I've just upgraded to GNOME 2.6, GTK+2.4, and GAIM 0.76 and suddenly now I can't chat with anyone. I'm using the MSN plugin; I haven't tried other protocols yet. I know there are a lot of similar bugs around but none had the same crash address or type, so I thought this looked like a new one, or maybe it's just an alpha-specific variant of one of the others. Reproducible: Always Steps to Reproduce: 1. Log in to MSN using GAIM 2. Initiate a chat with someone 3. Receive an IM from the other person Actual Results: Program received signal SIGFPE, Arithmetic exception. 0x00000200007a0760 in g_param_spec_override () from /usr/lib/libgobject-2.0.so.0 Expected Results: Not crash! (gdb) bt #0 0x00000200007a0760 in g_param_spec_override () from /usr/lib/libgobject-2.0.so.0 #1 0x000002000079b484 in g_param_value_validate () from /usr/lib/libgobject-2.0.so.0 #2 0x00000200007945d8 in g_object_set_valist () from /usr/lib/libgobject-2.0.so.0 #3 0x000002000038cef0 in gtk_text_buffer_create_tag () from /usr/lib/libgtk-x11-2.0.so.0 #4 0x00000001200c5fc0 in insert_cb (buffer=0x120247e08, iter=0x11fff2f10, text=0x3c <Address 0x3c out of bounds>, len=1, imhtml=0x12042cd70) at gtkimhtml.c:2551 #5 0x00000200002ed588 in _gtk_marshal_VOID__BOXED_STRING_INT () from /usr/lib/libgtk-x11-2.0.so.0 #6 0x000002000078fc98 in g_closure_invoke () from /usr/lib/libgobject-2.0.so.0 #7 0x00000200007aa6d8 in g_signal_emit_by_name () from /usr/lib/libgobject-2.0.so.0 #8 0x00000200007a93a0 in g_signal_emit_valist () from /usr/lib/libgobject-2.0.so.0 #9 0x00000200007a9844 in g_signal_emit () from /usr/lib/libgobject-2.0.so.0 #10 0x00000200003882b4 in gtk_text_buffer_set_text () from /usr/lib/libgtk-x11-2.0.so.0 #11 0x00000001200c3d90 in gtk_imhtml_append_text_with_images (imhtml=0x12042cd70, text=0x0, options=0, images=0x0) at gtkimhtml.c:2010 #12 0x00000001200ae118 in gaim_gtkconv_write_conv (conv=0x10, who=0x1204b1d20 "�,K \001", message=0x1204b18a0 "<FONT FACE=\"Modern No. 20\"><FONT COLOR=\"#000080\">asdasd</FONT></FONT>", flags=537940514, mtime=1081547673) at gtkconv.c:4627 #13 0x0000000120038d5c in gaim_conversation_write (conv=0x120408ab0, who=0x1203f8bd0 "a_catania", message=0x1204ae590 "<FONT FACE=\"Modern No. 20\"><FONT COLOR=\"#000080\">asdasd</FONT></FONT>", flags=GAIM_MESSAGE_RECV, mtime=1081547673) at conversation.c:1490 #14 0x00000001200ad8ec in gaim_gtkconv_write_im (conv=0x120408ab0, who=0x0, message=0x1204ae590 "<FONT FACE=\"Modern No. 20\"><FONT COLOR=\"#000080\">asdasd</FONT></FONT>", flags=GAIM_MESSAGE_RECV, mtime=1081547673) at gtkconv.c:4401 #15 0x0000000120039548 in gaim_conv_im_write (im=0x60000, who=0x200007e7b28 "", message=0xf <Address 0xf out of bounds>, flags=GAIM_MESSAGE_SEND, mtime=2199031515600) at conversation.c:1712 #16 0x000000012005f318 in serv_got_im (gc=0x1203a2410, who=0x0, msg=0x0, imflags=0, mtime=1081547673) at server.c:1083 #17 0x0000020001324900 in plain_msg (servconn=0x120479760, msg=0x1204adf90) at switchboard.c:373 #18 0x0000020001321f40 in process_multi_line (servconn=0x120479760, buffer=0x1204adf90 "\001") at servconn.c:49 #19 0x00000200013215f4 in msn_servconn_parse_data (data=0x120479760, source=20079112, cond=541774053) at servconn.c:605 #20 0x00000001200bd078 in gaim_gtk_io_invoke (source=0x120247e08, condition=536815072, data=0x3c) at gtkeventloop.c:61 #21 0x0000020000870e24 in g_io_channel_unix_get_fd () from /usr/lib/libglib-2.0.so.0 #22 0x0000020000840900 in g_idle_remove_by_data () from /usr/lib/libglib-2.0.so.0 #23 0x000002000083d968 in g_main_context_dispatch () from /usr/lib/libglib-2.0.so.0 #24 0x0000020000840fa4 in g_idle_remove_by_data () from /usr/lib/libglib-2.0.so.0 #25 0x000002000083e234 in g_main_loop_run () from /usr/lib/libglib-2.0.so.0 #26 0x00000200002e68ac in gtk_main () from /usr/lib/libgtk-x11-2.0.so.0 #27 0x00000001200fca18 in main (argc=1, argv=0x11ffff8c8) at main.c:907
This happens to me with -r2, but only if the other person attempts to send an encrypted message (Yahoo protocol). Gaim-encryption 2.23 supposedly fixed the issues with encryption (though perhaps they introduced new ones inadvertently).
throw_away_2002@yahoo.com: Yes I've noticed some gaim-encryption crashing and have stopped using it. Allen are you using gaim-encryption?
