Linux-Development-Sys Digest #101, Volume #7 Tue, 24 Aug 99 19:14:19 EDT
Contents:
Re: Subj: Redirecting console to serial port for a SBC (Miquel van Smoorenburg)
Re: upgrade linux kernel (Philip Armstrong)
Re: upgrade linux kernel (Konrad Mierendorff)
Bug in glibc-2.1.1 (initgroups) ([EMAIL PROTECTED])
Re: update_vm_cache (Peter Samuelson)
Re: C++ templates: More than Turing Complete? (Nathan Myers)
Re: TAO: the ultimate OS (John Simpson)
Subj: Redirecting console to serial port for a SBC (W. Tucker)
The hashes sendmail uses... Berkeley DB? ("Ted Pavlic")
Re: TAO: the ultimate OS (Jonathan Guthrie)
upgrade linux kernel ("Hung P. Tran")
Re: Shared Libraries: what is the linux equivalent of "dllimport" and "dllexport"
(Peter Mutsaers)
increasing process limits ("Donald Setlur")
Re: Subj: Redirecting console to serial port for a SBC (Torsten Poulin)
Re: How can I make device driver module to support many version of kernel?
(Marcus Sundberg)
----------------------------------------------------------------------------
From: [EMAIL PROTECTED] (Miquel van Smoorenburg)
Subject: Re: Subj: Redirecting console to serial port for a SBC
Date: 24 Aug 1999 20:42:24 +0200
In article <[EMAIL PROTECTED]>,
W. Tucker <[EMAIL PROTECTED]> wrote:
>I am trying to install Linux on a Single Board Computer (SBC) that
>has a floppy disk and a serial connection, but no keyboard or
>monitor. I plan to connect a null modem serial cable between the SBC
>and another computer and boot the SBC to Linux from a floppy. Then I
>would like to redirect the SBC console so I can talk to it from the
>other computer.
You need to enable serial console support. It's now standard in 2.2.x
and up. See linux/Documentation/serial-console.txt
>
>If I were using DOS on the SBC, I would include the command "ctty COM1"
>in the autoexec.bat file to perform the console redirection. Is
>there a similar command that I can use from Linux? Or is there a
>better way to approach the problem?
See above. If there is no graphic card installed, the serial console
will take over automatically (if you compiled it in ofcourse).
>As an experiment, I tried to redirect the console of my current Linux
>box using the command "setconsole cua1". This failed to run and gave
>an error message complaining that /proc/openprom/options does not
>exist.
Weird, what architecture is this? That's not an i386 error message..
I don't even have the command on my distribtion (Debian).
>I am using a 2.0.36 kernel from the Red Hat 5.2 distribution.
If it's an Intel processor, you could either install a 2.2.x kernel
and enable serial console support, or get the patch for 2.0.x from
ftp://ftp.cistron.nl/pub/people/miquels/kernel/v2.0/
I'm not sure how far along the serial console code is for other
architectures. I think that in 2.2.x it works on Atari and Sparc as well.
Mike.
--
... somehow I have a feeling the hurting hasn't even begun yet
-- Bill, "The Terrible Thunderlizards"
------------------------------
From: [EMAIL PROTECTED] (Philip Armstrong)
Crossposted-To: comp.os.linux.misc
Subject: Re: upgrade linux kernel
Date: 24 Aug 1999 19:48:01 GMT
In article <[EMAIL PROTECTED]>,
Hung P. Tran <[EMAIL PROTECTED]> wrote:
>When I try to recompile the kernel using "make zImage",
>I receive an error of duplicate variable from
>/usr/src/linux/arch/i386/lib/checksum.c. Further inspection
>reveals that the variables are not duplicate. It's just that they
>are very long and have the same 1st 17 characters, and so
>appear the same to the compiler when they are truncated.
>How do I overcome this problem ? Do I need to update my
>gcc compiler (curently ver 2.7.2.1) or can I modify a makefile
>somewhere to add a switch ??? Please advice ...
gcc 2.7.2.1 is *very* old. You should at least upgrade to 2.7.2.3
which should compile the kernel without problems.
