[ sorry I've been out of town... ] On Mon, Nov 14, 2011 at 3:39 PM, Tejun Heo <tj@xxxxxxxxxx> wrote: > Hello, Guennadi. > > On Tue, Nov 15, 2011 at 12:33:29AM +0100, Guennadi Liakhovetski wrote: >> No, this isn't a correct fix. > > I've been looking at the code and am scratching my head about what > happens after timeout. If test timed out (tmo == 0), it prints out > error message and continues to the next iteration of the test loop, > which will reinitialize the on-stack completion. This essentially > makes the previous test run's callback_param pointer dangling. The > completion needs to be either detached from the callback or waited > upon even if it timed out. Am I missing something? If the completion times out then it is almost certainly a hardware or driver bug. There is no facility to get a dma driver to forget a queued operation. The test should probably abort at that point, but it isn't intended to be a recoverable condition. -- Dan -- To unsubscribe from this list: send the line "unsubscribe linux-next" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
