Linux-Development-Sys Digest #686, Volume #6 Fri, 7 May 99 00:13:55 EDT
Contents:
Re: kernel nfsd (Martin Pottendorfer)
Re: lp and kernel 2.2 (Jwbat)
Re: kernel nfsd (Todd L. Cohen)
ISDN PCI Cards with Linux & other bits ("Steve Stairs")
Re: [ANN] CodeWarrior for Red Hat Linux, GNU ed. Shipping (Eric Hegstrom)
Re: Help Please!!! (Dan Conti)
polling to interrupt mode (Surya P Kommareddy)
Re: Trying to detect PnP Monitors (James Youngman)
Re: Mac-emulation on Linux? (Chris Johnson)
Kernel Mode Driver Reentrancy ("Patrick L. McGuire")
PCI Physical Memory Access ("Patrick L. McGuire")
Re: Change from gcc to egcs gives seg fault in make (Clifford T. Matthews)
Re: New Kernel not booting (Greg de Freitas)
Re: PCI Physical Memory Access (DAVID JOHNSON)
Mount multi-track CD ROMs? (Keith Wright)
----------------------------------------------------------------------------
From: Martin Pottendorfer <[EMAIL PROTECTED]>
Subject: Re: kernel nfsd
Date: 06 May 1999 17:00:18 +0200
[EMAIL PROTECTED] (Todd L. Cohen) writes:
> Hi,
> I've installed RedHat 6.0 and then updated to kernel 2.2.7 from
> source. Since I want to NFS export space, I chose the experimental NFSD
> built into the kernel since the old nfsd isn't shipped with redhat
> anymore. Well, the old one worked fine, but with the new one I get a
> "protocol not supported" error on my client, and a "svc: unknown version
> (3) " on the linux box.. any ideas?
I encountered the same (but even with the RH 6.0 shipped kernel:
2.2.5-15).
I found that the client side (in my case a Sun Solaris-2.5.1) box
tries to use NFS Version 3, which is not yet supported by Linux (when
will this be?), but although Linux tries to negotiate version 2 of NFS
the Sun retries NFS version 3.
Here's what `snoop' on the SUN told me:
[...]
mylinux -> mysun PORTMAP R GETPORT port=820
mysun -> mylinux MOUNT3 C Null
mylinux -> mysun MOUNT3 R Null
mysun -> mylinux MOUNT3 C Mount /
mylinux -> mysun MOUNT3 R Mount OK FH=6AD0 Auth=none,unix
mysun -> mylinux PORTMAP C GETPORT prog=100003 (NFS) vers=3 proto=TCP
mylinux -> mysun PORTMAP R GETPORT port=0
mysun -> mylinux PORTMAP C GETPORT prog=100003 (NFS) vers=3 proto=UDP
mylinux -> mysun PORTMAP R GETPORT port=2049
mysun -> mylinux NFS C NULL3
mylinux -> mysun RPC R (#48) XID=926670748 Program number mismatch (low=2,
high=2)
mysun -> mylinux PORTMAP C GETPORT prog=100005 (MOUNT) vers=3 proto=UDP
[...]
where `mylinux' is the Linux-2.2.5-15 box, and `mysun' the Solaris
2.5.1 box.
Explicitly forcing version 2 by:
mount -F nfs -o vers=2 ...
on the Solaris client side was somwhow successfull, although I found
the following in the Linux syslog:
[...]
May 6 16:34:08 mylinux mountd[431]: authenticated mount request from ...
May 6 16:34:08 mylinux kernel: svc: unknown program 100227 (me 100003)
May 6 16:34:08 mylinux last message repeated 5 times
[...]
I upgraded a RH5.2 System to 6.0, maybe there's some old package
around which interferes somehow?
What's the proper way to get NFS servers running on a 2.2.7 kernel?
Currently I have the following installed:
$ rpm -q -f /usr/sbin/showmount
knfsd-clients-1.2.2-4
$ rpm -q -f /usr/sbin/rpc.nfsd
knfsd-1.2.2-4
Should I switch to the `userland' version of NFS?
