From f9e040efcc28309e5c592f7e79085a9a52e31f58 Mon Sep 17 00:00:00 2001 From: Mikulas Patocka Date: Mon, 24 Aug 2020 11:09:47 -0400 Subject: dm writecache: handle DAX to partitions on persistent memory correctly The function dax_direct_access doesn't take partitions into account, it always maps pages from the beginning of the device. Therefore, persistent_memory_claim() must get the partition offset using get_start_sect() and add it to the page offsets passed to dax_direct_access(). Signed-off-by: Mikulas Patocka Fixes: 48debafe4f2f ("dm: add writecache target") Cc: stable@vger.kernel.org # 4.18+ Signed-off-by: Mike Snitzer --- drivers/md/dm-writecache.c | 12 ++++++++++-- 1 file changed, 10 insertions(+), 2 deletions(-) (limited to 'drivers') diff --git a/drivers/md/dm-writecache.c b/drivers/md/dm-writecache.c index 86dbe0c8b45c..6271d1e741cf 100644 --- a/drivers/md/dm-writecache.c +++ b/drivers/md/dm-writecache.c @@ -231,6 +231,7 @@ static int persistent_memory_claim(struct dm_writecache *wc) pfn_t pfn; int id; struct page **pages; + sector_t offset; wc->memory_vmapped = false; @@ -245,9 +246,16 @@ static int persistent_memory_claim(struct dm_writecache *wc) goto err1; } + offset = get_start_sect(wc->ssd_dev->bdev); + if (offset & (PAGE_SIZE / 512 - 1)) { + r = -EINVAL; + goto err1; + } + offset >>= PAGE_SHIFT - 9; + id = dax_read_lock(); - da = dax_direct_access(wc->ssd_dev->dax_dev, 0, p, &wc->memory_map, &pfn); + da = dax_direct_access(wc->ssd_dev->dax_dev, offset, p, &wc->memory_map, &pfn); if (da < 0) { wc->memory_map = NULL; r = da; @@ -269,7 +277,7 @@ static int persistent_memory_claim(struct dm_writecache *wc) i = 0; do { long daa; - daa = dax_direct_access(wc->ssd_dev->dax_dev, i, p - i, + daa = dax_direct_access(wc->ssd_dev->dax_dev, offset + i, p - i, NULL, &pfn); if (daa <= 0) { r = daa ? daa : -EINVAL; -- cgit From c322ee9320eaa4013ca3620b1130992916b19b31 Mon Sep 17 00:00:00 2001 From: Mike Snitzer Date: Mon, 24 Aug 2020 14:19:55 -0400 Subject: dm mpath: fix racey management of PG initialization Commit 935fcc56abc3 ("dm mpath: only flush workqueue when needed") changed flush_multipath_work() to avoid needless workqueue flushing (of a multipath global workqueue). But that change didn't realize the surrounding flush_multipath_work() code should also only run if 'pg_init_in_progress' is set. Fix this by only doing all of flush_multipath_work()'s PG init related work if 'pg_init_in_progress' is set. Otherwise multipath_wait_for_pg_init_completion() will run unconditionally but the preceeding flush_workqueue(kmpath_handlerd) may not. This could lead to deadlock (though only if kmpath_handlerd never runs a corresponding work to decrement 'pg_init_in_progress'). It could also be, though highly unlikely, that the kmpath_handlerd work that does PG init completes before 'pg_init_in_progress' is set, and then an intervening DM table reload's multipath_postsuspend() triggers flush_multipath_work(). Fixes: 935fcc56abc3 ("dm mpath: only flush workqueue when needed") Cc: stable@vger.kernel.org Reported-by: Ben Marzinski Signed-off-by: Mike Snitzer --- drivers/md/dm-mpath.c | 22 +++++++++++++++------- 1 file changed, 15 insertions(+), 7 deletions(-) (limited to 'drivers') diff --git a/drivers/md/dm-mpath.c b/drivers/md/dm-mpath.c index e3283d35c7fd..de4da825ade6 100644 --- a/drivers/md/dm-mpath.c +++ b/drivers/md/dm-mpath.c @@ -1287,17 +1287,25 @@ static void multipath_wait_for_pg_init_completion(struct multipath *m) static void flush_multipath_work(struct multipath *m) { if (m->hw_handler_name) { - set_bit(MPATHF_PG_INIT_DISABLED, &m->flags); - smp_mb__after_atomic(); + unsigned long flags; + + if (!atomic_read(&m->pg_init_in_progress)) + goto skip; + + spin_lock_irqsave(&m->lock, flags); + if (atomic_read(&m->pg_init_in_progress) && + !test_and_set_bit(MPATHF_PG_INIT_DISABLED, &m->flags)) { + spin_unlock_irqrestore(&m->lock, flags); - if (atomic_read(&m->pg_init_in_progress)) flush_workqueue(kmpath_handlerd); - multipath_wait_for_pg_init_completion(m); + multipath_wait_for_pg_init_completion(m); - clear_bit(MPATHF_PG_INIT_DISABLED, &m->flags); - smp_mb__after_atomic(); + spin_lock_irqsave(&m->lock, flags); + clear_bit(MPATHF_PG_INIT_DISABLED, &m->flags); + } + spin_unlock_irqrestore(&m->lock, flags); } - +skip: if (m->queue_mode == DM_TYPE_BIO_BASED) flush_work(&m->process_queued_bios); flush_work(&m->trigger_event); -- cgit From 7785a9e4c228db6d01086a52d5685cd7336a08b7 Mon Sep 17 00:00:00 2001 From: Damien Le Moal Date: Mon, 31 Aug 2020 14:55:55 +0900 Subject: dm crypt: Initialize crypto wait structures Use the DECLARE_CRYPTO_WAIT() macro to properly initialize the crypto wait structures declared on stack before their use with crypto_wait_req(). Fixes: 39d13a1ac41d ("dm crypt: reuse eboiv skcipher for IV generation") Fixes: bbb1658461ac ("dm crypt: Implement Elephant diffuser for Bitlocker compatibility") Cc: stable@vger.kernel.org Signed-off-by: Damien Le Moal Signed-off-by: Mike Snitzer --- drivers/md/dm-crypt.