On Wed, Mar 04, 2020 at 09:00:31PM +0000, Al Viro wrote: > On Wed, Mar 04, 2020 at 09:59:46PM +1100, Aleksa Sarai wrote: > > > > FWIW, I'm putting together some litmus tests for pathwalk semantics - > > > one of the things I'd like to discuss at LSF; quite a few codepaths > > > are simply not touched by anything in xfstests. > > > > I won't be at LSF unfortunately, but this is something I would be very > > interested in helping with -- one of the things I've noticed is the lack > > of a test-suite for some of the more generic VFS bits (such as namei). > > BTW, has anyone tried to run tests with oprofile and see how much of the > core kernel gets exercised? That looks like an obvious thing to try - > at least the places outside of spin_lock_irq() ought to get lit after > a while... > > Have any CI folks tried doing that, or am I missing some obvious reason > why that is not feasible? I don't know about oprofile, but LTP got their gcov patches merged into 2.6.31: http://ltp.sourceforge.net/coverage/gcov.php