Thanx for any information,
martin
------------------------------
From: [EMAIL PROTECTED] (Jwbat)
Subject: Re: lp and kernel 2.2
Date: 06 May 1999 17:27:58 GMT
In article <[EMAIL PROTECTED]>, Greg White <[EMAIL PROTECTED]>
writes:
>
>2. Make sure parallel support is selected in _both_ places it appears (
>General setup and Character devices). I burned myself on that one the
>first time I compiled a 2.1.xx kernel.
>
>
This (to my embarassment) turned out to be the case. Parallel support was NOT
set in the chracter devices. Hopefully, this will save somebody else the same
"gotcha". In my weak defense, the option DOES come AFTER all the serial card
selections.
------------------------------
From: [EMAIL PROTECTED] (Todd L. Cohen)
Subject: Re: kernel nfsd
Date: 6 May 1999 17:41:00 GMT
they don't seem to have shipped the userland one.. any idea where to get
it? metalab?
:
: What's the proper way to get NFS servers running on a 2.2.7 kernel?
: Currently I have the following installed:
: $ rpm -q -f /usr/sbin/showmount
: knfsd-clients-1.2.2-4
: $ rpm -q -f /usr/sbin/rpc.nfsd
: knfsd-1.2.2-4
:
: Should I switch to the `userland' version of NFS?
:
: Thanx for any information,
: martin
--
===================================================
http://www.clarkson.edu/~cohentl
"You can use divine intervention to solve the problem if you need to."
- Coleman, 1998
"Sometimes crazy things happen in circuits." - Muku, 1999
===================================================
------------------------------
From: "Steve Stairs" <[EMAIL PROTECTED]>
Subject: ISDN PCI Cards with Linux & other bits
Date: Thu, 6 May 1999 19:44:07 +0100
Hi all I am new to Linux and really want to ditch Windows. I use a Mac at
work and love the way Linux works. However I have a couple of queries...I
hope someone can help.
I can't get an internet setup...I am using Redhat 5.1 and I use X Windows. I
have a BT Speedway ISDN Internal PCI Card and as such am unsure of how to
get Linux to recognise it
Also I have a VooDoo II card form Diamond and am wondering if i can use that
with Quake Arena when I get downloadinf from the net.
I am a true novice (only installed last night) but am in need of assistance.
Thanks all in advance
Steve (Cable)
http://www.btinternet.com/~xtc.cable/XTC.html
------------------------------
From: Eric Hegstrom <[EMAIL PROTECTED]>
Crossposted-To: comp.os.linux.development.apps,comp.unix.programmer
Subject: Re: [ANN] CodeWarrior for Red Hat Linux, GNU ed. Shipping
Date: Thu, 06 May 1999 12:15:46 -0700
Andy Law wrote:
> And will Metrowerks (or anyone else) _force_ you to buy an IDE?
Yea they came by my office and held a gun to my head till I ordered it.
Actually I did order it up last night. I used their environments many
years ago at Apple (they had one of the first really good power-pc
compilers). I liked their IDE (even though I still spent a lot of time
in the venerable command line oriented MPW). Sorry! Sure, I may not use
it on my big development projects, but I'm not ashamed to say I am
always looking for shortcuts and even if the company wasn't paying for
it, $80 is not what I would call a big tool investment. Call me a whimp,
but I sometimes I miss having a car with an automatic transmission.
--
Eric Hegstrom .~.
Senior Software Engineer /V\
Sonoran Scanners, Inc. // \\ L I N U X
[EMAIL PROTECTED] /( )\ >don't fear the penguin<
520-617-0072 x402 ^^-^^
------------------------------
From: Dan Conti <[EMAIL PROTECTED]>
Subject: Re: Help Please!!!
Date: Thu, 06 May 1999 12:37:08 -0700
Igor Zlatkovic wrote:
> Process Synchronization
> Considering the statements made on inter-process communication, there
> is only one thing to differ from the real life here. Children should never
> live longer than their parents.
>
> Handling of Deadlocks
> When you drive in a tight street and another car comes from the
> opposite direction, one of you will drive reverse sooner or later. There are
> better things to do than sit in a car all day long. In the process world
> situation is totally different. there is no way to handle a deadlock. You
> simply must not let it happen.
