2018-06-06 05:42:14 +03:00
// SPDX-License-Identifier: GPL-2.0
2013-08-12 14:49:39 +04:00
/*
* Copyright ( c ) 2000 - 2005 Silicon Graphics , Inc .
* Copyright ( c ) 2013 Red Hat , Inc .
* All Rights Reserved .
*/
# include "xfs.h"
# include "xfs_fs.h"
2013-10-23 03:36:05 +04:00
# include "xfs_shared.h"
2013-10-23 03:50:10 +04:00
# include "xfs_format.h"
# include "xfs_log_format.h"
# include "xfs_trans_resv.h"
2013-08-12 14:49:39 +04:00
# include "xfs_bit.h"
# include "xfs_mount.h"
2013-10-15 02:17:51 +04:00
# include "xfs_da_format.h"
2013-08-12 14:49:39 +04:00
# include "xfs_da_btree.h"
2013-10-23 03:51:50 +04:00
# include "xfs_inode.h"
2021-04-27 01:00:33 +03:00
# include "xfs_attr.h"
2013-08-12 14:49:39 +04:00
# include "xfs_attr_remote.h"
2013-10-23 03:50:10 +04:00
# include "xfs_trans.h"
2013-08-12 14:49:39 +04:00
# include "xfs_bmap.h"
# include "xfs_attr_leaf.h"
# include "xfs_quota.h"
2013-10-29 15:11:51 +04:00
# include "xfs_dir2.h"
2019-11-02 19:40:53 +03:00
# include "xfs_error.h"
2013-08-12 14:49:39 +04:00
/*
xfs: fix memory corruption during remote attr value buffer invalidation
While running generic/103, I observed what looks like memory corruption
and (with slub debugging turned on) a slub redzone warning on i386 when
inactivating an inode with a 64k remote attr value.
On a v5 filesystem, maximally sized remote attr values require one block
more than 64k worth of space to hold both the remote attribute value
header (64 bytes). On a 4k block filesystem this results in a 68k
buffer; on a 64k block filesystem, this would be a 128k buffer. Note
that even though we'll never use more than 65,600 bytes of this buffer,
XFS_MAX_BLOCKSIZE is 64k.
This is a problem because the definition of struct xfs_buf_log_format
allows for XFS_MAX_BLOCKSIZE worth of dirty bitmap (64k). On i386 when we
invalidate a remote attribute, xfs_trans_binval zeroes all 68k worth of
the dirty map, writing right off the end of the log item and corrupting
memory. We've gotten away with this on x86_64 for years because the
compiler inserts a u32 padding on the end of struct xfs_buf_log_format.
Fortunately for us, remote attribute values are written to disk with
xfs_bwrite(), which is to say that they are not logged. Fix the problem
by removing all places where we could end up creating a buffer log item
for a remote attribute value and leave a note explaining why. Next,
replace the open-coded buffer invalidation with a call to the helper we
created in the previous patch that does better checking for bad metadata
before marking the buffer stale.
Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com>
Reviewed-by: Christoph Hellwig <hch@lst.de>
2020-01-08 03:11:45 +03:00
* Invalidate any incore buffers associated with this remote attribute value
* extent . We never log remote attribute value buffers , which means that they
* won ' t be attached to a transaction and are therefore safe to mark stale .
* The actual bunmapi will be taken care of later .
2013-08-12 14:49:39 +04:00
*/
STATIC int
xfs: fix memory corruption during remote attr value buffer invalidation
While running generic/103, I observed what looks like memory corruption
and (with slub debugging turned on) a slub redzone warning on i386 when
inactivating an inode with a 64k remote attr value.
On a v5 filesystem, maximally sized remote attr values require one block
more than 64k worth of space to hold both the remote attribute value
header (64 bytes). On a 4k block filesystem this results in a 68k
buffer; on a 64k block filesystem, this would be a 128k buffer. Note
that even though we'll never use more than 65,600 bytes of this buffer,
XFS_MAX_BLOCKSIZE is 64k.
This is a problem because the definition of struct xfs_buf_log_format
allows for XFS_MAX_BLOCKSIZE worth of dirty bitmap (64k). On i386 when we
invalidate a remote attribute, xfs_trans_binval zeroes all 68k worth of
the dirty map, writing right off the end of the log item and corrupting
memory. We've gotten away with this on x86_64 for years because the
compiler inserts a u32 padding on the end of struct xfs_buf_log_format.
Fortunately for us, remote attribute values are written to disk with
xfs_bwrite(), which is to say that they are not logged. Fix the problem
by removing all places where we could end up creating a buffer log item
for a remote attribute value and leave a note explaining why. Next,
replace the open-coded buffer invalidation with a call to the helper we
created in the previous patch that does better checking for bad metadata
before marking the buffer stale.
Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com>
Reviewed-by: Christoph Hellwig <hch@lst.de>
2020-01-08 03:11:45 +03:00
xfs_attr3_rmt_stale (
2013-08-12 14:49:39 +04:00
struct xfs_inode * dp ,
xfs_dablk_t blkno ,
int blkcnt )
{
struct xfs_bmbt_irec map ;
int nmap ;
int error ;
/*
* Roll through the " value " , invalidating the attribute value ' s
* blocks .
*/
2020-01-15 01:31:49 +03:00
while ( blkcnt > 0 ) {
2013-08-12 14:49:39 +04:00
/*
* Try to remember where we decided to put the value .
*/
nmap = 1 ;
2020-01-15 01:31:49 +03:00
error = xfs_bmapi_read ( dp , ( xfs_fileoff_t ) blkno , blkcnt ,
2013-08-12 14:49:39 +04:00
& map , & nmap , XFS_BMAPI_ATTRFORK ) ;
xfs: fix memory corruption during remote attr value buffer invalidation
While running generic/103, I observed what looks like memory corruption
and (with slub debugging turned on) a slub redzone warning on i386 when
inactivating an inode with a 64k remote attr value.
