GNOME Bugzilla – Bug 153405
Terminal crashes often while working with gnopernicus
Last modified: 2005-11-01 17:22:47 UTC
Installed Cinnabar_18 on intel HP P4 machine. gnopernicus version is 0.9.10 1.Run Gnopernicus,enable accessibility,logout and login again. 2.Start gnopernicus, wait till you hear the speech output. 3.Open gnome terminal and keep using other applications normally. Observation: After sometime speech stopped and terminal crashed.This happened frequently when I did a 'pstack' on a big core file. Below is the pstack output of the core that we got after terminal crash ------------------------------------------------------------------------- core 'core' of 1555: /usr/lib/at-spi-registryd --oaf-activate- iid=OAFIID:Accessibility_Regi d1c92e3c strlen (d21f0ae7, 8046ef0, 80469e0, 0) + c d1cee5d5 vsnprintf (8046a2c, 1, d21f0ad0, 8046ef0, d27f0818, d1c60018) + 75 d1f59c96 g_printf_string_upper_bound () + 1a ----> end of pstack output.
There is not enough information here to evaluate. Can you provide more precise intructions to reproduce the problem. The stack trace provided is for at-spi-registryd and I do not think that it is related to this problem. You can use coreadm to avoid overwriting core files. This may be the same as bug #152645.
Steps to Reproduce: I tried the same again on an intel machine and found that it was crashing when I did a 'ls' and tried to move the vertical scroll bar. This happened twice. Following is the stack trace which I found now core 'core' of 1330: /usr/bin/gnome-terminal ----------------- lwp# 1 / thread# 1 -------------------- d175276b memcpy (8145fe8, 80be01c, 80be030, 80be044, 216, 8340c1d0) + db ----------------- lwp# 2 / thread# 2 -------------------- d17d2615 _read (1c, ceffdf98, 14) + 15 d1854710 child_watch_helper_thread (d12b8400) + 1e d17d1510 _lwp_start (d12b8400, 0, 0, d18af78c, 0, 1)
I am unabel to reproduce this on Solaris Sparc.
I believe that this is reproducible on Solaris x86 machines.
Got stack trace below from attaching dbx to the process. =>[1] _waitid(0x1f67, 0x8046198, 0x0, 0x0), at 0xd17e3035 [2] _waitpid(0x1f67, 0x8046198, 0x0), at 0xd1794563 [3] waitpid(0x1f67, 0x8046198, 0x0), at 0xd17d720a [4] libgnomeui_segv_handle(0xb, 0x0, 0x8046258, 0xd265c7c2), at 0xd265c871 [5] call_user_handler(0xb, 0x0, 0x8046258), at 0xd17d7caf [6] sigacthandler(), at 0xd17d7e3b ---- called from signal handler with signal 11 (SIGSEGV) ------ [7] _private_memcpy(0x813ea04, 0x80b6e08, 0x80b6e1c), at 0xd176278b [8] vte_terminal_accessible_get_text(), at 0xd2726191 (dbx)
Created attachment 32717 [details] [review] Proposed patch
*** Bug 150239 has been marked as a duplicate of this bug. ***
Please review. This fixes crashes for our a11y users.
Apologies for spam-- ensuring Sun a11y team are cc'ed on all current a11y bugs. Filter on "SUN A11Y SPAM" to ignore.
Padraig: where are the inverted start/end offsets coming from? From gnopernicus?
vte_text_get_offset was being called with a start_offset of 1 and an end_offset of 0. I assume that the call originated in gnopernicus.
Please commit.
Fixed in CVS. Thanks Kjartan for reviewing!