Recent changes to bugs

which 1.1.0.6?

2004-12-24T19:46:10Z

From: Ilya N. Golubev <gin@mo.msk.ru>

<http://sourceforge.net/project/showfiles.php?group_id=9521&package_id=9598>
contains 2 `freedce-1.1.0.6.tar.gz' items, and lists
different file
sizes for them. However, both link
<http://prdownloads.sourceforge.net/freedce/freedce-1.1.0.6.tar.gz?download>
that contains file of 1644803 bytes.

What is the purpose of `freedce-1.1.0.6.tar.gz' of
1698970 bytes and
how do we download it?

freedce: non-portable and non-fixable

2004-03-11T19:44:29Z

Also mailed by <gin@mo.msk.ru> as
<02604050908440-gin@mo.msk.ru> to maintainers.

package: freedce
version: 1.1.0.7

Its `include.mk' forces `-Werror' in `CFLAGS'. So if
compiler is just
a bit different from one originally used from building
it and issues
some additional warnings - it does not get built at
all. Let alone
using compiler other than gcc.

What is worse, `include.mk' does such a forcing
unconditionally, that
is, does not allow to change `CFLAGS' value by
`configure'.

What is even worse, changing it and rebuilding
`Makefile.in's is also
impossible. Attempting to do it fails like this:

automake --foreign --include-deps ncklib/auth/Makefile
ncklib/auth/Makefile.am:9: AUTH_DUMMY does not appear
in AM_CONDITIONAL

That is, `Makefile.am's use automake conditionals, so
appropriate
`AM_CONDITIONAL' must be in `autoconf' input, but they
are in `.m4'
files that are, as <0260404e1f5c50-gin@mo.msk.ru> says,
missing in
distribution.

Releasing a completed distribution is absolutely
required. Perhaps
missing `.m4' files may also be obtained from some
anonymous cvs.
Are such repositories exist? If yes, which one contain
most receint
version?

freedce & dcethreads for gcc3.2

2003-04-18T09:43:37Z

Hi,

I have ported the freedce project to newer linux
distributions (RH8 - RH9 gcc3.2) and have fixed some
bugs!!

I think this project is really dead - I tried to contact the
members of this project but their email doesn't work....

So if you want my new package - contact me
stocki@janfeatudssr.com

DCE Crash

2002-08-23T07:32:02Z

Hi,

I have a multithreaded RPC application onn Linux
which communicates with Other RPC application on
Windows. When the server on the Windows machine
crashes for some reason, the application cancels
the thread on timeout and aborts the transaction.

The problem is that atfer 3 or four such cancelled
transactions, the "assoc" structure passed to
the "receive_dispatch" function in cnrcvr.c file gets
corrupted or contains reference to invalid memory
locations. This leads to segmentation fault in the
application on Linux.

I am unable to find a reason for this behaviour.
Strangely, I see this behaviour only when the server
on the other side crashes when the transaction was
in progress.

I would really appreciate if somebody could provide
me with some hints.

regards
Sebastian J. Mathew
sebastianjm@integramicro.com
sebastianjm@vsnl.net

rpc_mgmt_is_server_listening

2002-07-05T09:21:02Z

the call to rpc_mgmt_is_server_listening does not
return, if the client is on the same station than the
server.

But if I run the same client on a remote machine, it
works.

Can you explain that ?