On a v5 filesystem, maximally sized remote attr values require one block
more than 64k worth of space to hold both the remote attribute value
header (64 bytes). On a 4k block filesystem this results in a 68k
buffer; on a 64k block filesystem, this would be a 128k buffer. Note
that even though we'll never use more than 65,600 bytes of this buffer,
XFS_MAX_BLOCKSIZE is 64k.
This is a problem because the definition of struct xfs_buf_log_format
allows for XFS_MAX_BLOCKSIZE worth of dirty bitmap (64k). On i386 when we
invalidate a remote attribute, xfs_trans_binval zeroes all 68k worth of
the dirty map, writing right off the end of the log item and corrupting
memory. We've gotten away with this on x86_64 for years because the
compiler inserts a u32 padding on the end of struct xfs_buf_log_format.
Fortunately for us, remote attribute values are written to disk with
xfs_bwrite(), which is to say that they are not logged. Fix the problem
by removing all places where we could end up creating a buffer log item
for a remote attribute value and leave a note explaining why. Next,
replace the open-coded buffer invalidation with a call to the helper we
created in the previous patch that does better checking for bad metadata
before marking the buffer stale.
Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com>
Reviewed-by: Christoph Hellwig <hch@lst.de>
2020-01-08 03:11:45 +03:00
if ( error )
2014-06-22 09:03:54 +04:00
return error ;
xfs: fix memory corruption during remote attr value buffer invalidation
While running generic/103, I observed what looks like memory corruption
and (with slub debugging turned on) a slub redzone warning on i386 when
inactivating an inode with a 64k remote attr value.
On a v5 filesystem, maximally sized remote attr values require one block
more than 64k worth of space to hold both the remote attribute value
header (64 bytes). On a 4k block filesystem this results in a 68k
buffer; on a 64k block filesystem, this would be a 128k buffer. Note
that even though we'll never use more than 65,600 bytes of this buffer,
XFS_MAX_BLOCKSIZE is 64k.
This is a problem because the definition of struct xfs_buf_log_format
allows for XFS_MAX_BLOCKSIZE worth of dirty bitmap (64k). On i386 when we
invalidate a remote attribute, xfs_trans_binval zeroes all 68k worth of
the dirty map, writing right off the end of the log item and corrupting
memory. We've gotten away with this on x86_64 for years because the
compiler inserts a u32 padding on the end of struct xfs_buf_log_format.
Fortunately for us, remote attribute values are written to disk with
xfs_bwrite(), which is to say that they are not logged. Fix the problem
by removing all places where we could end up creating a buffer log item
for a remote attribute value and leave a note explaining why. Next,
replace the open-coded buffer invalidation with a call to the helper we
created in the previous patch that does better checking for bad metadata
before marking the buffer stale.
Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com>
Reviewed-by: Christoph Hellwig <hch@lst.de>
2020-01-08 03:11:45 +03:00
if ( XFS_IS_CORRUPT ( dp - > i_mount , nmap ! = 1 ) )
return - EFSCORRUPTED ;
2013-08-12 14:49:39 +04:00
/*
xfs: fix memory corruption during remote attr value buffer invalidation
While running generic/103, I observed what looks like memory corruption
and (with slub debugging turned on) a slub redzone warning on i386 when
inactivating an inode with a 64k remote attr value.
On a v5 filesystem, maximally sized remote attr values require one block
more than 64k worth of space to hold both the remote attribute value
header (64 bytes). On a 4k block filesystem this results in a 68k
buffer; on a 64k block filesystem, this would be a 128k buffer. Note
that even though we'll never use more than 65,600 bytes of this buffer,
XFS_MAX_BLOCKSIZE is 64k.
This is a problem because the definition of struct xfs_buf_log_format
allows for XFS_MAX_BLOCKSIZE worth of dirty bitmap (64k). On i386 when we
invalidate a remote attribute, xfs_trans_binval zeroes all 68k worth of
the dirty map, writing right off the end of the log item and corrupting
memory. We've gotten away with this on x86_64 for years because the
compiler inserts a u32 padding on the end of struct xfs_buf_log_format.
Fortunately for us, remote attribute values are written to disk with
xfs_bwrite(), which is to say that they are not logged. Fix the problem
by removing all places where we could end up creating a buffer log item
for a remote attribute value and leave a note explaining why. Next,
replace the open-coded buffer invalidation with a call to the helper we
created in the previous patch that does better checking for bad metadata
before marking the buffer stale.
Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com>
Reviewed-by: Christoph Hellwig <hch@lst.de>
2020-01-08 03:11:45 +03:00
* Mark any incore buffers for the remote value as stale . We
* never log remote attr value buffers , so the buffer should be
* easy to kill .
2013-08-12 14:49:39 +04:00
*/
xfs: fix memory corruption during remote attr value buffer invalidation
While running generic/103, I observed what looks like memory corruption
and (with slub debugging turned on) a slub redzone warning on i386 when
inactivating an inode with a 64k remote attr value.
On a v5 filesystem, maximally sized remote attr values require one block
more than 64k worth of space to hold both the remote attribute value
header (64 bytes). On a 4k block filesystem this results in a 68k
buffer; on a 64k block filesystem, this would be a 128k buffer. Note
that even though we'll never use more than 65,600 bytes of this buffer,
XFS_MAX_BLOCKSIZE is 64k.
This is a problem because the definition of struct xfs_buf_log_format
allows for XFS_MAX_BLOCKSIZE worth of dirty bitmap (64k). On i386 when we
invalidate a remote attribute, xfs_trans_binval zeroes all 68k worth of
the dirty map, writing right off the end of the log item and corrupting
memory. We've gotten away with this on x86_64 for years because the
compiler inserts a u32 padding on the end of struct xfs_buf_log_format.
Fortunately for us, remote attribute values are written to disk with
xfs_bwrite(), which is to say that they are not logged. Fix the problem
by removing all places where we could end up creating a buffer log item
for a remote attribute value and leave a note explaining why. Next,
replace the open-coded buffer invalidation with a call to the helper we
created in the previous patch that does better checking for bad metadata
before marking the buffer stale.
Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com>
Reviewed-by: Christoph Hellwig <hch@lst.de>
2020-01-08 03:11:45 +03:00
error = xfs_attr_rmtval_stale ( dp , & map , 0 ) ;
if ( error )
return error ;
2013-08-12 14:49:39 +04:00
2020-01-15 01:31:49 +03:00
blkno + = map . br_blockcount ;
blkcnt - = map . br_blockcount ;
2013-08-12 14:49:39 +04:00
}
2014-06-22 09:03:54 +04:00
return 0 ;
2013-08-12 14:49:39 +04:00
}
/*
* Invalidate all of the " remote " value regions pointed to by a particular
* leaf block .
* Note that we must release the lock on the buffer so that we are not
* caught holding something that the logging code wants to flush to disk .
*/
STATIC int
xfs_attr3_leaf_inactive (
2020-01-15 01:31:49 +03:00
struct xfs_trans * * trans ,
struct xfs_inode * dp ,
struct xfs_buf * bp )
2013-08-12 14:49:39 +04:00
{
2020-01-15 01:31:49 +03:00
struct xfs_attr3_icleaf_hdr ichdr ;
struct xfs_mount * mp = bp - > b_mount ;
struct xfs_attr_leafblock * leaf = bp - > b_addr ;
struct xfs_attr_leaf_entry * entry ;
2013-08-12 14:49:39 +04:00
struct xfs_attr_leaf_name_remote * name_rmt ;
2020-01-23 18:54:09 +03:00
int error = 0 ;
2020-01-15 01:31:49 +03:00
int i ;
2013-08-12 14:49:39 +04:00
2015-04-13 04:26:02 +03:00
xfs_attr3_leaf_hdr_from_disk ( mp - > m_attr_geo , & ichdr , leaf ) ;
2013-08-12 14:49:39 +04:00
/*
2020-01-15 01:31:49 +03:00
* Find the remote value extents for this leaf and invalidate their
* incore buffers .
2013-08-12 14:49:39 +04:00
*/
entry = xfs_attr3_leaf_entryp ( leaf ) ;
for ( i = 0 ; i < ichdr . count ; entry + + , i + + ) {
2020-01-15 01:31:49 +03:00
int blkcnt ;
2013-08-12 14:49:39 +04:00
2020-01-15 01:31:49 +03:00
if ( ! entry - > nameidx | | ( entry - > flags & XFS_ATTR_LOCAL ) )
continue ;
2013-08-12 14:49:39 +04:00
2020-01-15 01:31:49 +03:00
name_rmt = xfs_attr3_leaf_name_remote ( leaf , i ) ;
if ( ! name_rmt - > valueblk )
continue ;
2013-08-12 14:49:39 +04:00
2020-01-15 01:31:49 +03:00
blkcnt = xfs_attr3_rmt_blocks ( dp - > i_mount ,
be32_to_cpu ( name_rmt - > valuelen ) ) ;
error = xfs_attr3_rmt_stale ( dp ,
be32_to_cpu ( name_rmt - > valueblk ) , blkcnt ) ;
if ( error )
goto err ;
2013-08-12 14:49:39 +04:00
}
2020-01-15 01:31:49 +03:00
xfs_trans_brelse ( * trans , bp ) ;
err :
2013-08-12 14:49:39 +04:00
return error ;
}
/*
* Recurse ( gasp ! ) through the attribute nodes until we find leaves .
* We ' re doing a depth - first traversal in order to invalidate everything .
*/
STATIC int
xfs_attr3_node_inactive (
2019-11-09 01:57:48 +03:00
struct xfs_trans * * trans ,
struct xfs_inode * dp ,
struct xfs_buf * bp ,
int level )
2013-08-12 14:49:39 +04:00
{
2019-11-20 20:46:05 +03:00
struct xfs_mount * mp = dp - > i_mount ;
2019-11-09 01:57:48 +03:00
struct xfs_da_blkinfo * info ;
xfs_dablk_t child_fsb ;
xfs_daddr_t parent_blkno , child_blkno ;
struct xfs_buf * child_bp ;
2013-08-12 14:49:39 +04:00
struct xfs_da3_icnode_hdr ichdr ;
2019-11-09 01:57:48 +03:00
int error , i ;
2013-08-12 14:49:39 +04:00
/*
* Since this code is recursive ( gasp ! ) we must protect ourselves .
