- l4-hackers - example.com Site

RE: l4-Linux and gasp
by Sebastian Schoenberg 27 Mar '01

27 Mar '01

The gasp .align directive should be written directly to the outfile. The warning is misleading, since some gas-versions (for alpha) use .align x as alignment to the power of two, other use the value given as argument. Sebastian --- PLEASE USE PGP !! http://os.inf.tu-dresden.de/~schoenbg/pgp-key.asc Dipl.-Inform. Sebastian Schoenberg University of Technology, Dresden Dept. of Computer Science Operating Systems Research Group Phone: +49 (351) 463 8401 Fax: +49 (351) 463 8284 > -----Original Message----- > From: Christian Stueble [mailto:stueble@cs.uni-sb.de] > Sent: Tuesday, March 27, 2001 3:11 PM > To: l4-hackers(a)kurt.inf.tu-dresden.de > Subject: l4-Linux and gasp > > > -----BEGIN PGP SIGNED MESSAGE----- > > Hello, > > due to problems with L4Linux/Fiasco I noticed that gasp > reports the warning: > > l4_idle.pre:1836 alignment must be one of 1, 2 or 4. > l4_idle.pre:1988 alignment must be one of 1, 2 or 4. > > when compiling l4_idle.pre. Could this cause trouble? > I use gcc 2.95.2, gasp 1.2 and as 2.9.5 > > Chris > - -- > - -=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=- > Christian Stueble......................stueble(a)cs.uni-sb.de > PubKey[BF7104F5].......fp=8678C5D3CAD9CD8C F1DDB8EC202F116A > > To be or not to be is true... (apocrypha of George Boole) > - -=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=- > > > -----BEGIN PGP SIGNATURE----- > Version: PGPfreeware 5.0i for non-commercial use > MessageID: uLWl6hG54Oh3oORs/VtBb5H+uikJFgkW > > iQCVAwUBOsCRefyJjJy/cQT1AQFTdgP+PdRya487EahMX7KzDYa/JQmLLT9gguJP > vQpd6sU6GpxmWpPr7bg4pNTN4L1QNj9WVCX6/IqW/2fJZKyeGMTfZiITLt6efnGm > PnZFH8RcOith2iROi8nyprSzv1GULEqJyrr1D7kQxVUodYmA3sxON4ToSxRi/AiU > 3k1XOab1N+o= > =MeHo > -----END PGP SIGNATURE----- > _______________________________________________ > l4-hackers mailing list > l4-hackers(a)os.inf.tu-dresden.de > http://os.inf.tu-dresden.de/mailman/listinfo/l4-hackers >

1 0

l4-Linux and gasp
by Christian Stueble 27 Mar '01

27 Mar '01

-----BEGIN PGP SIGNED MESSAGE----- Hello, due to problems with L4Linux/Fiasco I noticed that gasp reports the warning: l4_idle.pre:1836 alignment must be one of 1, 2 or 4. l4_idle.pre:1988 alignment must be one of 1, 2 or 4. when compiling l4_idle.pre. Could this cause trouble? I use gcc 2.95.2, gasp 1.2 and as 2.9.5 Chris - -- - -=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=- Christian Stueble......................stueble(a)cs.uni-sb.de PubKey[BF7104F5].......fp=8678C5D3CAD9CD8C F1DDB8EC202F116A To be or not to be is true... (apocrypha of George Boole) - -=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=- -----BEGIN PGP SIGNATURE----- Version: PGPfreeware 5.0i for non-commercial use MessageID: uLWl6hG54Oh3oORs/VtBb5H+uikJFgkW iQCVAwUBOsCRefyJjJy/cQT1AQFTdgP+PdRya487EahMX7KzDYa/JQmLLT9gguJP vQpd6sU6GpxmWpPr7bg4pNTN4L1QNj9WVCX6/IqW/2fJZKyeGMTfZiITLt6efnGm PnZFH8RcOith2iROi8nyprSzv1GULEqJyrr1D7kQxVUodYmA3sxON4ToSxRi/AiU 3k1XOab1N+o= =MeHo -----END PGP SIGNATURE-----