Actually, to be fair, there are methods of handling deadlocks, it's just that
they have not been implemented in Linux at all. VMS is the only operating system
that i know for a fact has deadlock handling. Also, in a multitasking system, you
can't really gaurantee that deadlocks wont happen.
-Dan
> --
> o
> O Cheers,
> ______O___
> \________/ Igor Zlatkovic
> \ o / mailto:[EMAIL PROTECTED]
> \ O /
> \ /
> \/
> || University of Applied Sciences
> ___||___ Frankfurt, Germany, EU.
--
Dan Conti, Computer Science, WWU
[EMAIL PROTECTED]
Life goes on long after the thrill of living is gone.
Live happy.
------------------------------
From: Surya P Kommareddy <[EMAIL PROTECTED]>
Subject: polling to interrupt mode
Date: Thu, 06 May 1999 16:40:33 -0700
Hi,
I am working on a data acquisition project using the parallel port hooked-up
to my Linux box. My box has RT1.1 patched onto RH5.2(Linux 2.0.36). I think the
parallel port device is using polling mode for data transfer (which I confirmed
from the /proc/log/messages file). I would like to change this from polling to
interrupt mode of data transfer. Can someone pass on information on how to do
this?
Thanks in anticipation.
Surya.
------------------------------
From: James Youngman <[EMAIL PROTECTED]>
Crossposted-To: comp.os.linux.development.apps
Subject: Re: Trying to detect PnP Monitors
Date: 06 May 1999 17:54:15 +0100
Menelik <[EMAIL PROTECTED]> writes:
> Hi
>
> I would like to write some code under Linux that will detect Plug-n-Play
> Monitors. My goal is to have it be similar to the Windows 98 monitor
> detection.
Work is under way on this (DCC support) in XFree86. Your best path is
to join XFree86 and help. See www.xfree96.org.
>
> I need some info (urls) on how to detect the Monitor ID, and also where
> I can find a database of Monitor specs so that the monitor
> resolution/Frequency is known already from the Monitor.
>
> Any ideas ? Any help is appreciate.
--
ACTUALLY reachable as @free-lunch.demon.(whitehouse)co.uk:james+usenet
------------------------------
From: [EMAIL PROTECTED] (Chris Johnson)
Crossposted-To: comp.os.linux.advocacy,comp.os.linux.misc,comp.os.linux.powerpc
Subject: Re: Mac-emulation on Linux?
Date: Thu, 06 May 1999 21:12:06 -0400
In article <01be9718$1947d5a0$[EMAIL PROTECTED]>, "FM"
<[EMAIL PROTECTED]> wrote:
>Just another related question. Do all macs have one
>mouse button or is it that it's so common that I've
>never encountered a mac with multiple mouse buttons.
>I'm quite used to 3-button Logitech mice that I
>find many 2-button mice very limiting and wonder if
>I could make use of multiple-button mice on a mac?
>If there are some available (which I assume), does
>the X windows (and window managers/desktop
>environments) make use of the extra buttons? How
>does it behave without the extra buttons?
>Dan
I find it works this way- having the single button mouse causes me to
access more functions on the keyboard. I run Window Maker and have mapped
many things to Fkeys- in particular, F1 is the applications menu, F2 is a
windows menu, F3 is the properties of a window menu and F4 is xkill :)
*kachik* *BLAM!*
One reason I like Window Maker so much is that it is cooperative in
this goal of bending X to my whims- it's more general and less complicated
than, say, KDE, and there is tremendous flexibility to tailor it to
situations such as using a one button mouse. You simply have keys
activating things- this is not so unthinkable. Fkeys are your friend if
you want the mouse to be mostly just a pointer :)
I also find X text handling to be extremely stupid, being used to the
Mac way of doing it (drag across stuff, drag selected stuff, delete gets
rid of stuff and cmd-X-C-V is cut-copy-paste). For that reason I don't
find anything admirable about the multibutton X text handling methods and
don't feel I'm missing out by not having them- instead I find I just type
everything, with an eye to keyboard-only methods, which strikes me as more
Unixlike anyhow. To me the mouse is mostly just for window-wrangling in X,
and one button is as good as three for that, if you are ready to customize
some keyboard mappings to your needs.
