Bug 218350

+++ This bug was initially created as a clone of Bug #217843 +++

Version-Release number of selected component (if applicable):
libswt3-gtk2-3.2.1-22.fc7

How reproducible:
All swt based applications (eclipse, azureus, ...) that try to open a file
dialog (this also reproduces it:
http://www.java2s.com/ExampleCode/SWT-JFace-Eclipse/FileDialogExample.htm).

I tried:
eclipse + jdk1.5.0
native eclipse + jdk1.5.0
eclipse + gcj
native eclipse + gcj

All combinations reproduce the bugs.

When I run strace on the simple example at the end I get this (killed it at the
end):

socket(PF_FILE, SOCK_STREAM, 0)         = 11
connect(11, {sa_family=AF_FILE, path=@/tmp/dbus-tJslKvVVOV}, 23) = 0
fcntl64(11, F_GETFL)                    = 0x2 (flags O_RDWR)
fcntl64(11, F_SETFL, O_RDWR|O_NONBLOCK) = 0
fcntl64(11, F_GETFD)                    = 0
fcntl64(11, F_SETFD, FD_CLOEXEC)        = 0
getuid32()                              = 500
rt_sigaction(SIGPIPE, {SIG_IGN}, {SIG_IGN}, 8) = 0
poll([{fd=11, events=POLLOUT, revents=POLLOUT}], 1, 0) = 1
write(11, "\0", 1)                      = 1
write(11, "AUTH EXTERNAL 353030\r\n", 22) = 22
poll([{fd=11, events=POLLIN, revents=POLLIN}], 1, -1) = 1
read(11, "OK 5ede5bb19c3a29b81bf7eb00456db"..., 2048) = 37
poll([{fd=11, events=POLLOUT, revents=POLLOUT}], 1, -1) = 1
write(11, "BEGIN\r\n", 7)               = 7
poll([{fd=11, events=POLLIN|POLLOUT, revents=POLLOUT}], 1, -1) = 1
writev(11, [{"l\1\0\1\0\0\0\0\1\0\0\0n\0\0\0\1\1o\0\25\0\0\0/org/fre"..., 128},
{"", 0}], 2) = 128
gettimeofday({1164821649, 503182}, NULL) = 0
poll([{fd=11, events=POLLIN, revents=POLLIN}], 1, 25000) = 1
read(11, "l\3\1\1\326\0\0\0\1\0\0\0]\0\0\0\4\1s\0\'\0\0\0org.fre"..., 2048) = 326
read(11, 0x892d858, 2048)               = -1 EAGAIN (Resource temporarily
unavailable)
close(11)                               = 0
futex(0x8929480, FUTEX_WAIT, 2, NULL)   = -1 EINTR (Interrupted system call)

It seems there is a problem with dbus.
--- SIGINT (Interrupt) @ 0 (0) ---

-- Additional comment from bkonrath on 2006-11-30 09:42 EST --
I get a problem when I run the example too. 

Here's what I have installed:

eclipse-platform-3.2.1-22.fc7.i386
dbus-1.0.1-2.fc7.i386

Here's what I get when I run that example:

% strace java -cp /usr/lib/eclipse/swt-gtk-3.2.jar:. FileDialogExample
...
open("/var/lib/dbus/machine-id", O_RDONLY) = -1 ENOENT (No such file or directory)
write(2, "process 21006: ", 15process 21006: )         = 15
write(2, "D-Bus library appears to be inco"..., 206D-Bus library appears to be
incorrectly set up; failed to read machine uuid: Failed to open
"/var/lib/dbus/machine-id": No such file or directory
See the manual page for dbus-uuidgen to correct this issue.
) = 206
write(2, "  D-Bus not built with -rdynamic"..., 64  D-Bus not built with
-rdynamic so unable to print a backtrace
) = 64
rt_sigprocmask(SIG_UNBLOCK, [ABRT], NULL, 8) = 0
tgkill(21006, 21006, SIGABRT)           = 0
--- SIGABRT (Aborted) @ 0 (0) ---
+++ killed by SIGABRT (core dumped) +++
Process 21006 detached

I'm going to update my rawhide box and test again before I punt this over to dbus.