*/
if ( level > XFS_DA_NODE_MAXDEPTH ) {
2020-03-11 20:37:54 +03:00
xfs_buf_mark_corrupt ( bp ) ;
2020-03-11 20:37:53 +03:00
xfs_trans_brelse ( * trans , bp ) ; /* no locks for later trans */
2019-10-29 02:12:34 +03:00
return - EFSCORRUPTED ;
2013-08-12 14:49:39 +04:00
}
2019-11-09 01:57:48 +03:00
xfs_da3_node_hdr_from_disk ( dp - > i_mount , & ichdr , bp - > b_addr ) ;
2021-08-19 04:47:05 +03:00
parent_blkno = xfs_buf_daddr ( bp ) ;
2013-08-12 14:49:39 +04:00
if ( ! ichdr . count ) {
xfs_trans_brelse ( * trans , bp ) ;
return 0 ;
}
2019-11-09 01:57:48 +03:00
child_fsb = be32_to_cpu ( ichdr . btree [ 0 ] . before ) ;
2013-08-12 14:49:39 +04:00
xfs_trans_brelse ( * trans , bp ) ; /* no locks for later trans */
xfs: fix use-after-free in xattr node block inactivation
The kernel build robot reported a UAF error while running xfs/433
(edited somewhat for brevity):
BUG: KASAN: use-after-free in xfs_attr3_node_inactive (fs/xfs/xfs_attr_inactive.c:214) xfs
Read of size 4 at addr ffff88820ac2bd44 by task kworker/0:2/139
CPU: 0 PID: 139 Comm: kworker/0:2 Tainted: G S 5.19.0-rc2-00004-g7cf2b0f9611b #1
Hardware name: Hewlett-Packard p6-1451cx/2ADA, BIOS 8.15 02/05/2013
Workqueue: xfs-inodegc/sdb4 xfs_inodegc_worker [xfs]
Call Trace:
<TASK>
dump_stack_lvl (lib/dump_stack.c:107 (discriminator 1))
print_address_description+0x1f/0x200
print_report.cold (mm/kasan/report.c:430)
kasan_report (mm/kasan/report.c:162 mm/kasan/report.c:493)
xfs_attr3_node_inactive (fs/xfs/xfs_attr_inactive.c:214) xfs
xfs_attr3_root_inactive (fs/xfs/xfs_attr_inactive.c:296) xfs
xfs_attr_inactive (fs/xfs/xfs_attr_inactive.c:371) xfs
xfs_inactive (fs/xfs/xfs_inode.c:1781) xfs
xfs_inodegc_worker (fs/xfs/xfs_icache.c:1837 fs/xfs/xfs_icache.c:1860) xfs
process_one_work
worker_thread
kthread
ret_from_fork
</TASK>
Allocated by task 139:
kasan_save_stack (mm/kasan/common.c:39)
__kasan_slab_alloc (mm/kasan/common.c:45 mm/kasan/common.c:436 mm/kasan/common.c:469)
kmem_cache_alloc (mm/slab.h:750 mm/slub.c:3214 mm/slub.c:3222 mm/slub.c:3229 mm/slub.c:3239)
_xfs_buf_alloc (include/linux/instrumented.h:86 include/linux/atomic/atomic-instrumented.h:41 fs/xfs/xfs_buf.c:232) xfs
xfs_buf_get_map (fs/xfs/xfs_buf.c:660) xfs
xfs_buf_read_map (fs/xfs/xfs_buf.c:777) xfs
xfs_trans_read_buf_map (fs/xfs/xfs_trans_buf.c:289) xfs
xfs_da_read_buf (fs/xfs/libxfs/xfs_da_btree.c:2652) xfs
xfs_da3_node_read (fs/xfs/libxfs/xfs_da_btree.c:392) xfs
xfs_attr3_root_inactive (fs/xfs/xfs_attr_inactive.c:272) xfs
xfs_attr_inactive (fs/xfs/xfs_attr_inactive.c:371) xfs
xfs_inactive (fs/xfs/xfs_inode.c:1781) xfs
xfs_inodegc_worker (fs/xfs/xfs_icache.c:1837 fs/xfs/xfs_icache.c:1860) xfs
process_one_work
worker_thread
kthread
ret_from_fork
Freed by task 139:
kasan_save_stack (mm/kasan/common.c:39)
kasan_set_track (mm/kasan/common.c:45)
kasan_set_free_info (mm/kasan/generic.c:372)
__kasan_slab_free (mm/kasan/common.c:368 mm/kasan/common.c:328 mm/kasan/common.c:374)
kmem_cache_free (mm/slub.c:1753 mm/slub.c:3507 mm/slub.c:3524)
xfs_buf_rele (fs/xfs/xfs_buf.c:1040) xfs
xfs_attr3_node_inactive (fs/xfs/xfs_attr_inactive.c:210) xfs
xfs_attr3_root_inactive (fs/xfs/xfs_attr_inactive.c:296) xfs
xfs_attr_inactive (fs/xfs/xfs_attr_inactive.c:371) xfs
xfs_inactive (fs/xfs/xfs_inode.c:1781) xfs
xfs_inodegc_worker (fs/xfs/xfs_icache.c:1837 fs/xfs/xfs_icache.c:1860) xfs
process_one_work
worker_thread
kthread
ret_from_fork
I reproduced this for my own satisfaction, and got the same report,
along with an extra morsel:
The buggy address belongs to the object at ffff88802103a800
which belongs to the cache xfs_buf of size 432
The buggy address is located 396 bytes inside of
432-byte region [ffff88802103a800, ffff88802103a9b0)
I tracked this code down to:
error = xfs_trans_get_buf(*trans, mp->m_ddev_targp,
child_blkno,
XFS_FSB_TO_BB(mp, mp->m_attr_geo->fsbcount), 0,
&child_bp);
if (error)
return error;
error = bp->b_error;
That doesn't look right -- I think this should be dereferencing
child_bp, not bp. Looking through the codebase history, I think this
was added by commit 2911edb653b9 ("xfs: remove the mappedbno argument to
xfs_da_get_buf"), which replaced a call to xfs_da_get_buf with the
current call to xfs_trans_get_buf. Not sure why we trans_brelse'd @bp
earlier in the function, but I'm guessing it's to avoid pinning too many
buffers in memory while we inactivate the bottom of the attr tree.
Hence we now have to get the buffer back.
I /think/ this was supposed to check child_bp->b_error and fail the rest
of the invalidation if child_bp had experienced any kind of IO or
corruption error. I bet the xfs_da3_node_read earlier in the loop will
catch most cases of incoming on-disk corruption which makes this check
mostly moot unless someone corrupts the buffer and the AIL pushes it out
to disk while the buffer's unlocked.
In the first case we'll never get to the bad check, and in the second
case the AIL will shut down the log, at which point there's no reason to
check b_error. Remove the check, and null out @bp to avoid this problem
in the future.
Cc: hch@lst.de
Reported-by: kernel test robot <oliver.sang@intel.com>
Fixes: 2911edb653b9 ("xfs: remove the mappedbno argument to xfs_da_get_buf")
Signed-off-by: Darrick J. Wong <djwong@kernel.org>
Reviewed-by: Dave Chinner <dchinner@redhat.com>
Reviewed-by: Christoph Hellwig <hch@lst.de>
2022-07-14 19:22:53 +03:00
bp = NULL ;
2013-08-12 14:49:39 +04:00
/*
* If this is the node level just above the leaves , simply loop
* over the leaves removing all of them . If this is higher up
* in the tree , recurse downward .
