On Thu, 2010-08-05 at 21:44 +1000, Nick Piggin wrote: > On Wed, Aug 04, 2010 at 09:04:03AM +0800, Zhang, Yanmin wrote: > > We ran lots of benchmarks on many machines. Below is something to > > share with you. > > > > Improvement: > > 1) We get about 30% improvement with kbuild workload on Nehalem > > machines. It's hard to improve kbuild performance. Your tree does. > > > > Issues: > > 1) Compiling fails on a couple of file systems, such like CONFIG_ISO9660_FS=y. > > 2) dbenchthreads has about 50% regression. We connect a JBOD of 12 disks to > > a machine. Start 4 dbench threads per disk. We run the workload under > > a regular user account. If we run it under root account, we get 22% > > improvement instead of regression. The root cause is ACL checking. > > With your patch, do_path_lookup firstly goes through rcu steps which > > including a exec permission checking. With ACL, the __exec_permission > > always fails. Then a later nameidata_drop_rcu often fails as > > dentry->d_seq is changed. > > Oh one other thing I wanted to ask about. d_seq changing should not > be too common. If the directory is renamed, or if it is turned negative > should be the only cases in which we should see a d_seq changes. > > Or unless there is a bug and it is checking the wrong sequence or > against the wrong dentry. Sorry for misleading you. It fails at the beginning in nameidata_drop_rcu because (nd->flags & LOOKUP_FIRST) is true. > How often would you say nameidata_drop_rcu > fails (without the following acl rcu patches)? I instrument kernel and find nameidata_drop_rcu always fails. -- To unsubscribe, send a message with 'unsubscribe linux-mm' in the body to majordomo@xxxxxxxxxx For more info on Linux MM, see: http://www.linux-mm.org/ . Don't email: <a href=mailto:"dont@xxxxxxxxx";> email@xxxxxxxxx </a>