Don't go to a 2.3 kernel. They're not at all stable and may eat your
system :)
Phil
--
nosig
------------------------------
From: Konrad Mierendorff <[EMAIL PROTECTED]>
Crossposted-To: comp.os.linux.misc
Subject: Re: upgrade linux kernel
Date: Tue, 24 Aug 1999 21:38:54 +0200
"Hung P. Tran" wrote:
> Should I just update the kernel to 2.2.5 or should I go all the way
> to the newest one available (which I believe is 2.3.14) ???
> Does newer necessarily mean better or that some version is
> more stable than other ???
Certainly you shouldn't upgrade to any of the 2.3.* Kernels if you don't
need any of the features which haven't gone into the stable kernels yet.
The parity of the minor(second) version-number indicates wether it is a
stable (even number) or a developer version (odd number). So if you
don't want to help bug-tracing you should stay away from these versions.
As to 2.2.5, I'd suggest to take 2.2.10 which is known to be very
stable. (I heard of some problems with 2.2.11 - nothing definite)
Maybe the compilations problems are also solved if you take 2.2.10
Good luck
- Konrad Mierendorff
------------------------------
From: [EMAIL PROTECTED]
Subject: Bug in glibc-2.1.1 (initgroups)
Date: Tue, 24 Aug 1999 19:46:14 GMT
There seems to be a bug in glibc-2.1.1-6.
After installing this this package I was�nt
able to su anymore. I always got the message
initgroups: invalid argument.
After uninstalling glibc-2.1.1-6 everything
works ok.
Ernst Braun
eb @ con.de
Sent via Deja.com http://www.deja.com/
Share what you know. Learn what you don't.
------------------------------
From: [EMAIL PROTECTED] (Peter Samuelson)
Subject: Re: update_vm_cache
Date: 24 Aug 1999 11:26:50 -0500
Reply-To: Peter Samuelson <[EMAIL PROTECTED]>
[newsseeker <[EMAIL PROTECTED]>]
> You have to admit that if a person reads the linux-kernel mailing
> list every other day, they can very easily miss a patch that should
> have been avoided.
Here's a shortcut: Richard Gooch's "Kernel Newsflash" page. I don't
know what's happening with it now that he is apparently moving to
Canada, but he tries to keep up-to-date with kernels that either don't
compile or have nasty bugs:
http://www.atnf.csiro.au/~rgooch/linux/docs/kernel-newsflash.html
Don't exaggerate my point, either. If you came and whined about the
masq-ftp module or packet forwarding or the tulip driver being broken,
no big deal. But here we're talking about a *major* redesign of the
Linux page caching scheme which took Linus and Ingo Molnar literally
*weeks* to get to the stage where they got at least ext2 somewhat
working and decided to merge it in (this was 2.3.7) and more weeks
while everyone is working the bugs out including having to make
significant updates to *all* filesystems. Meanwhile, 2.3.7
deliberately changed the VFS interface so that still-broken filesystems
would simply fail to compile rather than trashing your data later on.
Many filesystems including the FAT-based ones did not get updated for
several patchlevels, and things like hpfs and nfs are still not fixed.
This is easily *the* most involved change to Linux in the 2.3 cycle.
(Other big things have been filling out USB support, deserializing some
networking codepaths, several cleanups -- waitqueues, partitions,
`struct device' -- the big ISDN merge, a few new drivers and features
like khttpd. None of those fundamentally changed the Linux core very
much).
You see? If you were truly interested in following development,
running 2.3.x to try to fix bugs etc, you could not *possibly* have
missed the revamp of the pagecache. You must forgive me for therefore
assuming that you are *not* following development, at least not at all
closely.
> And, perhaps it is apparent to you as to where to find all the kernel
> change logs (much less putting together what works and what doesn't
> out of the change logs for each and every change),
Actually I read through pretty much every kernel patch I apply. Not a
substitute for keeping up with development, but it's a great way to
keep up on at least major features or changes. (Which is how I know,
without ever compiling or using hpfs, that it hasn't been updated yet.)
> I've yet to see a location mentioned for the new change-log, or
> collection of change-logs (freshmeat.net gives better info as to
> where the latest changelog is).
There is no official changelog. Alan Cox keeps a changelog on his -ac
patches which he emails to freshmeat.net and lwn.net among others, but
Linus doesn't. What he does is, for most patches, announces them on
linux-kernel along with usually a short explanation of what's in them
and (where applicable) some caveats. It's sort of hit-or-miss, but he
*often* does announce them.
