crash in boost/date_time/microsec_time_clock.hpp:70 v1.55
Hi, I am having crash in in local_time () at boost155_lin32_gcc480//include/boost/date_time/microsec_time_clock.hpp:70 I can call this in isolation and there is no issues, but in the production calls it crashes 100% reproducible with Program terminated with signal SIGABRT, Aborted. The only difference I could think about it is that the call is made from a dynamically loaded library. Is there any known issue about this? Thanks, George
On 20/05/2015 07:20, George Georgiev wrote:
I am having crash in
in local_time () at boost155_lin32_gcc480//include/boost/date_time/microsec_time_clock.hpp:70
I can call this in isolation and there is no issues, but in the production calls it crashes 100% reproducible with Program terminated with signal SIGABRT, Aborted.
The only difference I could think about it is that the call is made from a dynamically loaded library. Is there any known issue about this?
Are you compiling both to expect a shared libc, using the same compiler version on the same system? If not, do that.
On Tue, May 19, 2015 at 10:20 PM, George Georgiev
Hi,
I am having crash in
in local_time () at boost155_lin32_gcc480//include/boost/date_time/microsec_time_clock.hpp:70
I can call this in isolation and there is no issues, but in the production calls it crashes 100% reproducible with Program terminated with signal SIGABRT, Aborted.
The only difference I could think about it is that the call is made from a dynamically loaded library. Is there any known issue about this?
SIGABRT typically means an assertion failure or some fatal error in the runtime so that it calls abort(). I don't see any asserts in Boost code in these lines, and the runtime typically prints an error message to stderr before crashing, so you may want to look at the console output. Also, I assume you don't LD_PRELOAD anything that replaces the standard C functions.
On May 19, 2015 3:20:07 PM EDT, George Georgiev
Hi,
I am having crash in
in local_time () at boost155_lin32_gcc480//include/boost/date_time/microsec_time_clock.hpp:70
I can call this in isolation and there is no issues, but in the production calls it crashes 100% reproducible with Program terminated with signal SIGABRT, Aborted.
The only difference I could think about it is that the call is made from a dynamically loaded library. Is there any known issue about this?
Aside from the other information you've gotten to diagnose the problem, I'll ask one more question. Are there multiple threads involved? (I have no idea if there's a race in that code, but it's worth considering.) ___ Rob (Sent from my portable computation engine)
participants (4)
-
Andrey Semashev -
Gavin Lambert -
George Georgiev -
Rob Stewart