vs egcs vs PentiumPro/II (Juergen Heinzl)
Re: Redhat 6 & NFS ([EMAIL PROTECTED])
Re: Strange behaviour of LINUX SMP under pvm (Axel Farr)
select() vs. poll() ([EMAIL PROTECTED])
Re: Configuration Manager for Linux (Vassilis Virvilis)
Re: *Very* large physical buffers. (Marius Vollmer)
Re: the ultimate OS ("Selious")
How can i debugging with kernel? (lckun)
Re: Linux development tools... (Vassilis Virvilis)
Re: TAO: the ultimate OS ("Keith Peterson")
help on kernel configuration ([EMAIL PROTECTED])
Re: What are the differences between mySQL and mSQL? (Don Baccus)
can 2 linux machines talk to a dual port scsi raid system ?? (dan)
----------------------------------------------------------------------------
From: "Ausias" <[EMAIL PROTECTED]>
Subject: Modules aren't loaded after 2.2.5 or 2.2.9 kernel recompile
Date: Thu, 3 Jun 1999 12:49:21 -0700
I recompiled my kernel, both the one from original install (2.2.5) and 2.2.9
from download a few times (reinstalled a few times) but the modules
components aren't loaded on bootup. They can't be loaded with insmod
either. Both ppp0 and eth0 were compiled as modules and neither can be
loaded. They were working before (after fresh install from disk). The
modules-info link in /boot still points to the old kernel's, the
/etc/conf.modules file doesn't change either. I suspect this is the cause
of the modules not loading properly. I've followed the instructions in the
readme's that come with the kernels, the HOWTO for the kernel and the
mini-HOWTO for modules (both on disk and on-line). Are the documents
missing something for making modules work? Am I going to have to alter the
conf files manually? If so, they should mention in the manual that the
sequence of commands they give you to compile the kernel using modules isn't
complete. Is it just me?
Ausias
------------------------------
From: Craig Graham <[EMAIL PROTECTED]>
Subject: Framebuffer device?
Date: Fri, 04 Jun 1999 11:52:34 +0000
How do I configure a 2.2.x kernal to enable
the /dev/fb? framebuffer device?
I've tried config/rebuild but it still refuses
to appear...
Craig Graham.
------------------------------
From: Bill Anderson <[EMAIL PROTECTED]>
Subject: Re: Redhat 6 & NFS
Date: Thu, 03 Jun 1999 13:44:14 -0600
Mike Romberg wrote:
>
> >>>>> " " == James Linder <[EMAIL PROTECTED]> writes:
>
> > Hi The modular kernel that RH supplies with RH6.0 works with
> > NFS. Doing lsmod shows: nfs nfsd .. I'm trying to build a
> > kernel, but the nfsd will not run and the fs/nfsd directory is
> > not used. What do I do to build a kernel with NFS built in. I
> > already have the NFS stuff configured, but I cannot see how
> > CONFIG_NFSD gets to be set, or what I need to do to get nfs
> > working. Trying to run nfsd yields an error "nfssvc not
> > implemented"
>
> I had to enable CONFIG_EXPERIMENTAL to get CONFIG_NFSD. Once
> CONFIG_NFSD was set (as a module) the nfsd fired up and seemed to
> work. So, it looks like redhat has things setup to use the kernel
> nfsd.
>
> Mike Romberg ([EMAIL PROTECTED])
Which,IMNSHO, is a mistake in that it only supports linux nfs clients.
At least they could have shipped with a non-linux-only nfs daemon, even
if they wanted to default to knfs.
--
Bill Anderson Linux Administrator
MCS-Boise (ARC) [EMAIL PROTECTED]
My opinions are just that; _my_ opinions.
------------------------------
From: [EMAIL PROTECTED] (Miquel van Smoorenburg)
Crossposted-To: comp.os.linux.networking
Subject: Re: NFS Lock Deamon for Linux. Is there one?
Date: 4 Jun 1999 14:05:35 +0200
In article <[EMAIL PROTECTED]>,
Dann Church <[EMAIL PROTECTED]> wrote:
>RH6.0 includes a lockd daemon. If you want to get this without installing RH
>6.0, you could a new kernel (use 2.2.5 or better) and make sure that you answer
>"yes" to "emulate Sun NFS server."
