desktopcouch-service crashed with SIGSEGV

StacktraceTop:
 __mempcpy_ssse3 () from /lib/tls/i686/cmov/libc.so.6
 ?? ()
 _dbus_message_loader_queue_messages (loader=0xb72208c5)
 _dbus_transport_get_dispatch_status (transport=0x8c12d48)
 _dbus_transport_queue_messages (transport=0x8c12d48)

Source-package 'dbus' doesn't make "-dbg" packages. Had to rebuild.

Full stacktrace:

#0 0x00e39aeb in _dbus_list_unlink (list=0xb76bcb5c, link=0x88da83c) at dbus-list.c:517
#1 0x00e39de9 in _dbus_list_remove_link (list=0xb76bcb5c, link=0x88da83c) at dbus-list.c:537
#2 0x00e39e3a in _dbus_list_pop_last (list=0xb76bcb5c) at dbus-list.c:689
#3 0x00e28a63 in _dbus_validate_signature_with_reason (type_str=0xb76bcc1c, type_pos=0, len=1)
    at dbus-marshal-validate.c:218
#4 0x00e28d5c in validate_body_helper (reader=0xb76bcc98, byte_order=<value optimized out>, walk_reader_to_end=1,
    p=0x88dec32 "u", end=0x88df410 "", new_p=0xb76bcd24) at dbus-marshal-validate.c:558
#5 0x00e29022 in validate_body_helper (reader=0xb76bcda0, byte_order=<value optimized out>, walk_reader_to_end=0,
    p=0x88dec30 "\005\001u", end=0x88df410 "", new_p=0xb76bcdf4) at dbus-marshal-validate.c:623
#6 0x00e28fb7 in validate_body_helper (reader=<value optimized out>, byte_order=<value optimized out>,
    walk_reader_to_end=1, p=0x88dec2f "", end=0x88df410 "", new_p=0xb76bce2c) at dbus-marshal-validate.c:477
#7 0x00e2941a in _dbus_validate_body_with_reason (expected_signature=0xe49ca8, expected_signature_start=0,
    byte_order=108, bytes_remaining=0xb76bcf38, value_str=0x88be8dc, value_pos=0, len=2048)
    at dbus-marshal-validate.c:708
#8 0x00e251ff in _dbus_header_load (header=0x88c3394, mode=DBUS_VALIDATION_MODE_DATA_IS_UNTRUSTED,
    validity=0xb76bcfd8, byte_order=108, fields_array_len=45, header_len=64, body_len=102, str=0x88be8dc, start=0,
    len=2048) at dbus-marshal-header.c:987
#9 0x00e2cd37 in load_message (loader=0x88be8d8) at dbus-message.c:3533
#10 _dbus_message_loader_queue_messages (loader=0x88be8d8) at dbus-message.c:3683
#11 0x00e35b69 in _dbus_transport_get_dispatch_status (transport=0x86d9798) at dbus-transport.c:1080
#12 0x00e35cda in _dbus_transport_queue_messages (transport=0x86d9798) at dbus-transport.c:1107
#13 0x00e1fe27 in _dbus_connection_get_dispatch_status_unlocked (connection=0x88c32d8) at dbus-connection.c:4021
#14 0x00e212a0 in check_for_reply_and_update_dispatch_unlocked (connection=0x88c32d8, pending=<value optimized out>)
    at dbus-connection.c:2250
#15 0x00e22262 in _dbus_connection_block_pending_call (pending=0x88c06d0) at dbus-connection.c:2356
#16 0x00e2fa51 in dbus_pending_call_block (pending=0x88da83c) at dbus-pending-call.c:705
#17 0x00fcf2a4 in ?? () from /usr/lib/libgnome-keyring.so.0
#18 0x00fd87c5 in gnome_keyring_find_items_sync () from /usr/lib/libgnome-keyring.so.0
#19 0x003d36f4 in ?? () from /usr/lib/pymodules/python2.6/gtk-2.0/gnomekeyring.so
#20 0x080e0b13 in call_function (f=0x88ddc00, throwflag=0) at ../Python/ceval.c:3750
#21 PyEval_EvalFrameEx (f=0x88ddc00, throwflag=0) at ../Python/ceval.c:2412
#22 0x080e1dc4 in fast_function (f=0x88e0774, throwflag=0) at ../Python/ceval.c:3836
#23 call_function (f=0x88e0774, throwflag=0) at ../Python/ceval.c:3771
#24 PyEval_EvalFrameEx (f=0x88e0774, throwflag=0) at ../Python/ceval.c:2412
#25 0x080e1dc4 in fast_function (f=0xb6d0112c, throwflag=0) at ../Python/ceval.c:3836
#26 call_function (f=0xb6d0112c, throwflag=0) at ../Python/ceval.c:3771
#27 PyEval_EvalF...

