| Message ID | 20230715063647.2094989-4-leo.lilong@huawei.com (mailing list archive) |
|---|---|
| State | New, archived |
| Headers | show |
| Series | xfs: fix two problem when recovery intents fails | expand |
On Sat, Jul 15, 2023 at 02:36:47PM +0800, Long Li wrote: > KASAN report a uaf when recover intents fails: > > ================================================================== > BUG: KASAN: slab-use-after-free in xfs_cui_release+0xb7/0xc0 > Read of size 4 at addr ffff888012575e60 by task kworker/u8:3/103 > CPU: 3 PID: 103 Comm: kworker/u8:3 Not tainted 6.4.0-rc7-next-20230619-00003-g94543a53f9a4-dirty #166 > Workqueue: xfs-cil/sda xlog_cil_push_work > Call Trace: > <TASK> > dump_stack_lvl+0x50/0x70 > print_report+0xc2/0x600 > kasan_report+0xb6/0xe0 > xfs_cui_release+0xb7/0xc0 > xfs_cud_item_release+0x3c/0x90 > xfs_trans_committed_bulk+0x2d5/0x7f0 > xlog_cil_committed+0xaba/0xf20 > xlog_cil_push_work+0x1a60/0x2360 > process_one_work+0x78e/0x1140 > worker_thread+0x58b/0xf60 > kthread+0x2cd/0x3c0 > ret_from_fork+0x1f/0x30 > </TASK> > > Allocated by task 531: > kasan_save_stack+0x22/0x40 > kasan_set_track+0x25/0x30 > __kasan_slab_alloc+0x55/0x60 > kmem_cache_alloc+0x195/0x5f0 > xfs_cui_init+0x198/0x1d0 > xlog_recover_cui_commit_pass2+0x133/0x5f0 > xlog_recover_items_pass2+0x107/0x230 > xlog_recover_commit_trans+0x3e7/0x9c0 > xlog_recovery_process_trans+0x140/0x1d0 > xlog_recover_process_ophdr+0x1a0/0x3d0 > xlog_recover_process_data+0x108/0x2d0 > xlog_recover_process+0x1f6/0x280 > xlog_do_recovery_pass+0x609/0xdb0 > xlog_do_log_recovery+0x84/0xe0 > xlog_do_recover+0x7d/0x470 > xlog_recover+0x25f/0x490 > xfs_log_mount+0x2dd/0x6f0 > xfs_mountfs+0x11ce/0x1e70 > xfs_fs_fill_super+0x10ec/0x1b20 > get_tree_bdev+0x3c8/0x730 > vfs_get_tree+0x89/0x2c0 > path_mount+0xecf/0x1800 > do_mount+0xf3/0x110 > __x64_sys_mount+0x154/0x1f0 > do_syscall_64+0x39/0x80 > entry_SYSCALL_64_after_hwframe+0x63/0xcd > > Freed by task 531: > kasan_save_stack+0x22/0x40 > kasan_set_track+0x25/0x30 > kasan_save_free_info+0x2b/0x40 > __kasan_slab_free+0x114/0x1b0 > kmem_cache_free+0xf8/0x510 > xfs_cui_item_free+0x95/0xb0 > xfs_cui_release+0x86/0xc0 > xlog_recover_cancel_intents.isra.0+0xf8/0x210 > xlog_recover_finish+0x7e7/0x980 > xfs_log_mount_finish+0x2bb/0x4a0 > xfs_mountfs+0x14bf/0x1e70 > xfs_fs_fill_super+0x10ec/0x1b20 > get_tree_bdev+0x3c8/0x730 > vfs_get_tree+0x89/0x2c0 > path_mount+0xecf/0x1800 > do_mount+0xf3/0x110 > __x64_sys_mount+0x154/0x1f0 > do_syscall_64+0x39/0x80 > entry_SYSCALL_64_after_hwframe+0x63/0xcd > > The buggy address belongs to the object at ffff888012575dc8 > which belongs to the cache xfs_cui_item of size 432 > The buggy address is located 152 bytes inside of > freed 432-byte region [ffff888012575dc8, ffff888012575f78) > > The buggy address belongs to the physical page: > page:ffffea0000495d00 refcount:1 mapcount:0 mapping:0000000000000000 index:0xffff888012576208 pfn:0x12574 > head:ffffea0000495d00 order:2 entire_mapcount:0 nr_pages_mapped:0 pincount:0 > flags: 0x1fffff80010200(slab|head|node=0|zone=1|lastcpupid=0x1fffff) > page_type: 0xffffffff() > raw: 001fffff80010200 ffff888012092f40 ffff888014570150 ffff888014570150 > raw: ffff888012576208 00000000001e0010 00000001ffffffff 0000000000000000 > page dumped because: kasan: bad access detected > > Memory state around the buggy address: > ffff888012575d00: fb fb fb fb fb fb fb fb fb fb fb fc fc fc fc fc > ffff888012575d80: fc fc fc fc fc fc fc fc fc fa fb fb fb fb fb fb > >ffff888012575e00: fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb > ^ > ffff888012575e80: fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb > ffff888012575f00: fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb fc > ================================================================== > > If process intents fails, intent items left in AIL will be delete > from AIL and freed in error handling, even intent items that have been > recovered and created done items. After this, uaf will be triggered when > done item commited, because at this point the released intent item will > be accessed. > > xlog_recover_finish xlog_cil_push_work > ---------------------------- --------------------------- > xlog_recover_process_intents > xfs_cui_item_recover//cui_refcount == 1 > xfs_trans_get_cud > xfs_trans_commit > <add cud item to cil> > xfs_cui_item_recover > <error occurred and return> > xlog_recover_cancel_intents > xfs_cui_release //cui_refcount == 0 > xfs_cui_item_free //free cui > <release other intent items> > xlog_force_shutdown //shutdown > <...