Thanks
PAscal

freedce make error

2002-06-25T19:57:08Z

Making all in cn
make[3]: Entering directory
`/home/data/freedce-1.1.0.7/ncklib/transport/cn'
/bin/sh ../../../libtool --mode=compile gcc
-DPACKAGE=\&quot;freedce\&quot; -DVERSION=\&quot;1.1.0.7\&quot;
-DHAVE_DLFCN_H=1 -DYYTEXT_POINTER=1 -DHAVE_LIBPTHREAD=1
-DHAVE_FEATURES_H=1 -DHAVE_SYS_CDEFS_H=1 -D_REENTRANT=1
-D_GNU_SOURCE=1 -DDEBUG=1 -DHAVE_GETTIMEOFDAY=1
-DHAVE_SOCKET=1 -DPROT_NCACN=1 -DPROT_NCADG=1
-DNAF_IP=1 -DUSE_LOOPBACK=1 -DDEMO_PROGS=1 -I. -I.
-I../../../include -I/opt/dce/include
-I../../../ncklib/include
-I../../../ncklib/include/linux-gnu -g -Wall -W -O
-pipe -Werror -DNCK -D_POSIX_C_SOURCE=1 -c cncall.c
rm -f .libs/cncall.lo
gcc -DPACKAGE=\&quot;freedce\&quot; -DVERSION=\&quot;1.1.0.7\&quot;
-DHAVE_DLFCN_H=1 -DYYTEXT_POINTER=1 -DHAVE_LIBPTHREAD=1
-DHAVE_FEATURES_H=1 -DHAVE_SYS_CDEFS_H=1 -D_REENTRANT=1
-D_GNU_SOURCE=1 -DDEBUG=1 -DHAVE_GETTIMEOFDAY=1
-DHAVE_SOCKET=1 -DPROT_NCACN=1 -DPROT_NCADG=1
-DNAF_IP=1 -DUSE_LOOPBACK=1 -DDEMO_PROGS=1 -I. -I.
-I../../../include -I/opt/dce/include
-I../../../ncklib/include
-I../../../ncklib/include/linux-gnu -g -Wall -W -O
-pipe -Werror -DNCK -D_POSIX_C_SOURCE=1 -c cncall.c
-fPIC -DPIC -o .libs/cncall.lo
cc1: warnings being treated as errors
cncall.c: In function `rpc__cn_call_start':
cncall.c:197: warning: `cancel_timeout' might be used
uninitialized in this function
make[3]: *** [cncall.lo] Error 1
make[3]: Leaving directory
`/home/data/freedce-1.1.0.7/ncklib/transport/cn'
make[2]: *** [all-recursive] Error 1
make[2]: Leaving directory
`/home/data/freedce-1.1.0.7/ncklib/transport'
make[1]: *** [all-recursive] Error 1
make[1]: Leaving directory
`/home/data/freedce-1.1.0.7/ncklib'
make: *** [all-recursive] Error 1

pb with signals

2002-06-07T14:12:13Z

I am working on a program which forks and exec
other binaries, and then observes his children,
restarting them in case of premature stop, etc...
This program also uses rpc calls to inform
another program about the status of it's children.
For the detection of the death of a child, I use the
fact that the father receives a SIGCHLD signal
when one child dies..
The problem is that when I do the
rpc_string_binding_compose(...), it seems
that some signals are caught somewhere before
they arrive to my handlers...
this doesn't occurs when I comment the rpc calls

Have any idea ?
thanks.
Pascal

cross-compilation issue

2001-12-15T06:07:12Z

After fixed some simple problems, I have built DCE
threads and FreeDCE successfully on my RedHat7.1(intel
machine),
and it works great. However, when I tried to compile
the freeDCE for powerpc machine through cross
compilation in the intel machine, it fails.
Several points I have discovered:
(1) For DCE thread, some C header files have these
lines: #include </usr/include/pthread.h>.
The cross compiler will warn that it is not safe for
cross compilation using pthread.
So use the pthread.h which comes with cross compiler.
(2) For FreeDCE, it will make "dceidl"(I call it IDL
later) first, then make idllib, uuid, .... It will run
IDL in the middle of making. As we are using cross-
compiler, the built IDL will not work in intel
machine.
So I copied the built IDL(intel machine version) to
replace the non-worked IDL(powerpc). Then compile can
be done sucessfully. But this built version won't
work. The reason is that: IDL will generate little
endian stub file in the intel-machine, while powerpc
need big-endian stub file. This can be verified by
running IDL in these two kinds of machines, the
generated c source file is different. So I modified
the IDL source code such that IDL(intel machine
version) will generate the powerpc stub file format.
And then try to compile it again. Built can be done
successfully, it looks that rpcd works good as daemon.
But my test file still fails which runs good under
intel machine.

Can anyone give me some suggestions on cross
compilation.

Thanks,
Xiao
xiao@peakaudio.com

**************
Notes:
Machine environment:
intel machine + RedHat7.1; cross compiler; We need
target can be run under powerpc
**************

patches for build errors

2001-11-28T15:49:00Z

The packages (freedce and dcethreads) will not build
due to the -Werror in the build on Red Hat v7.2 as
shipped. I have some (trivial) patches that fix them.
The dcethreads bug was previously reported, but since
no patch was included with the report, I included it as
well.
The last fix was a Makefile change, as -Werror was
causing the build to fail, reporting a possible
variable clobber by longjump/vfork that didn't look
real to me.

Also, while I'm at it, I can't get the
rpc-internals.pdf file. The server
freedce.sourceforge.net doesn't seem to be up.

