Bernhard Walle | 5 Jun 11:35 2004
Picon
Picon

Build gdb for multiple targets

Hello,

I could not find any hints in the documentation for this question: 
Is it possible to build a gdb (i.e. one binary) which can handle more
than one target CPUs (in this case: PowerPC and M68000).

If yes: What are the options for ./configure?

Regards,
Bernhard

--

-- 
Ben Elliston | 5 Jun 12:23 2004
Picon

removing src/dejagnu, src/expect

At yesterday's GCC Summit testing BOF, there was a strong concensus
that these directories should be removed from the src tree and that
developers should install the latest releases of Expect and Dejagnu
from their distribution/packaging system or from a source tarball at
ftp.gnu.org.

I'm declaring that I intend to do this and no sooner than Friday June
11 at 00:00 UTC.  I will be ensuring that no local patches, if any,
are lost in this process.  If you have any reasonable objection,
please say so now .. :-)

Cheers, Ben

Michael Elizabeth Chastain | 5 Jun 16:39 2004
Picon

Re: [Gdb] removing src/dejagnu, src/expect

> At yesterday's GCC Summit testing BOF, there was a strong concensus
> that these directories should be removed from the src tree and that
> developers should install the latest releases of Expect and Dejagnu
> from their distribution/packaging system or from a source tarball at
> ftp.gnu.org.

Err, I'm kinda surprised that the "GCC Summit" has enough src/
people to make decisions about src/.  Are Andrew Cagney (gdb)
and Daniel Jacobowitz (binutils) there?

Nevertheless, I'm all for killing these packages in src/!
src/expect dates back to 1998, that's just wrong.

I test gdb with the stock released versions: tcl 8.4.6 + expect 5.41.0 +
dejagnu 1.4.4.  Here's my experience by host platform:

  i686-pc-linux-gnu
    run gdb test suite regularly.  occasionally, i run with both
    sourceware tcl+expect+dejagnu and stock tcl+expect+dejagnu
    and compare the results.  the results always match.

  hppa2.0w-hp-hpux11.11
    run gdb test suite occasionally.  works fine.

  alphaev68-dec-osf5.1
  i386-unknown-freebsd4.9
  ia64-hp-hpux11.22
  ia64-hp-hpux11.23
    built tcl+expect+dejagnu, haven't run them.

(Continue reading)

Albert Antony | 7 Jun 16:06 2004
Picon

symbol referencing errors while building cross gdb

Hi,

I am trying to build a sparc - sun solaris hosted cross gdb for i960-coff
target. I am getting the following two error messages:

Making all in po
gcc -DHAVE_CONFIG_H      -DWITH_DEFAULT_MODEL='"i960KA"'    -DPROFILE=1
-DWITH_PROFILE=-1      -DWITH_ALIGNMENT=NONSTRICT_ALIGNMENT   
-DWITH_TARGET_BYTE_ORDER=LITTLE_ENDIAN  -DWITH_ENVIRONMENT=ALL_ENVIRONMENT
     -DWITH_HOST_BYTE_ORDER=BIG_ENDIAN  -DDEFAULT_INLINE=0       
-DWITH_SCACHE=1024                  -I.
-I/usr/gnu/srcdir/gdb-5.2.1/sim/i960 -I../common
-I/usr/gnu/srcdir/gdb-5.2.1/sim/i960/../common  -I../../include
-I/usr/gnu/srcdir/gdb-5.2.1/sim/i960/../../include  -I../../bfd
-I/usr/gnu/srcdir/gdb-5.2.1/sim/i960/../../bfd  -I../../opcodes
-I/usr/gnu/srcdir/gdb-5.2.1/sim/i960/../../opcodes  -I../../intl
-I/usr/gnu/srcdir/gdb-5.2.1/sim/i960/../../intl -g -O2 -o run \
  nrun.o libsim.a ../../bfd/libbfd.a ../../opcodes/libopcodes.a 
../../libiberty/libiberty.a  -lnsl -lsocket
Undefined                       first referenced
 symbol                             in file
sim_parse_args                      libsim.a(sim-if.o)
sim_core_write_buffer               libsim.a(sim-hrw.o)
sim_post_argv_init                  libsim.a(sim-if.o)
sim_module_resume                   libsim.a(cgen-run.o)
sim_state_alloc                     libsim.a(sim-if.o)
sim_args_command                    libsim.a(sim-if.o)
sim_core_write_unaligned_2          libsim.a(cgen-utils.o)
sim_core_write_unaligned_4          libsim.a(cgen-utils.o)
sim_io_evprintf                     libsim.a(sim-engine.o)
(Continue reading)

Christopher Faylor | 8 Jun 22:52 2004

Re: [Gdb] removing src/dejagnu, src/expect

On Sat, Jun 05, 2004 at 10:39:42AM -0400, Michael Elizabeth Chastain wrote:
>Watch out for Cygwin.  Has anyone run the gdb test suite on Cygwin
>recently?  Can someone make two test runs, one with sourceware

The expect in src/expect is customized for cygwin.  Possibly dejagnu is
also.

cgf

Christopher Faylor | 9 Jun 03:28 2004

Re: removing src/dejagnu, src/expect

On Sat, Jun 05, 2004 at 08:23:40PM +1000, Ben Elliston wrote:
>At yesterday's GCC Summit testing BOF, there was a strong concensus
>that these directories should be removed from the src tree and that
>developers should install the latest releases of Expect and Dejagnu
>from their distribution/packaging system or from a source tarball at
>ftp.gnu.org.
>
>I'm declaring that I intend to do this and no sooner than Friday June
>11 at 00:00 UTC.  I will be ensuring that no local patches, if any,
>are lost in this process.  If you have any reasonable objection,
>please say so now .. :-)

When you say that you are going to "remove" these I hope you just mean
"cvs remove", right?