"emulate Sun NFS server" doesn't have anything to do with locking.
I wish they had given this option another name. What it does has only
to do with covering mountpoints, nothing more (see the help).
Mike.
--
Indifference will certainly be the downfall of mankind, but who cares?
------------------------------
From: [EMAIL PROTECTED] (ellis)
Subject: Re: Kernel Changes 2.0.x->2.2.x????
Date: 3 Jun 1999 22:15:04 GMT
In article <7ivgpi$rfq$[EMAIL PROTECTED]>, <[EMAIL PROTECTED]> wrote:
>Does anyone know what exactly change from 2.0.x to 2.2.x in the Kernel?
>
>I had a working driver in 2.0.36 that would mmap kernel buffer to user
>space but it is completely broken under 2.2.x. I made the following
>changes to the driver but I am still getting segmentation faults..
>
>modified 2.2.6 code: (this code is giving me segmentation fault)
>Any help is appreciated...
Try this link:
http://www.atnf.csiro.au/~rgooch/linux/docs/porting-to-2.2.html
--
http://www.fnet.net/~ellis/photo/linux.html
------------------------------
From: [EMAIL PROTECTED] (Juergen Heinzl)
Crossposted-To: comp.os.linux.development.apps
Subject: Re: kernel vs egcs vs PentiumPro/II
Date: Thu, 03 Jun 1999 21:01:38 GMT
In article <[EMAIL PROTECTED]>,
Horst von Brand wrote:
>On 3 Jun 1999 03:43:00 GMT, Conrad Sanderson <[EMAIL PROTECTED]> wrote:
>>Juergen Heinzl <[EMAIL PROTECTED]> wrote:
>>>All the best ... no problems here, usually using -Os instead
>>>of -O2 for all the stuff, so at least your chances to survive
>
>>what is -Os ?
>
>Optimize for space, not speed. This screws up parts of the kernel, as they
>assume certain functions to get inlined.
>
>>>I've seen and tried code that broke without -fno-strength-reduce
>>>up to and including egcs-1.1.2; -fschedule-insns and -fschedule-insns2
>>>are two options to avoid like hell too, out of experience.
>
>Could you be more specific? The strength reduction bug was supposed to be
>squashed as far back as egcs-1.0, at least (2.7.2.3 doesn't show the bug, as
>the optimization was just turned off wholesale). Others I don't know about.
Yes, I do know ... following the gcc bugs group too though and there were
and are source snippets that just showed that behaviour. Correct code,
compiles fine with all strange kind of optimisations, including none at
all ... as long as one does not use -fstrength-reduce. I often try that
one first to check. Say whoever tells me that bug is fixed ... fine with
me ... I am not going to use it even so 8) ... better safe than sorry.
IIRC at least one version of gcc fixed it by just disabling it anyway,
but correct me here. Minor note, I've got an Intel machine, so if you've
something else up and running all might be fine and well, beware.
>Note that new snapshots need -fno-strict-aliasing: The compiler is _too_
>eager to assume that a datum can only be accessed through a pointer to its
>own type.
I have never been using snapshots, simply no time to toy around with those
too. Mind this machine is 99 per cent self baked so I really mean it.
[...]
Cheers,
Juergen
--
\ Real name : Jürgen Heinzl \ no flames /
\ EMail Private : [EMAIL PROTECTED] \ send money instead /
------------------------------
From: [EMAIL PROTECTED]
Subject: Re: Redhat 6 & NFS
Date: 04 Jun 1999 09:04:12 -0400
>>>>> "Bill" == Bill Anderson <[EMAIL PROTECTED]> writes:
[...]
Bill> Which,IMNSHO, is a mistake in that it only supports linux nfs clients.
Bill> At least they could have shipped with a non-linux-only nfs daemon, even
Bill> if they wanted to default to knfs.
What does that mean? I have no problems that I know of interoperating
my RH6.0 + 2.2.7 as a server to some Solaris boxes.
