On Fri, Jul 19, 2019 at 11:07:29PM +0000, Luis Chamberlain wrote: > On Fri, Jul 19, 2019 at 07:30:32PM +0000, Luis Chamberlain wrote: > > [mcgrof: fixes kz#204223 ] > > This patch can be ingored for now for stable. It does not actually > fix the issue, just delays it a bit. Once I stress test over 1000 > runs with some other fixes I have I'll send a new set of stable > fixes. generic/388 is one of the tests we expect to uncover interesting failures over time. i.e. every time we fix a problem in these tests, it will expose another issue that we haven't been able to exercise until easier-to-hit failures have been fixed. The best you can do right now is minimise the occurence of failures by backporting fixes - this test (like generic/475) will continue to uncover new shutdown and recovery issues as they are exposed by new fixes. Expecting it to pass 1000 times without failure on an older stable kernel is, IMO, a somewhat unrealistic expectation... Cheers, Dave. -- Dave Chinner david@xxxxxxxxxxxxx
