- l4-hackers - example.com Site

first experiences with l4ka on a real pentium
by Farid Hajji 19 Oct '00

19 Oct '00

Hi Brent, [I Cc-ed this to contact(a)l4ka.org and l4-hackers. L4Ka reports from other users would be great too.] > I'm curious because I get similar problems to your > description when trying to boot the L4KA kernel > on a K6-2 chip. > > However, in my case the boot process seems to work > fine as far as the initial sigma0 task starts, then > rmgr in debug mode will print an error message > (or a "Hello from RMGR" message depending on the build > I use). > > Basically, the boot fails before getting into glinux > and if rmgr and the kernel are built for debugging > they will trap the error and stop. In "runnable" > mode, I just get a soft reboot. I just tried L4Ka on a real pentium box: CPU: Pentium/P54C (199.43-MHz 586-class CPU) Origin = "GenuineIntel" Id = 0x52c Stepping = 12 Features=0x1bf<FPU,VME,DE,PSE,TSC,MSR,MCE,CX8> real memory = 134217728 (131072K bytes) I was able to boot the provided root_task template, but not glinux (yet). I booted entirely from floppy, since I can't use the hd of this production FreeBSD-box for experiments. I compiled l4ka for 586 (not 686) with kernel debugger, sigma0, rmgr and the template root_task. The first thing to do was to disable the call to the kernel debugger on startup in l4-ka/kernel/src/init.c:init() #if defined CONFIG_DEBUG_KDB_ONSTART enter_kdebug("startup"); #endif I did this because I wanted to keep the kernel debugger but I didn't found out how to use it to resume booting. I then had to figure out, that the LINKBASE address in l4-ka/root_task/Makefile had to be changed to a real existing physical address: < LINKBASE=0x0c200000 > LINKBASE=0x07200000 If not, the following error message appears: RMGR: loading task (fd0)/l4ka/root_task from 0x159000_0x15a6bc to [0xc200000-0xc201504 cannot allocate page at 0xc200000 : owned by 0x89] Return reboots, "k" enters L4 kernel debugger. Trying to display the memory regions with the kernel debugger was easy. E.g. "d158fb0" shows a dump at address 00159000 which started like this: 00159000: 46 .... <after 46: ELF...> I wished the kdb was documented somewhere, not only in its sources :( The screen handling is also not perfect yet (see above with 'd' command). Dumping c200000 hangs the machine, probably because that address is outside of the physical range. After fixing LINKBASE, booting with the following menu.lst entry was successful: title=L4Ka without kernel debugger on start kernel=(fd0)/l4ka/rmgr -sigma0 module=(fd0)/l4ka/x86-kernel.nokdb module=(fd0)/l4ka/sigma0 module=(fd0)/l4ka/root_task.7200000 root_task is just a hello world, followed by an endless loop. This is the contents of the screen then: ----- cut here ---------- interrupt association failed (e0044800 irq: 2) interrupt association failed (e0046000 irq: 8) sigma0_main: kernel_info_page is at 0x00107000 sigma0: map 4M page at 5800000 to 4040001 sigma0: map 4M page at 5c00000 to 4040001 sigma0: map 4M page at 6000000 to 4040001 sigma0: map 4M page at 6400000 to 4040001 sigma0: map 4M page at 6800000 to 4040001 sigma0: map 4M page at 6c00000 to 4040001 sigma0: map 4M page at 7000000 to 4040001 sigma0: map 4M page at 7400000 to 4040001 RMGR: total RAM size = 130688 KB (reported by bootloader) received 122152 KB RAM from sigma0 204 KB reserved for RMGR RMGR: attached irqs = [<!0> 1 <!2> 3 4 5 6 7 <!8> 9 a b c d e f] RMGR: loading task (fd0)/l4ka/root_task.7200000 to [0x7200000-0x7201504] RMGR: starting task (fd0)/l4ka/root_task.7200000 at entry 0x7200000 via trampoline page code 0x3140. main: started. <<<-- appears over sigma0_main: ... <<<-- due to missing screen handling. ----- cut here ------------- So far, so good. I then tried to boot glinux with the following menu.lst entry: title=L4Ka with glinux kernel=(fd0)/l4ka/rmgr -sigma0 module=(fd0)/l4ka/x86-kernel.nokdb module=(fd0)/l4ka/sigma0 module=(fd0)/l4ka/glinux.gz -s knowing that I don't have any linux root partition on that FreeBSD box. Anyway, I just wanted to see glinux start and no more for now. As you can see from menu.lst, I had to gzip -9 glinux to fit it on one floppy. GRUB should be able to load gzipped modules. Booting this resulted in the following output screen: ----- cut here -------------- sigma0_main: kernel_info_page is at 0x00107000 sigma0: page 200002 requested twice, old=ff, new=4 sigma0: page 200002 requested twice, old=ff, new=4 sigma0: page 200002 requested twice, old=ff, new=4 sigma0: page 200002 requested twice, old=ff, new=4 sigma0: page 200002 requested twice, old=ff, new=4 sigma0: page 200002 requested twice, old=ff, new=4 sigma0: page 200002 requested twice, old=ff, new=4 sigma0: page 200002 requested twice, old=ff, new=4 sigma0: page 200002 requested twice, old=ff, new=4 sigma0: page 200002 requested twice, old=ff, new=4 ----- cut here -------------- The last line being constantly redrawn in an (endless?) loop. I didn't try anything else, because I was not able to compile Dresden's L4Linux on FreeBSD (I used l4ka.org's glinux binary). At least the root_task seems to work and this is very good for now. Based on root_task, L4 servers can be build, probably with flux' oskit (which compiles w/o problems on FreeBSD too!). The we can start thinking about libmom and porting the Hurd to L4Ka. I can't resume tests on a real pentium right now, so I'll have to look at bochs sources to add support for 4MB superpages. This will take its time though, since I'm not an x86 expert :-). Thanks, -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.