That's just my perspective, YMMV.
Chris Johnson
@airwindows.com
chrisj
------------------------------
From: "Patrick L. McGuire" <[EMAIL PROTECTED]>
Subject: Kernel Mode Driver Reentrancy
Date: Thu, 06 May 1999 18:07:52 -0700
Reply-To: [EMAIL PROTECTED]
This is a multi-part message in MIME format.
==============6AC29C09C089992A2C30F903
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit
This is probably the most basic question.
Aside from possible hardware interrupts, is it safe to assume that while
I am in the kernel driver, that another process will not call the driver
again?
That is, do I know that I have exclusive access to the hardware that I
am controlling, once I am in my kernel mode driver?
Of course, this is a single CPU system.
Thanks in advance.
==============6AC29C09C089992A2C30F903
Content-Type: text/x-vcard; charset=us-ascii;
name="plmcguire.vcf"
Content-Transfer-Encoding: 7bit
Content-Description: Card for Patrick L. McGuire
Content-Disposition: attachment;
filename="plmcguire.vcf"
begin:vcard
n:McGuire;Patrick
tel;fax:+1-510-887-7080
tel;work:+1-510-887-7070
x-mozilla-html:FALSE
url:http://www.epicom-inc.com
org:Epicom, Inc.;Engineering
adr:;;23588 Connecticut Street #1;Hayward;California;94545;USA
version:2.1
email;internet:[EMAIL PROTECTED]
title:Engineer
fn:Patrick L. McGuire
end:vcard
==============6AC29C09C089992A2C30F903==
------------------------------
From: "Patrick L. McGuire" <[EMAIL PROTECTED]>
Subject: PCI Physical Memory Access
Date: Thu, 06 May 1999 17:46:00 -0700
Reply-To: [EMAIL PROTECTED]
This is a multi-part message in MIME format.
==============807001564EC2774EF43B2B23
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit
I am trying to access a PCI device which is memory mapped.
Using pcibios_read_config_dword(), I am able to obtain the physical
address of the board, and I know this address is correct: 0xFC000000.
Subsequently, I use phys_to_virt() to translate this address into one
that I can use in my driver. Turns out, it's the same address anyway.
When I access this memory, I get a segmentation fault:
int addr, *p;
pcibios_read_config_dword(bus,function,PCI_BASE_ADDRESS_2,&addr);
addr &= PCI_BASE_MEM_MASK;
addr = phys_to_virt(addr);
p = (int *)addr;
x = *p; // causes segmentation fault
Any clues, or hints?
==============807001564EC2774EF43B2B23
Content-Type: text/x-vcard; charset=us-ascii;
name="plmcguire.vcf"
Content-Transfer-Encoding: 7bit
Content-Description: Card for Patrick L. McGuire
Content-Disposition: attachment;
filename="plmcguire.vcf"
begin:vcard
n:McGuire;Patrick
tel;fax:+1-510-887-7080
tel;work:+1-510-887-7070
x-mozilla-html:FALSE
url:http://www.epicom-inc.com
org:Epicom, Inc.;Engineering
adr:;;23588 Connecticut Street #1;Hayward;California;94545;USA
version:2.1
email;internet:[EMAIL PROTECTED]
title:Engineer
fn:Patrick L. McGuire
end:vcard
==============807001564EC2774EF43B2B23==
------------------------------
From: [EMAIL PROTECTED] (Clifford T. Matthews)
Subject: Re: Change from gcc to egcs gives seg fault in make
Date: 06 May 1999 15:33:32 -0600
>>>>> "DN" == Graffiti <[EMAIL PROTECTED]> writes:
DN> Wow, what a hack.
Thanks. I made it myself.
We're already on a first-name basis with the stack pointer and
frame-pointer. Sometimes we have them over for drinks, but we have to
be careful not to get carried away.
DN> Nice, though. Ever thought of statlically linking the app
That's what we originally did internally, but we would have been
running afoul of LGPL.