*/
for ( i = 0 ; i < ichdr . count ; i + + ) {
/*
* Read the subsidiary block to see what we have to work with .
* Don ' t do this in a transaction . This is a depth - first
* traversal of the tree so we may deal with many blocks
* before we come back to this one .
*/
2019-11-20 20:46:04 +03:00
error = xfs_da3_node_read ( * trans , dp , child_fsb , & child_bp ,
2017-10-18 00:16:28 +03:00
XFS_ATTR_FORK ) ;
2013-08-12 14:49:39 +04:00
if ( error )
2014-06-22 09:03:54 +04:00
return error ;
2013-08-12 14:49:39 +04:00
2017-10-18 00:16:28 +03:00
/* save for re-read later */
2021-08-19 04:46:57 +03:00
child_blkno = xfs_buf_daddr ( child_bp ) ;
2013-08-12 14:49:39 +04:00
2017-10-18 00:16:28 +03:00
/*
* Invalidate the subtree , however we have to .
*/
info = child_bp - > b_addr ;
switch ( info - > magic ) {
case cpu_to_be16 ( XFS_DA_NODE_MAGIC ) :
case cpu_to_be16 ( XFS_DA3_NODE_MAGIC ) :
error = xfs_attr3_node_inactive ( trans , dp , child_bp ,
level + 1 ) ;
break ;
case cpu_to_be16 ( XFS_ATTR_LEAF_MAGIC ) :
case cpu_to_be16 ( XFS_ATTR3_LEAF_MAGIC ) :
error = xfs_attr3_leaf_inactive ( trans , dp , child_bp ) ;
break ;
default :
2020-03-11 20:37:54 +03:00
xfs_buf_mark_corrupt ( child_bp ) ;
2017-10-18 00:16:28 +03:00
xfs_trans_brelse ( * trans , child_bp ) ;
2019-11-02 19:40:53 +03:00
error = - EFSCORRUPTED ;
2017-10-18 00:16:28 +03:00
break ;
2013-08-12 14:49:39 +04:00
}
2017-10-18 00:16:28 +03:00
if ( error )
return error ;
/*
* Remove the subsidiary block from the cache and from the log .
*/
2020-01-24 04:01:18 +03:00
error = xfs_trans_get_buf ( * trans , mp - > m_ddev_targp ,
2019-11-20 20:46:05 +03:00
child_blkno ,
2020-01-24 04:01:18 +03:00
XFS_FSB_TO_BB ( mp , mp - > m_attr_geo - > fsbcount ) , 0 ,
& child_bp ) ;
if ( error )
return error ;
2017-10-18 00:16:28 +03:00
xfs_trans_binval ( * trans , child_bp ) ;
xfs: fix use-after-free in xattr node block inactivation
The kernel build robot reported a UAF error while running xfs/433
(edited somewhat for brevity):
BUG: KASAN: use-after-free in xfs_attr3_node_inactive (fs/xfs/xfs_attr_inactive.c:214) xfs
Read of size 4 at addr ffff88820ac2bd44 by task kworker/0:2/139
CPU: 0 PID: 139 Comm: kworker/0:2 Tainted: G S 5.19.0-rc2-00004-g7cf2b0f9611b #1
Hardware name: Hewlett-Packard p6-1451cx/2ADA, BIOS 8.15 02/05/2013
Workqueue: xfs-inodegc/sdb4 xfs_inodegc_worker [xfs]
Call Trace:
<TASK>
dump_stack_lvl (lib/dump_stack.c:107 (discriminator 1))
print_address_description+0x1f/0x200
print_report.cold (mm/kasan/report.c:430)
kasan_report (mm/kasan/report.c:162 mm/kasan/report.c:493)
xfs_attr3_node_inactive (fs/xfs/xfs_attr_inactive.c:214) xfs
xfs_attr3_root_inactive (fs/xfs/xfs_attr_inactive.c:296) xfs
xfs_attr_inactive (fs/xfs/xfs_attr_inactive.c:371) xfs
xfs_inactive (fs/xfs/xfs_inode.c:1781) xfs
xfs_inodegc_worker (fs/xfs/xfs_icache.c:1837 fs/xfs/xfs_icache.c:1860) xfs
process_one_work
worker_thread
kthread
ret_from_fork
</TASK>
Allocated by task 139:
kasan_save_stack (mm/kasan/common.c:39)
__kasan_slab_alloc (mm/kasan/common.c:45 mm/kasan/common.c:436 mm/kasan/common.c:469)
kmem_cache_alloc (mm/slab.h:750 mm/slub.c:3214 mm/slub.c:3222 mm/slub.c:3229 mm/slub.c:3239)
_xfs_buf_alloc (include/linux/instrumented.h:86 include/linux/atomic/atomic-instrumented.h:41 fs/xfs/xfs_buf.c:232) xfs
xfs_buf_get_map (fs/xfs/xfs_buf.c:660) xfs
xfs_buf_read_map (fs/xfs/xfs_buf.c:777) xfs
xfs_trans_read_buf_map (fs/xfs/xfs_trans_buf.c:289) xfs
xfs_da_read_buf (fs/xfs/libxfs/xfs_da_btree.c:2652) xfs
xfs_da3_node_read (fs/xfs/libxfs/xfs_da_btree.c:392) xfs
xfs_attr3_root_inactive (fs/xfs/xfs_attr_inactive.c:272) xfs
xfs_attr_inactive (fs/xfs/xfs_attr_inactive.c:371) xfs
xfs_inactive (fs/xfs/xfs_inode.c:1781) xfs
xfs_inodegc_worker (fs/xfs/xfs_icache.c:1837 fs/xfs/xfs_icache.c:1860) xfs
process_one_work
worker_thread
kthread
ret_from_fork
Freed by task 139:
kasan_save_stack (mm/kasan/common.c:39)
kasan_set_track (mm/kasan/common.c:45)
kasan_set_free_info (mm/kasan/generic.c:372)
__kasan_slab_free (mm/kasan/common.c:368 mm/kasan/common.c:328 mm/kasan/common.c:374)
kmem_cache_free (mm/slub.c:1753 mm/slub.c:3507 mm/slub.c:3524)
xfs_buf_rele (fs/xfs/xfs_buf.c:1040) xfs
xfs_attr3_node_inactive (fs/xfs/xfs_attr_inactive.c:210) xfs
xfs_attr3_root_inactive (fs/xfs/xfs_attr_inactive.c:296) xfs
xfs_attr_inactive (fs/xfs/xfs_attr_inactive.c:371) xfs
