asmutils libc fun and games

Hello fellow asm heads,
I've discovered the joys of the asmutils boot disk and now want to make it possible to include statically linked code written in C and the asmutils libc library on the distro. But I can't get the static versions working. Here's the problem:

In the Makefile (in asmutils-0.16/lib/) I have:

BUILD_STATIC = y
BUILD_SHARED = y

But when building the static test proggie I get:

gcc -D__CPU__=386 -Wall -O1 -m386 -g -D__NO_STRING_INLINES -fno-builtin -nostartfiles -nostdlib -static -o test1-static test1.c libc_stub.o libc.a libc.a(libc.o): In function `_start':
libc.o(.text+0x4): multiple definition of `_start'
libc_stub.o(.text+0x0): first defined here
collect2: ld returned 1 exit status
make: *** [test1] Error 1

If I comment out STUB_OBJS and remake the static portion, then some of the static versions are compiled but output the __libc_banner file which reads:
A r e y ou s i c k ?

To which I reply yes! :) I would like a tiny libc written in asmutils that links statically so as to include some demo-style progs on the bootdisk. Am Imissing the point of this libc? Even the shared libraries that build correctly are dynamically linked to libNoVersion and GNU libc when you ldd them. What's up with that?

Paul.

I made a similar experience, -nostdlib &c just useless. found a workaround having "libc.a" first in the search path, which apparently keeps the GNU-ish stuff happy but, doesn't forcedly include such nonsense.

regards,
hp

=====
compiled, tests passed:

patched version 98.03 (with %xdefine and un-deleted listing output)

(always, as usual...)

both! ^^^^^^^^^^^^^^^^^^^^^^^^

Linux,Assembly,Forth: http://www.lxhp.in-berlin.de/index-lx.shtml en/de
NO abusive software patents http://petition.eurolinux.org/pr/pr17.html
-
To unsubscribe from this list: send the line "unsubscribe linux-assembly" in
the body of a message to [email protected]
More majordomo info at  http://vger.kernel.org/majordomo-info.html

hi...

found some funny differences w Linux 2.4.18 kernel wrt its predecessors where, for instance, the 'umount' system call doesn't work on devices and mountpoints, any more, but, only on the device names.

i.e, some systemcalls' names were silently changed, previously valid names assigned to different(ly operating) new/old syscalls.

I found this when trying the "eject" program from asmutils which was non functional w. 2.4.18. The patch cures the problem:
------
; patch eject.asm for Lx 2.4.18 sys_umount
.eject:
push ecx ; mountpoint
push ebx ; specialfile
%ifdef ONLY_CCD
%else
test byte[edi],-1
jnz .opn ;?; not -ccd, just the syscall
%endif; <= ONLY_CCD
%if __KERNEL__ < 24; new inconsistency in kernel 2.4.18! <== !
sys_umount ; eject won't work on a mounted device
%else
sys_umount dword[esp+4] ; 'oldumount' w/ #mountpoint#, only mov ebx,[esp]
%endif
.opn:
------

Further, the macro "_xchg" in asmutils' 'system.inc' is not correct (which I didn't recognize earlier because I'm using my own version;):

1) ---
%macro _xchg 2
-%if __OPTIMIZE__=__O_SPEED__
+%if __OPTIMIZE__=__O_SIZE__;
%assign ea 4
2) ---
@@ -721,7 +723,7 @@
; 2 clocks w. reg,reg
; (lea variant wasn't faster, may be cpu dependent?)
; 33 clocks w. reg,mem or mem,reg - either case
- xchg %{1},%{3}
+ xchg %{1},%{2};
-----
btw. the attempts to finding out about the actual addressing mode seem useless, "nasm" apparently providing no such means, at all.

regards,
hp

Linux,Assembly,Forth: http://www.lxhp.in-berlin.de/index-lx.shtml en/de
NO abusive software patents http://petition.eurolinux.org/pr/pr17.html
-
To unsubscribe from this list: send the line "unsubscribe linux-assembly" in
the body of a message to [email protected]
More majordomo info at  http://vger.kernel.org/majordomo-info.html

Hi guys,

I have been out for too long and now i see that there is no activity on this list. Is it like dead or something?

- halfdead

nope, still alive...
working hard!
(for info: if you want to boot a kernel from an already running kernel, you can use 'monte' project. Very good! (for x86 only))

Fred

Hey,
Its been rumoured for a while that certain vulnerabilities reside at kernel level in different unix flavours (solaris/bsd/linux). Some of them claim to be remotely exploitable (ABfrags.c, ABremkrnl.c, etc.). I have seen local kernel vulnerabilities and they seem to be pretty tricky to exploit hence i highly doubt that anyone would be able to work out a clean remote exploitation. Is any of you guys aware of any method or is this just FUD?

- halfdead

Wouldn't a venue such as comp.os.linux.security be a more appropriate venue for such a question?

Or are you trolling for help exploiting such vulnerabilities :)

Robin Miyagi<[email protected]>
http://www.geocities.com/SiliconValley/Ridge/2544/asm/assembler.html
The funny thing about brakes, is that when they break, they don't brake.

nice to know that you still seem just half of what this list is not. all questions answered, nothing new in the kernel which none understands.

best,
hp

Linux,Assembly,Forth: http://www.lxhp.in-berlin.de/index-lx.shtml en/de