Hi, 2016-04-05 14:41 GMT+01:00 Axel Beckert <a...@debian.org>: > Hi Manuel, > > Manuel A. Fernandez Montecelo wrote: >> I was trying to say that the functions whose name appear >> (vsnprintf_chk and the other _chk) are not part of >> aptitude/apt/cwidget, and other of the functions in the backtrace >> don't show names, so if you have of you have dbgsym of aptitude >> installed (as the message of "loading symbols" indicate), it's not >> part of aptitude. >> >> If you install libcwidget*-dbg, libapt*-dbg and any other -dbg of the >> "aptitude software stack" (sigc++, boost, libc6...) maybe something >> appears in that backtrace. > > aptitude-dbgsym and libc6-dbgsym was already installed. I've installed > libcwidget*-dbg, libapt*-dbg, libsigc++-*-dbgsym and libboost-dbg > today. > > But now I can't reproduce the segfault anymore, at least on amd64. And > the backtrace from the existing segfaults hasn't changed much -- if at > all -- since then: > > (gdb) bt > #0 0x00007fe2861e5973 in ?? () > #1 0x0000ffffffff0000 in ?? () > #2 0x0000000100001839 in ?? () > #3 0x0000000008000000 in ?? () > #4 0x00007fe287fa8b0c in ___vsprintf_chk (s=0x7ffd08eb4380 "", > flags=-1416311776, slen=140724753089664, format=0x564aab94cc10 > "\260R\266\252JV", > args=0x564aa764dc78, args@entry=0x7ffd08eb44c8) at vsprintf_chk.c:85 > #5 0x00007fe287fa8a5d in ___sprintf_chk (s=<optimized out>, flags=<optimized > out>, slen=<optimized out>, format=<optimized out>) at sprintf_chk.c:31 > #6 0x0000564aa764dc78 in ?? () > #7 0x0000564aab94cc20 in ?? () > #8 0x00007fe289d335d4 in cwidget::toplevel::eventq () from > /usr/lib/x86_64-linux-gnu/libcwidget.so.3 > #9 0x0000000000000080 in ?? () > #10 0x00007ffd08eb4b20 in ?? () > #11 0x0000564aab94cc10 in ?? () > #12 0x000000000000000d in ?? () > #13 0xfffffffffffffffc in ?? () > #14 0x00007fe288af204f in pthread_cond_wait@@GLIBC_2.3.2 () at > ../sysdeps/unix/sysv/linux/x86_64/pthread_cond_wait.S:183 > #15 0x0000000000000000 in ?? () > (gdb) > > Will try the same at home on the Raspberry Pi's console again, too.
Seeing that cwidget's events are involved, with the hairiness of cwidget's own threads implementation and sigc++ mixed, and seeing that libsigc++-2.0 was updated recently, I am wondering if it has something to do with that. I have the same versions, though, so I should be able to reproduce it if it's happening often -- altough this kind of weirdness is typical of issues involving threads. I very rarely use Linux virtual consoles and aptitude. I don't know if you use them enough to be able to tell if it's a behaviour that only started recently while not happening for years, or simply you don't use it often enough to see if it was only present since recently? Cheers. -- Manuel A. Fernandez Montecelo <manuel.montez...@gmail.com>