> > <push items in cil> > xlog_cil_committed > xfs_cud_item_release > xfs_cui_release // UAF > > Fix it by move log force forward to make sure done items committed before > cancel intents. > > Fixes: 2e76f188fd90 ("xfs: cancel intents immediately if process_intents fails") > Signed-off-by: Long Li <leo.lilong@huawei.com> > --- > fs/xfs/xfs_log_recover.c | 14 +++++++------- > 1 file changed, 7 insertions(+), 7 deletions(-) > > diff --git a/fs/xfs/xfs_log_recover.c b/fs/xfs/xfs_log_recover.c > index fdaa0ffe029b..c37031e64db5 100644 > --- a/fs/xfs/xfs_log_recover.c > +++ b/fs/xfs/xfs_log_recover.c > @@ -3444,6 +3444,13 @@ xlog_recover_finish( > int error; > > error = xlog_recover_process_intents(log); > + /* > + * Sync the log to get all the intents that have done item out of "...that have intent done items out of the AIL." > + * the AIL. This isn't absolutely necessary, but it helps in case > + * the unlink transactions would have problems pushing the intents > + * out of the way. > + */ > + xfs_log_force(log->l_mp, XFS_LOG_SYNC); Hmm. Why doesn't the shutdown force the (now dead) dead log items out of memory? /* * Flush all the completed transactions to disk before marking the log * being shut down. We need to do this first as shutting down the log * before the force will prevent the log force from flushing the iclogs * to disk. * * When we are in recovery, there are no transactions to flush, and * we don't want to touch the log because we don't want to perturb the * current head/tail for future recovery attempts. Hence we need to * avoid a log force in this case. * * If we are shutting down due to a log IO error, then we must avoid * trying to write the log as that may just result in more IO errors and * an endless shutdown/force loop. */ if (!log_error && !xlog_in_recovery(log)) xfs_log_force(log->l_mp, XFS_LOG_SYNC); Oh. We're in recovery, but we've passed the part where we've finished with the log/inode/dquot items that we read from the disk. IOWs, now we've gotten to the part of recovery where we're actually standing up new transactions to finish the work that was in the intents ... with new incore intents. I wonder, does the problem go away if you (hackishly) clear_bit(XLOG_ACTIVE_RECOVERY, &log->l_opstate); before the xlog_force_shutdown? --D > if (error) { > /* > * Cancel all the unprocessed intent items now so that we don't > @@ -3458,13 +3465,6 @@ xlog_recover_finish( > return error; > } > > - /* > - * Sync the log to get all the intents out of the AIL. This isn't > - * absolutely necessary, but it helps in case the unlink transactions > - * would have problems pushing the intents out of the way. > - */ > - xfs_log_force(log->l_mp, XFS_LOG_SYNC); > - > /* > * Now that we've recovered the log and all the intents, we can clear > * the log incompat feature bits in the superblock because there's no > -- > 2.31.1 >
On Sat, Jul 15, 2023 at 02:36:47PM +0800, Long Li wrote: > KASAN report a uaf when recover intents fails: .... > > If process intents fails, intent items left in AIL will be delete > from AIL and freed in error handling, even intent items that have been > recovered and created done items. After this, uaf will be triggered when > done item commited, because at this point the released intent item will > be accessed. > > xlog_recover_finish xlog_cil_push_work > ---------------------------- --------------------------- > xlog_recover_process_intents > xfs_cui_item_recover//cui_refcount == 1 > xfs_trans_get_cud > xfs_trans_commit > <add cud item to cil> > xfs_cui_item_recover > <error occurred and return> > xlog_recover_cancel_intents > xfs_cui_release //cui_refcount == 0 > xfs_cui_item_free //free cui > <release other intent items> > xlog_force_shutdown //shutdown > <...