From dc04db2aa7c9307e740d6d0e173085301c173b1a Mon Sep 17 00:00:00 2001 From: Dave Chinner Date: Wed, 4 May 2022 12:13:35 +1000 Subject: xfs: detect self referencing btree sibling pointers To catch the obvious graph cycle problem and hence potential endless looping. Signed-off-by: Dave Chinner Reviewed-by: Christoph Hellwig Reviewed-by: Darrick J. Wong Signed-off-by: Dave Chinner --- fs/xfs/libxfs/xfs_btree.c | 140 ++++++++++++++++++++++++++++++++++------------ 1 file changed, 105 insertions(+), 35 deletions(-) diff --git a/fs/xfs/libxfs/xfs_btree.c b/fs/xfs/libxfs/xfs_btree.c index a8c79e760d8a..2aa300f7461f 100644 --- a/fs/xfs/libxfs/xfs_btree.c +++ b/fs/xfs/libxfs/xfs_btree.c @@ -51,6 +51,52 @@ xfs_btree_magic( return magic; } +static xfs_failaddr_t +xfs_btree_check_lblock_siblings( + struct xfs_mount *mp, + struct xfs_btree_cur *cur, + int level, + xfs_fsblock_t fsb, + xfs_fsblock_t sibling) +{ + if (sibling == NULLFSBLOCK) + return NULL; + if (sibling == fsb) + return __this_address; + if (level >= 0) { + if (!xfs_btree_check_lptr(cur, sibling, level + 1)) + return __this_address; + } else { + if (!xfs_verify_fsbno(mp, sibling)) + return __this_address; + } + + return NULL; +} + +static xfs_failaddr_t +xfs_btree_check_sblock_siblings( + struct xfs_mount *mp, + struct xfs_btree_cur *cur, + int level, + xfs_agnumber_t agno, + xfs_agblock_t agbno, + xfs_agblock_t sibling) +{ + if (sibling == NULLAGBLOCK) + return NULL; + if (sibling == agbno) + return __this_address; + if (level >= 0) { + if (!xfs_btree_check_sptr(cur, sibling, level + 1)) + return __this_address; + } else { + if (!xfs_verify_agbno(mp, agno, sibling)) + return __this_address; + } + return NULL; +} + /* * Check a long btree block header. Return the address of the failing check, * or NULL if everything is ok. @@ -65,6 +111,8 @@ __xfs_btree_check_lblock( struct xfs_mount *mp = cur->bc_mp; xfs_btnum_t btnum = cur->bc_btnum; int crc = xfs_has_crc(mp); + xfs_failaddr_t fa; + xfs_fsblock_t fsb = NULLFSBLOCK; if (crc) { if (!uuid_equal(&block->bb_u.l.bb_uuid, &mp->m_sb.sb_meta_uuid)) @@ -83,16 +131,16 @@ __xfs_btree_check_lblock( if (be16_to_cpu(block->bb_numrecs) > cur->bc_ops->get_maxrecs(cur, level)) return __this_address; - if (block->bb_u.l.bb_leftsib != cpu_to_be64(NULLFSBLOCK) && - !xfs_btree_check_lptr(cur, be64_to_cpu(block->bb_u.l.bb_leftsib), - level + 1)) - return __this_address; - if (block->bb_u.l.bb_rightsib != cpu_to_be64(NULLFSBLOCK) && - !xfs_btree_check_lptr(cur, be64_to_cpu(block->bb_u.l.bb_rightsib), - level + 1)) - return __this_address; - return NULL; + if (bp) + fsb = XFS_DADDR_TO_FSB(mp, xfs_buf_daddr(bp)); + + fa = xfs_btree_check_lblock_siblings(mp, cur, level, fsb, + be64_to_cpu(block->bb_u.l.bb_leftsib)); + if (!fa) + fa = xfs_btree_check_lblock_siblings(mp, cur, level, fsb, + be64_to_cpu(block->bb_u.l.bb_rightsib)); + return