> In my case, I tried out 2.3.12 just to check it out...curiosity about
> the USB support, and wanted to see if I could see any differences in
> the 'improved' smp handling. Are you telling me I shouldn't run
> 2.3.12 just to check it out?
I'm telling you that if you do run 2.3.anything without taking steps to
figure out where they are in development, you run the risk of horrible
things happening to your system, and many people will consider you
foolish. At this point, asking questions about things that are broken
falls into the category of "open your mouth and remove all doubt".
> > The fact that you didn't know vfat was broken from 2.3.7 until
> > 2.3.13 tells me that you must have other reasons.
> You deduce wrong. I didn't know vfat was broken because I didn't
> read the changelog (much less trying to find the full thing) back to
> where it got broken.....
No, I was right. You did have other reasons. As I posted above, if
you had been following development *at all*, you couldn't have missed
the pagecache change. Ergo, you are *not* interested in following
development, only in trying a new release. Which is fine, except that
trying x.[13579].x releases without first knowing what is happening in
the scene isn't usually very smart.
> Don't worry, I won't be posting to the linux-kernel mailing list
> since I know 'outsiders' aren't welcome.
OK, now I'm going to have to take a mildly flaming tone. It really
bugs me when people go and do things like this and then try and play
martyr. Your sulk about elitism does not impress me; I hold that
linux-kernel is NOT elitist or unwelcoming of "outsiders". You only
get that impression because they are unwelcoming of people who post
questions without bothering to see if the answers are already out there
in places relatively easy to find. Basically by not doing your
homework you were wasting their time and bandwidth. (And if you don't
consider extra email to be "wasting bandwidth", ask someone elsewhere
in the world who has to pay per-K for their Internet connection.)
The same rules are honored ("more in the breach than the observance",
perhaps) on Usenet.
--
Peter Samuelson
<sampo.creighton.edu!psamuels>
------------------------------
From: [EMAIL PROTECTED] (Nathan Myers)
Crossposted-To: comp.os.linux.development.apps,comp.os.linux.networking
Subject: Re: C++ templates: More than Turing Complete?
Date: 24 Aug 1999 13:16:10 -0700
Ulrich Weigand <[EMAIL PROTECTED]> wrote:
>David Schwartz <[EMAIL PROTECTED]> writes:
>> If templates are Turing complete, it
>>is theoretically possible to create examples of C++ code that, while
>>legal, can never be compiled into valid code because the existence of
>>the valid code would be logically equivalent to a proof that the
>>compilation could not terminate.
This discussion has lost its referent.
The subject line is a bit of sarcasm from someone who didn't
understand the difference between computer science and engineering,
pretending to refute a claim that C++ templates allow libraries to
be written that cannot be written in any other lanugage. (Of course
Turing completeness doesn't address cleaving systems along library
boundaries.)
Later discussion was on the matter of what kind of computation
C++ templates can initiate at compile time. As that has already
been demonstrated to be Turing complete, the question is what
would be the effect of a non-halting compile-time metaprogram.
The result would be that either the compilation never finishes,
or (much more likely in practice) the compiler consumes all swap
space and dies, or just announces that its resource limits have
been exceeded. Not very interesting, really.
Of course none of this has the slightest to do with Linux,
so followups have been directed to comp.lang.c++.moderated.
--
Nathan Myers
[EMAIL PROTECTED] http://www.cantrip.org/
------------------------------
From: John Simpson <[EMAIL PROTECTED]>
Crossposted-To: alt.os.linux,comp.os.linux.advocacy,comp.os.misc,comp.unix.advocacy
Subject: Re: TAO: the ultimate OS
Date: Mon, 23 Aug 1999 13:31:26 -0700
Vladimir Z. Nuri wrote:
<snip>
>
> --
> ~^~^~^~^~^~^~^~^~^~^~^~^~^~^~^~^~^~^~^~^~^~^~^~^~^~^~^~^~^~^~^~^~^~^~^~^~^~^~^
> "in theory, there's no difference [EMAIL PROTECTED]
> between theory and practice, mad genius research lab
> but in practice there is!" http://www8.pair.com/mnajtiv/
Could you please provide a brief "diff" between this post and your
previous posts on "TAO" so that those of us who lived through the last
flamage don't have to go through it all again. Thanks.
