* The failtest module overrides various standard functions, and forks
* your unit test at those points to test failure paths. The failing
* child are expected to fail (eg. when malloc fails), but should not
- * leak memory or crash.
+ * leak memory or crash. After including failtest_override.h, you can
+ * include failtest_restore.h to return to non-failing versions.
*
* The unit test is a normal CCAN tap-style test, except it should
* start by calling failtest_init() and end by calling