From: "Darrick J. Wong" <darrick.wong@oracle.com>
To: tytso@mit.edu, darrick.wong@oracle.com
Cc: linux-ext4@vger.kernel.org
Subject: [PATCH 04/27] libext2fs: fix problems with LE<->BE conversions on BE platforms
Date: Sat, 16 Aug 2014 16:46:16 -0700 [thread overview]
Message-ID: <20140816234616.11171.78225.stgit@birch.djwong.org> (raw)
In-Reply-To: <20140816234550.11171.61585.stgit@birch.djwong.org>
Fix more problems that I found when testing on ppc64:
- Inode swap cut and paste error leads to immutable inodes being
detected as inlinedata inodes, leading to e2fsck incorrectly barfing
on i_block[] contents.
- Superblock csum/verify must be aware of the fs->super byte order
when checking for metadata_csum feature flag. (Hint: in _openfs(),
fs->super is in LE order for the first csum verification)
Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com>
---
lib/ext2fs/csum.c | 26 ++++++++++++++++++--------
lib/ext2fs/swapfs.c | 2 +-
2 files changed, 19 insertions(+), 9 deletions(-)
diff --git a/lib/ext2fs/csum.c b/lib/ext2fs/csum.c
index d7dbca8..12f6fe9 100644
--- a/lib/ext2fs/csum.c
+++ b/lib/ext2fs/csum.c
@@ -81,12 +81,17 @@ static __u32 ext2fs_superblock_csum(ext2_filsys fs EXT2FS_ATTR((unused)),
return ext2fs_crc32c_le(~0, (unsigned char *)sb, offset);
}
+/* NOTE: The input to this function MUST be in LE order */
int ext2fs_superblock_csum_verify(ext2_filsys fs, struct ext2_super_block *sb)
{
- __u32 calculated;
+ __u32 flag, calculated;
- if (!EXT2_HAS_RO_COMPAT_FEATURE(fs->super,
- EXT4_FEATURE_RO_COMPAT_METADATA_CSUM))
+ if (fs->flags & EXT2_FLAG_SWAP_BYTES)
+ flag = EXT4_FEATURE_RO_COMPAT_METADATA_CSUM;
+ else
+ flag = ext2fs_cpu_to_le32(EXT4_FEATURE_RO_COMPAT_METADATA_CSUM);
+
+ if (!EXT2_HAS_RO_COMPAT_FEATURE(fs->super, flag))
return 1;
calculated = ext2fs_superblock_csum(fs, sb);
@@ -94,13 +99,18 @@ int ext2fs_superblock_csum_verify(ext2_filsys fs, struct ext2_super_block *sb)
return ext2fs_le32_to_cpu(sb->s_checksum) == calculated;
}
+/* NOTE: The input to this function MUST be in LE order */
errcode_t ext2fs_superblock_csum_set(ext2_filsys fs,
struct ext2_super_block *sb)
{
- __u32 crc;
+ __u32 flag, crc;
- if (!EXT2_HAS_RO_COMPAT_FEATURE(fs->super,
- EXT4_FEATURE_RO_COMPAT_METADATA_CSUM))
+ if (fs->flags & EXT2_FLAG_SWAP_BYTES)
+ flag = EXT4_FEATURE_RO_COMPAT_METADATA_CSUM;
+ else
+ flag = ext2fs_cpu_to_le32(EXT4_FEATURE_RO_COMPAT_METADATA_CSUM);
+
+ if (!EXT2_HAS_RO_COMPAT_FEATURE(fs->super, flag))
return 0;
crc = ext2fs_superblock_csum(fs, sb);
@@ -756,8 +766,8 @@ __u16 ext2fs_group_desc_csum(ext2_filsys fs, dgrp_t group)
#ifdef WORDS_BIGENDIAN
if (save_size > ext4_bg_size)
crc32 = ext2fs_crc32c_le(crc32,
- (char *)save_desc + ext4_bg_size,
- save_size - ext4_bg_size);
+ (unsigned char *)save_desc + ext4_bg_size,
+ save_size - ext4_bg_size);
#endif
crc = crc32 & 0xFFFF;
goto out;
diff --git a/lib/ext2fs/swapfs.c b/lib/ext2fs/swapfs.c
index e2aa41d..ee7a455 100644
--- a/lib/ext2fs/swapfs.c
+++ b/lib/ext2fs/swapfs.c
@@ -243,7 +243,7 @@ void ext2fs_swap_inode_full(ext2_filsys fs, struct ext2_inode_large *t,
t->i_flags = ext2fs_swab32(f->i_flags);
if (!hostorder && (t->i_flags & EXT4_EXTENTS_FL))
has_extents = 1;
- if (!hostorder && (f->i_flags & EXT4_INLINE_DATA_FL))
+ if (!hostorder && (t->i_flags & EXT4_INLINE_DATA_FL))
has_inline_data = 1;
t->i_dir_acl = ext2fs_swab32(f->i_dir_acl);
/*
next prev parent reply other threads:[~2014-08-16 23:46 UTC|newest]
Thread overview: 42+ messages / expand[flat|nested] mbox.gz Atom feed top
2014-08-16 23:45 [PATCH 00/27] e2fsprogs Summer 2014 patchbomb, part 5.1 Darrick J. Wong