Here is the patch for dcethreads:

*** src/pthread_dce.c.orig Wed Nov 28 09:49:47 2001
--- src/pthread_dce.c Wed Nov 28 09:50:39 2001
***************
*** 55,60 ****
--- 55,61 ----

#include <dce/dcethreads_conf.h>
+ #include <stdlib.h> /* for abort() */

#ifndef lint
static const char rcsid[] __attribute__((__unused__))
= "$Id: pthread_dce.c,v 1.1 2001/04/12 20:11:25 wez Exp
$";

Here is the patch for freedce:

*** demos/echo_server/misc.c.orig Wed Nov 28 10:19:24
2001
--- demos/echo_server/misc.c Wed Nov 28 10:20:01 2001
***************
*** 2,7 ****
--- 2,8 ----
#include <string.h>
#include <dce/rpc.h>
#include <dce/dce_error.h>
+ #include <stdlib.h>

void
chk_dce_err(ecode, where, why, fatal)
*** demos/echo_server/echo_server.c.orig Wed Nov 28
10:20:21 2001
--- demos/echo_server/echo_server.c Wed Nov 28 10:20:37
2001
***************
*** 9,14 ****
--- 9,15 ----
#include <stdio.h>
#include <string.h>
#include <signal.h>
+ #include <stdlib.h>
#include <dce/rpc.h>
#include <dce/pthread_exc.h>
#include "echo.h"
*** demos/echo_server/echo_client.c.orig Wed Nov 28
10:20:55 2001
--- demos/echo_server/echo_client.c Wed Nov 28 10:21:10
2001
***************
*** 9,14 ****
--- 9,15 ----
#include <stdio.h>
#include <string.h>
#include <dce/rpc.h>
+ #include <stdlib.h>
#include <dce/pthread_exc.h>
#include "echo.h"
#include <misc.h>
*** ncklib/transport/cn/cncall.c.orig Wed Nov 28
09:57:34 2001
--- ncklib/transport/cn/cncall.c Wed Nov 28 09:59:24
2001
***************
*** 193,199 ****
rpc_iovector_elt_p_t iov_p;
rpc_cn_packet_p_t header_p;
unsigned32 cancel_count = 0;
! rpc_clock_t cancel_timeout;
unsigned32 temp_st;
pthread_t current_thread_id;
boolean resolving_thread;
--- 193,199 ----
rpc_iovector_elt_p_t iov_p;
rpc_cn_packet_p_t header_p;
unsigned32 cancel_count = 0;
! rpc_clock_t cancel_timeout = 0;
unsigned32 temp_st;
pthread_t current_thread_id;
boolean resolving_thread;
*** ncklib/transport/dg/dglsn.c.orig Wed Nov 28
10:00:49 2001
--- ncklib/transport/dg/dglsn.c Wed Nov 28 10:04:01
2001
***************
*** 2729,2735 ****
ready_to_go = rexmit_cnt + (xq->first_unsent ==
NULL ? 0 :
xq->next_fragnum - (signed16)
xq->first_unsent->fragnum);

! xq->blast_size = (ready_to_go <
(unsigned32)blast_size) ? ready_to_go : blast_size;

RPC_DBG_PRINTF(rpc_e_dbg_recv, 5, (
--- 2729,2735 ----
ready_to_go = rexmit_cnt + (xq->first_unsent ==
NULL ? 0 :
xq->next_fragnum - (signed16)
xq->first_unsent->fragnum);

! xq->blast_size = ((unsigned32) ready_to_go <
(unsigned32)blast_size) ? (signed32) ready_to_go :
blast_size;

RPC_DBG_PRINTF(rpc_e_dbg_recv, 5, (
*** rpcd/Makefile.orig Wed Nov 28 10:18:52 2001
--- rpcd/Makefile Wed Nov 28 10:19:00 2001
***************
*** 91,97 ****

dce_includes = -I$(top_srcdir)/include
$(DCETHREADINCLUDES)

! CFLAGS = -g -Wall -W -O -Werror -pipe -DNCK
-D_POSIX_SOURCE=1 -I$(top_srcdir)/ncklib/include
-I$(top_srcdir)/ncklib/include/$(target_os)
INCLUDES = $(dce_includes)

SUFFIXES = .idl
--- 91,97 ----

dce_includes = -I$(top_srcdir)/include
$(DCETHREADINCLUDES)

! CFLAGS = -g -Wall -W -O -pipe -DNCK -D_POSIX_SOURCE=1
-I$(top_srcdir)/ncklib/include
-I$(top_srcdir)/ncklib/include/$(target_os)
INCLUDES = $(dce_includes)

SUFFIXES = .idl

build error

2001-10-19T04:46:28Z

Hi, I have troubles try to build freedce-1.1.0.7 on
a RedHat 6.2 machine. It fails at "idl" directory.
Any suggestions?