-- Additional comment from bart.vanbrabant on 2006-11-30 09:51 EST --
At the moment I'm running an updated rawhide from 29/11 but I have the problem
for a weeks now but I didn't have the time to inspect it until now. I think it
happened when the final dbus was pushed.

The c gtk program's don't have this problem, I think there is something wrong in
the swt java bindings.

-- Additional comment from bkonrath on 2006-11-30 09:55 EST --
(In reply to comment #2)
> The c gtk program's don't have this problem, I think there is something wrong in
> the swt java bindings.

Ok, if you could narrow it down more that would be great. Thanks.



-- Additional comment from bkonrath on 2006-11-30 10:10 EST --
John, Have you ever seen anything like this before? Thanks.

-- Additional comment from bkonrath on 2006-11-30 10:42 EST --
Bart, I worked around my problem doing this:

* check if this file exists: 'ls -l /var/lib/dbus/machine-id'
* it didn't on my system so I ran 'dbus-uuidgen --ensure' as root
* the example app worked after that with the latest rawhide.

Could you try that?

-- Additional comment from bart.vanbrabant on 2006-11-30 10:51 EST --
That didn't work. The file existed, I moved it to some other place and I ran
dbus-uuidgen --ensure but that didn't change anything. :(
I move my old machine-id back. Did I need to reboot or something like that
before testing it?

-- Additional comment from bkonrath on 2006-11-30 10:58 EST --
You're not supposed to change that file in a live system (see man dbus-uuidgen
for details). I only ran the dbus-uuidgen because the file didn't exist for me.
But maybe try rebooting.

-- Additional comment from bart.vanbrabant on 2006-11-30 11:11 EST --
I tried that, but it didn't work.

-- Additional comment from bkonrath on 2006-11-30 11:24 EST --
Ok then there's nothing else I can tell you because it's working for me, sorry.
Maybe John has some ideas.

-- Additional comment from johnp on 2006-11-30 13:02 EST --
What version of D-Bus and version of gnome-vfs2 are you running?

-- Additional comment from bart.vanbrabant on 2006-11-30 13:08 EST --
$ rpm -qa dbus
dbus-1.0.1-2.fc7
$ rpm -qa gnome-vfs2
gnome-vfs2-2.16.2-2.fc7


-- Additional comment from johnp on 2006-11-30 17:23 EST --
Ok, I was finally able to update my box.  It looks like threading is not enabled
from the backtrace.  I need to check with Alex to make sure gnome-vfs is calling
dbus_threads_init_default.

-- Additional comment from johnp on 2006-11-30 17:27 EST --
Relevant backtrace (notice the (mutex=0x0) at #)5:


#0  0x0079c410 in __kernel_vsyscall ()
#1  0x0011b85e in __lll_mutex_lock_wait () from /lib/libpthread.so.0
#2  0x001176ec in _L_mutex_lock_83 () from /lib/libpthread.so.0
#3  0x0011723d in pthread_mutex_lock () from /lib/libpthread.so.0
#4  0x04e07efd in dbus_g_object_path_get_g_type ()
   from /usr/lib/libdbus-glib-1.so.2
#5  0x051827ff in _dbus_mutex_lock (mutex=0x0) at dbus-threads.c:150
#6  0x0516b87d in _dbus_bus_notify_shared_connection_disconnected_unlocked (
    connection=0x96ad0e8) at dbus-bus.c:353
#7  0x0516f53d in _dbus_connection_get_dispatch_status_unlocked (
    connection=0x96ad0e8) at dbus-connection.c:3788
#8  0x05170813 in _dbus_connection_close_possibly_shared_and_unlock (
    connection=0x96ad0e8) at dbus-connection.c:2685
#9  0x0516c2fd in internal_bus_get (type=DBUS_BUS_SESSION, private=0, 
    error=0xbf82a0b4) at dbus-bus.c:448
#10 0x04109f8b in gnome_vfs_daemon_message_iter_get_file_info ()
   from /usr/lib/libgnomevfs-2.so.0
#11 0x0412a74e in gnome_vfs_volume_monitor_client_shutdown_private ()
   from /usr/lib/libgnomevfs-2.so.0
#12 0x0412a309 in gnome_vfs_volume_monitor_client_get_type ()
   from /usr/lib/libgnomevfs-2.so.0
#13 0x03ceb64a in g_type_create_instance () from /lib/libgobject-2.0.so.0
#14 0x03cd2f32 in g_object_set () from /lib/libgobject-2.0.so.0
---Type <return> to continue, or q <return> to quit---
#15 0x03cd0b8b in g_object_newv () from /lib/libgobject-2.0.so.0
#16 0x03cd16ef in g_object_new_valist () from /lib/libgobject-2.0.so.0
#17 0x03cd18a0 in g_object_new () from /lib/libgobject-2.0.so.0
#18 0x0412b0c4 in gnome_vfs_volume_monitor_unref ()
   from /usr/lib/libgnomevfs-2.so.0
#19 0x0412b15a in gnome_vfs_get_volume_monitor ()
   from /usr/lib/libgnomevfs-2.so.0
#20 0x03f303b4 in gtk_file_system_gnome_vfs_new ()