------------------------------
From: Axel Farr <[EMAIL PROTECTED]>
Subject: Re: Strange behaviour of LINUX SMP under pvm
Date: Fri, 04 Jun 1999 14:59:45 +0200
Some more details:
Sometimes I can see a performance loss in the BIOS configuration with
MPP 1.1 specification, but it is only around 20 - 30 % instead of 50 -
100% with MPP 1.4 specification. And the system is by far less
"lethargic" than with the other configuration.
Run time between two outputs of my calculation jobs grows from 20
minutes when everything is ok to about 23 minutes, when partial blocking
occurs. That is about 15% loss in performance.
Greetings, Axel Farr
------------------------------
From: [EMAIL PROTECTED]
Subject: select() vs. poll()
Date: 4 Jun 1999 13:32:12 GMT
Hi there,
I was wondering what is the practical difference between the functions
'select()' and 'poll()'? I've used the former quite a lot, but stumbled
over the latter today. poll() seems to have entered Linux quite recently,
and I think the interface it provides is nicer then select().
Is there something I should be aware of if I intend to use poll() instead
of select()? Performance? Thread-safety (I use threads)? Stability?
I'd appreciate in opinions/warnings/suggestions about this issue. Have you
used poll() without problems?
Regards,
Jan Ekholm
--
=====================+======================================================
Jan 'Chakie' Ekholm | CS at Åbo Akademi University, Turku, Finland
Linux Inside | I'm the blue screen of death, no-one hears you scream
------------------------------
From: Vassilis Virvilis <[EMAIL PROTECTED]>
Subject: Re: Configuration Manager for Linux
Date: Fri, 04 Jun 1999 16:37:24 +0300
Karl-Heinz Zimmer wrote:
>
[snip almost everything]
> I would like to discuss the following ideas:
> - The programs should be completely independent from
> distribution or package format.
> - They should run either in textmode or under X.
> - They would provide an online-help by right click
> on a special part of the dialog box.
> - All the user will have to do is pick items from
> Dropdown lists, select checkboxes, enter few
> keywords (such as the POP3 password) . . .
> - At each step there would be the possibility to
> get additional information by just pressing F1.
> - The agents would perform checks
> - concerning package dependencies
> - whether the chosen parameters will fit together
> - is hardware is at the port it is expected to be
> - ...
>
> At my opinion those two projects should be merged into one plan:
>
> How to make Linux configuration as simple as possible
> even for those who do NOT WANT to read fine explanations.
>
> We will have to think about that problem anyway as there is not way to
> change it: many(!) people are looking at Linux now and planning to
> take this fine OS - only few of them can afford the time to study
> HOWTOs: they just want it to run now!
>
As everybody who responds to this thread, me too, I had the idea of the
absolute configuration system. I gave some thought to this about 2 years ago and
I made some research. You should be interested in the fact that there have been
a lot of projects with that holy grail in mind.
I am just posting the references in case you are not familiar with the relative
projects :)
http://www.imada.ou.dk/~blackie/dotfile/
http://www.solucorp.qc.ca/linuxconf/
http://www.webmin.com/webmin/
http://www.foxnet.net/~apenwarr/figurine/
http://www.cs.vu.nl/~bernsti/conf/index.html
http://perso.wanadoo.fr/claud.maille/gaa.html
By that time a lot of people were advocating 2 possible solutions
-The configurator knows about all (via modules) programs internals (defaults,
config file syntax, inter dependencies and such...)
-This can be avoided by the usage of a "UNIVERSAL"/"TRUE"/"ONE % ONLY" libConfig
which will replace all the different file formats with a universal one. (All
programs need rewrite of course.)
Both opinions have their strong and weak points and I wouldn't like you to think
that I am advocating either of two.
.Bill
------------------------------
From: Marius Vollmer <[EMAIL PROTECTED]>
Subject: Re: *Very* large physical buffers.
Date: 04 Jun 1999 16:17:50 +0200
[EMAIL PROTECTED] (Robert Kaiser) writes:
> In article <[EMAIL PROTECTED]>,
> Marius Vollmer <[EMAIL PROTECTED]> writes:
> > For convenience, we would like to have a 8MB large DMA buffer in
> > physical memory (that is, it should have continues physical
> > addresses). Our WindowsNT box wont give it to us, of course.
> >
>
> Have you considered using scatter/gather DMA ?
