l4-hackers August 2003

l4-hackers@os.inf.tu-dresden.de

14 participants
12 discussions

New cli/sti semantics in Fiasco 1.0?
by Jacob Gorm Hansen 30 Aug '03

30 Aug '03

hi, as far as I recall, Fiasco used to re-enable interrupts upon returning from a syscall, even if they had been disabled before the syscall. Some of my old code (my patched version of L4Linux22 which I am using with nomadbios) seems to have relied on this behaviour and now fails with a "page fault in cli mode" on the first intruction after an IPC-operation. Am I correct that the behaviour has been changed like this, or do I just have to configure my Fiasco differently? The old Fiasco I am using is from CVS about a year ago. thanks, Jacob

1 1

l4/pkg/con etc
by David Denny 30 Aug '03

30 Aug '03

Hello Hackers. Having got to the stage of compiling successfully my second (debian) hello+fiasco+glinux set, I'm now looking at cool next step (read: "easy wins for the temporally challenged"). UX and CON seem to be my candidates. OK, UX looks more like the world I'm used to, so I won't ask about that yet. But I /do/ have a Q. about the drops con, which has wider learning implications, so a few hints here won't go amiss. My guess is that to yield up a con binary, I cd to l4/pkg/con, "make", and /no/ that isn't it because it falls over... my predecessor instructions were per website, CVSco, compile hello and fiasco using gcc3.2, compile linux22 using gcc2.95. so: how do I compile "con" and other packages, please? (or: a generalised rule to cover this) ;-) TIA David D

2 1

L4Linux boot stop "param_word bits 16-19 must be 0. Fix your bindings!"
by David Denny 28 Aug '03

28 Aug '03

Hello Hackers booting l4linux stops with: KDB: thread_schedule: param_word bits 16-19 must be 0. Fix your bindings! My grub menu contains: +-------------------------------------------------------------------[0]---+ | 0: kernel (hd0,2)/denny/Dresden/rmgr -sigma0 | | 1: modaddr 0x02000000 | | 2: module (hd0,2)/denny/Dresden/main -nokdb -nowait -comspeed 9600 -com| | 3: module (hd0,2)/denny/Dresden/sigma0 | | 4: module (hd0,2)/denny/Dresden/glinux root=/dev/hda3 | | 5: module (hd0,2)/denny/Dresden/hello umm, pasting from the remote serial console.. line 2 ends in "-comport 1" OK let's reboot the big system & paste the whole thing. BTW your hackked grub is now sitting in the mbr of (hd0) AND it WORKS ! 8-) title failsafe chavin:/boot/grub # less menu.lst # Modified by YaST2. Last modification on Tue Aug 26 06:06:59 2003 color white/blue black/light-gray # default 0 # timeout 8 serial --unit=0 --speed=9600 --word=8 --parity=no --stop=1 #terminal --timeout=10 serial console terminal serial console title SuSE Linux 8.2 Kernel 2.4.20 kernel (hd0,2)/boot/vmlinuz root=/dev/hda3 vga=0x31a splash=silent showopts initrd (hd0,2)/boot/initrd title (Remote) SuSE Linux 8.2 Kernel 2.4.20 kernel (hd0,2)/boot/vmlinuz root=/dev/hda3 console=tty0 console=ttyS0,9600n8 initrd (hd0,2)/boot/initrd title Debian Linux testing/unstable Kernel 2.2.20 kernel (hd0,1)/boot/vmlinuz-2.2.20 root=/dev/hda2 splash=silent vga=0x31a title (Remote) Debian Linux testing/unstable Kernel 2.2.20 kernel (hd0,1)/boot/vmlinuz-2.2.20 root=/dev/hda2 console=tty0 console=ttyS0,9600n8 title floppy root (fd0) chainloader +1 title failsafe kernel (hd0,2)/boot/vmlinuz.shipped root=/dev/hda3 showopts ide=nodma apm=of f acpi=off vga=normal nosmp noapic maxcpus=0 3 initrd (hd0,2)/boot/initrd.shipped title L4Linux -+- Hello World kernel (hd0,2)/denny/Dresden/rmgr -sigma0 modaddr 0x02000000 module (hd0,2)/denny/Dresden/main -nokdb -nowait -comspeed 9600 -comport 1 module (hd0,2)/denny/Dresden/sigma0 module (hd0,2)/denny/Dresden/glinux root=/dev/hda3 module (hd0,2)/denny/Dresden/hello title L4Linux-------------------><8------------------------ Apols I have not gone the netboot road, as "all"! my boxes are totally different there are too many risks for cross compiling, I compile for the OS and box where it loads from. Host OS is SuSE8.2 with linux22 compiled under 2.95.3 and l4 + the rest under 3.2 (i.e. reverted), all compiles terminated OK, all libraries & dependencies found to the best of my ability. The same error has stopped me for some 24 hours; I have been round the houses tidying up all I can find here so I await your advice with eagerness. In the meantime, to get a bit nearer, I have the Debian system ready for a similar assault when I have my energy restored... TIA David Denny