-- Additional comment from alexl on 2006-12-01 04:00 EST --
Well, we call dbus_g_thread_init()


-- Additional comment from bart.vanbrabant on 2006-12-04 09:56 EST --
I have some extra information, when I run it from a root shell I don't have this
problem. Maybe because I don't have a session bus then.

-- Additional comment from johnp on 2006-12-04 10:01 EST --
Yes it is because gnome-vfs doesn't go through the D-Bus path.  One thing I
can't understand is why this happens only in Java and not other apps.  Perhaps
some other library is using D-Bus without knowing it needs to initialize threads
and since Java is highly threaded it is causing problems.  Does eclipse use
libraries like avahi or HAL?

-- Additional comment from bart.vanbrabant on 2006-12-04 10:22 EST --
I can also recreate it with a simple example app that only starts a window frame
and a menu item that "should" show a filedialog. So only plain java (gij and sun
java) and swt (from rawhide and swt from eclipse.org).

-- Additional comment from bart.vanbrabant on 2006-12-04 10:49 EST --
I tried the latest swt build from the eclipse website but the results stay the
same. 
I got a workaround now, starting eclipse like this: unset
DBUS_SESSION_BUS_ADDRESS && eclipse but it pretty ugly.

-- Additional comment from overholt on 2006-12-04 10:59 EST --
The Eclipse SDK itself uses gnome-vfs (via SWT) but not HAL or avahi AFAIK. 
Since SWT itself is using gnome-vfs and can be used independently of the Eclipse
platform, should it initialize D-Bus itself?  Either way, I think this is an
upstream bug and should probably be discussed at bugs.eclipse.org to involve the
SWT developers.  But I guess that would mean John and/or Alex have to create BZ
accounts there <sigh>.

-- Additional comment from johnp on 2006-12-04 12:36 EST --
Interesting note.  This is not happening in rawhide anymore.  I am not sure what
update might have fixed it here.

-- Additional comment from bart.vanbrabant on 2006-12-04 12:51 EST --
I'm running an up to date rawhide and I still have the problem :(

-- Additional comment from johnp on 2006-12-04 13:12 EST --
Ok, finally got the test program to run.  It isn't crashing because for some
reason dialog doesn't seem to be running the gnome-vfs plugin (I don't see my
remote volumes).  It is running it on C apps though.  I'm going to update again
and reboot to see if I can get this working. 

-- Additional comment from bkonrath on 2006-12-04 13:21 EST --
(In reply to comment #22)
> Ok, finally got the test program to run.  It isn't crashing because for some
> reason dialog doesn't seem to be running the gnome-vfs plugin (I don't see my
> remote volumes).

Maybe that's why it was working for me too.