> > <push items in cil> > xlog_cil_committed > xfs_cud_item_release > xfs_cui_release // UAF Huh. The log stores items in the AIL without holding a reference to them, then on shutdown takes the intent done reference away because it assumes the intent has not been processed as it is still in the AIL. Ok, that's broken. > Fix it by move log force forward to make sure done items committed before > cancel intents. That doesn't fix the fact we have a reference counted object that is being accessed by code that doesn't actually own a reference to the object. Intent log items are created with a reference count of 2 - one for the creator, and one for the intent done object. Look at xlog_recover_cui_commit_pass2(): /* * Insert the intent into the AIL directly and drop one reference so * that finishing or canceling the work will drop the other. */ xfs_trans_ail_insert(log->l_ailp, &cuip->cui_item, lsn); xfs_cui_release(cuip); return 0; } Log recovery explicitly drops the creator reference after it is inserted into the AIL, but it then processes the log item as if it also owns the intent-done reference. The moment we call ->iop_recover(), the intent-done reference should be owned by the log item. The recovery of the BUI, RUI and EFI all do the same thing. I suspect that these references should actually be held by log recovery until it is done processing the item, at which point it should be removed from the AIL by xlog_recover_process_intents(). The code in ->iop_recovery should assume that it passes the reference to the done intent, but if that code fails before creating the done-intent then it needs to release the intent reference itself. That way when we go to cancel the intent, the only intents we find in the AIL are the ones we know have not been processed yet and hence we can safely drop both the creator and the intent done reference from xlog_recover_cancel_intents(). Cheers, Dave.
On Wed, Jul 19, 2023 at 04:41:41PM +1000, Dave Chinner wrote: > On Sat, Jul 15, 2023 at 02:36:47PM +0800, Long Li wrote: > > KASAN report a uaf when recover intents fails: > .... > > > > If process intents fails, intent items left in AIL will be delete > > from AIL and freed in error handling, even intent items that have been > > recovered and created done items. After this, uaf will be triggered when > > done item commited, because at this point the released intent item will > > be accessed. > > > > xlog_recover_finish xlog_cil_push_work > > ---------------------------- --------------------------- > > xlog_recover_process_intents > > xfs_cui_item_recover//cui_refcount == 1 > > xfs_trans_get_cud > > xfs_trans_commit > > <add cud item to cil> > > xfs_cui_item_recover > > <error occurred and return> > > xlog_recover_cancel_intents > > xfs_cui_release //cui_refcount == 0 > > xfs_cui_item_free //free cui > > <release other intent items> > > xlog_force_shutdown //shutdown > > <...> > > <push items in cil> > > xlog_cil_committed > > xfs_cud_item_release > > xfs_cui_release // UAF > > Huh. The log stores items in the AIL without holding a reference to > them, then on shutdown takes the intent done reference away because > it assumes the intent has not been processed as it is still in the > AIL. > > Ok, that's broken. > > > Fix it by move log force forward to make sure done items committed before > > cancel intents. > > That doesn't fix the fact we have a reference counted object that is > being accessed by code that doesn't actually own a reference to the > object. Intent log items are created with a reference count of 2 - > one for the creator, and one for the intent done object. > > Look at xlog_recover_cui_commit_pass2(): > > /* > * Insert the intent into the AIL directly and drop one reference so > * that finishing or canceling the work will drop the other. > */ > xfs_trans_ail_insert(log->l_ailp, &cuip->cui_item, lsn); > xfs_cui_release(cuip); > return 0; > } > > Log recovery explicitly drops the creator reference after it is > inserted into the AIL, but it then processes the log item as if it > also owns the intent-done reference. The moment we call > ->iop_recover(), the