3 3

Driver using DDE
by Chris Lambert 26 Aug '03

26 Aug '03

Hello, we try to implement drivers for DROPS using the DDE, but we only have the sounddriver described in the diploma thesis of Christian Helmuth. Are there other sample drivers (smaller) available for us to see how to use DDE? It's part of a project at Operating Systems Chair at University of Saarland. Chris Lambert Markus Pister Marc Schlickling

2 1

Thanks all
by David Denny 25 Aug '03

25 Aug '03

Well we now have fiasco booting Hello, and more or less satisfactory boots of cpufreq and pingpong, on SuSE8.2 with reverted compiler (3.2). Thanks ALL for your assistance. My next steps are to get a L4Linux working on SuSE8,2 and then repeat for Debian. The side effects are of course swapping compilers and libraries in and out of the host OS. I should like to see after that if I can get the DoPE demo working properly. I am guessing so far that it may be some underlying library or compilation incompatibility. I have checked that vbeset, vbeprobe and testvbe work in grub, so it /doesn't/ seem to be the machine, and it's likely therefore that the issue has to be somewhere in the compiled binaries on the demo diskette. I'm wondering therefore if at some time the sources for vscrtest and its prereqs might be made available? (I will speak to Norman about DoPE). Thanks again. I hope to be back when I have got the next bits working successfully. Kind Regards & thanks all once again. David D

1 0

DROPS make in Debian Testing, failed from linux22 drivers
by David Denny 23 Aug '03

23 Aug '03

Hi Hackers After failing to "make hello" from the website, I tried "make" Drops from the docs page, however this has also crashed... see snip line. gcc3.3.1 again. debian/testing TIA David ----8><---------------------------------------------------------------------------------- In file included from /home/drops/linux22/include/asm/l4arch/floppy.h:4, from /home/drops/linux22/include/asm/floppy.h:4, from floppy.c:201: /home/drops/linux22/include/asm-i386/floppy.h:78:8: missing terminating " character /home/drops/linux22/include/asm-i386/floppy.h:99:9: missing terminating " character In file included from /home/drops/linux22/include/linux/blkdev.h:6, from /home/drops/linux22/include/linux/blk.h:4, from floppy.c:206: /home/drops/linux22/include/linux/genhd.h: In function `ptype': /home/drops/linux22/include/linux/genhd.h:83: warning: deprecated use of label at end of compound st atement make[5]: *** [floppy.o] Error 1 make[5]: Leaving directory `/home/drops/linux22/drivers/block' make[4]: *** [first_rule] Error 2 make[4]: Leaving directory `/home/drops/linux22/drivers/block' make[3]: *** [_subdir_block] Error 2 make[3]: Leaving directory `/home/drops/linux22/drivers' make[2]: *** [_dir_drivers] Error 2 make[2]: Leaving directory `/home/drops/linux22' make[1]: *** [vmlinux] Error 2 make[1]: Leaving directory `/home/drops/linux22' make: *** [../linux22] Error 2 dresden@chavin:/home/drops/l4$