xfs_inactive (fs/xfs/xfs_inode.c:1781) xfs
xfs_inodegc_worker (fs/xfs/xfs_icache.c:1837 fs/xfs/xfs_icache.c:1860) xfs
process_one_work
worker_thread
kthread
ret_from_fork
I reproduced this for my own satisfaction, and got the same report,
along with an extra morsel:
The buggy address belongs to the object at ffff88802103a800
which belongs to the cache xfs_buf of size 432
The buggy address is located 396 bytes inside of
432-byte region [ffff88802103a800, ffff88802103a9b0)
I tracked this code down to:
error = xfs_trans_get_buf(*trans, mp->m_ddev_targp,
child_blkno,
XFS_FSB_TO_BB(mp, mp->m_attr_geo->fsbcount), 0,
&child_bp);
if (error)
return error;
error = bp->b_error;
That doesn't look right -- I think this should be dereferencing
child_bp, not bp. Looking through the codebase history, I think this
was added by commit 2911edb653b9 ("xfs: remove the mappedbno argument to
xfs_da_get_buf"), which replaced a call to xfs_da_get_buf with the
current call to xfs_trans_get_buf. Not sure why we trans_brelse'd @bp
earlier in the function, but I'm guessing it's to avoid pinning too many
buffers in memory while we inactivate the bottom of the attr tree.
Hence we now have to get the buffer back.
I /think/ this was supposed to check child_bp->b_error and fail the rest
of the invalidation if child_bp had experienced any kind of IO or
corruption error. I bet the xfs_da3_node_read earlier in the loop will
catch most cases of incoming on-disk corruption which makes this check
mostly moot unless someone corrupts the buffer and the AIL pushes it out
to disk while the buffer's unlocked.
In the first case we'll never get to the bad check, and in the second
case the AIL will shut down the log, at which point there's no reason to
check b_error. Remove the check, and null out @bp to avoid this problem
in the future.
Cc: hch@lst.de
Reported-by: kernel test robot <oliver.sang@intel.com>
Fixes: 2911edb653b9 ("xfs: remove the mappedbno argument to xfs_da_get_buf")
Signed-off-by: Darrick J. Wong <djwong@kernel.org>
Reviewed-by: Dave Chinner <dchinner@redhat.com>
Reviewed-by: Christoph Hellwig <hch@lst.de>
2022-07-14 19:22:53 +03:00
child_bp = NULL ;
2013-08-12 14:49:39 +04:00
/*
* If we ' re not done , re - read the parent to get the next
* child block number .
*/
if ( i + 1 < ichdr . count ) {
2019-11-09 01:57:48 +03:00
struct xfs_da3_icnode_hdr phdr ;
2019-11-20 20:46:04 +03:00
error = xfs_da3_node_read_mapped ( * trans , dp ,
parent_blkno , & bp , XFS_ATTR_FORK ) ;
2013-08-12 14:49:39 +04:00
if ( error )
return error ;
2019-11-09 01:57:48 +03:00
xfs_da3_node_hdr_from_disk ( dp - > i_mount , & phdr ,
bp - > b_addr ) ;
child_fsb = be32_to_cpu ( phdr . btree [ i + 1 ] . before ) ;
2013-08-12 14:49:39 +04:00
xfs_trans_brelse ( * trans , bp ) ;
xfs: fix use-after-free in xattr node block inactivation
The kernel build robot reported a UAF error while running xfs/433
(edited somewhat for brevity):
BUG: KASAN: use-after-free in xfs_attr3_node_inactive (fs/xfs/xfs_attr_inactive.c:214) xfs
Read of size 4 at addr ffff88820ac2bd44 by task kworker/0:2/139
CPU: 0 PID: 139 Comm: kworker/0:2 Tainted: G S 5.19.0-rc2-00004-g7cf2b0f9611b #1
Hardware name: Hewlett-Packard p6-1451cx/2ADA, BIOS 8.15 02/05/2013
Workqueue: xfs-inodegc/sdb4 xfs_inodegc_worker [xfs]
Call Trace:
<TASK>
dump_stack_lvl (lib/dump_stack.c:107 (discriminator 1))
print_address_description+0x1f/0x200
print_report.cold (mm/kasan/report.c:430)
kasan_report (mm/kasan/report.c:162 mm/kasan/report.c:493)
xfs_attr3_node_inactive (fs/xfs/xfs_attr_inactive.c:214) xfs
xfs_attr3_root_inactive (fs/xfs/xfs_attr_inactive.c:296) xfs
xfs_attr_inactive (fs/xfs/xfs_attr_inactive.c:371) xfs
xfs_inactive (fs/xfs/xfs_inode.c:1781) xfs
xfs_inodegc_worker (fs/xfs/xfs_icache.c:1837 fs/xfs/xfs_icache.c:1860) xfs
process_one_work
worker_thread
kthread
ret_from_fork
</TASK>
Allocated by task 139:
kasan_save_stack (mm/kasan/common.c:39)
__kasan_slab_alloc (mm/kasan/common.c:45 mm/kasan/common.c:436 mm/kasan/common.c:469)
kmem_cache_alloc (mm/slab.h:750 mm/slub.c:3214 mm/slub.c:3222 mm/slub.c:3229 mm/slub.c:3239)
_xfs_buf_alloc (include/linux/instrumented.h:86 include/linux/atomic/atomic-instrumented.h:41 fs/xfs/xfs_buf.c:232) xfs
xfs_buf_get_map (fs/xfs/xfs_buf.c:660) xfs
xfs_buf_read_map (fs/xfs/xfs_buf.c:777) xfs
xfs_trans_read_buf_map (fs/xfs/xfs_trans_buf.c:289) xfs