--
John H. Simpson Phone: 503-450-2667 FAX: 503-450-3629
CNF AdTech Center. (CNF Transportation, Inc.)
1717 NW 21st St.
Portland, OREGON 97209 USA
for e-mail, remove _not_oj_ from address
------------------------------
From: [EMAIL PROTECTED] (W. Tucker)
Subject: Subj: Redirecting console to serial port for a SBC
Date: 24 Aug 1999 16:54:33 GMT
I am trying to install Linux on a Single Board Computer (SBC) that
has a floppy disk and a serial connection, but no keyboard or
monitor. I plan to connect a null modem serial cable between the SBC
and another computer and boot the SBC to Linux from a floppy. Then I
would like to redirect the SBC console so I can talk to it from the
other computer.
If I were using DOS on the SBC, I would include the command "ctty COM1"
in the autoexec.bat file to perform the console redirection. Is
there a similar command that I can use from Linux? Or is there a
better way to approach the problem?
As an experiment, I tried to redirect the console of my current Linux
box using the command "setconsole cua1". This failed to run and gave
an error message complaining that /proc/openprom/options does not
exist. I am using a 2.0.36 kernel from the Red Hat 5.2 distribution.
Any help or advice is greatly appreciated.
Thank you,
Wendy Tucker
Symmetric Research
[EMAIL PROTECTED]
------------------------------
From: "Ted Pavlic" <[EMAIL PROTECTED]>
Subject: The hashes sendmail uses... Berkeley DB?
Date: Tue, 24 Aug 1999 13:03:52 -0400
What type of hashes does sendmail use?
On my old RedHat 5.2 Linux systems, running sendmail-8.8.7, I had written
scripts in perl that used perl's "dbmopen" and "dbmclose" routines
(basically just tied hashes to DB_File databases) to read and write sendmail
hashes. (like /etc/aliases, for example)
Now that I have RedHat 6.0 Linux systems with sendmail 8.9.3, none of my old
scripts work anymore. It's as if the hashes makemap creates are not standard
DB_File. (Berkeley DB1, right?)
Any pointers? How can I read/write these things in perl?
All the best --
Ted
------------------------------
From: Jonathan Guthrie <[EMAIL PROTECTED]>
Crossposted-To: alt.os.linux,comp.os.linux.advocacy,comp.os.misc,comp.unix.advocacy
Subject: Re: TAO: the ultimate OS
Date: 24 Aug 1999 21:10:27 GMT
In comp.os.misc Vladimir Z. Nuri <[EMAIL PROTECTED]> wrote:
> ah, but surely we can assert that the more brilliant & visionary
> the design, the more difficult it is to pull it off, typically..
Well, you can assert it, but I won't believe you.
Truly brilliant stuff is cake to implement. If it's hard to do, then
perhaps it isn't as brilliant as you would have others believe.
--
Jonathan Guthrie ([EMAIL PROTECTED])
Brokersys +281-895-8101 http://www.brokersys.com/
12703 Veterans Memorial #106, Houston, TX 77014, USA
------------------------------
From: "Hung P. Tran" <[EMAIL PROTECTED]>
Crossposted-To: comp.os.linux.misc
Subject: upgrade linux kernel
Date: Tue, 24 Aug 1999 17:20:23 GMT
I am using Suse Linux 5.3 and Redhat linux 6.0. I am trying
to upgrade the Suse Linux kernel from 2.0.35 to 2.2.5 to
be compatible with the RedHat kernel. I download
the kernel image from an ftp site an attempt the upgrade.
When I try to recompile the kernel using "make zImage",
I receive an error of duplicate variable from
/usr/src/linux/arch/i386/lib/checksum.c. Further inspection
reveals that the variables are not duplicate. It's just that they
are very long and have the same 1st 17 characters, and so
appear the same to the compiler when they are truncated.
How do I overcome this problem ? Do I need to update my
gcc compiler (curently ver 2.7.2.1) or can I modify a makefile
somewhere to add a switch ??? Please advice ...
Should I just update the kernel to 2.2.5 or should I go all the way
to the newest one available (which I believe is 2.3.14) ???
Does newer necessarily mean better or that some version is
more stable than other ???
