Software crashes are severe manifestations of software faults. Especially, software crashes in production software usually result in bad user experiences. Therefore, crashing faults mostly are required to be fixed with a high priority. Diagnosing crashing faults on production software is non-trivial, due to the characteristics of production environment. In general, it is required to address two major challenges. First, crash reports in production software are usually numerous, since production software is used by a large number of end users in various environments and configurations. Especially, a single fault may manifest as different crash reports, which makes the prioritizing debugging and understanding faults difficult. Second, deployed software is required to run with minimal overhead and cannot afford a heavyweight instrumentation approach to collect program execution information. Furthermore, end users require that the logged information should not reveal sensitive production data. This thesis contributes for developing crashing fault diagnosis tools that can be used in production environment.