On Fri, 09 Sep 2005 17:30:35 EDT, Shailabh Nagar wrote: > Shailabh Nagar wrote: > > > Paul Menage wrote: > > > >> There seem to be some problems with refcounting of inode and dentry > >> objects in RCFS. With the single e18 patch I find that running > >> > >> while true; do mkdir /rcfs/taskclass/foo; rmdir /rcfs/taskclass/foo; > >> done > >> > >> causes the size of rcfs_inode_cache to grow steadily; other > >> instrumentation shows that the class directory inodes never get freed. > > > > Yes, we'd seen this too (for the SLES9 kernel code base). I have a > > patch that fixes it...will send out shortly (after checking it for e18). > > > Here are two patches that fix the inode leak problem. They apply over > 2.6.12 + e18 + magic_c.patch which is our interim version till e19 comes > out. > > > The series file is > > ckrm_e18_2612_single.patch > magic_c.patch > rcfs-inode-leak-create-internal.patch > rcfs-inode-leak-delete-internal.patch > > > -- Shailabh
I'll review and queue these up for the release I'm working on right now. Yeah, I'm already a day late. Hopefully later tonight I'll have a nearly untested or lightly tested version to release of the full e19 patch set. Shailabh, do you think the IO controller components are ready to be included in the core set? If so, I'll look at merging those in as well. gerrit ------------------------------------------------------- SF.Net email is Sponsored by the Better Software Conference & EXPO September 19-22, 2005 * San Francisco, CA * Development Lifecycle Practices Agile & Plan-Driven Development * Managing Projects & Teams * Testing & QA Security * Process Improvement & Measurement * http://www.sqe.com/bsce5sf _______________________________________________ ckrm-tech mailing list https://lists.sourceforge.net/lists/listinfo/ckrm-tech