3 4

RE: first experiences with l4ka on a real pentium
by Brent Fulgham 19 Oct '00

19 Oct '00

Farid, > [I Cc-ed this to contact(a)l4ka.org and l4-hackers. L4Ka > reports from other users would be great too.] > (I'll also CC the L4 groups) > I then had to figure out, that the LINKBASE address in > l4-ka/root_task/Makefile > had to be changed to a real existing physical address: > > < LINKBASE=0x0c200000 > > LINKBASE=0x07200000 > > If not, the following error message appears: > > RMGR: loading task (fd0)/l4ka/root_task from 0x159000_0x15a6bc > to [0xc200000-0xc201504 cannot allocate page at > 0xc200000 : owned by 0x89] > Return reboots, "k" enters L4 kernel debugger. > Yes, I saw this as well. > So far, so good. > I then tried to boot glinux with the following menu.lst entry: > > title=L4Ka with glinux > kernel=(fd0)/l4ka/rmgr -sigma0 > module=(fd0)/l4ka/x86-kernel.nokdb > module=(fd0)/l4ka/sigma0 > module=(fd0)/l4ka/glinux.gz -s > > knowing that I don't have any linux root partition on that > FreeBSD box. > Anyway, I just wanted to see glinux start and no more for now. As you > can see from menu.lst, I had to gzip -9 glinux to fit it on > one floppy. > GRUB should be able to load gzipped modules. Booting this resulted in > the following output screen: > In my case, for a split second I actually did see the start of a standard Linux 2.2-series boot sequence. But it almost immediately reboot. > I didn't try anything else, because I was not able to compile > Dresden's L4Linux on FreeBSD (I used l4ka.org's glinux binary). > At least the root_task seems to work and this is very good for now. > Based on root_task, L4 servers can be build, probably with flux' oskit > (which compiles w/o problems on FreeBSD too!). The we can start thinking > about libmom and porting the Hurd to L4Ka. > Yes, this is excellent news. But why do you feel the OsKit is necessary? It seems that with a working L4KA it should be possible to follow the example of the hello-world template to produce servers of our own. Or perhaps it's just easier with OsKit? I must admit I haven't taken the time to master that beast :) > I can't resume tests on a real pentium right now, so I'll have to look > at bochs sources to add support for 4MB superpages. This will take its > time though, since I'm not an x86 expert :-). > Me neither. Luckily, I do have a second machine for working with this. Unfortunately, it's another AMD K6-2 so it may suffer from problems we don't forsee yet. But it will probably be helpful data for the L4 hackers. After all, it should work on all Intel-ish processors. Thanks, -Brent

2 1

[Patch] Bochs with 4M superpages runs L4Ka
by Farid Hajji 19 Oct '00

19 Oct '00

Hi, [Kevin: would you please have a closer look at this patch? I'm not an x86 expert and this is at best a premature first try at it. -Thanks] the following patch: http://home.kamp.net/home/farid.hajji/l4ka-bochs/ enables 4 MB superpages in bochs. With this hack, you can boot (among others) L4Ka+root_task without problems on your favorite workstation ;-) (at least I hope so!). Please refer to the file SUPERPAGES in the patched bochs sources for a quick overview. General instructions and screen shots can be found on the above URL. BEWARE: That hack is very alpha and not tested at all! Don't rely on it too much for now. At least, it won't destroy your system, since everything is emulated in software. Bochs may be slow (it is!) but at least it's more secure than plex86 or vmware and it runs on non- intel boxes too. If you find any bugs (e.g. bochs panics()), please try to track them down and fix the sources as far as possible. It is easier than it sounds because bochs sources are very clear and because you can insert printf()s everywhere. Try this with a real processor! Thanks, -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

Re: Task generation & Fiasco scheduler
by Michael Hohmuth 17 Oct '00

17 Oct '00

"kargahi"<kargahi(a)moe.or.ir> writes: > 1. I tried more than a few times to create a 'task' with the l4_task_new > system call but i couldn't. If possible, please send me some examples and > describtion on creating tasks except the L4 manuals. In order to create a new task using l4_task_new, you need to own a task-creation right first. You can get one from Rmgr using the librmgr.a library call rmgr_get_task(). After that, l4_task_new() should work. Also, please don't forget that the new task needs a pager. I'll append a simple example below. > 2. I am looking for the manager of resource CPU( scheduler ). Please guide > me that where is the scheduler and how i can change it with my algorithm. Most of it is implemented in thread_t::schedule(). Regards, Michael -- hohmuth(a)innocent.com, hohmuth(a)sax.de http://www.sax.de/~hohmuth/

1 0

Re: Real-time Fiasco
by Michael Hohmuth 17 Oct '00

17 Oct '00

"kargahi"<kargahi(a)moe.or.ir> writes: > Would you please say me why Fiasco is a real-time u-kernel & what > are the real-time properties of Fiasco? For Fiasco, "Real-Time" means that Fiasco implements static ``hard'' thread priorities, and the worst-case time to react to an event such as an interrupt or a timeout are bound. To keep the bound low, Fiasco is is fully preemtible in kernel mode, and it guarantees that ``independent thread groups'' are always executable no matter what is going on in the kernel. > After that, please say me how the priorities affect the timeslice of a task > and where the timeslice of a task is allocated to it. The scheduler uses two variables: Priorities determine the order in which threads are executed (thread_t::schedule()). The time-slice length determines when the kernel considers running a different thread of the same priority. The kernel has the concept of a "current time slice" which is the number of clock ticks the current thread can execute without being preempted. The current time slice (timeslice_ticks_left) is checked in thread_timer_interrupt(), and it is initialized in thread_t::schedule(). Please not that it is not set on every kernel-context switch -- only when a thread is preempted or voluntarily goes to sleep (that is, only in thread_t::schedule(), not in thread_t::switch_to()). Different L4 kernels handle scheduling differently. Jean Wolter has a short paper describing the current state of the art for some L4 implementations that I'm sure he will make available for the asking. Michael -- hohmuth(a)innocent.com, hohmuth(a)inf.tu-dresden.de http://home.pages.de/~hohmuth/