DN> or shipping it w/ the shared libs you use, and providing a
DN> wrapper script that uses LD_PRELOAD?
I hadn't thought of that when we first started shipping for Linux, but
I did read about that technique recently.
I never expected someone to so totally break the shared libraries*.
There are a couple of drawbacks doing it that way, including code
bloat and to some extent it defeats the purpose of the LGPL, even if
it doesn't break the letter of it. If someone finds and fixes a bug
in the db libary reinstalling a new db shared library in /lib won't
fix the bug in Executor if Executor is calling its own version of db.
Sure, the end-user could try to compile his own version of db that
used whatever calling conventions we used when we prepared our own
version. Since we're not devious people, it wouldn't be hard to do
for someone who understands all the subtleties; but if the issues are
tricky enough for major distributions like Red Hat 6.0 and SuSE 6.0 to
botch, just what percentage of end-users are likely to benefit from
this?
I don't know what we'll do for 2.1 final. Our hack really does work,
and the two issues raised above are not academic concerns.
--Cliff
[EMAIL PROTECTED]
__________________
*With the existing arrangement it's possible that some code will function
properly except for the rare times when a signal is processed at a critical
point when the stack is 4 bytes higher than it should be and the data that's
on the top of the stack hasn't yet been copied to its eventual destination.
I'd guess this actually happens in practice, since -O turns on -defer-pop,
which makes it possible for the code to appear to work properly as it
continues to access the stack through the frame-pointer which is correct.
------------------------------
From: Greg de Freitas <[EMAIL PROTECTED]>
Reply-To: [EMAIL PROTECTED]
Subject: Re: New Kernel not booting
Date: Thu, 06 May 1999 22:44:39 GMT
I had this too, until I pushed my BIOS chip back in ! (2.5 yrs, on and off, -
hot and cold, = thermal creep !
Just a guess...
--
Ciao 4 now, Greg.
# Email : mailto:[EMAIL PROTECTED] #
# Email : mailto:[EMAIL PROTECTED] #
# To Live, To Love, To Learn, To Leave A Legacy. #
------------------------------
From: DAVID JOHNSON <[EMAIL PROTECTED]>
Subject: Re: PCI Physical Memory Access
Date: Fri, 07 May 1999 03:32:38 GMT
You have to use ioremap (vremap with kernels 2.0.x).
addr = ioremap (addr, size);
or
addr = vremap(addr, size)
When you are done you should unmap it by
iounmap(addr, size)
or
vfree(addr, size)
David
"Patrick L. McGuire" wrote:
> I am trying to access a PCI device which is memory mapped.
>
> Using pcibios_read_config_dword(), I am able to obtain the physical
> address of the board, and I know this address is correct: 0xFC000000.
>
> Subsequently, I use phys_to_virt() to translate this address into one
> that I can use in my driver. Turns out, it's the same address anyway.
>
> When I access this memory, I get a segmentation fault:
>
> int addr, *p;
> pcibios_read_config_dword(bus,function,PCI_BASE_ADDRESS_2,&addr);
> addr &= PCI_BASE_MEM_MASK;
> addr = phys_to_virt(addr);
> p = (int *)addr;
>
> x = *p; // causes segmentation fault
>
> Any clues, or hints?
------------------------------
From: Keith Wright <[EMAIL PROTECTED]>
Subject: Mount multi-track CD ROMs?
Date: 06 May 1999 23:49:25 -0400
I have just gotten into burning CD's. The 'cdrecord' program
seems to make multi-track CD, but I can't figure out how to
mount any but the first track.
Is it possible to mount other than the first track of a CD?
Is anybody working on making it possible?
Does anybody have thoughts on how it should work?
Like
mt -f /dev/cdrom fsf 2
mount /dev/cdrom
What about getting the data with a verbatim copy if it
can't be mounted? Is that possible?
Where can I find more information on the mythical
Orange Book?
--
-- Keith Wright <[EMAIL PROTECTED]>
Programmer in Chief, Free Computer Shop <http://www.free-comp-shop.com>
--- Food, Shelter, Source code. ---
------------------------------
** 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
******************************
