kernel/linux.git/drivers/md, branch v5.2.16

bcache: fix race in btree_flush_write()

2019-09-16T06:23:21+00:00

There is a race between mca_reap(), btree_node_free() and journal code btree_flush_write(), which results very rare and strange deadlock or panic and are very hard to reproduce. Let me explain how the race happens. In btree_flush_write() one btree node with oldest journal pin is selected, then it is flushed to cache device, the select-and-flush is a two steps operation. Between these two steps, there are something may happen inside the race window, - The selected btree node was reaped by mca_reap() and allocated to other requesters for other btree node. - The slected btree node was selected, flushed and released by mca shrink callback bch_mca_scan(). When btree_flush_write() tries to flush the selected btree node, firstly b->write_lock is held by mutex_lock(). If the race happens and the memory of selected btree node is allocated to other btree node, if that btree node's write_lock is held already, a deadlock very probably happens here. A worse case is the memory of the selected btree node is released, then all references to this btree node (e.g. b->write_lock) will trigger NULL pointer deference panic. This race was introduced in commit cafe56359144 ("bcache: A block layer cache"), and enlarged by commit c4dc2497d50d ("bcache: fix high CPU occupancy during journal"), which selected 128 btree nodes and flushed them one-by-one in a quite long time period. Such race is not easy to reproduce before. On a Lenovo SR650 server with 48 Xeon cores, and configure 1 NVMe SSD as cache device, a MD raid0 device assembled by 3 NVMe SSDs as backing device, this race can be observed around every 10,000 times btree_flush_write() gets called. Both deadlock and kernel panic all happened as aftermath of the race. The idea of the fix is to add a btree flag BTREE_NODE_journal_flush. It is set when selecting btree nodes, and cleared after btree nodes flushed. Then when mca_reap() selects a btree node with this bit set, this btree node will be skipped. Since mca_reap() only reaps btree node without BTREE_NODE_journal_flush flag, such race is avoided. Once corner case should be noticed, that is btree_node_free(). It might be called in some error handling code path. For example the following code piece from btree_split(), 2149 err_free2: 2150 bkey_put(b->c, &n2->key); 2151 btree_node_free(n2); 2152 rw_unlock(true, n2); 2153 err_free1: 2154 bkey_put(b->c, &n1->key); 2155 btree_node_free(n1); 2156 rw_unlock(true, n1); At line 2151 and 2155, the btree node n2 and n1 are released without mac_reap(), so BTREE_NODE_journal_flush also needs to be checked here. If btree_node_free() is called directly in such error handling path, and the selected btree node has BTREE_NODE_journal_flush bit set, just delay for 1 us and retry again. In this case this btree node won't be skipped, just retry until the BTREE_NODE_journal_flush bit cleared, and free the btree node memory. Fixes: cafe56359144 ("bcache: A block layer cache") Signed-off-by: Coly Li Reported-and-tested-by: kbuild test robot Cc: stable@vger.kernel.org Signed-off-by: Jens Axboe

bcache: add comments for mutex_lock(&b->write_lock)

2019-09-16T06:23:21+00:00

When accessing or modifying BTREE_NODE_dirty bit, it is not always necessary to acquire b->write_lock. In bch_btree_cache_free() and mca_reap() acquiring b->write_lock is necessary, and this patch adds comments to explain why mutex_lock(&b->write_lock) is necessary for checking or clearing BTREE_NODE_dirty bit there. Signed-off-by: Coly Li Signed-off-by: Jens Axboe

bcache: only clear BTREE_NODE_dirty bit when it is set

2019-09-16T06:23:20+00:00

In bch_btree_cache_free() and btree_node_free(), BTREE_NODE_dirty is always set no matter btree node is dirty or not. The code looks like this, if (btree_node_dirty(b)) btree_complete_write(b, btree_current_write(b)); clear_bit(BTREE_NODE_dirty, &b->flags); Indeed if btree_node_dirty(b) returns false, it means BTREE_NODE_dirty bit is cleared, then it is unnecessary to clear the bit again. This patch only clears BTREE_NODE_dirty when btree_node_dirty(b) is true (the bit is set), to save a few CPU cycles. Signed-off-by: Coly Li Signed-off-by: Jens Axboe

dm zoned: fix potential NULL dereference in dmz_do_reclaim()

2019-08-29T06:30:27+00:00

[ Upstream commit e0702d90b79d430b0ccc276ead4f88440bb51352 ] This function is supposed to return error pointers so it matches the dmz_get_rnd_zone_for_reclaim() function. The current code could lead to a NULL dereference in dmz_do_reclaim() Fixes: b234c6d7a703 ("dm zoned: improve error handling in reclaim") Signed-off-by: Dan Carpenter Reviewed-by: Dmitry Fomichev Signed-off-by: Mike Snitzer Signed-off-by: Sasha Levin