1 0

Re: L4ka and Bochs (was: L4ka and plex86?)
by Farid Hajji 16 Oct '00

16 Oct '00

Hi Volkmar and Kevin, trying to get l4ka running under bochs would probably require slight modifications to the bochs sources, so that 4 MB pages (bit 4 PSE of CR4) can be supported (please see summary below). > ------------------ BoSnip ------------------- > > // Only allow writes of 0 to CR4 for now. > > // Writes to bits in CR4 should not be 1s as CPUID > > // returns not-supported for all of these features. > ------------------ EoSnip ------------------- > > So what happens is that we initialize 4MB pages (located in the first level > page dirs). > If bochs does not support 4MB pages - you are pretty much out of luck - we > currently do not support > 4K pages only. The kernels virtual view of physical memory is mapped as 4MB > pages. Furthermore, > RMGR and L4Linux assumes 4MB/Super-pages. So if you have absolutely no > possibility to get a normal > Pentium System - we have to do some mid-major :) changes of the startup > code. I don't hope that it would be necessary to tweak the l4ka sources so much (at least not for that 4MB problem). Looking at the sources of bochs, I think that it would be _much_ easier to add support for the PSE bit (bit 4 of CR4) to bochs itself at two places, rather than modifying l4ka. This is what I've found out so far (sorry for lengthy message): ---------- cut here ----------- cut here ------------- cut here -------------- Summary: ======== 1. About superpages, PSE bit etc... - - - - - - - - - - - - - - - - - - Intel Architecture Software Developer's Manual Volume 3: System Programming Order-Nr: 243192 http://developer.intel.com/design/pentiumii/manuals/24319202.pdf Section 3.6 Paging (Virtual Memory): PG-Flag (Bit 31 of CR0) enables paging translation, >= 386 PSE-Flag (Bit 4 of CR4) 4k or 4MB pages, >= 586 PAE-Flag (Bit 5 of CR4) 36Bit physical addr, >= 686 Section 3.6.2.1 Linear Address Translation (4 kByte Pages) Section 3.6.2.2 Linear Address Translation (4 MByte Pages) Section 3.6.2.3 Mixing 4kByte and 4MByte pages Section 3.8.2 Linear Address Translation With Extended Adressing Enabled (2 MByte or 4 MByte Pages). ====> PSE bit is only used while translating linear addresses to physical ====> addresses. paging directory entries and page table entries have slighly ====> different format (?) when 4 MB pages are involved. See figures ====> 3-14, 3-15, 3-18, 3-19, 3-20, 3-21. ====> In bochs, this is mainly located in the file: ====> bochs-2000_0325a/cpu/paging.cc 2. L4ka sources - - - - - - - - 2.1: l4-ka/kernel/src/x86/init.c:init_paging() .............................................. #if defined(CONFIG_ARCH_X86_I586) || defined(CONFIG_ARCH_X86_I686) /* Turn on super pages. */ enable_super_pages(); #endif ====> I configured with CONFIG_ARCH_X86_I586 to avoid having to: ====> #if defined(CONFIG_ARCH_X86_I686) ====> enable_global_pages(); ====> setup_sysenter_msrs(); ====> #endif 2.2: l4-ka/kernel/include/x86/cpu.h:enable_super_pages() ........................................................ INLINE void enable_super_pages() { __asm__ __volatile__ ("mov %%cr4, %%eax\n" "orl $0x10, %%eax\n" "mov %%eax, %%cr4\n" : : : "eax"); } ====> Turning on bit 4 of CR4 (PSE) enables 4 MB pages. ====> Volkmar: In Section 3.6.2.2 (at the End), the note says ====> Volkmar: that the TLBs _must_ be flushed (invalidated) after setting ====> Volkmar: or clearing the PSE bit. Forgot to do it? ====> This is the instruction that crashes bochs. 3. Bochs sources: - - - - - - - - - 3.1 output of bochs.out logfile: ................................ WBINVD: (ignoring) MOV_RdCd: read of CR4 MOV_CdRd: ignoring write to CR4 of 0x00000010 bochs: panic, MOV_CdRd: (CR4) write of 0x00000010 >> mov CR4, EAX ====> Bochs panic()s because it doesn't support CR4 bits yet (see below). ====> l4ka (enable_super_pages()) tries to set the PSE bit in CR4 because ====> it needs 4 MB superpages. 3.2 bochs-2000_0325a/cpu/proc_ctrl.cc: ...................................... This file contains the implementation of the x86-instructions. Let's look at the missing/buggy parts that were reported in bochs.out above: 3.2.1: WBINVD: (ignoring): . . . . . . . . . . . . . void BX_CPU_C::WBINVD(BxInstruction_t *i) { bx_printf("WBINVD: (ignoring)\n"); #if BX_CPU_LEVEL >= 4 invalidate_prefetch_q(); if (BX_CPU_THIS_PTR cr0.pe) { if (CPL!=0) { bx_printf("WBINVD: CPL!=0\n"); exception(BX_GP_EXCEPTION, 0, 0); } } BX_INSTR_CACHE_CNTRL(BX_INSTR_WBINVD); #else UndefinedOpcode(i); #endif } ====> BX_CPU_LEVEL was defined to be 5, so invalidate_prefetch_q() ====> was called. WBINVD (invalidates TLB etc...) seems to be okay. 3.2.2: read of CR4: . . . . . . . . . . void BX_CPU_C::MOV_RdCd(BxInstruction_t *i) { // mov control register data to register <snip> case 4: // CR4 #if BX_CPU_LEVEL == 3 val_32 = 0; bx_printf("MOV_RdCd: read of CR4 causes #UD\n"); UndefinedOpcode(i); #else bx_printf("MOV_RdCd: read of CR4\n"); val_32 = BX_CPU_THIS_PTR cr4; #endif break; <snip> } ====> This seems to be okay once more. The contents of the variable ====> cr4 can be read on this BX_CPU_LEVEL of 5. 3.2.3: ignoring write to CR4 of 0x00000010....., then panic() . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . void BX_CPU_C::MOV_CdRd(BxInstruction_t *i) { // mov general register data to control register <snip> case 4: // CR4 #if BX_CPU_LEVEL == 3 bx_panic("MOV_CdRd: write to CR4 of 0x%08x on 386\n", val_32); UndefinedOpcode(i); #else // Protected mode: #GP(0) if attempt to write a 1 to // any reserved bit of CR4 bx_printf("MOV_CdRd: ignoring write to CR4 of 0x%08x\n", val_32); if (val_32) { bx_panic("MOV_CdRd: (CR4) write of 0x%08x\n", val_32); } // Only allow writes of 0 to CR4 for now. // Writes to bits in CR4 should not be 1s as CPUID // returns not-supported for all of these features. BX_CPU_THIS_PTR cr4 = 0; #endif break; <snip> } ====> This has to be tweaked, perhaps by adding a call to a new ====> CR4_change() function (like CR3_change()), which can ====> then be added to paging.cc. If the change is minimal, ====> modifying variable cr4 directly may be enough. Anyway, ====> in paging.cc, the translation functions must take into ====> account the PSE bit it cr4!!! 3.3 bochs-2000_0325a/cpu/paging.cc .................................. // Translate a linear address to a physical address, for // a data access (D) Bit32u BX_CPU_C::dtranslate_linear(Bit32u laddress, unsigned pl, unsigned rw) { <snip long code which uses cr0, cr3 but not cr4 (yet)> } // Translate a linear address to a physical address, for // an instruction fetch access (I) Bit32u BX_CPU_C::itranslate_linear(Bit32u laddress, unsigned pl) { <snip long code which uses cr0, cr3 but not cr4 (yet)> } ====> these functions will have to additionally take into ====> account the value in cr4 (e.g. PSE bit), so that ====> 4 MB pages are supported. ---------- cut here ----------- cut here ------------- cut here -------------- Thanks, -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.

