l4-hackers April 2006

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

15 participants
16 discussions

Getting Started...
by Colin Burgess 25 Apr '06

25 Apr '06

Hi All, I tried following the build instructions for building L4-Linux 2.6 but to no avail. I am running Suse Linux 2.6.13-15.8-default I checked out l4linux-2.6-env, and typed make config, and selected a DROPS_DIR of /home/cburgess/drops I then typed make and it eventually failed to find l4/sys/types.h trying to build log-server Are there other modules I need to checkout? Do I also need the l4linux-2.6 module, and should it be in the same toplevel dir as the l4linux-2.6-env module? Thanks, Colin

3 4

About l4.sec schedule
by Wei Shen 24 Apr '06

24 Apr '06

Hi, Is there any schedule for the first release of l4.sec kernel? Maybe, in several months? Or before the end of this year? How do you consider the compatibility of l4.sec and l4env? Thanks, Wei

1 0

fiasco-ux does not build
by Stefan Scheler 21 Apr '06

21 Apr '06

Hello, I get the following error building fiasco-ux: ==> Linking kernel.image /usr/lib/gcc-lib/i686-pc-linux-gnu/3.3.6/../../../../i686-pc-linux-gnu/bin/ld: errno: TLS definition in /usr/lib/gcc-lib/i686-pc-linux-gnu/3.3.6/../../../libc.a(errno.o) section .tbss mismatches non-TLS reference in boot_info-ux.o /usr/lib/gcc-lib/i686-pc-linux-gnu/3.3.6/../../../libc.a: could not read symbols: Bad value collect2: ld returned 1 exit status make[2]: *** [kernel.image] Error 1 make[2]: Leaving directory `/home/sts/hellomodule/l4/kernel/fiasco/build' make[1]: *** [all] Error 2 make[1]: Leaving directory `/home/sts/hellomodule/l4/kernel/fiasco/build' make: *** [all] Error 2 Real: 91.12s User: 74.26s System: 11.10s Percent: 93% Cmd: make So is this an gcc, glibc or binutils issue in the first place? And which versions are supposed to be able to build fiasco-ux? I'm using binutils 2.16.92 and glibc 2.4 and various gcc versions btw. Thanks in advance Stefan

2 5

L4Linux booting failed
by chenggh04＠st.lzu.edu.cn 18 Apr '06

18 Apr '06

Hi: I compile the L4Linux with plain mode! But when I boot L4Linux it is wrong. The information is as following: RNGR: Stage1 loading (hd0,1)/boot/fiasco_L4/sigma0 loading kernel (hd0,1)/boot/fiasco_L4/main -nowait -nokdb -serial_esc -compseed The module 00:00300000-00390408: (hd0,1)/boot/fiasco_L4/main -nowait -nokdb -serial overlaps with module loaded by bootloader at 00100000-001caec0: RMGR 00:0020f000-002661ac: (hd0,1)/boot/fiasco_L4/main -nowait 01:00267000-00275925: (hd0,1)/boot/fiasc_L4/sigma0 ==>02:00276000-004d8950: (hd0,1)/boot/fiasco_L4/vmlinuz.V2 no-scroll no 03:00409000-014d9000:(hd0,1)/boot/fiasco_L4/drops-rd.rd The reason for this error message is that the modules as loaded by GRUB Interface with their final position as defined by the linker. Most often you can solve this problem by inserting the command "modaddr 0x02000000" immediately as next line after the "kernel=" line of you menu.lst. Note that you have to use a special version of GRUB which is avaiable from http://os.inf.tu-dresden.de/drops/download.html The grub about L4Linux is : title fiasco_L4Linux root (hd0,1) kernel (hd0,1)/boot/fiasco_L4/rmgr -sigma0 task modname "vmlinuz" module modaddr 0x02000000 module (hd0,1)/boot/fiasco_L4/main -nowait -nokdb -serial_esc -comspee d 115200 -comport 1 module (hd0,1)/boot/fiasco_L4/sigma0 module (hd0,1)/boot/fiasco_L4/vmlinuz.V2 no-scroll no-hlt l4irqack=lin ux load_ramdisk=1 ramdisk_size=16384 root=/dev/ram module (hd0,1)/boot/fiasco_L4/drops-rd.rd I don't know what is wrong! Please help me! Thank you!