xfs_da_read_buf (fs/xfs/libxfs/xfs_da_btree.c:2652) xfs
xfs_da3_node_read (fs/xfs/libxfs/xfs_da_btree.c:392) xfs
xfs_attr3_root_inactive (fs/xfs/xfs_attr_inactive.c:272) xfs
xfs_attr_inactive (fs/xfs/xfs_attr_inactive.c:371) xfs
xfs_inactive (fs/xfs/xfs_inode.c:1781) xfs
xfs_inodegc_worker (fs/xfs/xfs_icache.c:1837 fs/xfs/xfs_icache.c:1860) xfs
process_one_work
worker_thread
kthread
ret_from_fork
Freed by task 139:
kasan_save_stack (mm/kasan/common.c:39)
kasan_set_track (mm/kasan/common.c:45)
kasan_set_free_info (mm/kasan/generic.c:372)
__kasan_slab_free (mm/kasan/common.c:368 mm/kasan/common.c:328 mm/kasan/common.c:374)
kmem_cache_free (mm/slub.c:1753 mm/slub.c:3507 mm/slub.c:3524)
xfs_buf_rele (fs/xfs/xfs_buf.c:1040) xfs
xfs_attr3_node_inactive (fs/xfs/xfs_attr_inactive.c:210) xfs
xfs_attr3_root_inactive (fs/xfs/xfs_attr_inactive.c:296) xfs
xfs_attr_inactive (fs/xfs/xfs_attr_inactive.c:371) xfs
xfs_inactive (fs/xfs/xfs_inode.c:1781) xfs
xfs_inodegc_worker (fs/xfs/xfs_icache.c:1837 fs/xfs/xfs_icache.c:1860) xfs
process_one_work
worker_thread
kthread
ret_from_fork
I reproduced this for my own satisfaction, and got the same report,
along with an extra morsel:
The buggy address belongs to the object at ffff88802103a800
which belongs to the cache xfs_buf of size 432
The buggy address is located 396 bytes inside of
432-byte region [ffff88802103a800, ffff88802103a9b0)
I tracked this code down to:
error = xfs_trans_get_buf(*trans, mp->m_ddev_targp,
child_blkno,
XFS_FSB_TO_BB(mp, mp->m_attr_geo->fsbcount), 0,
&child_bp);
if (error)
return error;
error = bp->b_error;
That doesn't look right -- I think this should be dereferencing
child_bp, not bp. Looking through the codebase history, I think this
was added by commit 2911edb653b9 ("xfs: remove the mappedbno argument to
xfs_da_get_buf"), which replaced a call to xfs_da_get_buf with the
current call to xfs_trans_get_buf. Not sure why we trans_brelse'd @bp
earlier in the function, but I'm guessing it's to avoid pinning too many
buffers in memory while we inactivate the bottom of the attr tree.
Hence we now have to get the buffer back.
I /think/ this was supposed to check child_bp->b_error and fail the rest
of the invalidation if child_bp had experienced any kind of IO or
corruption error. I bet the xfs_da3_node_read earlier in the loop will
catch most cases of incoming on-disk corruption which makes this check
mostly moot unless someone corrupts the buffer and the AIL pushes it out
to disk while the buffer's unlocked.
In the first case we'll never get to the bad check, and in the second
case the AIL will shut down the log, at which point there's no reason to
check b_error. Remove the check, and null out @bp to avoid this problem
in the future.
Cc: hch@lst.de
Reported-by: kernel test robot <oliver.sang@intel.com>
Fixes: 2911edb653b9 ("xfs: remove the mappedbno argument to xfs_da_get_buf")
Signed-off-by: Darrick J. Wong <djwong@kernel.org>
Reviewed-by: Dave Chinner <dchinner@redhat.com>
Reviewed-by: Christoph Hellwig <hch@lst.de>
2022-07-14 19:22:53 +03:00
bp = NULL ;
2013-08-12 14:49:39 +04:00
}
/*
* Atomically commit the whole invalidate stuff .
*/
2017-08-28 20:21:03 +03:00
error = xfs_trans_roll_inode ( trans , dp ) ;
2013-08-12 14:49:39 +04:00
if ( error )
return error ;
}
return 0 ;
}
/*
* Indiscriminately delete the entire attribute fork
*
* Recurse ( gasp ! ) through the attribute nodes until we find leaves .
* We ' re doing a depth - first traversal in order to invalidate everything .
*/
2016-06-01 10:38:15 +03:00
static int
2013-08-12 14:49:39 +04:00
xfs_attr3_root_inactive (
struct xfs_trans * * trans ,
struct xfs_inode * dp )
{
2019-11-20 20:46:05 +03:00
struct xfs_mount * mp = dp - > i_mount ;
2013-08-12 14:49:39 +04:00
struct xfs_da_blkinfo * info ;
struct xfs_buf * bp ;
xfs_daddr_t blkno ;
int error ;
/*
* Read block 0 to see what we have to work with .
* We only get here if we have extents , since we remove
* the extents in reverse order the extent containing
* block 0 must still be there .
*/
2019-11-20 20:46:04 +03:00
error = xfs_da3_node_read ( * trans , dp , 0 , & bp , XFS_ATTR_FORK ) ;
2013-08-12 14:49:39 +04:00
if ( error )
return error ;
2021-08-19 04:47:05 +03:00
blkno = xfs_buf_daddr ( bp ) ;
2013-08-12 14:49:39 +04:00
/*
* Invalidate the tree , even if the " tree " is only a single leaf block .
* This is a depth - first traversal !