3 2

Trying L4ka under Bochs/FreeBSD (the saga continues...).
by Farid Hajji 15 Oct '00

15 Oct '00

Hi, [Kevin: please have a look at the bochs panic below: What's wrong with writing non-zero to CR4 when --enable-cpu-level=5 is mentioned? Is there a quick workaround to this problem? Thanks.] I'm trying to run l4ka under bochs but I'm still fighting with some bochs panic()s. This is what I've found out so far: 1. 3 modules _must_ be attached to rmgr, so that the assertion in startup() about the number of modules doesn't bite. 2. I used the following grub startup commands thru all following tests (bochs simulates disk C: from a 112M bytes regular file): grub> root (hd0,0) Filesystem type is ext2fs, partition type 0x83 grub> kernel (hd0,0)/l4ka/rmgr.debug -sigma0 [Multiboot-elf, <0x111000:0x20734:0x0>, <0x132740:0x48c:0xfcc8>, entry=0x111000] grub> module (hd0,0)/l4ka/x86-kernel [Multiboot-module @ 0x143000, 0x12769 bytes] grub> module (hd0,0)/l4ka/sigma0 [Multiboot-module @ 0x156000, 0x287a bytes] grub> module (hd0,0)/l4ka/glinux [Multiboot-module @ 0x159000, 0x143dd4 bytes] grub> boot rmgr.debug is just original rmgr binary, with another name. I tried to slightly modify rmgr in later steps... rmgr[.debug], x86-kernel and sigma0 were compiled on FreeBSD without problems. Note: compiled for 586, not 686!. They are not stripped. I was not able to compile linux22 from Dresden on FreeBSD, so I just used the binary 'glinux' from your distribution. 3. Without any modifications to the l4-ka sources, bochs panic()s right after displaying the message: check_l4_version 0x00100000 to 0x0011041c with the following error message: MOV_RdCd: read of CR4 MOV_CdRd: ignoring write to CR4 of 0x00000010 bochs: panic, MOV_CdRd: (CR4) write of 0x00000010 protected mode CS.d_b = 32 bit SS.d_b = 32 bit | EAX=00000010 EBX=ff000000 ECX=0010f40f EDX=0010a000 | ESP=0010f3f7 EBP=00109000 ESI=f0109000 EDI=00109000 | IOPL=0 NV UP DI PL NZ NA PO NC | SEG selector base limit G D | SEG sltr(index|ti|rpl) base limit G D | DS:0010( 0002| 0| 0) 00000000 000fffff 1 1 | ES:0010( 0002| 0| 0) 00000000 000fffff 1 1 | FS:0010( 0002| 0| 0) 00000000 000fffff 1 1 | GS:0010( 0002| 0| 0) 00000000 000fffff 1 1 | SS:0010( 0002| 0| 0) 00000000 000fffff 1 1 | CS:0008( 0001| 0| 0) 00000000 000fffff 1 1 | EIP=0010fa55 (0010fa52) >> 77875018 0f77875018 2277875018 e077875018 : mov CR4, EAX bochs was not able to grok the instruction: mov CR4, EAX (that is EAX -> CR4). I compiled bochs-2000_0325a with the following options: ./configure --enable-cpu-level=5 \ --enable-disasm --enable-vga --enable-fpu \ --enable-x86-debugger Support for x86-debugger (DR? registers etc...) is therefore enabled. Looking at the source code of bochs-2000_0325a, file cpu/proc_ctrl.cc says in function void BX_CPU_C::MOV_CdRd(BxInstruction_t *i): [...] case 4: // CR4 #if BX_CPU_LEVEL == 3 bx_panic("MOV_CdRd: write to CR4 of 0x%08x on 386\n", val_32); UndefinedOpcode(i); #else // Protected mode: #GP(0) if attempt to write a 1 to // any reserved bit of CR4 bx_printf("MOV_CdRd: ignoring write to CR4 of 0x%08x\n", val_32); if (val_32) { bx_panic("MOV_CdRd: (CR4) write of 0x%08x\n", val_32); } // Only allow writes of 0 to CR4 for now. // Writes to bits in CR4 should not be 1s as CPUID // returns not-supported for all of these features. BX_CPU_THIS_PTR cr4 = 0; #endif break; [...] So bochs doesn't allow writing 0x00000010 (non-zero) to CR4 for now. To Kevin: Why? To Uwe: Can we avoid having to write to CR4 for now? 4. I tried to localize the exact position of that instruction in l4-ka by adding the following MYDEBUG macro to halt execution before bochs panic()s: #ifndef MYDEBUG #define MYDEBUG(str) { printf(str); for(;;) ; } #endif this helped to track down at least portions of the code that were definitely executed by bochs, as well as to see the output. This is what I've found out so far: 4.1. control definitely reaches the end of l4-ka/apps/rmgr/src/startup.c:startup(), at least just before the final: asm volatile ("pushl $" SYMBOL_NAME_STR(exit)"\n" "pushl %2\n" "ret\n" : /* no output */ : "a" (MULTIBOOT_VALID), "b" (&l4_mbi), "r" (exec_info.entry)); /*NORETURN*/ This is the output in bochs: RMGR: loading (hd0,0)/l4ka/sigma0 check_l4_version 0x00100000 to 0x0011041c RMGR: detected L4-Karlsruhe find kernel info page... found kernel info page at 0x00107000 configuring s0: (0x00200000, 0x00213000), start: 0x00200000 s0 config: 0x00200000,0x00000000,0x00200000,0x00213000 RMGR: reserve modules memory range: 156000 to 29cdd4 RMGR: starting (hd0,0)/l4ka/x86-kernel @ 0x0010f000 >>>> rmgr.debug:startup() just before asm volatile... at end of function. 4.2. control doesn't reach even the very beginning of l4-ka/apps/rmgr/src/init.c:init() where I've put the next MYDEBUG call. bochs panic()s with the same error message as before (ignoring write to CR4 of 0x00000010: mov CR4, EAX). Uwe, where does startup() exactly jumps to? I thought it was rmgr's init()??? Thanks, -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

