From mboxrd@z Thu Jan 1 00:00:00 1970 From: Milosz Tanski Subject: [PATCH 0/2] Fscache cleanup and fix Date: Tue, 27 Aug 2013 19:08:31 -0400 Message-ID: Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Return-path: Received: from mail-qe0-f48.google.com ([209.85.128.48]:43433 "EHLO mail-qe0-f48.google.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1751661Ab3H0XI1 (ORCPT ); Tue, 27 Aug 2013 19:08:27 -0400 Received: by mail-qe0-f48.google.com with SMTP id 1so3009983qec.35 for ; Tue, 27 Aug 2013 16:08:27 -0700 (PDT) Content-Disposition: inline Sender: ceph-devel-owner@vger.kernel.org List-ID: To: ceph-devel@vger.kernel.org Cc: sage@inktank.com, zheng.z.yan@intel.com Sage, These are the changes you requested. I eliminated the use of ifdefs in the C code. It's all self contained in cache.[ch] and in the case of it being disabled becomes a nop. Also, I found a small bug where we would not free the invalidate queue on unmount. It's split into it's own commit. Finally, I saw that you had a hard time building it without the kernel flag enabled. So I made sure to test a config with fscache for Ceph disabled right before sping things. The commits are from my branch (which is based on top of the main testing branch) from my repo: https://adfin@bitbucket.org/adfin/linux-fs.git branch: wip-ceph-fscache Best, - Milosz Milosz Tanski (2): ceph: Do not leak fscache workqueue ceph: fscache cleanup fs/ceph/cache.c | 90 ++++++++++++++++++++++++++++++++++++++++++++++++++------- fs/ceph/cache.h | 68 +++++++++++++++++++------------------------ fs/ceph/inode.c | 61 +------------------------------------- fs/ceph/super.c | 22 ++++---------- fs/ceph/super.h | 1 - 5 files changed, 115 insertions(+), 127 deletions(-) -- 1.8.1.2