From: "Darrick J. Wong" Subject: Re: [PATCH] ext4: initialize multi-block allocator before checking block descriptors Date: Mon, 17 Mar 2014 10:23:57 -0700 Message-ID: <20140317172357.GH9070@birch.djwong.org> References: <20140315214029.GA18873@thunk.org> <20140316023841.GA902@thunk.org> <20140316161637.GA19885@azat> <20140316184634.GA14162@thunk.org> <20140316205659.GC19885@azat> <20140317020029.GC14162@thunk.org> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Cc: a3at.mail@gmail.com, "open list:EXT4 FILE SYSTEM" To: tytso@mit.edu Return-path: Received: from aserp1040.oracle.com ([141.146.126.69]:25497 "EHLO aserp1040.oracle.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1752083AbaCQRYC (ORCPT ); Mon, 17 Mar 2014 13:24:02 -0400 Content-Disposition: inline In-Reply-To: <20140317020029.GC14162@thunk.org> Sender: linux-ext4-owner@vger.kernel.org List-ID: On Sun, Mar 16, 2014 at 10:00:30PM -0400, tytso@mit.edu wrote: > On Mon, Mar 17, 2014 at 12:56:59AM +0400, a3at.mail@gmail.com wrote: > > After I tested ext4 dev branch (eb3e7abb161ad5), without any xfs-tests complaints, > > I understand what goes wrong, you have not last version of > > this patch, the latest is v3. > > (Actually you have description from last patch, but not the latest changes.) > > What I did was to take your v3 version of the patch, and then since > that patch removed the label failed_mount5, I changed instances of > failed_mount4a to failed_mount5, just for aesthetic reasons. /me wonders if it would be easier just to do something like this: ext4: don't crash when validating block bitmap If EXT4FS_DEBUG is defined, ext4_validate_block_bitmap is called via ext4_count_free_clusters before sb->s_group_info is set up. When this happens, the kernel crashes because group info hasn't been loaded. Forego marking the group corrupt for now; not setting BH_Verified means we'll revisit the bitmap if something went wrong. Signed-off-by: Darrick J. Wong --- fs/ext4/balloc.c | 13 ++++++++++--- 1 file changed, 10 insertions(+), 3 deletions(-) diff --git a/fs/ext4/balloc.c b/fs/ext4/balloc.c index 2f35689..b375768 100644 --- a/fs/ext4/balloc.c +++ b/fs/ext4/balloc.c @@ -355,7 +355,10 @@ void ext4_validate_block_bitmap(struct super_block *sb, struct buffer_head *bh) { ext4_fsblk_t blk; - struct ext4_group_info *grp = ext4_get_group_info(sb, block_group); + struct ext4_group_info *grp = NULL; + + if (EXT4_SB(sb)->s_group_info) + grp = ext4_get_group_info(sb, block_group); if (buffer_verified(bh)) return; @@ -366,14 +369,18 @@ void ext4_validate_block_bitmap(struct super_block *sb, ext4_unlock_group(sb, block_group); ext4_error(sb, "bg %u: block %llu: invalid block bitmap", block_group, blk); - set_bit(EXT4_GROUP_INFO_BBITMAP_CORRUPT_BIT, &grp->bb_state); + if (grp) + set_bit(EXT4_GROUP_INFO_BBITMAP_CORRUPT_BIT, + &grp->bb_state); return; } if (unlikely(!ext4_block_bitmap_csum_verify(sb, block_group, desc, bh))) { ext4_unlock_group(sb, block_group); ext4_error(sb, "bg %u: bad block bitmap checksum", block_group); - set_bit(EXT4_GROUP_INFO_BBITMAP_CORRUPT_BIT, &grp->bb_state); + if (grp) + set_bit(EXT4_GROUP_INFO_BBITMAP_CORRUPT_BIT, + &grp->bb_state); return; } set_buffer_verified(bh);