Comment 1 Andre Klapper 2008-10-09 11:15:00 UTC

Thanks for reporting.
You didn't set the hardware field - N810, N800?

Can you please run Filemanager in a terminal window by using the command
"maemo-summoner /usr/bin/ossofilemanager.launch > log 2>&1" and hit the Enter
key) and attach the resulting "log" file here after it crashed?

Having a core dump could be even more useful. It requires installing the
"sp-rich-core" package from the tools repository[1] and creating a "core-dumps"
directory to the memory card with at least few megabytes of free space. You can
get core dumps also without the rich-core package, but they are much less
useful (and without it pre-installed applications core dumps are then named
maemo-launcher-*).

[1] http://maemo.org/development/tools/

Created an attachment (id=963) [details]
log, per request from aklapper@openismus.com

Created an attachment (id=964) [details]
sp-rich-core dump, per request from aklapper@openismus.com

Kicked off the filemanager crash with the "maemo-summoner
/usr/bin/ossofilemanager.launch > log 2>&1" command.
In filemanager, navigated to an SMB share.

Set HW field to N810

Comment 5 Andre Klapper 2008-10-10 14:26:10 UTC

(In reply to comment #2)
> Created an attachment (id=963) [details] [details]
> log, per request from aklapper@openismus.com


I'm surprised - the log says for example
> Reading settings for file:///root/MyDocs

So you are trying to run the filemanager as root?! This directory doesn't exist
on my device. Using the default user instead of root is not possible
because...?

No reason.  filemanager bombs the same way when i run it as 'user'.

Comment 7 Andre Klapper 2008-10-10 16:02:29 UTC

Can you also provide that log? I hope it removes all that root noise in there.
Why did you run this as root at all?

Created an attachment (id=969) [details]
log, executed as user, per request from aklapper@openismus.com

executed as user

Created an attachment (id=970) [details]
sp-rich-core dump, executed as user, per request from aklapper@openismus.com

executed as user

Comment 10 Andre Klapper 2008-10-11 03:30:52 UTC

Thanks, that
   CALLBACK: fill-authentication!!!
looks somehow suspicious...

dunno.  the windows SMB share i am trying to access is open to "Everyone".  in
any case, the segment fault is not a very useful outcome.

Comment 12 Eero Tamminen 2008-10-13 16:59:00 UTC

(In reply to comment #9)
> Created an attachment (id=970) [details] [details]
> sp-rich-core dump, executed as user, per request from aklapper@openismus.com
> 
> executed as user

Thanks.  The backtrace is:
Signal 11 (SIGSEGV)
0     0x4009fcf0     _start () at rtld.c:611
1     0x4106fe68     ?? () from unknown
2     0x4106fe68     ?? () from unknown

I.e. it crashes in the system C-library dynamic loader which unfortunately
isn't very useful.  We've had crashes like this now and then (in modest,
browser or file manager), but only few times a year and never repeatable. 
Investigations about the dynamic linker code never produced any reason for this
and there's plenty of free RAM, shared memory file descriptors, disk, OK
looking WLAN connection etc i.e. those were not the reasons either.

So... Having a test-case that is easily reproducible by others would be useful,
otherwise we probably would have to mark this as worksforme (not reproducible
by us and not enough information to continue otherwise).  Andre, could you look
whether you can get this reproduced in your setup and what that requires?

Comment 13 Andre Klapper 2008-10-14 00:06:49 UTC

(In reply to comment #12)
> Andre, could you look
> whether you can get this reproduced in your setup and what that requires?

Hmm, I'm able to crash File manager repeatedly when accessing the Windows Vista
network share of my flatmate, but no "clear" steps to reproduce. :-/

1) Start File manager
2) Click on "Shared folders"
3) Click on the name of the network share
4) Expand subfolders in the left pane
5) Quickly click around on several folders until File manager crashes.

Tried this 4 times and always succeeded to crash it (sometimes I had to click
like 30 times).

gdb only says "SIGSEGV, Segmentation fault" in raise () from /lib/libc.so.6
Not helpful.

Attaching two of the coredumps.

Comment 14 Andre Klapper 2008-10-14 00:09:49 UTC

Created an attachment (id=973) [details]
coredump 1

Comment 15 Andre Klapper 2008-10-14 00:11:13 UTC

Created an attachment (id=974) [details]
coredump 2

Comment 16 Eero Tamminen 2008-10-14 15:18:25 UTC

Thanks, both core dumps have the same same backtrace, which is different from
the previously reported one, but maybe they're somehow related:
Signal 11 (SIGSEGV)
0     0x400d7cf0     strcmp () from /lib/libc.so.6
1     0x41517e68     is_desktop_file (vfs_info=0x41525258) at
gtkfilesystemgnomevfs.c:921
2     0x41519e2c     info_from_vfs_info (system_vfs=0xa7150, uri=0x1e1228,
vfs_info=0x21c3a8, types=GTK_FILE_INFO_ALL, error=0x41525344) at
gtkfilesystemgnomevfs.c:3012
3     0x4151a238     gtk_file_folder_gnome_vfs_get_info (folder=0x20fea0,
path=0x1e1228, error=0xbea9ccac) at gtkfilesystemgnomevfs.c:2921
4     0x40656cb0     IA__gtk_file_folder_get_info (folder=0x4ea50,
path=0x1e1228, error=0xbea9ccac) at gtkfilesystem.c:1166