Real-time Fiasco
by kargahi 11 Oct '00

11 Oct '00

Hi L4 Hackers Would you please say me why Fiasco is a real-time u-kernel & what are the real-time properties of Fiasco? After that, please say me how the priorities affect the timeslice of a task and where the timeslice of a task is allocated to it. Thanks.

1 0

Task generation & Fiasco scheduler
by kargahi 18 Sep '00

18 Sep '00

Hi L4 hackers I am trying to write some servers on the L4 MK but i have two problems, please guide me with enough details and simple examples. my problems are here : 1. I tried more than a few times to create a 'task' with the l4_task_new system call but i couldn't. If possible, please send me some examples and describtion on creating tasks except the L4 manuals. 2. I am looking for the manager of resource CPU( scheduler ). Please guide me that where is the scheduler and how i can change it with my algorithm. thanks so much M. KARGAHI

1 0

Compiling L4 on FreeBSD (Part 2)
by Farid Hajji 15 Sep '00

15 Sep '00

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. -- 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.

2 3

CVS Server broken? + compiling L4 on FreeBSD
by Farid Hajji 14 Sep '00

14 Sep '00

Hi Michael, I already asked this on l4-hackers, but I didn't get any replies (yet): -------- begin forwarded message ----------- begin forwarded message -------- 1. I'm missing some files, like e.g. in l4 module the file pkg/log/lib/include/l4/liblog.h (among others). [oskit10 is missing _a lot_ of files too, not only .html's] 2. In l4-hackers archive http://www.mail-archive.com/l4-hackers@os.inf.tu-dresden.de/msg00399.html Michael refers to an additional "log" package. This is fine, but this package (included in l4) doesn't include liblog.h (cvs server says: nothing known about pkg/log/lib/include/l4/liblog.h) --------- end forwarded message ------------ end forwarded message -------- I did a complete "checkout l4" again in a fresh directory today, but the problem remained (the same with cvs update -d). One guess may be that the files are not present in the repository or that they have wrong permissions. Could you please check that the cvs server is running correctly (at least for l4 and oskit10)? I'm trying to compile l4 on a FreeBSD machine. I was able to compile oskit (_not_ oskit10) with gcc-2.7.2.3 and binutils-2.9.1 without problems (gcc-2.95.2 couldn't grok some asm directives because of forbidden use of certain x86 regs) except for the usual warnings. Compiling l4 with gcc-2.95.2 and binutils-2.9.1 went well, up to the point where liblog.h is required. I was not able to proceed any further at this point (liblog.h was not present either in l4-990428.tar.gz). BTW, I had to modify l4/Makeconf slightly, so that the native gcc(-2.95.2) of FreeBSD be used instead of the cross-compiler i486-linux-gcc. This is the modified part of l4/Makeconf: ifdef CROSS CPPFLAGS+= -DCROSS ifeq ($(SYSTEM),Linux) SYSTEM_TARGET= FLOPPY=/dev/fd0 else ifeq ($(SYSTEM),FreeBSD) SYSTEM_TARGET= FLOPPY=/dev/rfd0.1440 else SYSTEM_TARGET=i386-linux- endif endif endif I also tried to compile oskit10, but to no avail. Probably due to the missing files problem reported earlier (I'm not sure though...), even oskit10/configure bombed at a very early stage with the following error message: checking if compiler groks __attribute__((packed)) on structs... yes checking if compiler groks __attribute__((pure))... yes checking if compiler groks __attribute__((noreturn))... yes checking if compiler groks __attribute__((constructor))... yes configure: error: OSKit modules file ./modules.x86.pc was not found. It didn't matter wether I used gcc-2.95.2 or gcc-2.7.2.3: env CC=gcc27 MAKE=gmake ./configure env CC=gcc MAKE=gmake ./configure This is my compile environment: gcc -v Using builtin specs. gcc version 2.95.2 19991024 (release) gcc27 -v Reading specs from /usr/local/lib/gcc-lib/i386-portbld-freebsd5.0/2.7.2.3/specs gcc version 2.7.2.3 as -v GNU assembler version 2.9.1 (i386-unknown-freebsdelf), using BFD version 2.9.1 ld -v GNU ld version 2.9.1 (with BFD 2.9.1) gmake -v GNU Make version 3.79, by Richard Stallman and Roland McGrath. Built for i386--freebsd5.0 uname -rs FreeBSD 5.0-20000506-CURRENT Please help! Thanks, -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

CVS server problems?
by Farid Hajji 12 Sep '00

12 Sep '00

Hi, is the CVS server currently broken? 1. I'm missing some files, like e.g. in l4 module the file pkg/log/lib/include/l4/liblog.h (among others). [oskit10 is missing _a lot_ of files too] 2. In l4-hackers archive http://www.mail-archive.com/l4-hackers@os.inf.tu-dresden.de/msg00399.html Michael refers to an additional "log" package. This is fine, but this package (included in l4) doesn't include liblog.h (cvs server says: nothing known about pkg/log/lib/include/l4/liblog.h) Thanks, -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