intent-done reference should be owned by the > log item. Hi, Dave Thanks for the reply. Yes, your analysis seems reasonable, it helped me a lot to understand the intent lifecycle. > > The recovery of the BUI, RUI and EFI all do the same thing. I > suspect that these references should actually be held by log > recovery until it is done processing the item, at which point it > should be removed from the AIL by xlog_recover_process_intents(). Why do we need to remove the intent from the AIL at this point, shouldn't it be removed from the AIL when the done intent is committed? Or is there any way to ensure that the intents are removed from the AIL when they are processed. > > The code in ->iop_recovery should assume that it passes the > reference to the done intent, but if that code fails before creating > the done-intent then it needs to release the intent reference > itself. > > That way when we go to cancel the intent, the only intents we find > in the AIL are the ones we know have not been processed yet and > hence we can safely drop both the creator and the intent done > reference from xlog_recover_cancel_intents(). Yes, if the processed intent has been removed from the AIL, then only the unprocessed intent remains in the current AIL, and it is safe to cancel the intent. Best regards Long Li > > Cheers, > > Dave. > -- > Dave Chinner > david@fromorbit.com
On Tue, Jul 18, 2023 at 07:19:14PM -0700, Darrick J. Wong wrote: > On Sat, Jul 15, 2023 at 02:36:47PM +0800, Long Li wrote: > > KASAN report a uaf when recover intents fails: > > > > ================================================================== > > BUG: KASAN: slab-use-after-free in xfs_cui_release+0xb7/0xc0 > > Read of size 4 at addr ffff888012575e60 by task kworker/u8:3/103 > > CPU: 3 PID: 103 Comm: kworker/u8:3 Not tainted 6.4.0-rc7-next-20230619-00003-g94543a53f9a4-dirty #166 > > Workqueue: xfs-cil/sda xlog_cil_push_work > > Call Trace: > > <TASK> > > dump_stack_lvl+0x50/0x70 > > print_report+0xc2/0x600 > > kasan_report+0xb6/0xe0 > > xfs_cui_release+0xb7/0xc0 > > xfs_cud_item_release+0x3c/0x90 > > xfs_trans_committed_bulk+0x2d5/0x7f0 > > xlog_cil_committed+0xaba/0xf20 > > xlog_cil_push_work+0x1a60/0x2360 > > process_one_work+0x78e/0x1140 > > worker_thread+0x58b/0xf60 > > kthread+0x2cd/0x3c0 > > ret_from_fork+0x1f/0x30 > > </TASK> > > > > Allocated by task 531: > > kasan_save_stack+0x22/0x40 > > kasan_set_track+0x25/0x30 > > __kasan_slab_alloc+0x55/0x60 > > kmem_cache_alloc+0x195/0x5f0 > > xfs_cui_init+0x198/0x1d0 > > xlog_recover_cui_commit_pass2+0x133/0x5f0 > > xlog_recover_items_pass2+0x107/0x230 > > xlog_recover_commit_trans+0x3e7/0x9c0 > > xlog_recovery_process_trans+0x140/0x1d0 > > xlog_recover_process_ophdr+0x1a0/0x3d0 > > xlog_recover_process_data+0x108/0x2d0 > > xlog_recover_process+0x1f6/0x280 > > xlog_do_recovery_pass+0x609/0xdb0 > > xlog_do_log_recovery+0x84/0xe0 > > xlog_do_recover+0x7d/0x470 > > xlog_recover+0x25f/0x490 > > xfs_log_mount+0x2dd/0x6f0 > > xfs_mountfs+0x11ce/0x1e70 > > xfs_fs_fill_super+0x10ec/0x1b20 > > get_tree_bdev+0x3c8/0x730 > > vfs_get_tree+0x89/0x2c0 > > path_mount+0xecf/0x1800 > > do_mount+0xf3/0x110 > > __x64_sys_mount+0x154/0x1f0 > > do_syscall_64+0x39/0x80 > > entry_SYSCALL_64_after_hwframe+0x63/0xcd > > > > Freed by task 531: > > kasan_save_stack+0x22/0x40 > > kasan_set_track+0x25/0x30 > > kasan_save_free_info+0x2b/0x40 > > __kasan_slab_free+0x114/0x1b0 > > kmem_cache_free+0xf8/0x510 > > xfs_cui_item_free+0x95/0xb0 > > xfs_cui_release+0x86/0xc0 > > xlog_recover_cancel_intents.isra.0+0xf8/0x210 > > xlog_recover_finish+0x7e7/0x980 > > xfs_log_mount_finish+0x2bb/0x4a0 > > xfs_mountfs+0x14bf/0x1e70 > > xfs_fs_fill_super+0x10ec/0x1b20 > > get_tree_bdev+0x3c8/0x730 > > vfs_get_tree+0x89/0x2c0 > > path_mount+0xecf/0x1800 > > do_mount+0xf3/0x110 > > __x64_sys_mount+0x154/0x1f0 > > do_syscall_64+0x39/0x80 > > entry_SYSCALL_64_after_hwframe+0x63/0xcd > > > > The buggy address belongs to the object at ffff888012575dc8 > > which belongs to the cache xfs_cui_item of size 432 > > The buggy address is located 152 bytes inside of > > freed 432-byte region [ffff888012575dc8, ffff888012575f78) > > > > The buggy address belongs to the physical page: > > page:ffffea0000495d00 refcount:1 mapcount:0 mapping:0000000000000000 index:0xffff888012576208 pfn:0x12574 > > head:ffffea0000495d00 order:2 entire_mapcount:0 nr_pages_mapped:0 pincount:0 > > flags: 0x1fffff80010200(slab|head|node=0|zone=1|lastcpupid=0x1fffff) > > page_type: 0xffffffff() > > raw: 001fffff80010200 ffff888012092f40 ffff888014570150 ffff888014570150 > > raw: ffff888012576208 00000000001e0010 00000001ffffffff 0000000000000000 > > page dumped because: kasan: bad access detected > > > > Memory state around the buggy address: > > ffff888012575d00: fb fb fb fb fb fb fb fb fb fb fb fc fc fc fc fc > > ffff888012575d80: fc fc fc fc fc fc fc fc fc fa fb fb fb fb fb fb > > >ffff888012575e00: fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb > > ^ > > ffff888012575e80: fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb > > ffff888012575f00: fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb fc > > ================================================================== > > > > If process intents fails, intent items left in AIL will be delete > > from AIL and freed in error handling, even intent items that have been > > recovered and created done items. After this, uaf will be triggered when > > done item commited, because at this point the released intent item will > > be accessed. > > > > xlog_recover_finish xlog_cil_push_work > > ---------------------------- --------------------------- > > xlog_recover_process_intents > > xfs_cui_item_recover//cui_refcount == 1 > > xfs_trans_get_cud > > xfs_trans_commit > > <add cud item to cil> > > xfs_cui_item_recover > > <error occurred and return> > > xlog_recover_cancel_intents > > xfs_cui_release //cui_refcount == 0 > > xfs_cui_item_free //free cui > > <release other intent items> > > xlog_force_shutdown //shutdown > > <...> > > <push items in cil> > > xlog_cil_committed > > xfs_cud_item_release > > xfs_cui_release // UAF > > > > Fix it by move log force forward to make sure done items committed before > > cancel intents. > > > > Fixes: 2e76f188fd90 ("xfs: cancel intents immediately if process_intents fails") > > Signed-off-by: Long Li <leo.lilong@huawei.com> > > --- > > fs/xfs/xfs_log_recover.c | 14 +++++++------- > > 1 file changed, 7 insertions(+), 7 deletions(-) > > > > diff --git a/fs/xfs/xfs_log_recover.c b/fs/xfs/xfs_log_recover.c > > index fdaa0ffe029b..c37031e64db5 100644 > > --- a/fs/xfs/xfs_log_recover.c > > +++ b/fs/xfs/xfs_log_recover.c > > @@ -3444,6 +3444,13 @@ xlog_recover_finish( > > int error; > > > > error = xlog_recover_process_intents(log); > > + /* > > + * Sync the log to get all the intents that have done item out of > > "...that have intent done items out of the AIL." > > > + * the AIL. This isn't absolutely necessary, but it helps in case > > + * the unlink transactions would have problems pushing the intents > > + * out of the way. > > + */ > > + xfs_log_force(log->l_mp, XFS_LOG_SYNC); > > Hmm. Why doesn't the shutdown force the (now dead) dead log items out > of memory? > > /* > * Flush all the completed transactions to disk before marking the log > * being shut down. We need to do this first as shutting down the log > * before the force will prevent the log force from flushing the iclogs > * to disk. > * > * When we are in recovery, there are no transactions to flush, and > * we don't want to touch the log because we don't want to perturb the > * current head/tail for future recovery attempts. Hence we need to > * avoid a log force in this case. > * > * If we are shutting down due to a log IO error, then we must avoid > * trying to write the log as that may just result in more IO errors and > * an endless shutdown/force loop. > */ > if (!log_error && !xlog_in_recovery(log)) > xfs_log_force(log->l_mp, XFS_LOG_SYNC); > > Oh. We're in recovery, but we've passed the part where we've finished > with the log/inode/dquot items that we read from the disk. IOWs, now > we've gotten to the part of recovery where we're actually standing up > new transactions to finish the work that was in the intents ... with new > incore