On Fri, 10 Jul 2015 17:06:59 +0200 William Dauchy <william@xxxxxxxxx> wrote: > On Jul10 10:39, Jeff Layton wrote: > > So if a file happened to have some flock locks on it, then we could > > be taking a new reference to a file that has already had its refcount > > go to zero. > > do you have any test case which may confirm that? Not right offhand, at least not with a stock kernel. I was planning to turn that get_file call into an atomic_long_inc_return and then test the return value and WARN if it was 1. Then, set a flock lock on a file, and close it without explicitly unlocking. I suspect that would make the WARN pop. > in the commit messages, you also were refering to some tests; I wonder > if you could share it. Jean? Are those tests something you could share? -- Jeff Layton <jlayton@xxxxxxxxxxxxxxx>
Attachment:
pgplRZLGFe28H.pgp
Description: OpenPGP digital signature