Re: How to run Linux 2.2
by Jean Wolter 09 Aug '00

09 Aug '00

Fabio Riccardi <fabio.riccardi(a)free.fr> writes: > Hi, > > I have downloaded the last CVS sources of fiasco, oskit and linux. > > I managed to get fiasco + oskit (old and new version, all example > applications) to run, and I have succesfully recompiled Linux 2.2. I can > even boot Linux, but as soon as it frees the unused kernel memory I get > an infinite sequence of messages like this: > > root: rcvd 4bb5bc, 4bb5bc, src: a.0 > unknown request from a.0 > root: rcvd 4bb5bc, 4bb5bc, src: a.0 > unknown request from a.0 > root: rcvd 4bb5bc, 4bb5bc, src: a.0 > unknown request from a.0 > etc... This is a bug in L4Linux. It took a while to figure out what happens since the bug only happens if you say no to "Use flat clan structure". The following fix will correct this: Index: l4_pager.c =================================================================== RCS file: /home/cvs/linux22/arch/l4/x86/lib/l4_pager.c,v retrieving revision 1.2 retrieving revision 1.4 diff -u -r1.2 -r1.4 --- l4_pager.c 2000/05/12 12:29:48 1.2 +++ l4_pager.c 2000/08/09 12:58:23 1.4 @@ -77,6 +77,7 @@ ping_pong_id = template_id; ping_pong_id.id.task = n; + ping_pong_id.id.lthread = 0; pager.id.lthread = LTHREAD_NO_ROOT_PAGER; Jean -- I get up each morning, gather my wits. Pick up the paper, read the obits. if I'm not there I know I'm not dead. So I eat a good breakfast and go back to bed. Peete Seeger

1 0

How to run Linux 2.2
by Fabio Riccardi 07 Aug '00

07 Aug '00

Hi, I have downloaded the last CVS sources of fiasco, oskit and linux. I managed to get fiasco + oskit (old and new version, all example applications) to run, and I have succesfully recompiled Linux 2.2. I can even boot Linux, but as soon as it frees the unused kernel memory I get an infinite sequence of messages like this: root: rcvd 4bb5bc, 4bb5bc, src: a.0 unknown request from a.0 root: rcvd 4bb5bc, 4bb5bc, src: a.0 unknown request from a.0 root: rcvd 4bb5bc, 4bb5bc, src: a.0 unknown request from a.0 etc... Do I have to configure the rmgr in a special way? I found a configuration file on the diskette image, but it doesn't work either... What am I doing wrong? Thank you very much in advance for your help, ciao, - Fabio

1 0

Sigma0 bug?
by Christian Stueble 25 Jul '00

25 Jul '00

-----BEGIN PGP SIGNED MESSAGE----- Hello again, Last week I had some problems using fiasco on a 196 MB machine. Suzuki told me to modify l4/pkg/rmgr/server/src/memmap.h, but this does not help. (I set MEM_MAX to 256MB and the RMGR told me: "WARNING: All memory above 256 is wasted", on a 196MB machine!!). I noticed that sigma0 also defines MEM_MAX, and after setting it to 256MB, too, it works now. But, isn't it wrong that sigma0 returns memory pages up to RMGR's MEM_MAX value, even if only 196MB are available? 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: wEaiE9zGKl756mb1+UDwYiIm1DkcNdUa iQCVAwUBOX2vsfyJjJy/cQT1AQGyxQQArkMBqWKaIOditHQIuoh+so+7mL0GK3i7 uPYdvQ5Ys6QDw12xhR1zhtPBA1KZcg4DE5CHzqiTY4RfKMx9X5wovIOVu1thujUv Z1OyvqjNTbCwinOoijq+yH53W7hz3aNA+ljbuJ8JghQE5lbFlHjJMkl5aFLwX93u knN6nHmSqPA= =e4UE -----END PGP SIGNATURE-----

1 0

L4 and grub difficulties
by James Riordan 20 Jul '00

20 Jul '00

I have been trying to launch L4 (Fiasco) and the hello server from a version of grub other than that supplied on the grub-ext2fs-floppy.gz image. When I take grub from the gnu server, it compiles but when I boot I get hung at: RMGR: loading (hd1,1)/fiasco/sigma0 RMGR: detected new-style DD-L4 RMGR: starting (hd1,1)/fiasco/main -nokdb proto=0x10128c Welcome to Fiasco! DD-L4/x86 microkernel (c) 1998 TU Dresden - Jul 4 2000 SIGMA0: Hello! RMGR: Hi there! RMGR running on L4/Pentium (hang) When I take grub from the CVS server at Dresen, I fail in compilation: gcc -DHAVE_CONFIG_H -I. -I. -I.. -I../stage1 -Wall -Wmissing-prototypes -Wunused -Os -fno-builtin -nostdinc -DFSYS_EXT2FS=1 -DFSYS_FAT=1 -DFSYS_FFS=1 -DFSYS_MINIX=1 -g -c -o pre_stage2_exec-asm.o asm.S /tmp/cc0xSXW3.s: Assembler messages: /tmp/cc0xSXW3.s:1189: Error: operands given don't match any known 386 instruction /tmp/cc0xSXW3.s:2109: Error: operands given don't match any known 386 instruction this is using the latest versions of gcc and binutils (to the best of my knowledge) montepulciano 502 # gcc -v Reading specs from /usr/local/lib/gcc-lib/i686-pc-linux-gnu/egcs-2.91.66/specs gcc version egcs-2.91.66 19990314 (egcs-1.1.2 release) montepulciano 503 # as --version GNU assembler 2.9.5 Might anybody have any advice on how to proceed? Thanks, james -- Dr. James Riordan email: rij(a)zurich.ibm.com IBM Forschungslaboratorium notes: James Riordan/Zurich/IBM Säumerstrasse 4 Tel +41 (0) 1 724 89 81 CH-8803 Rüschlikon, Switzerland Fax +41 (0) 1 724 89 53

4 9

L4-Linux bug
by Christian Stueble 20 Jul '00

20 Jul '00

-----BEGIN PGP SIGNED MESSAGE----- Hello, the file linux22/kernel/setup.c includes the "../arch/l4-i386/include/fake_interrupt.h" but I think it should include <l4linux/fake_interrupt.h>. Right? 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: uFwTpm7e3g3EJ1l+XzVtymp/RbDe9Usn iQCVAwUBOXQZ2/yJjJy/cQT1AQHkZAQArm3523HYKSy3RbLiBFygHXyiC9PIDN8J O7jHy6edu+xotsW531deas8aWPKSakumU5UeHqGDi8FLNAsvvLj9k1nHivpq0F55 sfGmrJFaLDsWYneinB29qFJYnmWy1YkH2CymVFuzRnMPxN8IWEZwD1hkH9bQdjCo eDc/W9iNelI= =UdXf -----END PGP SIGNATURE-----