intents. > > I wonder, does the problem go away if you (hackishly) > clear_bit(XLOG_ACTIVE_RECOVERY, &log->l_opstate); before the > xlog_force_shutdown? I used the patch below and tested it to confirm that the uaf problem goes away. diff --git a/fs/xfs/xfs_log_recover.c b/fs/xfs/xfs_log_recover.c index 82c81d20459d..5d9f84ce2b9b 100644 --- a/fs/xfs/xfs_log_recover.c +++ b/fs/xfs/xfs_log_recover.c @@ -3452,9 +3452,10 @@ xlog_recover_finish( * (inode reclaim does this) before we get around to * xfs_log_mount_cancel. */ + clear_bit(XLOG_ACTIVE_RECOVERY, &log->l_opstate); + xlog_force_shutdown(log, SHUTDOWN_CORRUPT_INCORE); xlog_recover_cancel_intents(log); xfs_alert(log->l_mp, "Failed to recover intents"); - xlog_force_shutdown(log, SHUTDOWN_LOG_IO_ERROR); return error; } I'm wondering if a shutdown with SHUTDOWN_CORRUPT_INCORE is appropriate, even though there is no log io error at this point. Best regards Long Li > > --D > > > if (error) { > > /* > > * Cancel all the unprocessed intent items now so that we don't > > @@ -3458,13 +3465,6 @@ xlog_recover_finish( > > return error; > > } > > > > - /* > > - * Sync the log to get all the intents out of the AIL. This isn't > > - * absolutely necessary, but it helps in case the unlink transactions > > - * would have problems pushing the intents out of the way. > > - */ > > - xfs_log_force(log->l_mp, XFS_LOG_SYNC); > > - > > /* > > * Now that we've recovered the log and all the intents, we can clear > > * the log incompat feature bits in the superblock because there's no > > -- > > 2.31.1 > >
On Sat, Jul 22, 2023 at 09:19:09AM +0800, Long Li wrote: > On Wed, Jul 19, 2023 at 04:41:41PM +1000, Dave Chinner wrote: > > On Sat, Jul 15, 2023 at 02:36:47PM +0800, Long Li wrote: > > > KASAN report a uaf when recover intents fails: > > .... > > > > > > If process intents fails, intent items left in AIL will be delete > > > from AIL and freed in error handling, even intent items that have been > > > recovered and created done items. After this, uaf will be triggered when > > > done item commited, because at this point the released intent item will > > > be accessed. > > > > > > xlog_recover_finish xlog_cil_push_work > > > ---------------------------- --------------------------- > > > xlog_recover_process_intents > > > xfs_cui_item_recover//cui_refcount == 1 > > > xfs_trans_get_cud > > > xfs_trans_commit > > > <add cud item to cil> > > > xfs_cui_item_recover > > > <error occurred and return> > > > xlog_recover_cancel_intents > > > xfs_cui_release //cui_refcount == 0 > > > xfs_cui_item_free //free cui > > > <release other intent items> > > > xlog_force_shutdown //shutdown > > > <...> > > > <push items in cil> > > > xlog_cil_committed > > > xfs_cud_item_release > > > xfs_cui_release // UAF > > > > Huh. The log stores items in the AIL without holding a reference to > > them, then on shutdown takes the intent done reference away because > > it assumes the intent has not been processed as it is still in the > > AIL. > > > > Ok, that's broken. > > > > > Fix it by move log force forward to make sure done items committed before > > > cancel intents. > > > > That doesn't fix the fact we have a reference counted object that is > > being accessed by code that doesn't actually own a reference to the > > object. Intent log items are created with a reference count of 2 - > > one for the creator, and one for the intent done object. > > > > Look at xlog_recover_cui_commit_pass2(): > > > > /* > > * Insert the intent into the AIL directly and drop one reference so > > * that finishing or canceling the work will drop the other. > > */ > > xfs_trans_ail_insert(log->l_ailp, &cuip->cui_item, lsn); > > xfs_cui_release(cuip); > > return 0; > > } > > > > Log recovery explicitly drops the creator reference after it is > > inserted into the AIL, but it then processes the log item as if it > > also owns the intent-done reference. The moment we call > > ->iop_recover(), the intent-done reference should be owned by the > > log item. > > Hi, Dave > > Thanks for the reply. Yes, your analysis seems reasonable, it helped me a > lot to understand the intent