fa; } /* Check a long btree block header. */ @@ -130,6 +178,9 @@ __xfs_btree_check_sblock( struct xfs_mount *mp = cur->bc_mp; xfs_btnum_t btnum = cur->bc_btnum; int crc = xfs_has_crc(mp); + xfs_failaddr_t fa; + xfs_agblock_t agbno = NULLAGBLOCK; + xfs_agnumber_t agno = NULLAGNUMBER; if (crc) { if (!uuid_equal(&block->bb_u.s.bb_uuid, &mp->m_sb.sb_meta_uuid)) @@ -146,16 +197,18 @@ __xfs_btree_check_sblock( if (be16_to_cpu(block->bb_numrecs) > cur->bc_ops->get_maxrecs(cur, level)) return __this_address; - if (block->bb_u.s.bb_leftsib != cpu_to_be32(NULLAGBLOCK) && - !xfs_btree_check_sptr(cur, be32_to_cpu(block->bb_u.s.bb_leftsib), - level + 1)) - return __this_address; - if (block->bb_u.s.bb_rightsib != cpu_to_be32(NULLAGBLOCK) && - !xfs_btree_check_sptr(cur, be32_to_cpu(block->bb_u.s.bb_rightsib), - level + 1)) - return __this_address; - return NULL; + if (bp) { + agbno = xfs_daddr_to_agbno(mp, xfs_buf_daddr(bp)); + agno = xfs_daddr_to_agno(mp, xfs_buf_daddr(bp)); + } + + fa = xfs_btree_check_sblock_siblings(mp, cur, level, agno, agbno, + be32_to_cpu(block->bb_u.s.bb_leftsib)); + if (!fa) + fa = xfs_btree_check_sblock_siblings(mp, cur, level, agno, + agbno, be32_to_cpu(block->bb_u.s.bb_rightsib)); + return fa; } /* Check a short btree block header. */ @@ -4271,6 +4324,21 @@ xfs_btree_visit_block( if (xfs_btree_ptr_is_null(cur, &rptr)) return -ENOENT; + /* + * We only visit blocks once in this walk, so we have to avoid the + * internal xfs_btree_lookup_get_block() optimisation where it will + * return the same block without checking if the right sibling points + * back to us and creates a cyclic reference in the btree. + */ + if (cur->bc_flags & XFS_BTREE_LONG_PTRS) { + if (be64_to_cpu(rptr.l) == XFS_DADDR_TO_FSB(cur->bc_mp, + xfs_buf_daddr(bp))) + return -EFSCORRUPTED; + } else { + if (be32_to_cpu(rptr.s) == xfs_daddr_to_agbno(cur->bc_mp, + xfs_buf_daddr(bp))) + return -EFSCORRUPTED; + } return xfs_btree_lookup_get_block(cur, level, &rptr, &block); } @@ -4445,20 +4513,21 @@ xfs_btree_lblock_verify( { struct xfs_mount *mp = bp->b_mount; struct xfs_btree_block *block = XFS_BUF_TO_BLOCK(bp); + xfs_fsblock_t fsb; + xfs_failaddr_t fa; /* numrecs verification */ if (be16_to_cpu(block->bb_numrecs) > max_recs) return __this_address; /* sibling pointer verification */ - if (block->bb_u.l.bb_leftsib != cpu_to_be64(NULLFSBLOCK) && - !xfs_verify_fsbno(mp, be64_to_cpu(block->bb_u.l.bb_leftsib))) - return __this_address; - if (block->bb_u.l.bb_rightsib != cpu_to_be64(NULLFSBLOCK) && - !xfs_verify_fsbno(mp, be64_to_cpu(block->bb_u.l.bb_rightsib))) - return __this_address; - - return NULL; + fsb = XFS_DADDR_TO_FSB(mp, xfs_buf_daddr(bp)); + fa = xfs_btree_check_lblock_siblings(mp, NULL, -1, fsb, + be64_to_cpu(block->bb_u.l.bb_leftsib)); + if (!fa) + fa = xfs_btree_check_lblock_siblings(mp, NULL, -1, fsb, + be64_to_cpu(block->bb_u.l.bb_rightsib)); + return