From patchwork Tue Feb 28 07:30:00 2012 Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: 7bit X-Patchwork-Submitter: Lukas Czerner X-Patchwork-Id: 143368 Return-Path: X-Original-To: patchwork-incoming@ozlabs.org Delivered-To: patchwork-incoming@ozlabs.org Received: from vger.kernel.org (vger.kernel.org [209.132.180.67]) by ozlabs.org (Postfix) with ESMTP id A41E2B6EE7 for ; Tue, 28 Feb 2012 18:30:33 +1100 (EST) Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S932335Ab2B1HaN (ORCPT ); Tue, 28 Feb 2012 02:30:13 -0500 Received: from mx1.redhat.com ([209.132.183.28]:55624 "EHLO mx1.redhat.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S932221Ab2B1HaM (ORCPT ); Tue, 28 Feb 2012 02:30:12 -0500 Received: from int-mx12.intmail.prod.int.phx2.redhat.com (int-mx12.intmail.prod.int.phx2.redhat.com [10.5.11.25]) by mx1.redhat.com (8.14.4/8.14.4) with ESMTP id q1S7UAVE008953 (version=TLSv1/SSLv3 cipher=DHE-RSA-AES256-SHA bits=256 verify=OK); Tue, 28 Feb 2012 02:30:10 -0500 Received: from dhcp-27-109.brq.redhat.com (dhcp-1-248.brq.redhat.com [10.34.1.248]) by int-mx12.intmail.prod.int.phx2.redhat.com (8.14.4/8.14.4) with ESMTP id q1S7U8rY022441; Tue, 28 Feb 2012 02:30:09 -0500 From: Lukas Czerner To: linux-ext4@vger.kernel.org Cc: tytso@mit.edu, Lukas Czerner Subject: [PATCH 1/2 v3] e2fsck: Discard only unused parts of inode table Date: Tue, 28 Feb 2012 08:30:00 +0100 Message-Id: <1330414201-31788-1-git-send-email-lczerner@redhat.com> X-Scanned-By: MIMEDefang 2.68 on 10.5.11.25 Sender: linux-ext4-owner@vger.kernel.org Precedence: bulk List-ID: X-Mailing-List: linux-ext4@vger.kernel.org When calling e2fsck with '-E discard' option it might happen that valid inodes are discarded accidentally. This is because we just discard the part of inode table which lies past the free inode count. This is terribly wrong (sorry!). This patch fixes it so only the free parts of an inode table is discarded, leaving used inodes intact. This was tested with highly fragmented inode tables with block size 4k and 1k. Signed-off-by: Lukas Czerner Reported-by: Phillip Susi --- v2: reworked so that we comply with inode number counting and adjust start in the e2fsck_discard_inodes(). Add some comments v3: Update commit description e2fsck/pass5.c | 82 ++++++++++++++++++++++++++++++++++++++++--------------- 1 files changed, 59 insertions(+), 23 deletions(-) diff --git a/e2fsck/pass5.c b/e2fsck/pass5.c index 1e836e3..f70928f 100644 --- a/e2fsck/pass5.c +++ b/e2fsck/pass5.c @@ -94,6 +94,43 @@ static void e2fsck_discard_blocks(e2fsck_t ctx, io_manager manager, ctx->options &= ~E2F_OPT_DISCARD; } +/* + * This will try to discard number 'count' inodes starting at + * inode number 'start'. Note that 'start' is a inode number and + * henc it is counted from 1, it means that we need to adjust it + * by -1 in this function to compute right offset in the + * inode table. + */ +static void e2fsck_discard_inodes(e2fsck_t ctx, int group, + int start, int count) +{ + ext2_filsys fs = ctx->fs; + blk64_t blk, num; + int orig = count; + + if ((ctx->options & E2F_OPT_NO) || !(ctx->options & E2F_OPT_DISCARD)) + return; + + /* + * Start is inode number which starts counting from 1, + * so we need to adjust it. + */ + start -= 1; + + /* + * We can discard only blocks containing only unused + * inodes in the table. + */ + blk = DIV_ROUND_UP(start, + EXT2_INODES_PER_BLOCK(fs->super)); + count -= (blk * EXT2_INODES_PER_BLOCK(fs->super) - start); + blk += ext2fs_inode_table_loc(fs, group); + num = count / EXT2_INODES_PER_BLOCK(fs->super); + + if (num > 0) + e2fsck_discard_blocks(ctx, fs->io->manager, blk, num); +} + #define NO_BLK ((blk64_t) -1) static void print_bitmap_problem(e2fsck_t ctx, int problem, @@ -435,6 +472,7 @@ static void check_inode_bitmaps(e2fsck_t ctx) int skip_group = 0; int redo_flag = 0; io_manager manager = ctx->fs->io->manager; + unsigned long long first_free = fs->super->s_inodes_per_group + 1; clear_problem_context(&pctx); free_array = (int *) e2fsck_allocate_memory(ctx, @@ -497,6 +535,7 @@ redo_counts: * are 0, count the free inode, * skip the current block group. */ + first_free = 1; inodes = fs->super->s_inodes_per_group - 1; group_free = inodes; free_inodes += inodes; @@ -561,50 +600,47 @@ redo_counts: ctx->options &= ~E2F_OPT_DISCARD; do_counts: + inodes++; if (bitmap) { if (ext2fs_test_inode_bitmap2(ctx->inode_dir_map, i)) dirs_count++; + if (inodes > first_free) { + e2fsck_discard_inodes(ctx, group, first_free, + inodes - first_free); + first_free = fs->super->s_inodes_per_group + 1; + } } else if (!skip_group || csum_flag) { group_free++; free_inodes++; + if (first_free > inodes) + first_free = inodes; } - inodes++; if ((inodes == fs->super->s_inodes_per_group) || (i == fs->super->s_inodes_count)) { - - free_array[group] = group_free; - dir_array[group] = dirs_count; - - /* Discard inode table */ - if (ctx->options & E2F_OPT_DISCARD) { - blk64_t used_blks, blk, num; - - used_blks = DIV_ROUND_UP( - (EXT2_INODES_PER_GROUP(fs->super) - - group_free), - EXT2_INODES_PER_BLOCK(fs->super)); - - blk = ext2fs_inode_table_loc(fs, group) + - used_blks; - num = fs->inode_blocks_per_group - - used_blks; - e2fsck_discard_blocks(ctx, manager, blk, num); - } - + /* + * If the last inode is free, we can discard it as well. + */ + if (inodes >= first_free) + e2fsck_discard_inodes(ctx, group, first_free, + inodes - first_free + 1); /* * If discard zeroes data and the group inode table * was not zeroed yet, set itable as zeroed */ if ((ctx->options & E2F_OPT_DISCARD) && - (io_channel_discard_zeroes_data(fs->io)) && + !(ctx->options & E2F_OPT_NO) && + io_channel_discard_zeroes_data(fs->io) && !(ext2fs_bg_flags_test(fs, group, - EXT2_BG_INODE_ZEROED))) { + EXT2_BG_INODE_ZEROED))) { ext2fs_bg_flags_set(fs, group, EXT2_BG_INODE_ZEROED); ext2fs_group_desc_csum_set(fs, group); } + first_free = fs->super->s_inodes_per_group + 1; + free_array[group] = group_free; + dir_array[group] = dirs_count; group ++; inodes = 0; skip_group = 0;