1 0

Little bug in L4-Linux?
by Christian Stueble 21 Feb '01

21 Feb '01

-----BEGIN PGP SIGNED MESSAGE----- Hello, When I compile linux22, I get the follwing error message in linux22/arch/l4/x86/emulib: ------------------------- ld -m elf_i386 -N -Ttext 0xa0008000 -o emulib int_entry.o user.o ld: warning: cannot find entry symbol _start; defaulting to a0008000 cp emulib emulib-s strip emulib-s ld -m elf_i386 -r \ -defsym _user_task_init=0x`nm emulib | grep user_task_init | cut -f1 -d' '` \ -defsym user_level_pager=0x`nm emulib | grep user_level_pager | cut -f1 -d' '` \ -defsym emulib_start=0x`nm emulib | grep -w real_sys_call | cut -f1 -d' '` \ -defsym emulib_end=0x`nm emulib | grep [_L]etext | cut -f1 -d' '` \ -defsym idt_table=0x`nm emulib | grep -w idt_table | cut -f1 -d' '` \ -defsym debug_idt_table=0x`nm emulib | grep -w debug_idt_table | cut -f1 -d' '` \ -defsym emulib_idt_descr=0x`nm emulib | grep emulib_idt_descr | cut -f1 -d' '` \ -defsym emulib_debug_idt_descr=0x`nm emulib | grep emulib_debug_idt_descr | cut -f1 -d' '` \ -T link-r.ld -o emulib.o emulib-s ld: cannot open a00087c2: Datei oder Verzeichnis nicht gefunden ------------------------- The problem seems to be the command "nm emulib | grep [_L]etext | cut -f1 ..." which returns two lines instead of one: ------------- a00087c2 a00087c2 ------------- Replacing "[_L]etext" by "_etext" works fine... Is this a bug or a feature? Ciao, Chris -- -=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=- Christian Stueble......................stueble(a)cs.uni-sb.de PubKey[BF7104F5].......fp=8678C5D3CAD9CD8C F1DDB8EC202F116A To be or not to be is true... (apocrypha of George Boole) -=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-

2 1

Problems compiling CVS Version 02/20/2001
by Christian Stueble 20 Feb '01

20 Feb '01

-----BEGIN PGP SIGNED MESSAGE----- Hello, I tried to compile the l4 CVS version from today, but I have some oskit/oskit10 problems: 1) l4/Makeconf seems to add oskit10 includes even if OSKIT10 is not set. Maybe the line "ifeq ($(OSKIT10),) " is wrong? 2) When I fixed that, l4/pkg/log/lib/src/doprnt.c produces an error, because oskit/boolean.h is not there. Do you mean flux/boolean.h, or should oskit10 be used instead of oskit? 3) I also tried to use oskit10, which produces errors in oskit_support and some services because of missing headers, e.g., lmm/lmm.h or flux/machine/pc/reset.h Some files seem to require oskit and some oskit10?? Ciao, Chris -- -=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=- Christian Stueble......................stueble(a)cs.uni-sb.de PubKey[BF7104F5].......fp=8678C5D3CAD9CD8C F1DDB8EC202F116A To be or not to be is true... (apocrypha of George Boole) -=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-

2 1

do i need oskit for working with L4?
by Subbaraj, Thirumurugan 22 Jan '01

22 Jan '01

Hello, I have downloaded the GRUB and OSkit. How do i load the kernel/build the image with L4? Any suggestions? regards -thiru

1 0

problem starting the hello server
by Christian Rehn 11 Jan '01

11 Jan '01

Hi, I,ve got a problem starting the hello server on top of fiasco. I checked out l4-fiasco from the cvs repository and compiled it. My menu.lst looks as follows: kernel=(nd)/boot/tftpboot/rehn/rmgr -sigma0 module=(nd)/boot/tftpboot/rehn/main -nojdb module=(nd)/boot/tftpboot/rehn/sigma0 module=(nd)/boot/tftpboot/rehn/hello When I try to boot l4 rmgr starts, loads the l4 kernel and the sigma0 server but does not load the hello server (I'm not sure if it is loaded but not executed or if it is not loaded but I can not find the code that should load the hello server). Where should the hello server be loaded? Why does the rmgr "exec_load" main and the sigma0 server but not the hello server in function startup() in startup.c? Thanks, Christian

