Hi, is Fiasco-UX suited for debugging with GDB? I noticed the lines generic.inc: [ -n "$RUN_GDB" ] && gdb_wrap="gdb $GDB_ARGS --args" generic.inc: $gdb_wrap $FIASCODIR/$FIASCO in l4/tools/runux/generic.inc and tried to run GDB by setting the RUN_GDB variable to something when calling one of the predefined ux-scripts, but it always results in segmentation faults concerning the calls to the cli and sti instructions.
GDB output: Welcome to Fiasco(ux)! DD-L4(v2)/x86 microkernel (C) 1998-2006 TU Dresden Rev: Thu Aug 3 18:16:37 2006 compiled with gcc 4.0.2 for Intel Pentium 4
Detaching after fork from child process 5989.
Program received signal SIGSEGV, Segmentation fault. Dirq::alloc (this=0xce93f, t=0x639f5000, ack_in_kernel=802816) at auto/processor.h:128 128 asm volatile ("cli" : : : "memory"); (gdb)
$ gdb --version GNU gdb Red Hat Linux (6.3.0.0-1.84rh)
Regards, Andi
On Mon, 07 Aug 2006 19:16:06 +0200 Andreas Niederl (AN) wrote:
AN> is Fiasco-UX suited for debugging with GDB?
Yes
AN> [...], but it always results in segmentation faults concerning the AN> calls to the cli and sti instructions.
You must tell GDB to ignore SIGSEGV signals and pass them on to Fiasco-UX. The following at the GDB prompt will do:
handle 11 nostop noprint
Of course you could also do this from .gdbinit. Note that when using Fiasco-UX with GDB you'll lose the ability to use Fiasco's built-in debugger JDB as SIGTRAP is used by GDB.
- Udo
Hi,
Udo A. Steinberg wrote:
On Mon, 07 Aug 2006 19:16:06 +0200 Andreas Niederl (AN) wrote:
[...]
AN> [...], but it always results in segmentation faults concerning the AN> calls to the cli and sti instructions.
You must tell GDB to ignore SIGSEGV signals and pass them on to Fiasco-UX. The following at the GDB prompt will do:
handle 11 nostop noprint
Thanks for the information.
Regards, Andi
l4-hackers@os.inf.tu-dresden.de