I have shortly thought about this, but I don't know in what way NT
supports this. And I don't think we should discuss this here... ;-)
Anyway, it's not that important. We want to transfer data to our DSP
board for purely academic purposes and we can do this without DMA.
It's slow, but it works. The DMA chip sits on the board, so we would
like have to write our own driver or something when the support
software does not do scatter/gather already, etc, pp.
> There are several approaches. The one that seems to best fit your
> requirements is Matt Walsh's "bigphysarea" patch.
Yes, that thing should really be all that is needed.
Thanks!
- Marius
------------------------------
From: "Selious" <[EMAIL PROTECTED]>
Crossposted-To: alt.os.linux,comp.os.linux.advocacy,comp.os.misc,comp.unix.advocacy
Subject: Re: the ultimate OS
Date: Fri, 4 Jun 1999 16:27:08 +0200
>: How can I say this, without allowing denyal ??
>
>denial? who is in denial? I acknowlege there are many
>active development efforts going right now.
Yeah, but I mean, how can I claim it is not a GOOD idea, untill you solved
the problems like drivers, development IDE's, etc.
>
>: Ehh, let's get back on you're progress in 6 months !!
>
>I see, a passive bystander.
>maybe I will inquire on YOUR progress in 6 months<g>
Ehh, I set deadlines of achievements for each month... If I fail to achieve
them, I quit my project. The current project timeline would bring me in 6
months to a console based networking OS.. I hope to be able to port
linux/unix/bsd appz by then...if not, I'll forget it (and I did not announce
my project jet, my project's design took 4 years (before I used PC, I
already wanted a more high filosofy OS) and it's hard NOT to write programs
for linux, especially with this free LDAP, Corba, KDE, etc. stuff.
Why becomes a issue more and more every month !!
>and be careful not to be influenced by what I wrote<g>
I only browsed it, because it's a huge pile of stuff.. But it simply sais,
let's base a OS on objects directly, instead of simulating them with
difficult API's and all !!
Anyway, that's two ways if we share our deeper secrets ! You should look at
the internet for a list of OS's. There are sites already filled with GREAT
designs for OO-OS, but they are all either in design state, or updated last
on 7 june 1996 !!).
Anyway, we COULD share hardware specs... I have a lot of them (mouses, MCA,
TokenRing, and all more familiar), but I miss the NE2000 specs and many
other things.
Anyway, all I wanted to say is, from wanting to doing is almost as big a
step as from console to GUI, etc... I took the first step already (BIOS
indepandant OS), and I must admit I thank linux for the example source code
!!
------------------------------
From: lckun <[EMAIL PROTECTED]>
Subject: How can i debugging with kernel?
Date: Fri, 04 Jun 1999 23:58:22 +0900
Hi all,
Can anyone tell me how can i debugging with the kernel??
thanks
------------------------------
From: Vassilis Virvilis <[EMAIL PROTECTED]>
Subject: Re: Linux development tools...
Date: Fri, 04 Jun 1999 17:22:08 +0300
Selious wrote:
>
> >I come from the Microsoft Windows world and I think Visual C++ has
> >me spoiled. :) Please help so I can break away from the Microsoft grasp.
>
> No, don't !! Simply install Samba on your linux system, and let Visual
> Studio create a empty project in a shared directory. Add files as normal,
> and add a Makefile (you can even edit it in Visual Studio).
>
> When done editing, press Save All, and do a make on your linux system !!
>
> I even made a linux kernel project in Visual Studio !!
>
> The next step is to make a small VB-Script macro that allows you to execute
> make remotely....
>
> Anyway, when Linux offers a good C/C++ parsing and hierarchical listing
> development enviroment, it will be time to use that !!. As long as they
> allow extensions (through API, Perl, TCL, whatever.)
How is this possible...?
MSVC uses makefiles which are processed by nmake which is __NOT__ compatible
with GNU make or any kind of Open Source make that I am aware of...
Any way you are losing the ability of on line debugging that MS VC offers so It
looks to me that you are using MSVC (150 MB) as an editor only.