-- Additional comment from johnp on 2006-12-04 14:49 EST --
Ok, reproducing the bug with D-Bus compiled with -O0 seems to point to a
deadlock and not a crash on a NULL mutex:

#0  0x00b29410 in __kernel_vsyscall ()
#1  0x0032585e in __lll_mutex_lock_wait () from /lib/libpthread.so.0
#2  0x003216ec in _L_mutex_lock_83 () from /lib/libpthread.so.0
#3  0x0032123d in pthread_mutex_lock () from /lib/libpthread.so.0
#4  0x4e6bfefd in dbus_gmutex_lock (mutex=0x8ebaac8) at dbus-gthread.c:95
#5  0x007fd762 in _dbus_mutex_lock (mutex=0x8ebaac8) at dbus-threads.c:150
#6  0x007d35bc in _dbus_bus_notify_shared_connection_disconnected_unlocked (
    connection=0x8ebe1b8) at dbus-bus.c:353
#7  0x007dcba6 in notify_disconnected_unlocked (connection=0x8ebe1b8)
    at dbus-connection.c:3788
#8  0x007dcd33 in _dbus_connection_get_dispatch_status_unlocked (
    connection=0x8ebe1b8) at dbus-connection.c:3865
#9  0x007d9c46 in _dbus_connection_close_possibly_shared_and_unlock (
    connection=0x8ebe1b8) at dbus-connection.c:2685
#10 0x007d83f7 in _dbus_connection_close_possibly_shared (
    connection=0x8ebe1b8) at dbus-connection.c:1803
#11 0x007d397e in internal_bus_get (type=DBUS_BUS_SESSION, private=0, 
    error=0xbf9a1650) at dbus-bus.c:448
#12 0x007d3a6a in dbus_bus_get (type=DBUS_BUS_SESSION, error=0xbf9a1650)
    at dbus-bus.c:515
#13 0x4e851f8b in _gnome_vfs_get_main_dbus_connection ()
---Type <return> to continue, or q <return> to quit---
    at gnome-vfs-dbus-utils.c:151
#14 0x4e87274e in setup_dbus_connection (client=0x8e97a60)
    at gnome-vfs-volume-monitor-client.c:380
#15 0x4e872309 in gnome_vfs_volume_monitor_client_init (
    volume_monitor_client=0x8e97a60) at gnome-vfs-volume-monitor-client.c:252
#16 0x4f16664a in g_type_create_instance () from /lib/libgobject-2.0.so.0
#17 0x4f14df32 in g_object_set () from /lib/libgobject-2.0.so.0
#18 0x4f14bb8b in g_object_newv () from /lib/libgobject-2.0.so.0
#19 0x4f14c6ef in g_object_new_valist () from /lib/libgobject-2.0.so.0
#20 0x4f14c8a0 in g_object_new () from /lib/libgobject-2.0.so.0
#21 0x4e8730c4 in _gnome_vfs_get_volume_monitor_internal (create=1)
    at gnome-vfs-volume-monitor.c:317
#22 0x4e87315a in gnome_vfs_get_volume_monitor ()
    at gnome-vfs-volume-monitor.c:342
#23 0x0018b3b4 in gtk_file_system_gnome_vfs_new ()
   from /usr/lib/gtk-2.0/2.10.0/filesystems/libgnome-vfs.so
#24 0x4f16664a in g_type_create_instance () from /lib/libgobject-2.0.so.0
#25 0x4f14df32 in g_object_set () from /lib/libgobject-2.0.so.0
#26 0x4f14bb8b in g_object_newv () from /lib/libgobject-2.0.so.0
#27 0x4f14c6ef in g_object_new_valist () from /lib/libgobject-2.0.so.0
#28 0x4f14c8a0 in g_object_new () from /lib/libgobject-2.0.so.0
---Type <return> to continue, or q <return> to quit---
#29 0x0018b0d9 in gtk_file_system_gnome_vfs_new ()
   from /usr/lib/gtk-2.0/2.10.0/filesystems/libgnome-vfs.so
#30 0x001911a5 in fs_module_create ()
   from /usr/lib/gtk-2.0/2.10.0/filesystems/libgnome-vfs.so
#31 0x4f32339d in _gtk_file_system_module_create (
    fs_module=<value optimized out>) at gtkfilesystem.c:1358