Rhetorical question: Are you taking cygwin into account in this
decision?  If you've looked at the expect directory you would have seen
a few cygwin patches.

cgf

Chen Qi-A18263 | 11 Jun 14:50 2004

Can I use gdbserver without BDM device to debug ColdFire applicat ion


> Hi, all:
> 	I have encountered a problem while I am using the GDB server to debug a coldfire application:
> 
> 	I use the following steps for my 5272 application ( uClinux kernel 2.4, gcc3.3.3 ):
> 	1. on the target board, run gdbserver first
> 		
> 		 gdbserver :2000 test.flt
> 
> 	2. on host pc:
> 
> 		./m68k-bdm-elf-gdb test.flt.gdb
> 		(gdb) target remote 200.2.1.51:2000
> 			Remote debugging using 200.2.1.51:2000
> 			0x00f00048 in _start ()
> 		(gdb) n
> 			Single stepping until exit from function _start,
> 			which has no line number information.
> 			Software breakpoints not implemented for this target.	
> 
> 	It told me that "Software breakpoints not implemented for this target",
> 	Can any one give me some suggestion of how to deal this problems?
> 
> Thanks and Best Regards
> ChenQi(v)	
Chen Qi-A18263 | 14 Jun 16:01 2004

RE: Can I use gdbserver without BDM device to debug ColdFire appl ication


Hi, all:

	It seems that the m68k cross tools ( both gcc3.3.3 and gcc2.95 ) only have m68k-bdm-elf-gdb ( the
m68k-elf-gdb is linked to m68k-bdm-elf-gdb )
	Does it means that It only support debugging with BDM device?

Best Regards
ChenQi(v)
>  -----Original Message-----
> From: 	Chen Qi-A18263  
> Sent:	2004年6月11日 20:50
> To:	'gdb <at> gnu.org'; 'uclinux-dev <at> uclinux.org'
> Subject:	Can I use gdbserver without BDM device to debug ColdFire application
> 
> 
> 
> 	Hi, all:
> 		I have encountered a problem while I am using the GDB server to debug a coldfire application:
> 
> 		I use the following steps for my 5272 application ( uClinux kernel 2.4, gcc3.3.3 ):
> 		1. on the target board, run gdbserver first
> 			
> 			 gdbserver :2000 test.flt
> 
> 		2. on host pc:
> 
> 			./m68k-bdm-elf-gdb test.flt.gdb
> 			(gdb) target remote 200.2.1.51:2000
> 				Remote debugging using 200.2.1.51:2000
(Continue reading)

David McCullough | 15 Jun 01:40 2004

Re: [uClinux-dev] Can I use gdbserver without BDM device to debug ColdFire applicat ion


Jivin Chen Qi-A18263 lays it down ...
> > Hi, all:
> > 	I have encountered a problem while I am using the GDB server to debug a coldfire application:
> > 
> > 	I use the following steps for my 5272 application ( uClinux kernel 2.4, gcc3.3.3 ):
> > 	1. on the target board, run gdbserver first
> > 		
> > 		 gdbserver :2000 test.flt
> > 
> > 	2. on host pc:
> > 
> > 		./m68k-bdm-elf-gdb test.flt.gdb
> > 		(gdb) target remote 200.2.1.51:2000
> > 			Remote debugging using 200.2.1.51:2000
> > 			0x00f00048 in _start ()
> > 		(gdb) n
> > 			Single stepping until exit from function _start,
> > 			which has no line number information.
> > 			Software breakpoints not implemented for this target.	
> > 
> > 	It told me that "Software breakpoints not implemented for this target",
> > 	Can any one give me some suggestion of how to deal this problems?

Are you using XIP ?  Sounds like you may be running from ROM, and you
cannot set SW break points in ROM (the default breakpoint type). Check
out the following,  it might help:

	http://www.ucdot.org/article.pl?sid=03/01/30/0548223

(Continue reading)

linux | 26 Jun 06:44 2004

sigwait exits in gdb on pthread_exit call

 When I run a multithreaded program in the debugger, it exists on my 
sigwait call (EINTR) when a thread calls pthread_exit(NULL).  If I 
change the call to a return 0, then the problem goes away.  I don't have 
this problem if I don't run in gdb.  I was able to recreate the problem 
with a very small program.  To compile, I use g++ -g -O0 -o mymain 
mymain.cpp -lpthread -lrt

I am using gdb 6.1.1 - Red Hat Fedora Core 1 - Kernel 2.4.22-1.2115 with 
my setup. GCC 3.3.2

#include <signal.h>
#include <iostream>

void *MyThread(void *classInt)
{
   std::cout << "My thread runs\n";
   //If I replace with return 0; then the problem goes away
   pthread_exit(NULL);
}

int main(int argc,char *argv[])
{
   sigset_t set;
   int signum;
   int results;
   pthread_attr_t threadAttribute;
   pthread_t      threadId;
   struct sched_param param;

   sigemptyset(&set);
(Continue reading)


Gmane