On 7/5/20 9:34 AM, Greg Kroah-Hartman wrote: > On Sun, Jul 05, 2020 at 03:41:48AM +0200, Heinrich Schuchardt wrote: >> On 7/4/20 4:02 PM, Greg Kroah-Hartman wrote: >>> Test the functionality of readfile(2) in various ways. >> >> Hello Greg, >> >> I expect readfile() to generate fanotify events FAN_OPEN_PERM, FAN_OPEN, >> FAN_ACCESS_PERM, FAN_ACCESS, FAN_CLOSE_NOWRITE in this sequence. > > Yes, it should, I don't think I do anything unique here when it comes to > vfs accesses that would go around those events. > >> Looking at patch 1/3 you took care of notifications. Would this deserve >> testing here? > > Possibly, do we have other in-tree tests of syscalls that validate those > events properly being created? There is an inotify test in tools/testing/selftests/cgroup/test_freezer.c There is no fanotify test in tree test. An fanotify test will require running with CAP_SYS_ADMIN. The kselftest documentation does not describe that tests should be run as root. So it may be preferable to test that the inotify events IN_OPEN, IN_ACCESS, IN_CLOSE_NOWRITE are created for readfile(). Example coding is included in the inotify.7 and fanotify.7 manpages. Best regards Heinrich