dm zoned: properly handle backing device failure

2019-08-29T06:30:25+00:00

commit 75d66ffb48efb30f2dd42f041ba8b39c5b2bd115 upstream. dm-zoned is observed to lock up or livelock in case of hardware failure or some misconfiguration of the backing zoned device. This patch adds a new dm-zoned target function that checks the status of the backing device. If the request queue of the backing device is found to be in dying state or the SCSI backing device enters offline state, the health check code sets a dm-zoned target flag prompting all further incoming I/O to be rejected. In order to detect backing device failures timely, this new function is called in the request mapping path, at the beginning of every reclaim run and before performing any metadata I/O. The proper way out of this situation is to do dmsetup remove and recreate the target when the problem with the backing device is resolved. Fixes: 3b1a94c88b79 ("dm zoned: drive-managed zoned block device target") Cc: stable@vger.kernel.org Signed-off-by: Dmitry Fomichev Reviewed-by: Damien Le Moal Signed-off-by: Mike Snitzer Signed-off-by: Greg Kroah-Hartman

dm zoned: improve error handling in i/o map code

2019-08-29T06:30:25+00:00

commit d7428c50118e739e672656c28d2b26b09375d4e0 upstream. Some errors are ignored in the I/O path during queueing chunks for processing by chunk works. Since at least these errors are transient in nature, it should be possible to retry the failed incoming commands. The fix - Errors that can happen while queueing chunks are carried upwards to the main mapping function and it now returns DM_MAPIO_REQUEUE for any incoming requests that can not be properly queued. Error logging/debug messages are added where needed. Fixes: 3b1a94c88b79 ("dm zoned: drive-managed zoned block device target") Cc: stable@vger.kernel.org Signed-off-by: Dmitry Fomichev Reviewed-by: Damien Le Moal Signed-off-by: Mike Snitzer Signed-off-by: Greg Kroah-Hartman

dm zoned: improve error handling in reclaim

2019-08-29T06:30:25+00:00

commit b234c6d7a703661b5045c5bf569b7c99d2edbf88 upstream. There are several places in reclaim code where errors are not propagated to the main function, dmz_reclaim(). This function is responsible for unlocking zones that might be still locked at the end of any failed reclaim iterations. As the result, some device zones may be left permanently locked for reclaim, degrading target's capability to reclaim zones. This patch fixes these issues as follows - Make sure that dmz_reclaim_buf(), dmz_reclaim_seq_data() and dmz_reclaim_rnd_data() return error codes to the caller. dmz_reclaim() function is renamed to dmz_do_reclaim() to avoid clashing with "struct dmz_reclaim" and is modified to return the error to the caller. dmz_get_zone_for_reclaim() now returns an error instead of NULL pointer and reclaim code checks for that error. Error logging/debug messages are added where necessary. Fixes: 3b1a94c88b79 ("dm zoned: drive-managed zoned block device target") Cc: stable@vger.kernel.org Signed-off-by: Dmitry Fomichev Reviewed-by: Damien Le Moal Signed-off-by: Mike Snitzer Signed-off-by: Greg Kroah-Hartman

dm table: fix invalid memory accesses with too high sector number

2019-08-29T06:30:25+00:00

commit 1cfd5d3399e87167b7f9157ef99daa0e959f395d upstream. If the sector number is too high, dm_table_find_target() should return a pointer to a zeroed dm_target structure (the caller should test it with dm_target_is_valid). However, for some table sizes, the code in dm_table_find_target() that performs btree lookup will access out of bound memory structures. Fix this bug by testing the sector number at the beginning of dm_table_find_target(). Also, add an "inline" keyword to the function dm_table_get_size() because this is a hot path. Fixes: 512875bd9661 ("dm: table detect io beyond device") Cc: stable@vger.kernel.org Reported-by: Zhang Tao Signed-off-by: Mikulas Patocka Signed-off-by: Mike Snitzer Signed-off-by: Greg Kroah-Hartman

dm space map metadata: fix missing store of apply_bops() return value

2019-08-29T06:30:25+00:00

commit ae148243d3f0816b37477106c05a2ec7d5f32614 upstream. In commit 6096d91af0b6 ("dm space map metadata: fix occasional leak of a metadata block on resize"), we refactor the commit logic to a new function 'apply_bops'. But when that logic was replaced in out() the return value was not stored. This may lead out() returning a wrong value to the caller. Fixes: 6096d91af0b6 ("dm space map metadata: fix occasional leak of a metadata block on resize") Cc: stable@vger.kernel.org Signed-off-by: ZhangXiaoxu Signed-off-by: Mike Snitzer Signed-off-by: Greg Kroah-Hartman

dm raid: add missing cleanup in raid_ctr()

2019-08-29T06:30:25+00:00

commit dc1a3e8e0cc6b2293b48c044710e63395aeb4fb4 upstream. If rs_prepare_reshape() fails, no cleanup is executed, leading to leak of the raid_set structure allocated at the beginning of raid_ctr(). To fix this issue, go to the label 'bad' if the error occurs. Fixes: 11e4723206683 ("dm raid: stop keeping raid set frozen altogether") Cc: stable@vger.kernel.org Signed-off-by: Wenwen Wang Signed-off-by: Mike Snitzer Signed-off-by: Greg Kroah-Hartman