On Sun, 16 May 2010, Helge Deller wrote: > I wish could report success as well.... > But I still do see segfaults (although they seem to happen not as often as before). Thanks again for testing. I'm convinced that we have various race conditions in doing cache user and kernel flushes that are not easily fixed. For example, in pte_wrprotect, it difficult to ensure that the cache is clean in a SMP environment. Think there are similar problems with copy_user_page. At the moment, I'm fairly pessimistic about finding a solution with our current implementation of copy_user_page. I've done more than a hundred kernel builds, and while things are better, there are still problems that look like cache corruption. Often, it takes more than a day for problems to appear. > Which kernel do you test? I'm on 2.6.33.3 + your patches. I'm on to 2.6.33.4. Dave -- J. David Anglin dave.anglin@xxxxxxxxxxxxxx National Research Council of Canada (613) 990-0752 (FAX: 952-6602) -- To unsubscribe from this list: send the line "unsubscribe linux-parisc" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
