- l4-hackers - example.com Site

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

Re: Building the environment fails
by Alexander Boettcher 05 Apr '06

05 Apr '06

1 0

Re: Building the environment fails
by Alexander Boettcher 05 Apr '06

05 Apr '06

1 0

Building the environment fails
by Stefan Schulz 05 Apr '06

05 Apr '06

Sorry for using this ML as my personal helpdesk, but i believe there is a problem in the build process for l4env. I checked the current version out from CVS, and used "make config" to configure. While "make" was running, the process failed with ... Compiling ARCH-all/libc/misc/internals/__uClibc_main.o LD_PRELOAD=/l4/tmp/l4/tool/gendep/libgendep.so GENDEP_TARGET=ARCH-all/libc/misc/internals/__uClibc_main.o GENDEP_BINARY=cc1 gcc -c -nostdinc -include ../../../lib/uclibc/ARCH-all/include/libc-symbols.h -DNDEBUG -D_LIBC -DSTATIC -DSYSTEM_x86_586 -DARCH_x86 -DCPUTYPE_586 -DL4API_ -DCONFIG_L4_CALL_SYSCALLS -DCONFIG_L4_ABS_SYSCALLS -I../ARCH-x86/include -I../ARCH-all/include -I../../../../../include/x86 -I../../../../../include -I/usr/lib/gcc/i686-pc-linux-gnu/3.4.5/include -fno-builtin -gstabs+ -g -O2 -fno-strict-aliasing -march=i586 -Wall -Wstrict-prototypes -fno-common /l4/tmp/l4/pkg/uclibc/lib/uclibc/ARCH-all/libc/misc/internals/__uClibc_main.c -o ARCH-all/libc/misc/internals/__uClibc_main.o /l4/tmp/l4/pkg/uclibc/lib/uclibc/ARCH-all/libc/misc/internals/__uClibc_main.c:2:49: l4/sys/consts.h: No such file or directory /l4/tmp/l4/pkg/uclibc/lib/uclibc/ARCH-all/libc/misc/internals/__uClibc_main.c:15: error: `L4_PAGESIZE' undeclared here (not in a function) make[4]: *** [ARCH-all/libc/misc/internals/__uClibc_main.o] Error 1 make[4]: Leaving directory `/l4/tmp/l4/pkg/uclibc/lib/uclibc/OBJ-x86_586' make[3]: *** [OBJ-x86_586] Error 2 make[3]: Leaving directory `/l4/tmp/l4/pkg/uclibc/lib/uclibc' make[2]: *** [virt_dep] Error 2 make[2]: Leaving directory `/l4/tmp/l4/pkg/uclibc' make[1]: *** [uclibc.headers] Error 2 make[1]: Leaving directory `/l4/tmp/l4/pkg' make: *** [pkg] Error 2 Even if i "fixed" this error (by manually defining L4_PAGESIZE", similar problems crept up all over the build, far too many to manually fix. I tried this build on an IBM T43, using current sources, gcc3.4.5 and make 3.80 Is the build process really broken, or am i doing something wrong?

1 0

Make hello stops with an error
by Ricard de Benito Panillo 05 Apr '06

05 Apr '06

Hi, the following error arises making 'hello': cc1: note: obsolete option -I- used, please use -iquote instead In file included from ../libkern/x86/base_cpu_init.c:22: ../flux/x86/base_trap.h:105: error: array type has incomplete element type make[3]: *** [base_cpu_init.o] Error 1 gcc version 4.0.2 20051125 (Red Hat 4.0.2-8) on Fedora Core 4 Any hint? it seems that the compiler versions is newer than expected. Best regards, Ricard de Benito

2 1