2 1

Executable Subset
by David Denny 23 Aug '03

23 Aug '03

Hello Hackers again.. I've reconstituted my Fiasco box with a Debian (Stable) and a SuSE (8.0) dual boot. My aim is to have a workable approximation to the Dresden configurations using Debian, while the SuSE version will enable me to help test on that platform. I have acquired gcc3.2 and binutils to match and those will be installed shortly. I have a question for the group. My disk space is limited. I can build on this computer but it would be ideal if after building I could move the bulk of the l4, ($BUILDDIR), oskit, and the rest onto another machine, retaining on my "Fiasco" box only those files needed for a running installation, i.e. kernel modules, libraries, and whatever else. (It goes without saying that I need to reserve space for any linux executables that I intend to run on top of Fiasco) I'm happy to experiment with this question, but if the group already possesses a list of executables, libraries, and so on, it would be nice to start from there. OTOH if you don't then perhaps I could make a contribution to the project by creating this list. -- ____________________________________________ http://www.operamail.com Get OperaMail Premium today - USD 29.99/year Powered by Outblaze

3 2

auth.idl
by Etienne Robillard 20 Aug '03

20 Aug '03

Hi Hackers, on implementing the auth module (currently running on mach_port_t and all) in idl4 grammar with L4 ipc/syscalls. Question is how the relationship between mach_port_t would affect l4_threadid_t and folks ? The following is not currently appropriate for any of the Hurd/Mach RPC device or L4 syscalls: module auth { /* this is the generic auth module implementation for Hurd */ interface auth_getids /* generic api for _ids */ { int l4_ids_t([out] static l4_threadid_t ids_t, [out] &idarray_t /* some foo here je je */); }; This is only a partial implementation of the auth.def interface - but since the module is so small; it may be easier to port with l4_* components. It does use the idl4 fast compiler and is expected to be loaded as a root_task/clan_ipc ;) The brackets are for compatibility with DCE/DICE. etienne ______________________________________________________________________ Post your free ad now! http://personals.yahoo.ca /* experimental (and partial too) reincarnation of the 'Hurd' auth server on the L4 microkernel. It should be loaded as the root_task .. I need to know however the exact translation (idl grammar of the auth server on L4) that is required in order to get the module functionning ( or at least to compile! ) please no flame. */ module auth { interface auth_getids { int get_ids([out] static l4_threadid_t ids_t, [out] &idarray_t); };

2 1

Video Issue with DROPS demo
by David Denny 14 Aug '03

14 Aug '03

Hello Hackers I have a video problem with this machine running the Fiasco demo and my compiled fiasco modules. The latter hang on boot, or appear to do so. the demo appears to run, but black screen, and 1-pixel deep band of pretty coloured dots right at the top. Compaq pII400 deskpro sff. ATI3D Rage Pro AGP 1x/2x I will appreciate knowing which parts of which files might be worth looking at. My belief is that it's a vesa/framebuffer setting. Normal Linux runs X perfectly (both Deb and SuSE), and the (sorry) Pistachio demo boots and runs properly. SO, it has to be something simple. Grub settings and menu taken from Dresden website. TIA David -- ____________________________________________ http://www.operamail.com Get OperaMail Premium today - USD 29.99/year Powered by Outblaze

2 1

memory descriptor overflow was RE: [RMGR] kernel entry point and pistachio
by Etienne Robillard 11 Aug '03

11 Aug '03

--- Volkmar Uhlig <volkmar(a)ira.uka.de> wrote: > You can't use RMGR with Pistachio--why would you...? O.K - maybe it was not a very good idea.. you decide ;) I'm getting a memory descriptor overflow (m=200; n=200), while trying to _boot_ the kernel. 'esquisse' of whats going on: RMGR -- configuring s0 / roottask RMGR -- can't detect L4 kernel -- assuming UKA kernel. RMGR -- booting /boot/ia32-kernel... pistachio on foo@bar ... init_arch->boot_mem_init->kmem_init->activate_gdt(); several remaps() occurs... entering KDB, locked, blood shed ;) You can see in init.cc (pistachio-0.1) the following line : i < get_kip()->mem_info.get_num_descriptors(); i++ (...) hint: It would be great if the kernel could return to RMGR but its too early to tell - for now Volkmar is right and thus no RMGR can be used with X.2 ;) etienne ===== iD ______________________________________________________________________ Post your free ad now! http://personals.yahoo.ca

2 1

2024

2023

2022

2021

2020

2019

2018

2017

2016

2015

2014

2013

2012

2011

2010

2009

2008

2007

2006

2005

2004

2003

2002

2001

2000

1999

1998

l4-hackers August 2003