3 2

Strange L4Linux behavior
by Jugwan Eom 14 Apr '06

14 Apr '06

Hi, L4Linux fails to run correctly with the SCSI and network device under heavy load. My system has a Adaptec AIC7902 SCSI adapter (one SCSI hard disk is connected) and a E1000 network adapter. When the two devices is used at the same time under heavy load (some benchmarks to SCSI disk and transferring files using ftp), SCSI adapter driver prints error message and dumps device's status. and then system is not work. Actually, this error doesn't happen immediately after using the two devices. It works during some seconds (about 10~20secs). I could not capture all the message by SCSI adapter driver printed on console because it is very fast and vast. Anyway, the message means "PCI interrupt error". In L4Linux, the two devices share same IRQ number. IMHO, this is one of the source of this behavior. To use SCSI in L4Linux, I use dm_phys with '--isa=0x00800000' option. Two different IRQ handling methods (one thread per IRQ and multiple IRQs) got the same result. Is there anyone can help me? Regards, -- Jugwan Eom

2 3

Fwd: attempted fix for asm in oskit cpuid.c
by spooq 12 Apr '06

12 Apr '06

*sigh* ---------- Forwarded message ---------- From: spooq <spoooq(a)gmail.com> Date: Apr 11, 2006 11:12 PM Subject: Re: attempted fix for asm in oskit cpuid.c To: Ronald Aigner <ra3(a)os.inf.tu-dresden.de> I see dietlibc's printf is not given a good review on http://www.and.org/vstr/printf_comparison... On 4/11/06, spooq <spoooq(a)gmail.com> wrote: > Ok, I'll start using one of those. Is there any real difference between the two? > > Id also like to know what would need to be done to port L4Env to a new > arm-based arch, eg video card driver (not supported by existing linux > driver), touchscreen, mmu. Is there a list somewhere of interfaces > that need to be implemented? > > btw, just spooq is fine > > -- > spooq > > On 4/11/06, Ronald Aigner <ra3(a)os.inf.tu-dresden.de> wrote: > > -----BEGIN PGP SIGNED MESSAGE----- > > Hash: SHA1 > > > > Hello spoo(o)q, > > > > spooq wrote on 10.04.2006 19:36 this: > > > I tried to fix some errors in cpuid.c ... wouldnt rely on it, my asms > > > not very good and I havent tested it beyond seeing that it compiles. > > > Basically it just makes the asm respect gcc using the ebx as a pic > > > reg. Not sure about the \n\t stuff, it made it work tho. > > Nice job. However, oskit is no longer supported as a libc for L4Env > > applications. Please use either dietlibc or uclibc. > > > > Thanks, keep it up. > > Ron. > > - -- > > Mit freundlichen Gruessen / with regards > > ra3 @ inf.tu-dresden.de > > http://os.inf.tu-dresden.de/~ra3/ > > -----BEGIN PGP SIGNATURE----- > > Version: GnuPG v1.4.2.2 (GNU/Linux) > > > > iD8DBQFEPCGivCdOf9l7ipgRAv9RAKCEMCFLEx1o0ehBLG+f5J+6I9/dIgCgwjdV > > SzC8KfKBWcPvexF8A8vzadw= > > =1wJw > > -----END PGP SIGNATURE----- > > >