#3 0x00e28a63 in _dbus_validate_signature_with_reason (type_str=0xb76bcc1c, type_pos=0, len=1)
    at dbus-marshal-validate.c:218
218 _DBUS_POINTER_TO_INT (_dbus_list_pop_last (&element_count_stack));
(gdb) l
213 if (*p != DBUS_TYPE_ARRAY &&
214 *p != DBUS_DICT_ENTRY_BEGIN_CHAR &&
215 *p != DBUS_STRUCT_BEGIN_CHAR)
216 {
217* element_count =
218 _DBUS_POINTER_TO_INT (_dbus_list_pop_last (&element_count_stack));
219
220 ++element_count;
221
222 if (!_dbus_list_append (&element_count_stack,
(gdb) print *p
$17 = 117 'u'

/** #DBUS_TYPE_UINT32 as a string literal instead of a int literal */
#define DBUS_TYPE_UINT32_AS_STRING "u"

#define _DBUS_POINTER_TO_INT(pointer) ((long)(pointer))

Running with "python -u", to unbuffer stdout and stderr, there's more:

*** glibc detected *** python: realloc(): invalid next size: 0x096ac1e8Segmentation fault (core dumped)

Oh wow, about 2 out of 8 times, it does *not* SEGV on my machine. On one of those, it gives an exception,

2010-03-19 10:33:15,154 - replication of services aborted
Traceback (most recent call last):
  File "/usr/lib/python2.6/dist-packages/desktopcouch/replication.py", line 144, in do_all_replication
    remote_oauth_data = oauth_info_for_service(sn)
  File "/usr/lib/python2.6/dist-packages/desktopcouch/replication.py", line 65, in oauth_info_for_service
    return getattr(mod, service_name).get_oauth_data()
  File "/usr/lib/python2.6/dist-packages/desktopcouch/replication_services/ubuntuone.py", line 35, in get_oauth_data
    'oauth-consumer-key': oauth_consumer_key})
IOError

and the other time, it runs properly.

Since the linked-list functions are unlikely to be the cause, I'm looking downward in the stack. In a fresh CD run, using "testdrive", there is no problem and all of desktopcouch packages are the same as here. Looking at dbus and gnome-keyring next.

New data point:

process 2980: arguments to dbus_message_unref() were incorrect, assertion "message->generation == _dbus_current_generation" failed in file dbus-message.c line 1396.
This is normally a bug in some application using the D-Bus library.
Segmentation fault (core dumped)

^Ccmiller@zippy:~/work/canonical/desktopcouch/segv$ PYTHONPATH=`pwd` bin/desktopcouch-service
2010-03-19 15:31:56,826 - Personal couchdb is already running at PID#2935.
python: malloc.c:3574: mremap_chunk: Assertion `((size + offset) & (mp_.pagesize-1)) == 0' failed.
Aborted (core dumped)

cmiller@zippy:~/work/canonical/desktopcouch/segv$ PYTHONPATH=`pwd` bin/desktopcouch-service
2010-03-19 15:41:24,526 - Personal couchdb is already running at PID#2935.
**
ERROR:gkr-operation.c:380:gkr_operation_block: code should not be reached