Thanks,
hung
------------------------------
From: Peter Mutsaers <[EMAIL PROTECTED]>
Crossposted-To: comp.os.linux.development.apps
Subject: Re: Shared Libraries: what is the linux equivalent of "dllimport" and
"dllexport"
Date: 23 Aug 1999 20:58:17 +0200
>> "BP" == Brian Poe <[EMAIL PROTECTED]> writes:
BP> Under VC++ in Win32, the lack of __declspec(dllexport) indeed
BP> does remove the function symbol from the dll.
AIX has a similar thing with its shared libraries: you have to list
every symbol that you intend to export. Personally I think that is a
horror.
In 9 out of 10 cases, using static is good enough (or better yet, when
using C++ you put all non-exported stuff in private and exported in
public sections). These are the generic language methods for selecting
whether to export something or not, and IMO shared libraries shouldn't
use too much extra magic from normal programs.
OK, C's way (static) is somewhat limited and crude; you have to
organize you program over source files so that everything which much
see each other is in a single file (module), that's just the way it is
with C. If you need more fine-grain control, use C++ classes.
--
Peter Mutsaers | Abcoude (Utrecht), | Trust me, I know
[EMAIL PROTECTED] | the Netherlands | what I'm doing.
------------------------------
From: "Donald Setlur" <[EMAIL PROTECTED]>
Subject: increasing process limits
Date: 25 Aug 1999 03:56:08 GMT
I have a program written to create multiple threads which will be
simultaneously active in the system.
I am experiencing problems with creating more than 251 concurrent threads
(probably the limit being 255 by default).
Another event that I also notice is that at the time the program has 251
threads, no other processes are allowed to exec.
Will increasing the process limit help ? If so how can a get a quick
cheatsheet to increase the number of allowable processes on the system ?
I am using Redhat 5.* for a OS. Please let me know if there is some more
details I need to supply to get an answer. Thanks in advance,
------------------------------
From: Torsten Poulin <[EMAIL PROTECTED]>
Subject: Re: Subj: Redirecting console to serial port for a SBC
Date: Tue, 24 Aug 1999 23:54:55 +0200
Miquel van Smoorenburg <[EMAIL PROTECTED]> wrote:
> In article <[EMAIL PROTECTED]>,
> W. Tucker <[EMAIL PROTECTED]> wrote:
>>As an experiment, I tried to redirect the console of my current Linux
>>box using the command "setconsole cua1". This failed to run and gave
>>an error message complaining that /proc/openprom/options does not
>>exist.
> Weird, what architecture is this? That's not an i386 error message..
> I don't even have the command on my distribtion (Debian).
It is in Red Hat 6.0 and is written by Eric Troan (it's a bash
script). The error message can be easily reproduced as there is no
/proc/openprom/options in RH 6.0. The man page does not say anything
about /proc/openproc/options; rather it mentions "/dev/openprom" in
the FILES section and describes it as "System interface to boot PROM"
which neither exists nor is used in the script, sigh.
-Torsten
------------------------------
From: Marcus Sundberg <[EMAIL PROTECTED]>
Subject: Re: How can I make device driver module to support many version of kernel?
Date: Tue, 24 Aug 1999 20:17:12 +0200
Peter Samuelson wrote:
> * Define a "compatibility layer" either now or whenever something
> changes, which is more or less guaranteed *not* to change. This is
> what libc provides (so that, for example, if the pre-2.2 version of
> the `umount' system call is ever dropped, programs that use it will
> still run fine so long as a semi-recent libc is installed). The idea
> here is that the libc layer would have a kernel-mode equivalent.
> Linus is adamantly opposed to this, though. He believes that such a
> layer is needlessly inefficient, complex at the source level, and
> still ends up tying your hands.
Linus doesn't have to like it.
People/companies wanting to provide binary only modules can write
their own (Open Source) compability layer and distribute it with
their binary drivers. Seems to work just fine for Vmware.
> so if you're reticent about giving out your driver source, you
> have to compile for individual releases and SMP/non-SMP. The AFS
> people complain about this occasionally.
Let them complain then. ;-)
http://www.stacken.kth.se/projekt/arla/
//Marcus
--
===============================+====================================
Marcus Sundberg | http://www.stacken.kth.se/~mackan/
Royal Institute of Technology | Phone: +46 707 295404
Stockholm, Sweden | E-Mail: [EMAIL PROTECTED]
------------------------------
** 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
******************************