#32 0x4f323523 in IA__gtk_file_system_create (
    file_system_name=0x8d933c8 "gnome-vfs") at gtkfilesystem.c:1397
#33 0x4f314074 in gtk_file_chooser_default_set_property (object=0x8eb1030, 
    prop_id=4097, value=0xbf9a1f34, pspec=0x8dd0f00)
    at gtkfilechooserdefault.c:5065
#34 0x4f14e10c in g_object_set () from /lib/libgobject-2.0.so.0
#35 0x4f3179f1 in gtk_file_chooser_default_constructor (type=149603648, 
    n_construct_properties=1, construct_params=0x8d93138)
    at gtkfilechooserdefault.c:4904
#36 0x4f14bb8b in g_object_newv () from /lib/libgobject-2.0.so.0
#37 0x4f14c799 in g_object_new_valist () from /lib/libgobject-2.0.so.0
#38 0x4f14c8a0 in g_object_new () from /lib/libgobject-2.0.so.0
#39 0x4f309808 in _gtk_file_chooser_default_new (file_system=0x0)
    at gtkfilechooserdefault.c:9372
#40 0x4f31c01c in gtk_file_chooser_widget_constructor (type=149613544, 
---Type <return> to continue, or q <return> to quit---
    n_construct_properties=1, construct_params=0x8eac518)
    at gtkfilechooserwidget.c:103
#41 0x4f14bb8b in g_object_newv () from /lib/libgobject-2.0.so.0
#42 0x4f14c6ef in g_object_new_valist () from /lib/libgobject-2.0.so.0
#43 0x4f14c8a0 in g_object_new () from /lib/libgobject-2.0.so.0
#44 0x4f3183bd in gtk_file_chooser_dialog_constructor (type=149531504, 
    n_construct_properties=3, construct_params=0x8eacb90)
    at gtkfilechooserdialog.c:374
#45 0x4f14bb8b in g_object_newv () from /lib/libgobject-2.0.so.0
#46 0x4f14c799 in g_object_new_valist () from /lib/libgobject-2.0.so.0
#47 0x4f14c8a0 in g_object_new () from /lib/libgobject-2.0.so.0
#48 0x4f317ebe in gtk_file_chooser_dialog_new_valist (
    title=<value optimized out>, parent=0x8cc0aa0, 
    action=GTK_FILE_CHOOSER_ACTION_OPEN, backend=0x0, 
    first_button_text=0x56f4db "gtk-cancel", varargs=0xbf9a26ac "������V")
    at gtkfilechooserdialog.c:569
#49 0x4f317f6c in IA__gtk_file_chooser_dialog_new (title=0x768958 "Open", 
    parent=0x8cc0aa0, action=GTK_FILE_CHOOSER_ACTION_OPEN, 
    first_button_text=0x56f4db "gtk-cancel") at gtkfilechooserdialog.c:614
#50 0x0056f3cf in
Java_org_eclipse_swt_internal_gtk_OS__1gtk_1file_1chooser_1dialog_1new () from
/usr/lib/libswt-pi-gtk-3235.so
---Type <return> to continue, or q <return> to quit--- 
#51 0x05d42e6b in
org::eclipse::swt::internal::gtk::OS::_gtk_file_chooser_dialog_new ()
   from /usr/lib/gcj/eclipse/org.eclipse.swt.gtk.linux.x86_3.2.1.v3235.jar.so
#52 0x05d42f0f in
org::eclipse::swt::internal::gtk::OS::gtk_file_chooser_dialog_new ()
   from /usr/lib/gcj/eclipse/org.eclipse.swt.gtk.linux.x86_3.2.1.v3235.jar.so
#53 0x05e333d5 in org::eclipse::swt::widgets::FileDialog::openChooserDialog
    ()
   from /usr/lib/gcj/eclipse/org.eclipse.swt.gtk.linux.x86_3.2.1.v3235.jar.so
#54 0x05e332f3 in org::eclipse::swt::widgets::FileDialog::open ()
   from /usr/lib/gcj/eclipse/org.eclipse.swt.gtk.linux.x86_3.2.1.v3235.jar.so
#55 0x01f616b7 in _Jv_strtod_r () from /usr/lib/libgcj.so.7rh
#56 0x01f61454 in _Jv_strtod_r () from /usr/lib/libgcj.so.7rh
#57 0x016c3541 in _Jv_InterpMethod::run () from /usr/lib/libgcj.so.7rh
