From: Matthew Wilcox Subject: [PATCH v3 0/3] Add XIP support to ext4 Date: Tue, 17 Dec 2013 14:18:25 -0500 Message-ID: Mime-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit Cc: Matthew Wilcox To: linux-ext4@vger.kernel.org, linux-fsdevel@vger.kernel.org Return-path: Received: from mga02.intel.com ([134.134.136.20]:51440 "EHLO mga02.intel.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1751884Ab3LQTSj (ORCPT ); Tue, 17 Dec 2013 14:18:39 -0500 Sender: linux-ext4-owner@vger.kernel.org List-ID: For v3, we've addressed the problem with unwritten extents that Dave Chinner pointed out. Rather than testing bh_unwritten() as he suggested, I opted for checking !buffer_mapped() as block_read_full_page() in fs/buffer.c does. While I'm in that function I renamed the buffer_head from 'tmp' to 'bh' to follow normal usage. I also folded the "ext4: Add xip hole punching" patch into the main "Add XIP functionality" patch since otherwise we're introducing broken functionality in the middle of the series. Matthew Wilcox (2): Fix XIP fault vs truncate race xip: Add xip_zero_page_range Ross Zwisler (1): ext4: Add XIP functionality Documentation/filesystems/ext4.txt | 2 + Documentation/filesystems/xip.txt | 3 ++ fs/Kconfig | 2 +- fs/ext4/Kconfig | 11 +++++ fs/ext4/Makefile | 1 + fs/ext4/ext4.h | 4 +- fs/ext4/file.c | 17 ++++++++ fs/ext4/inode.c | 86 +++++++++++++++++++++++++++----------- fs/ext4/namei.c | 11 ++++- fs/ext4/super.c | 36 +++++++++++++++- fs/ext4/xip.c | 78 ++++++++++++++++++++++++++++++++++ fs/ext4/xip.h | 24 +++++++++++ include/linux/fs.h | 8 ++++ mm/filemap_xip.c | 55 ++++++++++++++++-------- 14 files changed, 289 insertions(+), 49 deletions(-) create mode 100644 fs/ext4/xip.c create mode 100644 fs/ext4/xip.h -- 1.8.4.rc3