2 1

Re: L4Linux hangs on boot
by Jean Wolter 19 Jul '00

19 Jul '00

James Riordan <rij(a)zurich.ibm.com> writes: > Continuing to try to get L4Linux to boot, I run into the problem that > it cannot open an initial console: > > Partition check: > hda: hda1 hda2 hda3 hda4 > hdb: hdb1 hdb2 hdb3 > VFS: Mounted root (ext2 filesystem) readonly. > Freeing unused kernel memory: 36k freed > Warning: unable to open an initial console > > and then goes into an infinite loop of: > > Might anybody have any pointers or suggestions of what to try? Can you send me an image of your boot disk (or if you boot over the network a tar file containing all your binaries)? I would like to try to reproduce the problem here. Jean -- I get up each morning, gather my wits. Pick up the paper, read the obits. if I'm not there I know I'm not dead. So I eat a good breakfast and go back to bed. Peete Seeger

2 1

L4Linux hangs on boot
by James Riordan 19 Jul '00

19 Jul '00

Continuing to try to get L4Linux to boot, I run into the problem that it cannot open an initial console: Partition check: hda: hda1 hda2 hda3 hda4 hdb: hdb1 hdb2 hdb3 VFS: Mounted root (ext2 filesystem) readonly. Freeing unused kernel memory: 36k freed Warning: unable to open an initial console and then goes into an infinite loop of: root: rcvd 4772a0, 4772a0, src: a.0 unknown request from a.0 root: rcvd 4772a0, 4772a0, src: a.0 unknown request from a.0 Actually this is the message of what happens under vmware; when I run natively, it seems to be the same but the messages go by too quickly to read. Might anybody have any pointers or suggestions of what to try? BTW: The build is from the modules "grub l4 linux22 oskit10" from the CVS server on 11 Jul 200. james -- Dr. James Riordan email: rij(a)zurich.ibm.com IBM Forschungslaboratorium notes: James Riordan/Zurich/IBM Säumerstrasse 4 Tel +41 (0) 1 724 89 81 CH-8803 Rüschlikon, Switzerland Fax +41 (0) 1 724 89 53

1 0

RH6.2 and Fiasco
by James Riordan 13 Jul '00

13 Jul '00

I'm still not having a great deal getting Fiasco + hello running. I opted for the slash and burn technique and reinstalled RedHat 6.2, downloaded fresh from the cvs server (10 July 2000) and tried to build from there. The only snag that I hit that I was not sure about was when compiling oskit10, the file oskit10/freebsd/src/lib/libc/i386/gen/alloca.S did not want to compile. There was, however, a comment in oskit mailing list suggesting that a fix ("jmp %edx " changes to "jmp *%edx"). I've appended a script at the end of this mail that make the other small changes that I needed to build. Now that it is built, it crashes at boot time: RMGR: loading (hd1,1)/fiasco/sigma0 RMGR: detected new-style DD-L4 RMGR: starting (hd1,1)/fiasco/main -nokdb proto=0x10128c Welcome to Fiasco! DD-L4/x86 microkernel (c) 1998 TU Dresden - Jul 13 2000 KDB: init L4KD -BP----------------------------------------------EIP: f00011c9 (hang) Any further suggestions? james -- Dr. James Riordan email: rij(a)zurich.ibm.com IBM Forschungslaboratorium notes: James Riordan/Zurich/IBM Säumerstrasse 4 Tel +41 (0) 1 724 89 81 CH-8803 Rüschlikon, Switzerland Fax +41 (0) 1 724 89 53 O / ----------- X Cut Here ----------------------------------------------- O \ #!/bin/bash -nolineediting # # This is a build script for the Fiasco distribution on a generic RedHat 6.2 platform # James Riordan (rij(a)zurich.ibm.com) # # easier to see in color but "alias comment=echo" will also work #function comment () {echo "[01;34m"$*" [0m"} alias comment=echo function backup () { if [ ! -f $1.orig ]; then cp $1{,.orig}; fi } BUILD_ROOT=$PWD export L4_DIR=${BUILD_ROOT}/l4 if [ -d grub -a -d l4 -a -d oskit -a -d oskit10 ]; then comment all directories found else comment you seem to be in the wrong directory exit -1 fi comment making oskit cd oskit ./configure comment edit oskit/flux/config.h to get rid of HAVE_CODE16 if [ ! -f flux/config.h.orig ]; then cp flux/config.h flux/config.h.orig; fi cat << EOF | ed flux/config.h 1 /CODE16 d w q EOF make # making oskit comment making oskit10 cd $BUILD_ROOT/oskit10 ./configure comment edit oskit/config.h to get rid of HAVE_CODE16 backup oskit/config.h cat << EOF | ed oskit/config.h 1 /CODE16 d w q EOF comment edit Makeconf to get rid of have_code16 backup Makeconf cat << EOF | ed Makeconf 1 /code16 s/have_code16// w q EOF comment removing in2000 from oskit/dev/linux_scsi.h backup oskit/dev/linux_scsi.h cat << EOF | ed oskit/dev/linux_scsi.h 1 /in2000 d w q EOF comment changing freebsd/src/lib/libc/i386/gen/alloca.S backup freebsd/src/lib/libc/i386/gen/alloca.S cat << EOF | ed freebsd/src/lib/libc/i386/gen/alloca.S 1 /jmp s/%/*%/ w q EOF make exit comment making fiasco cd $BUILD_ROOT/l4 comment killing pkg/names/man/Makefile as we do not have sdc backup pkg/names/man/Makefile echo all: > pkg/names/man/Makefile comment killing pkg/l4util/examples/perf/Makefile comment as we do not need it and it does not work backup pkg/l4util/examples/perf/Makefile echo all: > pkg/l4util/examples/perf/Makefile comment change egcc to gcc in kernel/fiasco/src/Makefile backup kernel/fiasco/src/Makefile cat << EOF | ed kernel/fiasco/src/Makefile 1 /egcc /CC s/egcc/gcc w q EOF comment create dependency files I am not sure how they are supposed comment to be created but this seems to work cat > Makedeps <<EOF ifeq (\${L4_DIR},) L4DIR = ../../.. else L4DIR = \${L4_DIR} endif ifeq (\${OSKIT_DIR},) OSKITDIR = \$(L4DIR)/../oskit else OSKITDIR = \${OSKIT_DIR} endif include \$(L4DIR)/Makeconf EOF allow_null_glob_expansion=t pushd kernel/fiasco/src/ /bin/rm .*.d for file in *.cc *.c *.S *.cpp; do CC=gcc make -f ${L4_DIR}/Makedeps .$file.d; done popd make pushd pkg/log make popd make