.Bill
------------------------------
From: "Keith Peterson" <[EMAIL PROTECTED]>
Crossposted-To: alt.os.linux,comp.os.linux.advocacy,comp.os.misc,comp.unix.advocacy
Subject: Re: TAO: the ultimate OS
Date: Fri, 4 Jun 1999 09:35:53 -0600
>Where is your code? Talk is cheap - unless you can demonstrate a code your
>words are worth nothing. Sorry, but you sound like a cross between manager
and
>salesweasel and those animals are, erm, not too good in producing things.
>Write something that would work and demonstrate it.
Geez... and, "When project X is done, it'll be terrific!" isn't the rallying
cry of most linux projects?
------------------------------
From: [EMAIL PROTECTED]
Subject: help on kernel configuration
Date: Fri, 04 Jun 1999 14:04:24 GMT
I want to config the kernel to use the serial port
as the system console, instead of the VGA display,
how can i do this?
in face, the system have no VGA and BIOS, can
linux kernel run over such a x86 platform?
thanks
[EMAIL PROTECTED]
Sent via Deja.com http://www.deja.com/
Share what you know. Learn what you don't.
------------------------------
Crossposted-To:
comp.os.linux.development.apps,comp.os.linux.misc,comp.os.linux.setup,comp.lang.java.databases
Subject: Re: What are the differences between mySQL and mSQL?
From: [EMAIL PROTECTED] (Don Baccus)
Date: 3 Jun 1999 16:44:13 PST
In article <_0E53.40$[EMAIL PROTECTED]>,
bryan <[EMAIL PROTECTED]> wrote:
>setup your data so that each record is pretty much an atomic entity.
It is often not possible to do this, that being the entire
point, of course.
>: How about cursor stability for the readers? Are the
>: readers sorting the data; what about an insert happening in the middle of
>: the sort?
>don't know how to answer these questions. I'm far from an sql expert
>but so far, in the netmgt apps I write and run, I've never run into
>these issues before.
Sort of like writing multi-threaded programs without proper synchronization.
"Gee, it always worked on the ground, can't imagine why the plane
crashed once we flew it!"
>: In a lightly loaded server your probability of encountering these problems
>: is low but not zero. Stress the server with lots of activity and these
>: problems will show themselves. Transactions and isolation levels address
>: these issues.
>you have a very valid set of points. perhaps I should consider the
>transaction thing a bit more closely. sometime before my servers get
>to the saturation point, I guess ;-)
It depends entirely on what you're doing. Slashdot runs on
MySQL, for instance. A lot of web sites using dbs can use a
very simple data model, and get away without transactions.
>still, it depends on the absolute accuracy of the info. if you're
^ need for? I assume you mean?
>using a freeware db to manage my money, I would be concerned.
Hmmm...sybase is freeware and nothing about it should cause
you worry.
And I have confidence that open-source Postgres will be there,
too, like in a year or so. They're making real progress.
(Postgres does transactions)
>for
>some snmp polled data (that is constantly being updated by new values
>by a poller), then even one erroneous entry won't take the network
>down. it probably won't even be noticed. and will be corrected by
>the next polled instance.
As long as you know your application, you're safe.
--
- Don Baccus, Portland OR <[EMAIL PROTECTED]>
Nature photos, on-line guides, at http://donb.photo.net
------------------------------
From: dan <[EMAIL PROTECTED]>
Subject: can 2 linux machines talk to a dual port scsi raid system ??
Date: Thu, 03 Jun 1999 19:09:50 -0500
i have a dual port scsi raid system with 2 linux systems connected to it
both run redhat 5.2
there is an ext2 filesystem on the raid box (~26Gbytes)
the problem i have is that if linux box A writes a file to the raid
filesystem,
linux box B doesn't see the file
if i reboot linux box B, it sees the file,
if i write the file with linux box A BEFORE the mounting the raid system
on linux box B for the FIRST time it sees the file
i tried unmounting and then mounting the raid system on linux box B,
but it still doesn't see the file
the first mount of the raid system takes much longer than later mounts,
i assume that linux is only "looking" at the raid system the first time
it
mounts it (and buffers the directory and file info into memory ?)
is there a way to get linux to "relook" at the disk ?
rebooting the system takes too long, i am looking for
a faster way of seeing the files
i hope this makes sense
thanks for any help in advance,
dan
------------------------------
** 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
******************************