2014-08-16 23:45 ` [PATCH 01/27] e2fuzz: fix pwrite64/pwrite usage Darrick J. Wong
2014-08-24 23:56 ` Theodore Ts'o
2014-08-16 23:46 ` [PATCH 02/27] misc: fix gcc warnings Darrick J. Wong
2014-08-24 23:56 ` Theodore Ts'o
2014-08-16 23:46 ` [PATCH 03/27] libext2fs: byteswap inode when performing the sanity scan Darrick J. Wong
2014-08-25 2:01 ` Theodore Ts'o
2014-08-16 23:46 ` Darrick J. Wong [this message]
2014-08-25 2:01 ` [PATCH 04/27] libext2fs: fix problems with LE<->BE conversions on BE platforms Theodore Ts'o
2014-08-16 23:46 ` [PATCH 05/27] e2fsck: on BE, re-swap everything after a damaged dirent so salvage works correctly Darrick J. Wong
2014-08-25 2:03 ` Theodore Ts'o
2014-08-16 23:46 ` [PATCH 06/27] libext2fs: create inlinedata symlinks Darrick J. Wong
2014-08-25 2:05 ` Theodore Ts'o
2014-08-16 23:46 ` [PATCH 07/27] contrib: add script to help resync journal code with kernel Darrick J. Wong
2014-08-27 3:44 ` Theodore Ts'o
2014-08-16 23:46 ` [PATCH 08/27] e2fsck: resync jbd2 recovery code from Linux 3.16 Darrick J. Wong
2014-08-27 3:44 ` Theodore Ts'o
2014-08-16 23:46 ` [PATCH 09/27] e2fsck: resync jbd2 revoke " Darrick J. Wong
2014-08-27 3:44 ` Theodore Ts'o
2014-08-16 23:46 ` [PATCH 10/27] e2fsck: fix infinite loop when recovering corrupt journal blocks Darrick J. Wong
2014-08-27 3:45 ` Theodore Ts'o
2014-08-16 23:47 ` [PATCH 11/27] e2fsck/debugfs: fix descriptor block size handling errors with journal_csum Darrick J. Wong
2014-08-28 1:15 ` Theodore Ts'o
2014-08-28 1:33 ` Darrick J. Wong
2014-08-16 23:47 ` [PATCH 12/27] debugfs: create journal handling routines Darrick J. Wong
2014-08-16 23:47 ` [PATCH 13/27] e2fsck: set journal superblock s_sequence to tail_sequence Darrick J. Wong
2014-08-16 23:47 ` [PATCH 14/27] debugfs: add the ability to write transactions to the journal Darrick J. Wong
2014-09-06 5:52 ` Darrick J. Wong
2014-08-16 23:47 ` [PATCH 15/27] tests: test writing and recovering checksum-free 32/64bit journals Darrick J. Wong
2014-08-16 23:47 ` [PATCH 16/27] tests: test writing and recovering 64bit csum_v3 journals Darrick J. Wong
2014-08-16 23:47 ` [PATCH 17/27] tests: test writing and recovering 32bit " Darrick J. Wong
2014-08-16 23:47 ` [PATCH 18/27] tests: write and replay blocks with the old journal checksum Darrick J. Wong
2014-08-16 23:47 ` [PATCH 19/27] tests: test recovery of 32 and 64-bit journals with checksum v2 Darrick J. Wong
2014-08-16 23:48 ` [PATCH 20/27] tests: test how e2fsck recovers from corrupt journal superblocks Darrick J. Wong
2014-08-16 23:48 ` [PATCH 21/27] tests: test e2fsck recovery of corrupt revoke blocks Darrick J. Wong
2014-08-16 23:48 ` [PATCH 22/27] tests: test e2fsck recovery with broken commit blocks Darrick J. Wong
2014-08-16 23:48 ` [PATCH 23/27] tests: test e2fsck recovery of corrupt descriptor blocks Darrick J. Wong
2014-08-16 23:48 ` [PATCH 24/27] mke2fs: set block_validity as a default mount option Darrick J. Wong
2014-08-16 23:48 ` [PATCH 25/27] ext2fs: add readahead method to improve scanning Darrick J. Wong
2014-08-16 23:48 ` [PATCH 26/27] libext2fs/e2fsck: provide routines to read-ahead metadata Darrick J. Wong
2014-08-16 23:48 ` [PATCH 27/27] e2fsck: read-ahead metadata during passes 1, 2, and 4 Darrick J. Wong
2014-08-17 17:23 ` [PATCH 28/27] libext2fs: write_journal_inode should check iterate return value Darrick J. Wong
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=20140816234616.11171.78225.stgit@birch.djwong.org \
--to=darrick.wong@oracle.com \
--cc=linux-ext4@vger.kernel.org \
--cc=tytso@mit.edu \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).