Comment 1 by Nikolaus Schaller, Oct 14, 2013
This was never observed in 2.6.32 or 3.7 kernels.
Comment 2 by Nikolaus Schaller, Nov 8, 2013
Here another sequence: 94294 393 94299 393 94293 393 94292 393 94289 393 94293 642 100710 393 94293 642 100710 642 100707 393 94296 642 100711 393 94292 393 94285 642 100701 393 94297 393 94295 393 94292 393 94291 642 100695 393 94296 393 94291 642 100690 393 94285 393 94289 393 94292 642 100693 393 94289 393 94299 642 100700 393 94290 393 94287 393 94294 393 94295 393 94290 642 100691 393 94291 393 94290 393 94297 642 100693 393 94283 393 94300 642 100697 393 94298 393 94295 393 94291 393
Comment 3 by Nikolaus Schaller, Nov 9, 2013
Fixed with commit d8c264c54d53f535c2d2071753417bbedb590511 What was the problem? The assumption that wait_for_completion_or_timeout guarantees a minimum timeout. Rather the timeout parameter specifies the number of jiffies (time ticks) to wait. For short delays (less than 1/HZ) this can result in no wait at all. So we have to add 1 to make sure that there is a minimum delay of 1/HZ.
Status:
Fixed
Sign in to reply to this comment.
Reported by Nikolaus Schaller, Oct 14, 2013