(gdb) bt
#0 0x003d7059 in link_before (list=0xbff5e66c, data=0x1) at dbus-list.c:121
#1 _dbus_list_prepend (list=0xbff5e66c, data=0x1) at dbus-list.c:277
#2 0x003d70b9 in _dbus_list_append (list=0xbff5e66c, data=0x1) at dbus-list.c:249
#3 0x003c5a75 in _dbus_validate_signature_with_reason (type_str=0xbff5e72c, type_pos=0, len=11)
    at dbus-marshal-validate.c:222
#4 0x003c5cec in validate_body_helper (reader=0xbff5e7c4, byte_order=<value optimized out>,
    walk_reader_to_end=0, p=0xb6c0563d "a{o(oayay)}", end=0xb6c05774 "", new_p=0xbff5e834)
    at dbus-marshal-validate.c:515
#5 0x003c6088 in validate_body_helper (reader=0xbff5e878, byte_order=<value optimized out>,
    walk_reader_to_end=1, p=0xb6c0563c "\va{o(oayay)}", end=0xb6c05774 "", new_p=0xbff5e904)
    at dbus-marshal-validate.c:594
#6 0x003c6022 in validate_body_helper (reader=0xbff5e980, byte_order=<value optimized out>,
    walk_reader_to_end=0, p=0xb6c05638 "\b\001g", end=0xb6c05774 "", new_p=0xbff5e9d4)
    at dbus-marshal-validate.c:623
#7 0x003c5fb7 in validate_body_helper (reader=<value optimized out>,
    byte_order=<value optimized out>, walk_reader_to_end=1, p=0xb6c05638 "\b\001g", end=0xb6c05774 "",
    new_p=0xbff5ea0c) at dbus-marshal-validate.c:477
#8 0x003c641a in _dbus_validate_body_with_reason (expected_signature=0x3e6ca8,
    expected_signature_start=0, byte_order=108, bytes_remaining=0xbff5eb18, value_str=0xb6c01dfc,
    value_pos=0, len=356) at dbus-marshal-validate.c:708
#9 0x003c21ff in _dbus_header_load (header=0x9a73bc4, mode=DBUS_VALIDATION_MODE_DATA_IS_UNTRUSTED,
    validity=0xbff5ebb8, byte_order=108, fields_array_len=61, header_len=80, body_len=276,
    str=0xb6c01dfc, start=0, len=356) at dbus-marshal-header.c:987
#10 0x003c9d37 in load_message (loader=0xb6c01df8) at dbus-message.c:3533
#11 _dbus_message_loader_queue_messages (loader=0xb6c01df8) at dbus-message.c:3683
#12 0x003d2b69 in _dbus_transport_get_dispatch_status (transport=0xb6c02c68) at dbus-transport.c:1080
#13 0x003d2cda in _dbus_transport_queue_messages (transport=0xb6c02c68) at dbus-transport.c:1107
#14 0x003d3cda in do_reading (transport=0xb6c02c68) at dbus-transport-socket.c:792
#15 0x003d4528 in socket_handle_watch (transport=0xb6c02c68, watch=0xb6c01dd0, flags=1)
    at dbus-transport-socket.c:873
#16 0x003d26c7 in _dbus_transport_handle_watch (transport=0xb6c02c68, watch=0xb6c01dd0, condition=1)
    at dbus-transport.c:865
#17 0x003be324 in _dbus_connection_handle_watch (watch=0xb6c01dd0, condition=1, data=0xb6c020f0)
    at dbus-connection.c:1440
#18 0x003d4cd9 in dbus_watch_handle (watch=0xb6c01dd0, flags=1) at dbus-watch.c:669
#19 0x00396239 in ?? () from /usr/lib/libgnome-keyring.so.0
#20 0x004c284b in g_io_unix_dispatch (source=0x0, callback=0xb6c020f0, user_data=0xb6c01fa0)
    at /build/buildd/glib2.0-2.23.5/glib/giounix.c:162
#21 0x00481395 in g_main_dispatch (context=0x983a1c0) at /build/buildd/glib2.0-2.23.5/glib/gmain.c:1960
#22 IA__g_main_context_dispatch (context=0x983a1c0) at /build/buildd/glib2.0-2.23.5/glib/gmain.c:2513
#23 0x00485088 in g_main_context_iterate (context=0x983a1c0, block=<value optimized out>, dispatch=1,
    self=0x96528a0) at /bui...

cmiller@zippy:~/work/canonical/desktopcouch/segv$ PYTHONPATH=`pwd` bin/desktopcouch-service
2010-03-19 17:37:40,904 - Personal couchdb is already running at PID#2935.
python: malloc.c:3096: sYSMALLOc: Assertion `(old_top == (((mbinptr) (((char *) &((av)->bins[((1) - 1) * 2])) - __builtin_offsetof (struct malloc_chunk, fd)))) && old_size == 0) || ((unsigned long) (old_size) >= (unsigned long)((((__builtin_offsetof (struct malloc_chunk, fd_nextsize))+((2 * (sizeof(size_t))) - 1)) & ~((2 * (sizeof(size_t))) - 1))) && ((old_top)->size & 0x1) && ((unsigned long)old_end & pagemask) == 0)' failed.
Segmentation fault (core dumped)

I begin to doubt these assertion failures will be meaningful.

