[fprint] Strange Crash in __pthread_mutex_unlock_usercnt

Bastien Nocera hadess at hadess.net
Thu Oct 4 05:16:55 PDT 2012


Hey,

On Wed, 2012-10-03 at 16:01 -0700, Keaton Mowery wrote:
> Hi all,
> 
> I'm seeing a very odd crash while using libfprint's async API. Here's
> a backtrace:
> 
> ---------
> Program received signal SIGSEGV, Segmentation fault.
> 0xb4e14211 in __pthread_mutex_unlock_usercnt () from
> /lib/i686/cmov/libpthread.so.0
> (gdb) bt
> #0  0xb4e14211 in __pthread_mutex_unlock_usercnt () from
> /lib/i686/cmov/libpthread.so.0
> #1  0xb4efd016 in pthread_mutex_unlock () from /lib/i686/cmov/libc.so.6
> #2  0xb4e04f0e in ?? () from /lib/libusb-1.0.so.0
> #3  0xb4e02e65 in ?? () from /lib/libusb-1.0.so.0
> #4  0xb4e03641 in libusb_handle_events_timeout () from /lib/libusb-1.0.so.0
> #5  0xb5137a7f in fp_handle_events_timeout (timeout=0xbffff5e4) at poll.c:231
> #6  0x08052800 in main (argc=1, argv=0xbffff6d4) at fpserv.cpp:27
> ---------
> 
> Something also sometimes dumps to dmesg:
> [   54.740378] fpserv[1714]: segfault at 9f2ac60 ip b453d211 sp
> bfee078c error 4 in libpthread-2.11.2.so[b4534000+15000]
> 
> We have a microsoft fingerprint reader, and have 156 enrolled fingerprints.
> 
> Any thoughts?

You should install the debug info, and then check with newer versions of
libusb (such as libusbx) or run your program under valgrind.

It's very likely you're doing something wrong in your program if the
problem occurs every time, seeing as that same model of hardware works
for others (including myself) using fprintd.

Cheers



More information about the fprint mailing list