26 Apr
2012
26 Apr
'12
4:37 p.m.
On Thu, 26 Apr 2012 11:26:08 -0500
"David C. Rankin" <drankinatty(a)suddenlinkmail.com> wrote:
On 04/26/2012 11:13 AM, David C. Rankin wrote:
In my glibc-2.11 binary the objdump output tells me that
___newselect_nocancel() always does a syscall (you can recognise them
by the "int $0x80" instruction), so you should be able to retrieve the
parameters with strace.
On 04/26/2012 08:01 AM, David C. Rankin wrote:
Tim,
This is much worse than we thought. It effects EVERYTHING! Even
simply trying to view a text file in konqueror file management (kate
preview) will lock konqueror. Here is that backtrace too:
(gdb) bt
#0 0xb777f416 in __kernel_vsyscall ()
#1 0xb65154ed in ___newselect_nocancel () from /lib/libc.so.6
#2 0xb7466f56 in KIO::SlaveBase::dispatchLoop (this=0xbfae15b8)
at /build/src/tdelibs/kio/kio/slavebase.cpp:286
#3 0xb7770fa6 in kdemain (argc=4, argv=0xbfae15b8) at
/build/src/tdelibs/kioslave/file/file.cc:128
#4 0x080503c1 in launch (argc=4, _name=0x8bc09b4 "kio_file",
args=0xbfae1784 "\300\216\367\266",
cwd=0x0, envc=0, envs=0x8bc0a2a "", reset_env=false, tty=0x0,
avoid_loops=false,
startup_id_str=0xfffffdfe <Address 0xfffffdfe out of bounds>,
startup_id_str@entry=0x4 <Address 0x4 out of bounds>) at
/build/src/tdelibs/kinit/kinit.cpp:673
#5 0x0805148c in handle_launcher_request (sock=<optimized out>,
sock@entry=-1) at /build/src/tdelibs/kinit/kinit.cpp:1240
#6 0x08051a0a in handle_requests (waitForPid=waitForPid@entry=0)
at /build/src/tdelibs/kinit/kinit.cpp:1443
#7 0x0804d775 in main (argc=5, argv=0xbfae1d54, envp=0xbfae1d6c)
at /build/src/tdelibs/kinit/kinit.cpp:1908
Bummer... Think about what other tests I can do that might
disclose exactly what it going south... I'll open a bug report shortly :)
Good Lord! Debug Packages are HUGE!!!
Nearly 600M of debugging symbols included in tqt3, tqtinterface,
tdelibs, tdebase. It looks like with will have to do this again on
x86_64, the i686 backtrace seems to stop early. However, it does
have line numbers for everything it caught:
(gdb) bt
#0 0xb610a959 in TQFontMetrics::charWidth (this=0x8fa3800, str=...,
pos=79) at kernel/qfont_x11.cpp:705
#1 0xb541a46b in width (tabWidth=8, italic=false, bold=false,
col=79, text=..., this=0x8fa37f0)
at /build/src/tdelibs/kate/part/katefont.h:76
#2 width (tabWidth=8, col=79, text=..., fs=..., this=<optimized
out>) at /build/src/tdelibs/kate/part/kateattribute.h:55
#3 KateRenderer::textWidth (this=0x933d3e8, textLine=...,
startcol=0, maxwidth=695,
needWrap=0xbf8e73a8, endX=0xbf8e739c) at
/build/src/tdelibs/kate/part/katerenderer.cpp:797
#4 0xb53f7d97 in KateViewInternal::range (this=0x9322170,
realLine=0, previous=0x0)
at /build/src/tdelibs/kate/part/kateviewinternal.cpp:1331
#5 0xb53f91ae in KateViewInternal::range (this=0x9322170,
realLine=0, viewLine=1)
at /build/src/tdelibs/kate/part/kateviewinternal.cpp:1418 #6
0x09322170 in ?? () Backtrace stopped: previous frame inner to this
frame (corrupt stack?)
I have opened a bug report and added this information to it:
http://bugs.trinitydesktop.org/show_bug.cgi?id=979
"select buffer paste in kate/kwrite causing line wrap causes crash
(X, qt3)"
Do you want me to begin a debug build on x86_64 as well? I can't
tell if this is enough of a backtrace to provide you with the
needed info. Let me know what else you need. Thanks!