*/
info = bp - > b_addr ;
switch ( info - > magic ) {
case cpu_to_be16 ( XFS_DA_NODE_MAGIC ) :
case cpu_to_be16 ( XFS_DA3_NODE_MAGIC ) :
error = xfs_attr3_node_inactive ( trans , dp , bp , 1 ) ;
break ;
case cpu_to_be16 ( XFS_ATTR_LEAF_MAGIC ) :
case cpu_to_be16 ( XFS_ATTR3_LEAF_MAGIC ) :
error = xfs_attr3_leaf_inactive ( trans , dp , bp ) ;
break ;
default :
2019-10-29 02:12:34 +03:00
error = - EFSCORRUPTED ;
2020-03-11 20:37:54 +03:00
xfs_buf_mark_corrupt ( bp ) ;
2013-08-12 14:49:39 +04:00
xfs_trans_brelse ( * trans , bp ) ;
break ;
}
if ( error )
return error ;
/*
* Invalidate the incore copy of the root block .
*/
2020-01-24 04:01:18 +03:00
error = xfs_trans_get_buf ( * trans , mp - > m_ddev_targp , blkno ,
XFS_FSB_TO_BB ( mp , mp - > m_attr_geo - > fsbcount ) , 0 , & bp ) ;
if ( error )
return error ;
2019-11-20 20:46:05 +03:00
error = bp - > b_error ;
if ( error ) {
xfs_trans_brelse ( * trans , bp ) ;
2013-08-12 14:49:39 +04:00
return error ;
2019-11-20 20:46:05 +03:00
}
2013-08-12 14:49:39 +04:00
xfs_trans_binval ( * trans , bp ) ; /* remove from cache */
/*
* Commit the invalidate and start the next transaction .
*/
2017-08-28 20:21:03 +03:00
error = xfs_trans_roll_inode ( trans , dp ) ;
2013-08-12 14:49:39 +04:00
return error ;
}
2015-05-29 00:40:08 +03:00
/*
* xfs_attr_inactive kills all traces of an attribute fork on an inode . It
* removes both the on - disk and in - memory inode fork . Note that this also has to
* handle the condition of inodes without attributes but with an attribute fork
* configured , so we can ' t use xfs_inode_hasattr ( ) here .
*
* The in - memory attribute fork is removed even on error .
*/
2013-08-12 14:49:39 +04:00
int
2015-05-29 00:40:08 +03:00
xfs_attr_inactive (
struct xfs_inode * dp )
2013-08-12 14:49:39 +04:00
{
2015-05-29 00:40:08 +03:00
struct xfs_trans * trans ;
struct xfs_mount * mp ;
int lock_mode = XFS_ILOCK_SHARED ;
int error = 0 ;
2013-08-12 14:49:39 +04:00
mp = dp - > i_mount ;
ASSERT ( ! XFS_NOT_DQATTACHED ( mp , dp ) ) ;
2015-05-29 00:40:08 +03:00
xfs_ilock ( dp , lock_mode ) ;
2022-07-09 20:56:06 +03:00
if ( ! xfs_inode_has_attr_fork ( dp ) )
2015-05-29 00:40:08 +03:00
goto out_destroy_fork ;
xfs_iunlock ( dp , lock_mode ) ;
2013-08-12 14:49:39 +04:00
2015-05-29 00:40:08 +03:00
lock_mode = 0 ;
2016-04-06 02:19:55 +03:00
error = xfs_trans_alloc ( mp , & M_RES ( mp ) - > tr_attrinval , 0 , 0 , 0 , & trans ) ;
2015-05-29 00:40:08 +03:00
if ( error )
2016-04-06 02:19:55 +03:00
goto out_destroy_fork ;
2015-05-29 00:40:08 +03:00
lock_mode = XFS_ILOCK_EXCL ;
xfs_ilock ( dp , lock_mode ) ;
2022-07-09 20:56:06 +03:00
if ( ! xfs_inode_has_attr_fork ( dp ) )
2015-05-29 00:40:08 +03:00
goto out_cancel ;
2013-08-12 14:49:39 +04:00
/*
* No need to make quota reservations here . We expect to release some
* blocks , not allocate , in the common case .
*/
xfs_trans_ijoin ( trans , dp , 0 ) ;
2015-06-23 01:47:20 +03:00
/*
* Invalidate and truncate the attribute fork extents . Make sure the
2022-07-09 20:56:06 +03:00
* fork actually has xattr blocks as otherwise the invalidation has no
2015-06-23 01:47:20 +03:00
* blocks to read and returns an error . In this case , just do the fork
* removal below .
*/
2022-07-09 20:56:06 +03:00
if ( dp - > i_af . if_nextents > 0 ) {
2015-05-29 00:40:08 +03:00
error = xfs_attr3_root_inactive ( & trans , dp ) ;
if ( error )
goto out_cancel ;
error = xfs_itruncate_extents ( & trans , dp , XFS_ATTR_FORK , 0 ) ;
if ( error )
goto out_cancel ;
2013-08-12 14:49:39 +04:00
}
2015-05-29 00:40:08 +03:00
/* Reset the attribute fork - this also destroys the in-core fork */
xfs_attr_fork_remove ( dp , trans ) ;
2013-08-12 14:49:39 +04:00
2015-06-04 06:48:08 +03:00
error = xfs_trans_commit ( trans ) ;
2015-05-29 00:40:08 +03:00
xfs_iunlock ( dp , lock_mode ) ;
2014-06-22 09:03:54 +04:00
return error ;
2013-08-12 14:49:39 +04:00
2015-05-29 00:40:08 +03:00
out_cancel :
2015-06-04 06:47:56 +03:00
xfs_trans_cancel ( trans ) ;
2015-05-29 00:40:08 +03:00
out_destroy_fork :
/* kill the in-core attr fork before we drop the inode lock */
2022-07-09 20:56:06 +03:00
xfs_ifork_zap_attr ( dp ) ;
2015-05-29 00:40:08 +03:00
if ( lock_mode )
xfs_iunlock ( dp , lock_mode ) ;
2014-06-22 09:03:54 +04:00
return error ;
2013-08-12 14:49:39 +04:00
}