lifecycle. > > > > > The recovery of the BUI, RUI and EFI all do the same thing. I > > suspect that these references should actually be held by log > > recovery until it is done processing the item, at which point it > > should be removed from the AIL by xlog_recover_process_intents(). > > Why do we need to remove the intent from the AIL at this point, Because we've processed the recovery of it - it is either completely done or we have a new intent in the CIL ready to continue operation. Either way, the next write to the journal will remove the item from the AIL when it completes. Intents don't need to be in the AIL, though - we can cancel them in memory (see the intent whiteout code) and so when we process the done item from journal IO completion the last reference goes away and they won't be in the AIL at this point in time. IOWs, the intent freeing code doesn't care if the intent is in the AIL or not, it does the right thing either way. Hence if we remove the intent from the list of intents that need to be recovered after we have done the initial recovery, we acheive two things: 1. the tail of the log can be moved forward with the commit of the done intent or new intent to continue the operation, and 2. We avoid the problem of trying to determine how many reference counts we need to drop from intent recovery cancelling because we never come across intents we've actually attempted recovery on. > shouldn't > it be removed from the AIL when the done intent is committed? Or is there > any way to ensure that the intents are removed from the AIL when they are > processed. THe reference counting ensures the right thing is done when the last reference goes away. If it is in the AIL, it will get removed, if it is not in the AIL, then AIL removal is a no-op and nothign bad happens. Cheers, Dave.
On Mon, Jul 24, 2023 at 11:05:48AM +1000, Dave Chinner wrote: > On Sat, Jul 22, 2023 at 09:19:09AM +0800, Long Li wrote: > > On Wed, Jul 19, 2023 at 04:41:41PM +1000, Dave Chinner wrote: > > > On Sat, Jul 15, 2023 at 02:36:47PM +0800, Long Li wrote: > > > > KASAN report a uaf when recover intents fails: > > > .... > > > > > > > > If process intents fails, intent items left in AIL will be delete > > > > from AIL and freed in error handling, even intent items that have been > > > > recovered and created done items. After this, uaf will be triggered when > > > > done item commited, because at this point the released intent item will > > > > be accessed. > > > > > > > > xlog_recover_finish xlog_cil_push_work > > > > ---------------------------- --------------------------- > > > > xlog_recover_process_intents > > > > xfs_cui_item_recover//cui_refcount == 1 > > > > xfs_trans_get_cud > > > > xfs_trans_commit > > > > <add cud item to cil> > > > > xfs_cui_item_recover > > > > <error occurred and return> > > > > xlog_recover_cancel_intents > > > > xfs_cui_release //cui_refcount == 0 > > > > xfs_cui_item_free //free cui > > > > <release other intent items> > > > > xlog_force_shutdown //shutdown > > > > <...> > > > > <push items in cil> > > > > xlog_cil_committed > > > > xfs_cud_item_release > > > > xfs_cui_release // UAF > > > > > > Huh. The log stores items in the AIL without holding a reference to > > > them, then on shutdown takes the intent done reference away because > > > it assumes the intent has not been processed as it is still in the > > > AIL. > > > > > > Ok, that's broken. > > > > > > > Fix it by move log force forward to make sure done items committed before > > > > cancel intents. > > > > > > That doesn't fix the fact we have a reference counted object that is > > > being accessed by code that doesn't actually own a reference to the > > > object. Intent log items are created with a reference count of 2 - > > > one for the creator, and one for the intent done object. > > > > > > Look at xlog_recover_cui_commit_pass2(): > > > > > > /* > > > * Insert the intent into the AIL directly and drop one reference so > > > * that finishing or canceling the work will drop the other. > > > */ > > > xfs_trans_ail_insert(log->l_ailp, &cuip->cui_item, lsn); > > > xfs_cui_release(cuip); > > > return 