On 10/25/19 9:40 AM, Pavel Begunkov wrote: > On 25/10/2019 18:13, Jens Axboe wrote: >> On 10/25/19 3:48 AM, Pavel Begunkov (Silence) wrote: >>> From: Pavel Begunkov <asml.silence@xxxxxxxxx> >>> >>> 1. test user_data integrity with cancelled links >>> 2. test the whole link is cancelled by sq_thread >>> 3. hunging io_uring based on koverflow and kdropped >>> >>> Be aware, that this test may leave unkillable user process, or >>> unstopped actively polling kthread. >> >> That's fine, that's what the test suite is for! Thanks, applied. > > Just found it "a bit uncomfortable", that after several runs > my CPU was doing nothing but polling in vain. Looks like it's testing one of the cases that isn't fixed that. We shouldn't do that until a known fix is available, at least. Basically the tests should test for things that used to be a problem, but are now fixed. But let's just get it fixed :-) -- Jens Axboe