1 0

IPC
by Jordan Curzon 03 Jan '01

03 Jan '01

Hello, What is the most common way for apps(or libs) to find the task id for a service(networking for example)? _________________________________________________________________ Get your FREE download of MSN Explorer at http://explorer.msn.com

1 0

L4 benchmarks
by Greg Law 12 Dec '00

12 Dec '00

Hi, I would like to get some benchmarks for L4 on Pentium. Specifically, I am after the times to create a process and execute a getpid (preferably in cycles). Does anyone know where I can get such info before I try and go through the pain of measuring it myself? Cheers, Greg -- Greg Law, email: gel(a)soi.city.ac.uk Dept. Of Computing, web: http://www.soi.city.ac.uk/~gel/ City University, phone: +44 20 7477 8341 London, UK. EC1V 0HB. fax: +44 20 7477 8587

1 0

Missing files in recent cvs versions
by David Welch 16 Nov '00

16 Nov '00

Hi, Some of the Makefiles in fiasco want to use Makeconf.lib.targets and Makeconf.lib.rules which don't seem to be in the cvs repository. Also the Fiasco bug tracking link gives a 404.

2 2

Ideas for some sample L4 root_task(s) [Learning L4].
by Farid Hajji 25 Oct '00

25 Oct '00

Hi Brent, this is what I'm planning to do next to get used to the L4-Ka interface. If you are interested, let's do this together. Basically it's about exercising misc. L4 syscalls in the root_task itself. Once we understand this, we can grab libmom and add L4 sysdeps to glibc and the Hurd. 01. map from sigma0 the kernel info page into the address space of the root_task. printf() its contents, especially its 'clock' field which is updated by the ukernel. sleep some useconds and printf clock again. [-> learn more about sigma0 rpc protocol, how to send/receive a message to/from sigma0, some memory managment issues in general and l4_sleep(). ] 02. activate some threads in root_task, and let each thread continouusly putc() a unique identifying character to the screen. look at the resulting pattern. A control thread could getc() chars from keyboard and use this char to start/stop the thread that prints this char. [-> learn about threads and scheduling.] Add a pager to a thread which just printf()s each page request, before forwarding it to sigma0. Alternatively, trap the request and handle it by mapping an own page to the requester. Finally, reclaim pages that are not needed anymore (how? and how to get notified about this condition?). [-> learn about pagers and the L4 application/pager protocol.] 03. create new (child) tasks and: * map/grant some pages * send/receive messages: short message, long message, with and without string dopes etc.. * synchronize threads via blocking l4-receive syscalls. [-> learn about messages, ipc calls.] 04. understand what rmgr really does, what its interfaces to the root_task and other tasks are (a.k.a. how to request/receive ressources from rmgr) and try to implement Okuji's idea about relocation of modules (glinux) so that we can boot glinux with vanilla GNU Grub (no modaddr). 05. Attach a root_task thread to one interrupt number (say com1, clock, keyboard or whatever) and printf/acknowledge each received interrupt. [-> Handling interrupts. Is it possible to get an interrupt that is already bound by rmgr? Hmmm... ] [-> What about chaining interrupts/events? ] Write a trivial device driver server (like sorta translator), e.g. a clock device, that gets clock interrupts and reads the clock page in the kernel to update its internal notion of time. Add protocol so that other threads/tasks can ask the time via RPC (simple message) from this clock server. [-> generalize this to device driver servers in general. Think about a Hurd abstraction to access such servers. Driver-Translators? Driver-IPC? ] 06. Grab the code of a real device driver from OSKit and try to turn it into a device driver server as in 05. 07. [...] Real world examples are located in RMGR source code, and in the sources of L4-Fiasco apps and L4Linux. I'll continue to use L4Ka right now though, simply because I can compile it without problems on my FreeBSD box ;-). Any suggestions? -Farid. -- Farid Hajji -- Unix Systems and Network Admin | Phone: +49-2131-67-555 Broicherdorfstr. 83, D-41564 Kaarst, Germany | farid.hajji(a)ob.kamp.net - - - - - - - - - - - - - - - - - - - - - - - + - - - - - - - - - - - - Murphy's Law fails only when you try to demonstrate it, and thus succeeds.

1 0