0; > > > } > > > > > > Log recovery explicitly drops the creator reference after it is > > > inserted into the AIL, but it then processes the log item as if it > > > also owns the intent-done reference. The moment we call > > > ->iop_recover(), the intent-done reference should be owned by the > > > log item. > > > > Hi, Dave > > > > Thanks for the reply. Yes, your analysis seems reasonable, it helped me a > > lot to understand the intent lifecycle. > > > > > > > > The recovery of the BUI, RUI and EFI all do the same thing. I > > > suspect that these references should actually be held by log > > > recovery until it is done processing the item, at which point it > > > should be removed from the AIL by xlog_recover_process_intents(). > > > > Why do we need to remove the intent from the AIL at this point, > > Because we've processed the recovery of it - it is either completely > done or we have a new intent in the CIL ready to continue operation. > Either way, the next write to the journal will remove the item from > the AIL when it completes. > > Intents don't need to be in the AIL, though - we can cancel them > in memory (see the intent whiteout code) and so when we process the > done item from journal IO completion the last reference goes away > and they won't be in the AIL at this point in time. > > IOWs, the intent freeing code doesn't care if the intent is in the > AIL or not, it does the right thing either way. > > Hence if we remove the intent from the list of intents that need to > be recovered after we have done the initial recovery, we acheive two > things: > > 1. the tail of the log can be moved forward with the commit of the > done intent or new intent to continue the operation, and > > 2. We avoid the problem of trying to determine how many reference > counts we need to drop from intent recovery cancelling because we > never come across intents we've actually attempted recovery on. > > > shouldn't > > it be removed from the AIL when the done intent is committed? Or is there > > any way to ensure that the intents are removed from the AIL when they are > > processed. > > THe reference counting ensures the right thing is done when the last > reference goes away. If it is in the AIL, it will get removed, if it > is not in the AIL, then AIL removal is a no-op and nothign bad > happens. Thank you for your detailed answer, it was great, I understand what you mean. The intent item stays in the AIL until the done item is committed, which prevents the tail lsn from moving forward. After the transaction is alloced in ->iop_recover(), we don't need to stop tail lsn from moving forward because there is no other thread committing the transaction at this point, the intent item will not be overwritten until the done intent is written to the log, so we can just remove the intent item from the AIL. I think we can remove the intent item from the AIL as soon as the done item is created, since the reference count has already been passed to the done item. Eventually the problem can be fixed. This is my understanding. Thanks Long Li
diff --git a/fs/xfs/xfs_log_recover.c b/fs/xfs/xfs_log_recover.c index fdaa0ffe029b..c37031e64db5 100644 --- a/fs/xfs/xfs_log_recover.c +++ b/fs/xfs/xfs_log_recover.c @@ -3444,6 +3444,13 @@ xlog_recover_finish( int error; error = xlog_recover_process_intents(log); + /* + * Sync the log to get all the intents that have done item out of + * the AIL. This isn't absolutely necessary, but it helps in case + * the unlink transactions would have problems pushing the intents + * out of the way. + */ + xfs_log_force(log->l_mp, XFS_LOG_SYNC); if (error) { /* * Cancel all the unprocessed intent items now so that we don't @@ -3458,13 +3465,6 @@ xlog_recover_finish( return error; } - /* - * Sync the log to get all the intents out of the AIL. This isn't - * absolutely necessary, but it helps in case the unlink transactions - * would have problems pushing the intents out of the way. - */ - xfs_log_force(log->l_mp, XFS_LOG_SYNC); - /* * Now that we've recovered the log and all the intents, we can clear * the log incompat feature bits in the superblock because there's no