> On Thu, Apr 10, 2014 at 01:52:09AM +0900, Masatake YAMATO wrote: >> In unwind part of tcb finalization, tcp->s_ent >> was accessed to decide whether pending stacktrace in tcp->queue >> should be printed or not. >> >> However, s_ent can be NULL if invalid pid is given to strace >> as an trace target. Here invalid pid means that the pid is >> not associated with any process. > > tcp->s_ent can be NULL iff either trace_syscall_entering hasn't been > called for this tcp yet, or the process disappeared before the first > trace_syscall_entering call for this tcp. > > Another approach to fix this bug is to move unwind_tcb_init call from > alloctcb to the point where its result is going to be used for the first > time (unwind_stacktrace_capture and unwind_stacktrace_print).
In the new patch tcp->s_ent is never referred in finalization stage. So the this issue is gone now. [PATCH v4 08/37] unwind: introduce queue_t for capturing stacktrace Thanks. Masatake YAMATO > > -- > ldv ------------------------------------------------------------------------------ Learn Graph Databases - Download FREE O'Reilly Book "Graph Databases" is the definitive new guide to graph databases and their applications. Written by three acclaimed leaders in the field, this first edition is now available. Download your free book today! http://p.sf.net/sfu/NeoTech _______________________________________________ Strace-devel mailing list [email protected] https://lists.sourceforge.net/lists/listinfo/strace-devel
