Hi Michael,
I tried again to compile l4 and I'm a step further: cd l4 mkdir include/l4/log cd include/l4/log ln -s ../../../pkg/log/include/l4/log/l4log.h .
Now, gmake bombs with gmake[4]: Entering directory `/tmp/l4-devel/l4/pkg/l4util/examples/perf' gmake[4]: *** No rule to make target `main.ld', needed by `perf_p6'. Stop.
How can a main.ld be generated? The two other main.ld I've found: ./pkg/log/server/src/main.ld ./pkg/names/server/src/main.ld were quite old (nov 1999) and certainly not generated by gmake!
I'm still suspecting an incomplete l4 distribution, probably due to CVS.
Trying to compile oskit10, I'm also missing files (I already told you this). I generated a oskit10/modules.x86.pc manually by putting the name of each subdirectory of oskit10 by itself on a line. Using gcc-2.95.2, configure resumes but misses the following files:
sed: ./unsupported/scripts/sym-trace.in: No such file or directory sed: ./boot/GNUmakefile.in: No such file or directory sed: ./examples/GNUmakefile.in: No such file or directory sed: ./freebsd/GNUmakefile.in: No such file or directory sed: ./fs/GNUmakefile.in: No such file or directory sed: ./fsnamespace/GNUmakefile.in: No such file or directory sed: ./linux/GNUmakefile.in: No such file or directory sed: ./netbsd/GNUmakefile.in: No such file or directory sed: ./posix/GNUmakefile.in: No such file or directory sed: ./video/GNUmakefile.in: No such file or directory sed: ./wimp/GNUmakefile.in: No such file or directory sed: ./x11/GNUmakefile.in: No such file or directory
Here again, I'm suspecting an incomplete distribution due to cvs problems.
Thank you,
-Farid.
Hello Mr. Hajji,
I'm still suspecting an incomplete l4 distribution, probably due to CVS.
Also I'm wondering if there's any touble in CVS server. I remember I could build Fiasco & L4-Linux on 2000/July/6th on GNU/Linux, but building oskit10 of the latest CVS source tree is stopped by missing file, as Mr. Hajji reported. I tried to track when the critical commit was done, but I could not. I suppose some files are lost from CVS server. For example, the tarball I downloaded on 2000/July/6th from CVS has such entries:
% cat oskit10/flask/CVS/Entries /GNUmakefile.in/1.1.1.1/Thu Sep 23 13:07:27 1999// /GNUmakerules/1.1.1.1/Thu Sep 23 13:07:27 1999// /file.av/1.1.1.1/Thu Sep 23 13:07:27 1999// /file.common/1.1.1.1/Thu Sep 23 13:07:27 1999// /mkaccess_vector.sh/1.1.1.1/Thu Sep 23 13:07:27 1999// /mksecurity.sh/1.1.1.1/Thu Sep 23 13:07:27 1999// /net.av/1.1.1.1/Thu Sep 23 13:07:27 1999// /net.common/1.1.1.1/Thu Sep 23 13:07:27 1999// /security.prefix/1.1.1.1/Thu Sep 23 13:07:27 1999// /security.suffix/1.1.1.1/Thu Sep 23 13:07:27 1999// /security_classes/1.1.1.1/Thu Sep 23 13:07:27 1999// D
But, I tried to download today, the result is % cvs -z3 -d :pserver:guest@os.inf.tu-dresden.de:/home/remote-cvs co -D "2000-06-01 00:00" oskit10 ... % cat oskit10/flask/CVS/Entries /GNUmakefile.in/1.1.1.1/Thu Sep 23 13:07:27 1999//D2000.05.31.15.00.00 /GNUmakerules/1.1.1.1/Thu Sep 23 13:07:27 1999//D2000.05.31.15.00.00 /mkaccess_vector.sh/1.1.1.1/Thu Sep 23 13:07:27 1999//D2000.05.31.15.00.00 /security_classes/1.1.1.1/Thu Sep 23 13:07:27 1999//D2000.05.31.15.00.00 D
7 files could not be downloaded. What absurd mistake I did?
Best Wishes,
suzuki
P.S. To Mr. Hajji: I don't have FreeBSD-5.0 pc. But I can try on NetBSD-1.4.2 pc, are you interested in?