Well, I was using it before but when I started getting this problem I deactivated it. That didn't make any difference in terms of gaim crashing. I'll try unmerging it and I'll let you know if that makes a difference. I've also just rebuilt gtk+ so maybe that will fix something.
This looks like it might be upstream. Can you ignore font faces and try to duplicte the problem? 22:56 < LSchiere> rizzo: does he have "Modern No. 20" ? 22:56 < rizzo> LSchiere: hmm... if buddy uses a font he doesn't have, it might crash? 22:57 < LSchiere> rizzo: its not supposed to, but i've seen reports of it doing so. its a pango bug 22:57 < LSchiere> pango does very wierd things when it can't find a font 22:57 < khc> I think it crashes if gtk cannot find a fallback font. 22:58 < LSchiere> and since gtk uses pango for that, i blame pango
Perhaps try this as well: 23:02 < khc> rizzo: I've seen people report that running fc-cache as root fixes it.
Ok. I've tried unmerging gaim-encryption and I still get the crash. I also rebuilt gtk+ and gaim with the latest ebuild (0.76-r2 I think) and I still get the crash. How do I ignore font-faces exactly? I checked my Gaim font preferences and "Use custom font" is checked. Is that what you are talking about?
I tried setting all the font options including font, size, and color and that doesn't fix the crash.
I also tried changing the font that the other party to the conversation uses to one that I have on my Linux box. I changed the font to terminal and still got the crash.
Tools -> Preferences Interface -> Message Text You want to [x] Ignore Font Faces. It sounds like you've found it but I'm stating just to be sure. When it crashes now do you get the same backtrace?
Oh Ok. I hadn't tried that option before. But I still get the same crash, same backtrace. So do you think the problem is with the null pointer in the call to gtk_imhtml_append_text_with_images ? Or is it in the libraries that are called by that call? What libraries are those -- gtk and glib?? I can try building those as debug if we think that would be helpful.
I also tried running fc-cache as root and I still get the crash. I did some searching around and found this: http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=203722 where people have reported frequent SIGFPE's with alpha machines. It says gcc has a patch to fix this problem, but that you can try compiling with -mieee to fix it. If that doesn't work maybe I'll try changing my optimization flags. Here are my current CFLAGS: CFLAGS="-mcpu=ev56 -O3 -pipe"
Um... well neither adding -mieee or getting rid of my optimization flags made any difference. I'm totally out of ideas now.
While it doesn't look protocol related to me, does it only happen with MSN? Do you have other protocol accounts?
I have an AIM account too but I don't have a way of testing it. (My wife has a MSN account so I can just test from hers...) Maybe I can try opening up a new account.. I did just confirm that this is definitely 0.76 related --- I rolled back to 0.75-r10 and now I can chat again. So either it's gaim 0.76 or perhaps the different sets of patches that are included with the 0.75 vs. 0.76 ebuilds?
My AIM id is "dtseiler". Feel free to try and catch me if I'm online. Although I probably won't be until Monday again.
I'm not aware of any changes that address your issue, but please see if this still happens in gaim-0.77.
It does appear to be fixed now. Although just after I was able to chat successfully I did something and got this: Program received signal SIGSEGV, Segmentation fault. 0x00000200008384f0 in g_list_remove () from /usr/lib/libglib-2.0.so.0 (gdb) bt #0 0x00000200008384f0 in g_list_remove () from /usr/lib/libglib-2.0.so.0 #1 0x000002000130ac94 in msn_servconn_destroy (servconn=0x1204e99c0) at servconn.c:206 #2 0x000002000130b764 in read_cb (data=0x60, source=542022080, cond=19993411) at servconn.c:620 #3 0x00000001200c24c8 in gaim_gtk_io_invoke (source=0x60, condition=542022080, data=0x20001311343) at gtkeventloop.c:61 #4 0x0000020000870e24 in g_io_channel_unix_get_fd () from /usr/lib/libglib-2.0.so.0 #5 0x0000020000840900 in g_idle_remove_by_data () from /usr/lib/libglib-2.0.so.0 #6 0x000002000083d968 in g_main_context_dispatch () from /usr/lib/libglib-2.0.so.0 #7 0x0000020000840fa4 in g_idle_remove_by_data () from /usr/lib/libglib-2.0.so.0 #8 0x000002000083e234 in g_main_loop_run () from /usr/lib/libglib-2.0.so.0 #9 0x00000200002e68ac in gtk_main () from /usr/lib/libgtk-x11-2.0.so.0 #10 0x0000000120106718 in main (argc=1, argv=0x11ffff8b8) at main.c:911 (gdb) quit Not sure if I can reproduce that...
Exactly when does this error occur? During a chat? I'm not sure what you mean by "just after". When you close the convo window/tab?
OK I got confirmation from a gaim patch writer that this is upstream in gaim with MSN chats/conferences. I don't see a bug in gaim's tracker. I'm told that the gaim developer in charge of MSN is aware, but I'm going to ask you to file a new bug with them for posterity.
You mean file a bug about this other new crash I just mentioned? I got busy with other things for a while and forgot about this bug...
Yes file a bug on the last crash in MSN chat closing. As I said it is already known and may have already been fixed, I would be sure to search their bug tracker again before doing so.
I have confirmation that this is already fixed in CVS upstream. I'll try to get a patch. no need to file your bug.
Added a patch to gaim-0.77-r1 to fix this. rsync after midnight CST and you should be good to grow.