3 2

Re: Fwd: attempted fix for asm in oskit cpuid.c
by spooq 12 Apr '06

12 Apr '06

I think this list needs to be fixing its reply-tos... I wasnt the only one caught by it... On 4/12/06, Steffen Liebergeld <s1010824(a)inf.tu-dresden.de> wrote: > On Tue, 11 Apr 2006 23:14:08 +0100, spooq <spoooq(a)gmail.com> said: > > *sigh* ---------- Forwarded message ---------- From: spooq > > <spoooq(a)gmail.com> Date: Apr 11, 2006 11:12 PM Subject: Re: > > attempted fix for asm in oskit cpuid.c To: Ronald Aigner > > <ra3(a)os.inf.tu-dresden.de> > > > > I see dietlibc's printf is not given a good review on > > http://www.and.org/vstr/printf_comparison... > > That comparison is very old, and they tested version 0.22 of > dietlibc. The current version is 0.29 (have a look at > http://www.fefe.de/dietlibc/). > > -- > Microsoft is simply one example of a proprietary software developer, a > software developer that tries to subjugate users to keep them divided > and helpless. -- Richard M. Stallman >

2 1

attempted fix for asm in oskit cpuid.c
by spooq 11 Apr '06

11 Apr '06

Hello list, I tried to fix some errors in cpuid.c ... wouldnt rely on it, my asms not very good and I havent tested it beyond seeing that it compiles. Basically it just makes the asm respect gcc using the ebx as a pic reg. Not sure about the \n\t stuff, it made it work tho. --- virgin_code/oskit/libkern/x86/cpuid.c 2006-04-02 15:41:55.000000000 +0000 +++ cpuid.c 2006-04-10 17:14:57.000000000 +0000 @@ -39,12 +39,17 @@ unsigned i; cicount++; - asm volatile("cpuid" + asm volatile( + "movl %%ebx, %%edi\n\t" + "cpuid\n\t" + "movl %1, %%ebx\n\t" + "movl %%edi, %%ebx\n\t" : "=a" (ci[0]), - "=b" (ci[1]), + "=r" (ci[1]), "=c" (ci[2]), "=d" (ci[3]) - : "a" (2)); + : "a" (2) + : "edi"); for (i = 0; i < 4; i++) { @@ -90,21 +95,29 @@ * CPUID is supported, so use it. * First get the vendor ID string. */ - asm volatile("cpuid" + asm volatile( + "movl %%ebx, %%edi\n\t" + "cpuid\n\t" + "movl %1, %%ebx\n\t" + "movl %%edi, %%ebx\n\t" : "=a" (highest_val), - "=b" (*((int*)(out_id->vendor_id+0))), + "=r" (*((int*)(out_id->vendor_id+0))), "=d" (*((int*)(out_id->vendor_id+4))), "=c" (*((int*)(out_id->vendor_id+8))) - : "a" (0)); + : "a" (0) + : "edi"); /* Now the feature information. */ if (highest_val >= 1) { - asm volatile("cpuid" - : "=a" (*((int*)out_id)), - "=d" (out_id->feature_flags) - : "a" (1) - : "ebx", "ecx"); + asm volatile( + "movl %%ebx, %%edi\n\t" + "cpuid\n\t" + "movl %%edi, %%ebx\n\t" + : "=a" (*((int*)out_id)), + "=d" (out_id->feature_flags) + : "a" (1) + : "ecx","edi"); } /* Cache and TLB information. */ -- spooq

2 1

Static Linux kernel, anyone?
by Stefan Schulz 09 Apr '06

09 Apr '06

Hi there I am currently trying to build a static linux kernel (one usable without libloader). I have read the earlier thread in this mailing list, but have run into some problems. I tried adding libraries to fix undefined references, since merely adding the static dependencies of libloader.s.so did not work. Finally, i hit upon references such as __l4sys_abs_ipc_fixup, which frankly looks like something i should not have to care about. ;) A quick scan through the libraries also seemed to indicate that it isn't even exported by any library. So, what am i doing so fundamentally wrong? Is there a way to build a static linux binary, or is it just intrinsically tied to the loader?

2 1

L4Linux and AMD64
by Zephiris 09 Apr '06

09 Apr '06

I haven't been looking around at things very long, but I was noticing, while Fiasco largely supports AMD64 processors in 64-bit mode, L4Linux appears to specifically avoid it. I couldn't find any references for this, will it ever be supported? It's something I was excited about/meaning to try, but as my system is setup with 64-bit binaries in mind, it looks like the only way to test it would be for L4Linux to support that.

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 April 2006