Comment 17 Andre Klapper 2008-10-14 15:50:43 UTC

(In reply to comment #16)
> Thanks, both core dumps have the same same backtrace, which is different from
> the previously reported one, but maybe they're somehow related:
> Signal 11 (SIGSEGV)
> 0       0x400d7cf0      strcmp () from /lib/libc.so.6
> 1       0x41517e68      is_desktop_file (vfs_info=0x41525258) at
> gtkfilesystemgnomevfs.c:921
> 2       0x41519e2c      info_from_vfs_info (system_vfs=0xa7150, uri=0x1e1228,
> vfs_info=0x21c3a8, types=GTK_FILE_INFO_ALL, error=0x41525344) at
> gtkfilesystemgnomevfs.c:3012
> 3       0x4151a238      gtk_file_folder_gnome_vfs_get_info (folder=0x20fea0,
> path=0x1e1228, error=0xbea9ccac) at gtkfilesystemgnomevfs.c:2921
> 4       0x40656cb0      IA__gtk_file_folder_get_info (folder=0x4ea50,
> path=0x1e1228, error=0xbea9ccac) at gtkfilesystem.c:1166

My one looks a bit similar to
http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=498212 .

Comment 18 Andre Klapper 2008-11-07 13:17:23 UTC

"My" crasher (comment 14 and on) has been fixed in package osso-gnomevfs-extra
1.9.6-2 which is part of the internal build version diablo build x.2008.45
(Note that 2008 is the year and the number after is the week.)

bjhellstrom, I do not know if this might also fix or improve the issues that
you run into...
Would be nice if you could check again after the next public update has been
released with or after this build version (x.2008.45-x).

Comment 19 Quim Gil 2008-12-23 15:27:26 UTC

Any news with the latest Diablo update?

fwiw Samba support is being dropped in Fremantle...

Comment 20 Andre Klapper 2008-12-23 16:00:14 UTC

> Any news with the latest Diablo update?

Problem is that the coredump here is "one of the mystery backtraces":
Signal 11 (SIGSEGV)
0       0x4009fcf0      _start () at rtld.c:611
1       0x4106fe68      ?? () from unknown
2       0x4106fe68      ?? () from unknown

...and if the fix for "my" crash also fixes this original one cannot be tested
as the fix was included in week 45 and is hence not yet available for public in
the last SSU (week 43).

Putting this into moreinfo mode.

@bjhellstrom: Can you please test this when the next official update is
available?

Now running version 5.2008.43-7.  Installed something that boinked my all my
USB drivers and had to send the whole damn thing back to Nokia.  Just got it
back.

Filemanager doesnt see any public shares at all so there is nothing to try to
open.

(In reply to comment #20)

> @bjhellstrom: Can you please test this when the next official update is
> available?
> 
I am seeing a similar problem on my N810.  In File Manager, under Shared
folders, I can see the shared folder names of the two machines running XP.  But
if I select the machine that is running Windows ME, File Manager displays its
shared folder names, pauses a second and evaporates.  This is 100% repeatable
so far.

I'm running 5.2008.43-7.  Is there a later version I should try?

I will upload a log file.  I couldn't find a link to install sp-rich-core, but
if given one will try to create a memory dump.

Created an attachment (id=1130) [details]
Log file per Comment #22

Comment 24 Andre Klapper 2009-03-11 16:07:24 UTC

Even though it's a crasher issue, Nokia currently will not put manpower on
fixing samba issues in Diablo, especially as samba support will be dropped for
Fremantle.
Closing this as WONTFIX as it's more honest than just leaving it open and
ignoring it.

(Yes, 5.2008.43-7 is the latest available version.)

*** Bug 4869 has been marked as a duplicate of this bug. ***

(In reply to comment #18)
> "My" crasher (comment 14 and on) has been fixed in package osso-gnomevfs-extra
> 1.9.6-2 which is part of the internal build version diablo build x.2008.45

Andre, can you confirm that osso-gnomevfs-extra is the right package that fixed
this in the internal build?  It seems to only contain OBEX VFS, but if it does
fix the SMB crash I wonder if it's worth trying to build and test 1.9.6-3 (from
Fremantle beta2).

Comment 27 Andre Klapper 2009-08-03 14:04:55 UTC

(In reply to comment #26)
> Andre, can you confirm that osso-gnomevfs-extra is the right package that fixed
> this in the internal build?  It seems to only contain OBEX VFS, but if it does
> fix the SMB crash I wonder if it's worth trying to build and test 1.9.6-3 (from
> Fremantle beta2).

I'm not sure. Hopefully the following stuff is a bit helpful:

For the initial issue reported here see comment 12.

Regrading comments 13 to 15: This crash happens both in SMB and BT in Diablo.

Debian papered over this just by checking whether
gnome_vfs_file_info_get_mime_type() returns NULL (which it shouldn't), see
http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=498212 .

This will not be fixed properly in GNOME in
http://bugzilla.gnome.org/show_bug.cgi?id=512555 - GnomeVFS is dead, in favor
of GVFS.

Created an attachment (id=1289) [details]
osso-gnomevfs-extra package with Fremantle changes backported

(In reply to comment #27)
> I'm not sure.

Oh well.  The diff is minimal, and according to the changelog 1.9.6-3 /is/
1.9.6-2 with just a version bump so it seems worth a shot anyway.  I can't see
SMB shares in my filemanager but if anyone wants to try the attachment and
report back on whether it changes anything it'd be great.

The only real changelog entry references a different internal bug and seems
truncated though:

+osso-gnomevfs-extra (1.9.6-2) unstable; urgency=low
+
+  * F i x e s : NB#88054, File Manager crashes when user tries to browse
through
+

> Hopefully the following stuff is a bit helpful:
> 
> Regrading comments 13 to 15: This crash happens both in SMB and BT in Diablo.

BT works without crashing here...

Marking patches of interest to Diablo (Maemo4) community updates, please excuse
the noise.