Hi,
On Thu, Jul 28, 2011 at 07:23:34PM +0900, Nobuhiro Iwamatsu wrote:
> I checked on sid/amd64, I dont have problem. If you can, please send
> log of GDB and config file of bluez.
bluez config is just the default one - nothing changed.
I had to rebuild bluez and libcap-ng with debugging symbols enabled
- would be cool to have a bluez-dbg package, no?
gdb output is:
(gdb) run
Starting program: /usr/sbin/bluetoothd
[Thread debugging using libthread_db enabled]
Program received signal SIGSEGV, Segmentation fault.
0x4002d020 in init () at cap-ng.c:152
152 memset(&m.hdr, 0, sizeof(m.hdr));
(gdb) bt full
#0 0x4002d020 in init () at cap-ng.c:152
No locals.
#1 0x4002d234 in capng_clear (set=CAPNG_SELECT_BOTH) at cap-ng.c:176
No locals.
#2 0x401637e8 in main (argc=1, argv=0xbe96ed94) at src/main.c:377
context = 0x0
err = 0x0
sa = {__sigaction_handler = {sa_handler = 0, sa_sigaction = 0}, sa_mask
= {__val = {0, 0, 0, 0, 72, 0, 1074594976, 1075752960, 0, 1074594536,
1074057455, 1075752896, 1075752888, 0, 1075752912, 0, 0, 8,
1074432404, 0, 4131212846, 12, 1075640784, 1077110064, 1074429952,
1075752392, 129100401, 4294966728, 1077140016, 3197561888,
1075711004, 0}}, sa_flags = 1074694400, sa_restorer = 0x401ab134}
mtu = 0
config = 0x0
__FUNCTION__ = "main"
(gdb)
So I investigated further and I found out that this is indeed NOT a
regression in bluez as I initially thought but a regression in
libcap-ng.
Because with libcap-ng0 0.6.5-1+b1, bluez 4.94-2 as well as bluez 4.94-3
work fine but as soon as libcap-ng0 0.6.6-1 is installed I get the
segmentation fault from above.
Can you reproduce this with the newer version of libcap-ng0?
cheers, josch
--
To UNSUBSCRIBE, email to debian-bugs-dist-requ...@lists.debian.org
with a subject of "unsubscribe". Trouble? Contact listmas...@lists.debian.org