2 1

Re: L4 and grub difficulties
by Lars Reuther 12 Jul '00

12 Jul '00

Christian Stueble <stueble(a)cs.uni-sb.de> writes: > Another question: > I think oskit10 is required now to compile the pkg package, but some > subdirectories also use oskit/flux/c/*.h include files. Do I really need both > oskit versions or should the makefiles be modified to use oskit10/oskit/c/*.h? Which packages? All packages available in our remote CVS should still use the old OSKit (version 0.6), except the Fiasco kernel. We are currently working on a support package for the latest OSKit, but aren't finished yet. Lars -- Dipl.-Inf. Lars Reuther Dresden University of Technology Department of Computer Science Operating Systems Research Group Phone: +49 (351) 463-8401 Fax: +49 (351) 463-8284 Email: reuther(a)os.inf.tu-dresden.de

1 0

Re: Problems compiling Fiasco (OSKIT)
by Pete Buechler 07 Jul '00

07 Jul '00

On Fri, 07 Jul 2000, Christian Stueble wrote: > I had the same problems. Use an older compiler. The > gcc-2.7.2.3 works for me. To compile Fiasco, > use e.g. egcs-2.91.60. I used egcs-2.91.66 for Fiasco and gcc-2.7.2.3 for oskit. Then I had to follow the directions by Edmund GRIMLEY EVANS <edmundo(a)rano.org> - remove HAVE_CODE16 from oskit/config.h - remove have_code16 from Makeconf - remove in2000 from oskit/dev/linux_scsi.h Then I had all kinds of path problems in the Makefiles, having to search for various paths and add them to the include files, plus moving header files around. And when the dust all settled, Fiasco came up and printed some stuff, but the hello server never managed to print hello to my screen. -- Pete Buechler : SuSE Linux Developer Work e-mail : peterb(a)suse.com Work web page : http://www.suse.com/~peterb Personal e-mail : peter.buechler(a)home.com

1 0

Adapting L4 to automake
by Juho Östman 09 Jun '00

09 Jun '00

I think it would be useful to adapt L4 building system to GNU automake and autoconf as I have written earlier. I have now done preliminary modification to the source tree and I'd like to know how should I continue. I already have a source tree which builds the Fiasco kernel, rmgr, sigma0 and hello server using automake generated Makefiles. There is also a lot of need of finishing up. For example it doesn't create distribution packages completely using 'make dist'. The source tree package was so large that I didn't want to post it to this mailing list but I could send it anyone who is willing to look at it.

2 1

build problems
by Edmund GRIMLEY EVANS 23 May '00

23 May '00

I've just spent some time trying to build Fiasco from the CVS source. (1) It would be nice if the build instructions and FAQ could be put with the source in CVS for the benefit of people who have to dial up. (2) The build instructions don't tell you to build oskit10 first. (3) I found I needed gcc-2.7.2.3 to build oskit10, and I had to make the following changes to make it compile: - remove HAVE_CODE16 from oskit/config.h - remove have_code16 from Makeconf - remove in2000 from oskit/dev/linux_scsi.h Which assember are other people using to build this? I'm using: GNU assembler version 2.9.5 (i386-linux) using BFD version 2.9.5.0.37 (4) I used egcs-1.1.1 to build l4, and had a series of makefile problems which might be very easy for someone else to fix. A complete log of the errors I got and how I worked around them is given below. (5) The build instructions then tell you to cd to kernel/fiasco/src and make again, but this was already done by make in l4/. (I'll tell you about my run-time problems when I try it ...) Edmund make[4]: *** No rule to make target `../../../../include/l4/log/l4log.h', needed by `.p6.c.d'. Stop. make[4]: Leaving directory `/home/ege/aferoj/l4-cvs/l4/pkg/l4util/examples/perf' $ mkdir include/l4/log ; cp -i pkg/log/include/l4/log/l4log.h include/l4/log ld -N -Ttext 0x430000 -o perf_p6 p6.o crt0.o -L../../../../lib -L../../../../../oskit/lib -L/home/drops/lib -llog -ll4util -lpci -lmc -lnames -loskit_support -lkern -lmc -lrmgr -llmm ld: cannot find -llog make[4]: *** [perf_p6] Error 1 make[4]: Leaving directory `/home/ege/aferoj/l4-cvs/l4/pkg/l4util/examples/perf' $ ( cd lib ; ln -s ../pkg/log/lib/src/liblogserver.a ../pkg/log/lib/src/liblog.a .) ld -N -Ttext 0x430000 -o perf_p6 p6.o crt0.o -L../../../../lib -L../../../../../oskit/lib -L/home/drops/lib -llog -ll4util -lpci -lmc -lnames -loskit_support -lkern -lmc -lrmgr -llmm ld: cannot find -lpci make[4]: *** [perf_p6] Error 1 make[4]: Leaving directory `/home/ege/aferoj/l4-cvs/l4/pkg/l4util/examples/perf' $ touch pkg/l4util/examples/perf/perf_p6 # It finishes building now, but we don't have perf_p6, so: $ rm pkg/l4util/examples/perf/perf_p6 ld -N -Ttext 0x430000 -o perf_p6 p6.o crt0.o -L../../../../lib -L../../../../../oskit/lib -L/home/drops/lib -llog -ll4util -lpci -lmc -lnames -loskit_support -lkern -lmc -lrmgr -llmm ld: cannot find -lpci make[4]: *** [perf_p6] Error 1 make[4]: Leaving directory `/home/ege/aferoj/l4-cvs/l4/pkg/l4util/examples/perf' # cd to that directory and do the link without -lpci # now everything is built, I think

2 2

building linux22
by Edmund GRIMLEY EVANS 22 May '00

22 May '00

And could someone please tell me what symbolic links I should create to build linux22? I assume that's what missing after I did cvs update -d. In particular, what should linux22/include/asm point at, and where should linux22/include/asm-l4/param.h find <asm/l4arch/param.h>? Edmund

2 1