Ugh. Okay, after much GDBing, recompiling of DBus packages, and excising of code, I have it narrowed down to desktopcouch/replication_services/ubuntuone.py , where we retrieve credentials we need to authenticate to the user's couchdb in the cloud.

My first idea will be to move that outside the subthread, which entails moving that call up about 4 levels, outside replication proper and into the service, where we might have to be quick so as not to block DBus server calls.

This occurred immediately after setting up my existing Ubuntu One account from the Me menu on Lucid Beta 1.

Ok So i have no desktop environment. I was able to access the internet by remembering that ctrl + alt + T = open terminal window. I was lucky enough to remember that typing firefox on the terminal window will make my browser open and it did. I am not at all a computer linux literate so please forgive me if this is not the correct place for the bug. So in other words i have no access to the applications by clicking on icons, just by using terminal which scares me because i do not know how to do updates for when the bug is fixed.

I think I have it avoided. I was trying to use a singleton to cache some data that should not change often (if ever at all), and in the module, I create an instance of the singleton class.

I don't know the cause yet, but that seems to be the cause. If I use module variables to cache the data, there are no crashes. I wish I knew *why*, but SEGVs in multithreaded Python programs are not very easy to debug.

post patches as of the timestamp. amd64 system. Crash occurs when starting gwibber via 'broadcast'.
Gwibber continues to function after the crash.

I got the same issue here! hope it's solved soon.

Same issue affecting me on 64bit. Glad that it's being looked into.

Ah, I woke up this morning with some remembrance echoing in my head about Gtk and not making UI objects in subthreads. This is a promising avenue, I think.

Chad,

Based on your last comment, is the proposed branch still the avenue you want to take or is this a new road?

Crashed with: [ 80.980681] desktopcouch-se[1829]: segfault at 0 ip 00007f5e7efe6831 sp 00007fff141510a0 error 4 in libdbus-1.so.3.4.0[7f5e7efc2000+3d000]

This is affecting me as well, but lucid-beta continues running after the crash. Thanks for looking into this.

\

same for me: crashing on system-startup... no obvious effect to systemstability

same here, it crashs on logon.
but no problem with stability or something.

This bug also affects 64bit Ubuntu 10.04
(Linux les-studio 2.6.32-17-generic #26-Ubuntu SMP Sat Mar 20 02:23:45 UTC 2010 x86_64 GNU/Linux)

This continues to happen to be on my 32 bit Beta 1 install. The crash occurs less than a minute after boot.

I don't know if it's related to this bug or not, but each time I login, I have no button on top of my windows, so I can't move windows, close windows.... each time, I have to set "Visual effect" to normal or "extra" to see the 3 small buttons on the left corner.
BTW, is it possible to have them on the right ?

The Visual effect is not saved, and at each login, it's set to "none".

There's a patch to libgnome-keyring0 2.29.4git20100322-0ubuntu3 (previous is libgnome-keyring0 2.29.4git20100322-0ubuntu2) that causes this no longer to crash, but now it eats all CPU.

See also Bug#530605 .

Yup, I get CPU eating also. Single CPU only though.

This bug was fixed in the package desktopcouch - 0.6.3-0ubuntu3

---------------
desktopcouch (0.6.3-0ubuntu3) lucid; urgency=low

  * debian/patches/2-expire_log_files
    - Make log files rotate away so they don't grow out of control.
  * debian/patches/4-lp_524566_gui_in_main_thread
    - Execute GUI operations in the main execution thread only. (LP: #524566)
 -- Ken VanDine <email address hidden> Wed, 31 Mar 2010 09:46:47 -0400

I've got this too, a crash on log-in. "desktopcoach-service" closed unexpectedly, but my desktop icons are all there and the system is stable!!!

I've just downloaded and installed all updates this morning and the desktopcouch service is still crashing on startup.