fa; } /** @@ -4499,7 +4568,9 @@ xfs_btree_sblock_verify( { struct xfs_mount *mp = bp->b_mount; struct xfs_btree_block *block = XFS_BUF_TO_BLOCK(bp); - xfs_agblock_t agno; + xfs_agnumber_t agno; + xfs_agblock_t agbno; + xfs_failaddr_t fa; /* numrecs verification */ if (be16_to_cpu(block->bb_numrecs) > max_recs) @@ -4507,14 +4578,13 @@ xfs_btree_sblock_verify( /* sibling pointer verification */ agno = xfs_daddr_to_agno(mp, xfs_buf_daddr(bp)); - if (block->bb_u.s.bb_leftsib != cpu_to_be32(NULLAGBLOCK) && - !xfs_verify_agbno(mp, agno, be32_to_cpu(block->bb_u.s.bb_leftsib))) - return __this_address; - if (block->bb_u.s.bb_rightsib != cpu_to_be32(NULLAGBLOCK) && - !xfs_verify_agbno(mp, agno, be32_to_cpu(block->bb_u.s.bb_rightsib))) - return __this_address; - - return NULL; + agbno = xfs_daddr_to_agbno(mp, xfs_buf_daddr(bp)); + fa = xfs_btree_check_sblock_siblings(mp, NULL, -1, agno, agbno, + be32_to_cpu(block->bb_u.s.bb_leftsib)); + if (!fa) + fa = xfs_btree_check_sblock_siblings(mp, NULL, -1, agno, agbno, + be32_to_cpu(block->bb_u.s.bb_rightsib)); + return fa; } /* -- cgit v1.2.3 From 1eb70f54c445fcbb25817841e774adb3d912f3e8 Mon Sep 17 00:00:00 2001 From: Dave Chinner Date: Wed, 4 May 2022 12:13:53 +1000 Subject: xfs: validate inode fork size against fork format xfs_repair catches fork size/format mismatches, but the in-kernel verifier doesn't, leading to null pointer failures when attempting to perform operations on the fork. This can occur in the xfs_dir_is_empty() where the in-memory fork format does not match the size and so the fork data pointer is accessed incorrectly. Note: this causes new failures in xfs/348 which is testing mode vs ftype mismatches. We now detect a regular file that has been changed to a directory or symlink mode as being corrupt because the data fork is for a symlink or directory should be in local form when there are only 3 bytes of data in the data fork. Hence the inode verify for the regular file now fires w/ -EFSCORRUPTED because the inode fork format does not match the format the corrupted mode says it should be in. Signed-off-by: Dave Chinner Reviewed-by: Christoph Hellwig Reviewed-by: Darrick J. Wong Signed-off-by: Dave Chinner --- fs/xfs/libxfs/xfs_inode_buf.c | 35 ++++++++++++++++++++++++++--------- 1 file changed, 26 insertions(+), 9 deletions(-) diff --git a/fs/xfs/libxfs/xfs_inode_buf.c b/fs/xfs/libxfs/xfs_inode_buf.c index 74b82ec80f8e..3b1b63f9d886 100644 --- a/fs/xfs/libxfs/xfs_inode_buf.c +++ b/fs/xfs/libxfs/xfs_inode_buf.c @@ -357,21 +357,38 @@ xfs_dinode_verify_fork( { xfs_extnum_t di_nextents; xfs_extnum_t max_extents; + mode_t mode = be16_to_cpu(dip->di_mode); + uint32_t fork_size = XFS_DFORK_SIZE(dip, mp, whichfork); + uint32_t fork_format = XFS_DFORK_FORMAT(dip, whichfork); di_nextents = xfs_dfork_nextents(dip, whichfork); - switch (XFS_DFORK_FORMAT(dip, whichfork)) { + /* + * For fork types that can contain local data, check that the fork + * format matches the size of local data contained within the fork. + * + * For all types, check that when the size says the should be in extent + * or btree format, the inode isn't claiming it is in local format. + */ + if (whichfork == XFS_DATA_FORK) { + if (S_ISDIR(mode) || S_ISLNK(mode)) { + if (be64_to_cpu(dip->di_size) <= fork_size && + fork_format != XFS_DINODE_FMT_LOCAL) + return __this_address; + } + + if (be64_to_cpu(dip->di_size) > fork_size && + fork_format == XFS_DINODE_FMT_LOCAL) + return __this_address; + } + + switch (fork_format) { case XFS_DINODE_FMT_LOCAL: /* - * no local regular files yet + * No local regular files yet. */ - if (whichfork == XFS_DATA_FORK) { - if (S_ISREG(be16_to_cpu(dip->di_mode))) - return __this_address; - if (be64_to_cpu(dip->di_size) > - XFS_DFORK_SIZE(dip, mp, whichfork)) - return __this_address; - } + if (S_ISREG(mode) && whichfork == XFS_DATA_FORK) + return __this_address; if (di_nextents) return __this_address; break; -- cgit v1.2.3 From dd0d2f9755191690541b09e6385d0f8cd8bc9d8f Mon Sep 17 00:00:00 2001 From: Dave Chinner Date: Wed, 4 May 2022 12:14:13 +1000 Subject: xfs: set XFS_FEAT_NLINK correctly While xfs_has_nlink() is not used in kernel, it is used in userspace (e.g. by xfs_db) so we need to set the XFS_FEAT_NLINK flag correctly in xfs_sb_version_to_features(). Signed-off-by: Dave Chinner Reviewed-by: Christoph Hellwig Reviewed-by: Darrick J. Wong Signed-off-by: Dave Chinner --- fs/xfs/libxfs/xfs_sb.c | 2 ++ 1 file changed, 2 insertions(+) diff --git a/fs/xfs/libxfs/xfs_sb.c b/fs/xfs/libxfs/xfs_sb.c index cf9e5b9374c1..ec6eec5c0e02 100644 --- a/fs/xfs/libxfs/xfs_sb.c +++ b/fs/xfs/libxfs/xfs_sb.c @@ -70,6 +70,8 @@ xfs_sb_version_to_features( /* optional V4 features */ if (sbp->sb_rblocks > 0) features |= XFS_FEAT_REALTIME; + if (sbp->sb_versionnum & XFS_SB_VERSION_NLINKBIT) + features |= XFS_FEAT_NLINK; if (sbp->sb_versionnum & XFS_SB_VERSION_ATTRBIT) features |= XFS_FEAT_ATTR; if (sbp->sb_versionnum & XFS_SB_VERSION_QUOTABIT) -- cgit v1.2.3 From f0f5f658065a5af09126ec892e4c383540a1c77f Mon Sep 17 00:00:00 2001 From: Dave Chinner Date: Wed, 4 May 2022 12:17:18 +1000 Subject: xfs: validate v5 feature fields We don't check that the v4 feature flags taht v5 requires to be set are actually set anywhere. Do this check when we see that the filesystem is a v5 filesystem. Signed-off-by: Dave Chinner Reviewed-by: Christoph Hellwig Reviewed-by: Darrick J. Wong Signed-off-by: Dave Chinner --- fs/xfs/libxfs/xfs_sb.c | 68 ++++++++++++++++++++++++++++++++++++++++++-------- 1 file changed, 58 insertions(+), 10 deletions(-) diff --git a/fs/xfs/libxfs/xfs_sb.c b/fs/xfs/libxfs/xfs_sb.c index ec6eec5c0e02..a20cade590e9 100644 --- a/fs/xfs/libxfs/xfs_sb.c +++ b/fs/xfs/libxfs/xfs_sb.c @@ -30,6 +30,47 @@ * Physical superblock buffer manipulations. Shared with libxfs in userspace. */ +/* + * Check that all the V4 feature bits that the V5 filesystem format requires are + * correctly set. + */ +static bool +xfs_sb_validate_v5_features( + struct xfs_sb *sbp) +{ + /* We must not have any unknown V4 feature bits set */ + if (sbp->sb_versionnum & ~XFS_SB_VERSION_OKBITS) + return false; + + /* + * The CRC bit is considered an invalid V4 flag, so we have to add it + * manually to the OKBITS mask. + */ + if (sbp->sb_features2 & ~(XFS_SB_VERSION2_OKBITS | + XFS_SB_VERSION2_CRCBIT)) + return false; + + /* Now check all the required V4 feature flags are set. */ + +#define V5_VERS_FLAGS (XFS_SB_VERSION_NLINKBIT | \ + XFS_SB_VERSION_ALIGNBIT | \ + XFS_SB_VERSION_LOGV2BIT | \ + XFS_SB_VERSION_EXTFLGBIT | \ + XFS_SB_VERSION_DIRV2BIT | \ + XFS_SB_VERSION_MOREBITSBIT) + +#define V5_FEAT_FLAGS (XFS_SB_VERSION2_LAZYSBCOUNTBIT | \ + XFS_SB_VERSION2_ATTR2BIT | \ + XFS_SB_VERSION2_PROJID32BIT | \ + XFS_SB_VERSION2_CRCBIT) + + if ((sbp->sb_versionnum & V5_VERS_FLAGS) != V5_VERS_FLAGS) + return false; + if ((sbp->sb_features2 & V5_FEAT_FLAGS) != V5_FEAT_FLAGS) + return false; + return true; +} + /* * We support all XFS versions newer than a v4 superblock with V2 directories. */ @@ -37,9 +78,19 @@ bool xfs_sb_good_version( struct xfs_sb *sbp) { - /* all v5 filesystems are supported */ + /* + * All v5 filesystems are supported, but we must check that all the + * required v4 feature flags are enabled correctly as the code checks + * those flags and not for v5 support. + */ if (xfs_sb_is_v5(sbp)) - return true; + return xfs_sb_validate_v5_features(sbp); + + /* We must not have any unknown v4 feature bits set */ + if ((sbp->sb_versionnum & ~XFS_SB_VERSION_OKBITS) || + ((sbp->sb_versionnum & XFS_SB_VERSION_MOREBITSBIT) && + (sbp->sb_features2 & ~XFS_SB_VERSION2_OKBITS))) + return false; /* versions prior to v4 are not supported */ if (XFS_SB_VERSION_NUM(sbp) < XFS_SB_VERSION_4) @@ -51,12 +102,6 @@ xfs_sb_good_version( if (!(sbp->sb_versionnum & XFS_SB_VERSION_EXTFLGBIT)) return false; - /* And must not have any unknown v4 feature bits set */ - if ((sbp->sb_versionnum & ~XFS_SB_VERSION_OKBITS) || - ((sbp->sb_versionnum & XFS_SB_VERSION_MOREBITSBIT) && - (sbp->sb_features2 & ~XFS_SB_VERSION2_OKBITS))) - return false; - /* It's a supported v4 filesystem */ return true; } @@ -267,12 +312,15 @@ xfs_validate_sb_common( bool has_dalign; if (!xfs_verify_magic(bp, dsb->sb_magicnum)) { - xfs_warn(mp, "bad magic number"); + xfs_warn(mp, +"Superblock has bad magic number 0x%x. Not an XFS filesystem?", + be32_to_cpu(dsb->sb_magicnum)); return -EWRONGFS; } if (!xfs_sb_good_version(sbp)) { - xfs_warn(mp, "bad version"); + xfs_warn(mp, +"Superblock has unknown features enabled or corrupted feature masks."); return -EWRONGFS; } -- cgit v1.2.3