i have wrote a patch that let kvm share the ( for now 1024 ) mmu pages of all the guests.
the patch is wrote in a way that it take mmu pages from "idle" guest and move them to intensive working guest. i have design the patch from the beginning to not use locking at all. therefore when the intensive guest see that it need more pages, it set a global variable to 1 ( int request_bit ), when every guest see this varible set to 1, it check if it should free pages (in case it idle compared to the other guest). simple benchmark showed that it improved kernel compiling from 5:40 to 5:22 and it highly reduce the number of pf_fixed exits, the true potential is even bigger when more pages would be allocated to each guest. anyway look at the code and tell me what you think :) (note that i know that i have some cleanup things i didnt do, but forgive me for now) ------------------------------------------------------------------------- This SF.net email is sponsored by: Splunk Inc. Still grepping through log files to find problems? Stop. Now Search log events and configuration files using AJAX and a browser. Download your FREE copy of Splunk now >> http://get.splunk.com/ _______________________________________________ kvm-devel mailing list kvm-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/kvm-devel