#58 0x016c71cb in _Jv_InterpMethod::run_normal () from /usr/lib/libgcj.so.7rh
#59 0x01f6178d in _Jv_strtod_r () from /usr/lib/libgcj.so.7rh
#60 0x05e97b74 in org::eclipse::swt::widgets::TypedListener::handleEvent ()
   from /usr/lib/gcj/eclipse/org.eclipse.swt.gtk.linux.x86_3.2.1.v3235.jar.so
#61 0x05e2b74d in org::eclipse::swt::widgets::EventTable::sendEvent ()
   from /usr/lib/gcj/eclipse/org.eclipse.swt.gtk.linux.x86_3.2.1.v3235.jar.so
#62 0x05e9a23f in org::eclipse::swt::widgets::Widget::sendEvent ()
---Type <return> to continue, or q <return> to quit---
   from /usr/lib/gcj/eclipse/org.eclipse.swt.gtk.linux.x86_3.2.1.v3235.jar.so
#63 0x05e28ab0 in org::eclipse::swt::widgets::Display::runDeferredEvents ()
   from /usr/lib/gcj/eclipse/org.eclipse.swt.gtk.linux.x86_3.2.1.v3235.jar.so
#64 0x05e2791f in org::eclipse::swt::widgets::Display::readAndDispatch ()
   from /usr/lib/gcj/eclipse/org.eclipse.swt.gtk.linux.x86_3.2.1.v3235.jar.so
#65 0x01f616b7 in _Jv_strtod_r () from /usr/lib/libgcj.so.7rh
#66 0x01f61454 in _Jv_strtod_r () from /usr/lib/libgcj.so.7rh
#67 0x016c3541 in _Jv_InterpMethod::run () from /usr/lib/libgcj.so.7rh
#68 0x016c71cb in _Jv_InterpMethod::run_normal () from /usr/lib/libgcj.so.7rh
#69 0x01f6178d in _Jv_strtod_r () from /usr/lib/libgcj.so.7rh
#70 0x01f616b7 in _Jv_strtod_r () from /usr/lib/libgcj.so.7rh
#71 0x01f61454 in _Jv_strtod_r () from /usr/lib/libgcj.so.7rh
#72 0x016c3541 in _Jv_InterpMethod::run () from /usr/lib/libgcj.so.7rh
#73 0x016c711a in _Jv_InterpMethod::run_class () from /usr/lib/libgcj.so.7rh
#74 0x01f6178d in _Jv_strtod_r () from /usr/lib/libgcj.so.7rh
#75 0x016e0824 in gnu::java::lang::MainThread::call_main ()
   from /usr/lib/libgcj.so.7rh
#76 0x01745b07 in gnu::java::lang::MainThread::run ()
   from /usr/lib/libgcj.so.7rh
#77 0x016f060b in _Jv_ThreadRun () from /usr/lib/libgcj.so.7rh
#78 0x016ab5fc in _Jv_RunMain () from /usr/lib/libgcj.so.7rh
---Type <return> to continue, or q <return> to quit---
#79 0x0019fa98 in main () from /usr/lib/libgij.so.7rh
#80 0x001b7e5c in __libc_start_main () from /lib/libc.so.6
#81 0x08048481 in _Jv_RegisterClasses ()


-- Additional comment from johnp on 2006-12-04 15:04 EST --
I found the culprit.  It is an SELinux issue combined with an error path in
D-Bus that deadlocks.  I need to fix that error path but I also need to
investigate why SELinux is blocking Java from registering a connection on the bus. 

(gdb) print error->message
$10 = 0x9b5bca8 "An SELinux policy prevents this sender from sending this
message to this recipient (rejected message had interface
\"org.freedesktop.DBus\" member \"Hello\" error name \"(unset)\" destination
\"org.freedeskt"...

-- Additional comment from johnp on 2006-12-04 15:07 EST --
Dec  4 15:06:55 phuket dbus: Can't send to audit system: USER_AVC avc:  denied 
{ send_msg } for msgtype=method_call interface=org.freedesktop.DBus member=Hello
dest=org.freedesktop.DBus spid=19333 scontext=user_u:system_r:java_t:s0
tcontext=user_u:system_r:unconfined_t:s0 tclass=dbus : exe="/bin/dbus-daemon"
(sauid=500, hostname=?, addr=?, terminal=?)