On Thu, May 15, 2014 at 07:16:16PM +0800, Chen Yucong wrote: > We assume that there have three processes P1, P2, and P3 which share a > page frame PF0. PF0 have a multi-bit error that has not yet been detected. How likely is that? Did you see it in some real case? > As > a result, P1/P2 may raise the same MCE again. And how is that a problem? The memory error handling is always somewhat probabilistic. There are a lot of corner cases that could be be handled, but it would be even more complex than it already is, and most of them are unlikely to happen. The more complexity the more risk of unintended bugs. So the question is always how likely that case is, and is it worth handling. It's far better to focus on the common case. Another concern is always how to test this. Usually all explicit paths should have test cases in mce-test. But it's not clear to me the additional complexity here is justified. -andi -- To unsubscribe, send a message with 'unsubscribe linux-mm' in the body to majordomo@xxxxxxxxx. For more info on Linux MM, see: http://www.linux-mm.org/ . Don't email: <a href=mailto:"dont@xxxxxxxxx";> email@xxxxxxxxx </a>