Received: from iraun1.ira.uka.de (iraun1.ira.uka.de [129.13.10.90]) by kurt.inf.tu-dresden.de (AIX4.3/8.9.3/8.9.3) with ESMTP id CAA37458 for l4-hackers@os.inf.tu-dresden.de; Fri, 15 Sep 2000 02:51:07 +0200 Received: from i30s25.ira.uka.de by iraun1 (PP) with ESMTP; Fri, 15 Sep 2000 02:51:05 +0200 Received: from l4ka.org (actually i30s25) by i30s25 (PP) with ESMTP; Fri, 15 Sep 2000 02:51:01 +0200 Message-ID: 39C17275.A6722B5B@l4ka.org Date: Fri, 15 Sep 2000 02:51:01 +0200 From: L4KA Team info@l4ka.org Organization: University of Karlsruhe X-Mailer: Mozilla 4.75 [en] (X11; U; Linux 2.2.17 i686) X-Accept-Language: en MIME-Version: 1.0 To: info@l4ka.org Subject: L4KA out now Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Sender: l4-hackers-admin@os.inf.tu-dresden.de Errors-To: l4-hackers-admin@os.inf.tu-dresden.de X-BeenThere: l4-hackers@os.inf.tu-dresden.de X-Mailman-Version: 2.0beta2 Precedence: bulk List-Id: General L4 discussions <l4-hackers.os.inf.tu-dresden.de>
Hi L4-hackers,
we're proud to inform you that we've released a first preliminary version of the L4KA microkernel on September 15th.
L4KA is written in C++ and runs on x86 and ARM. We reach performance which is comparable to Jochens x86 assembly version (+7%). The kernel interface is compatible to L4 Version X.0 (lacking support for clans and chiefs and small address spaces). L4Linux runs stable on the x86 kernel.
For more information please visit (and register) at http://l4ka.org/.
---- University of Karlsruhe, Uwe Dannowski, Jochen Liedtke, Espen Skoglund, Volkmar Uhlig
Hello Mr. suzukis,
I'm still suspecting an incomplete l4 distribution, probably due to CVS.
Also I'm wondering if there's any touble in CVS server.
I've just downloaded the newest oskit from flux project [Version 0.97 - January 15, 1999 (Snapshot 20000901), directly off cvs]:
% # WARNING: don't clobber dresden's oskit; do it in another subdir: % mkdir flux-oskit % cd flux-oskit
% cvs -d :pserver:anoncvs@fluxcvs.cs.utah.edu:/cvs login <enter a non-empty password, e.g. 'blah'> % cvs -z9 -d :pserver:anoncvs@fluxcvs.cs.utah.edu:/cvs checkout oskit
The files that were missing in Dresden's oskit10 were present in flux (e.g. oskit/modules.x86.pc and the GNUmakefile.in in misc. dirs among others).
I was able to compile flux' oskit on FreeBSD-5.X out of the box without any problems (using gmake, gcc-2.95.2 and binutils-2.9.1). So there is definitely something wrong with the CVS server at Dresden :-(
BTW, did anyone tried to use flux' oskit instead of oskit10 with l4? If that was not possible, what are the _exact_ modifications w.r.t. flux' oskit that generated oskit10? I thought that oskit was designed in such a manner, that OS developers can override each oskit function separately (i.e. outside oskit's source tree). Is this the case with the l4 module at Dresden too?
P.S. To Mr. Hajji: I don't have FreeBSD-5.0 pc. But I can try on NetBSD-1.4.2 pc, are you interested in?
That would be great! The more OS build-environments we try, the better ;-). You'll probably want to check wether NetBSD-1.4.2's build environment uses gcc-2.95[-2]/binutils-2.9.1 already. FreeBSD switched only recently from native gcc-2.7.2.3 to gcc-2.95-2...
Thanks,
-Farid.
I wish if anybody in Dresden team comments such question...
did anyone tried to use flux' oskit instead of oskit10 with l4?
I think Fiasco is (although slowly) moving from oskit to oskit10. Towards to the newest Flux OSkit will be after it. In my previous building of Fiasco, it was still dependent on both of oskit & oskit10. So, replacing oskit by oskit10 seems to be long way. Is it so difficult work?
But I can try on NetBSD-1.4.2 pc, are you interested in?That would be great! The more OS build-environments we try, the better ;-). You'll probably want to check wether NetBSD-1.4.2's build environment uses gcc-2.95[-2]/binutils-2.9.1 already. FreeBSD switched only recently from native gcc-2.7.2.3 to gcc-2.95-2...
O.K., sooner I will try. I've ever compiled Fiasco on DEC OSF/1 :-).
Best Wishes,
suzuki
P.S., please don't send Cc: to suzukis@flavor1.ipc.hiroshima-u.ac.jp, this mail server talks only with the mailing list servers (of which I'm subscribing.) However, I opened to kamp.net too.
l4-hackers@os.inf.tu-dresden.de
-
Farid Hajji -
suzukis＠flavor1.ipc.hiroshima-u.ac.jp