The completeness of current test cases should also be analyzed and improved upon if necessary. We are currently looking at code coverage analysis tools to determine how much of the target kernel code is being executed by test cases in the LTP. As we find areas of kernel code that are not adequately covered by test cases in the LTP, test cases are written or modified to expand coverage to these areas.
Additional tests are of course critical to the test suite, but for the Linux Test Project to truly be effective, people must use it. It would be nice to see the LTP test suite run as part of the exit criteria for releasing new kernels in the stable and development trees. In addition to this, it would be useful for kernel developers to execute the test suite against patches before submitting them. The LTP test suite will not find all problems but it could reduce the number of errors in new code if used properly. If kernel developers and testers diligently submit tests for defects as they are found, the test suite could even help reduce the number of regressed defects found in Linux.