From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: from mail-wm0-f65.google.com ([74.125.82.65]:35371 "EHLO mail-wm0-f65.google.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1751373AbdATHXK (ORCPT ); Fri, 20 Jan 2017 02:23:10 -0500 Received: by mail-wm0-f65.google.com with SMTP id d140so4563620wmd.2 for ; Thu, 19 Jan 2017 23:22:18 -0800 (PST) Subject: Re: [PATCHv2 00/24] tree-log inode vs btrfs_inode cleanups To: dsterba@suse.cz, linux-btrfs@vger.kernel.org References: <20170117161550.GW12081@twin.jikos.cz> <1484692310-3987-1-git-send-email-n.borisov.lkml@gmail.com> <20170119182152.GA11951@twin.jikos.cz> From: Nikolay Borisov Message-ID: Date: Fri, 20 Jan 2017 09:22:14 +0200 MIME-Version: 1.0 In-Reply-To: <20170119182152.GA11951@twin.jikos.cz> Content-Type: text/plain; charset=windows-1252 Sender: linux-btrfs-owner@vger.kernel.org List-ID: On 19.01.2017 20:21, David Sterba wrote: > On Wed, Jan 18, 2017 at 12:31:26AM +0200, Nikolay Borisov wrote: >> So here is a new set of patches cleaning up tree-log function >> w.r.t inode vs btrfs_inode. There are still some which remain >> but I didn't find compelling arguments to cleaning them up, so >> I've left them unchanged. This time there are some size shrinkage: >> >> text data bss dec hex filename >> 2530598 174661 28288 2733547 29b5eb fs/btrfs/btrfs.ko - upstream master >> >> text data bss dec hex filename >> 2530774 174661 28288 2733723 29b69b fs/btrfs/btrfs.ko - before tree-log cleanup >> >> text data bss dec hex filename >> 2530163 174661 28288 2733112 29b438 fs/btrfs/btrfs.ko - both series applied >> >> So the net result of the 2 series is 435 bytes and I assume there >> will be further reduction in size once further cleanups are made >> >> Changes since v1: >> * Rebased all patche to latest master >> >> Nikolay Borisov (24): >> btrfs: Make btrfs_must_commit_transaction take btrfs_inode >> btrfs: Make btrfs_record_unlink_dir take btrfs_inode >> btrfs: Make btrfs_record_snapshot_destroy take btrfs_inode >> btrfs: Make btrfs_inode_in_log take btrfs_inode >> btrfs: Make btrfs_log_new_name take btrfs_inode >> btrfs: Make btrfs_del_dir_entries_in_log take btrfs_inode >> btrfs: Make btrfs_del_inode_ref take btrfs_inode >> btrfs: Make logged_inode_size take btrfs_inode >> btrfs: Make btrfs_check_ref_name_override take btrfs_inode >> btrfs: Make copy_items take btrfs_inode >> btrfs: Make btrfs_log_all_xattrs take btrfs_inode >> btrfs: Make btrfs_log_trailing_hole take btrfs_inode >> btrfs: Make btrfs_get_logged_extents take btrfs_inode >> btrfs: Make btrfs_log_changed_extents take btrfs_inode >> btrfs: Make log_dir_items take btrfs_inode >> btrfs: Make log_directory_changes take btrfs_inode >> btrfs: Make log_new_dir_dentries take btrfs_inode >> btrfs: Make btrfs_unlink_inode take btrfs_inode >> btrfs: Make drop_one_dir_item take btrfs_inode >> btrfs: Make __add_inode_ref take btrfs_inode >> btrfs: Make log_inode_item take btrfs_inode >> btrfs: Make btrfs_log_inode take btrfs_inode >> btrfs: Make count_inode_extrefs take btrfs_inode >> btrfs: Make count_inode_refs take btrfs_inode > > Added to 4.11 queue, thanks. There were several 80+ lines added by the > patches, I've updated the patches during review. Please be more careful > with changes like this > > From https://patchwork.kernel.org/patch/9522051/ > > - if (S_ISDIR(inode->i_mode) || > + if (S_ISDIR(inode->vfs_inode.i_mode) || > (!test_bit(BTRFS_INODE_NEEDS_FULL_SYNC, > - &BTRFS_I(inode)->runtime_flags) && > - inode_only >= LOG_INODE_EXISTS)) > + &inode->runtime_flags) && > + inode_only == LOG_INODE_EXISTS)) > > Notice the change from >= to == . Yeah, I did notice this after the rebase I did when you told me patch 6 wasn't applying cleanly, however I thought I had fixed it during the rebase. Apparently I do need to recheck things one more time for good measure. Anyway, thanks for pulling!