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) (limited to 'drivers') diff --git a/drivers/md/dm-crypt.c b/drivers/md/dm-crypt.c index 238cd80826a6..380386c36921 100644 --- a/drivers/md/dm-crypt.c +++ b/drivers/md/dm-crypt.c @@ -739,7 +739,7 @@ static int crypt_iv_eboiv_gen(struct crypt_config *cc, u8 *iv, u8 buf[MAX_CIPHER_BLOCKSIZE] __aligned(__alignof__(__le64)); struct skcipher_request *req; struct scatterlist src, dst; - struct crypto_wait wait; + DECLARE_CRYPTO_WAIT(wait); int err; req = skcipher_request_alloc(any_tfm(cc), GFP_NOIO); @@ -936,7 +936,7 @@ static int crypt_iv_elephant(struct crypt_config *cc, struct dm_crypt_request *d u8 *es, *ks, *data, *data2, *data_offset; struct skcipher_request *req; struct scatterlist *sg, *sg2, src, dst; - struct crypto_wait wait; + DECLARE_CRYPTO_WAIT(wait); int i, r; req = skcipher_request_alloc(elephant->tfm, GFP_NOIO); -- cgit From e27fec66f0a94e35a35548bd0b29ae616e62ec62 Mon Sep 17 00:00:00 2001 From: Mikulas Patocka Date: Mon, 31 Aug 2020 09:25:41 -0400 Subject: dm integrity: fix error reporting in bitmap mode after creation The dm-integrity target did not report errors in bitmap mode just after creation. The reason is that the function integrity_recalc didn't clean up ic->recalc_bitmap as it proceeded with recalculation. Fix this by updating the bitmap accordingly -- the double shift serves to rounddown. Signed-off-by: Mikulas Patocka Fixes: 468dfca38b1a ("dm integrity: add a bitmap mode") Cc: stable@vger.kernel.org # v5.2+ Signed-off-by: Mike Snitzer --- drivers/md/dm-integrity.c | 12 ++++++++++++ 1 file changed, 12 insertions(+) (limited to 'drivers') diff --git a/drivers/md/dm-integrity.c b/drivers/md/dm-integrity.c index 8c8d940e532e..3fc3757def55 100644 --- a/drivers/md/dm-integrity.c +++ b/drivers/md/dm-integrity.c @@ -2487,6 +2487,7 @@ next_chunk: range.logical_sector = le64_to_cpu(ic->sb->recalc_sector); if (unlikely(range.logical_sector >= ic->provided_data_sectors)) { if (ic->mode == 'B') { + block_bitmap_op(ic, ic->recalc_bitmap, 0, ic->provided_data_sectors, BITMAP_OP_CLEAR); DEBUG_print("queue_delayed_work: bitmap_flush_work\n"); queue_delayed_work(ic->commit_wq, &ic->bitmap_flush_work, 0); } @@ -2564,6 +2565,17 @@ next_chunk: goto err; } + if (ic->mode == 'B') { + sector_t start, end; + start = (range.logical_sector >> + (ic->sb->log2_sectors_per_block + ic->log2_blocks_per_bitmap_bit)) << + (ic->sb->log2_sectors_per_block + ic->log2_blocks_per_bitmap_bit); + end = ((range.logical_sector + range.n_sectors) >> + (ic->sb->log2_sectors_per_block + ic->log2_blocks_per_bitmap_bit)) << + (ic->sb->log2_sectors_per_block + ic->log2_blocks_per_bitmap_bit); + block_bitmap_op(ic, ic->recalc_bitmap, start, end - start, BITMAP_OP_CLEAR); + } + advance_and_next: cond_resched(); -- cgit From d16ff19e69ab57e08bf908faaacbceaf660249de Mon Sep 17 00:00:00 2001 From: Ye Bin Date: Tue, 1 Sep 2020 14:25:42 +0800 Subject: dm cache metadata: Avoid returning cmd->bm wild pointer on error Maybe __create_persistent_data_objects() caller will use PTR_ERR as a pointer, it will lead to some strange things. Signed-off-by: Ye Bin Cc: stable@vger.kernel.org Signed-off-by: Mike Snitzer --- drivers/md/dm-cache-metadata.c | 8 ++++++-- 1 file changed, 6 insertions(+), 2 deletions(-) (limited to 'drivers') diff --git a/drivers/md/dm-cache-metadata.c b/drivers/md/dm-cache-metadata.c index 151aa95775be..af6d4f898e4c 100644 --- a/drivers/md/dm-cache-metadata.c +++ b/drivers/md/dm-cache-metadata.c @@ -537,12 +537,16 @@ static int __create_persistent_data_objects(struct dm_cache_metadata *cmd, CACHE_MAX_CONCURRENT_LOCKS); if (IS_ERR(cmd->bm)) { DMERR("could not create block manager"); - return PTR_ERR(cmd->bm); + r = PTR_ERR(cmd->bm); + cmd->bm = NULL; + return r; } r = __open_or_format_metadata(cmd, may_format_device); - if (r) + if (r) { dm_block_manager_destroy(cmd->bm); + cmd->bm = NULL; + } return r; } -- cgit From 219403d7e56f9b716ad80ab87db85d29547ee73e Mon Sep 17 00:00:00 2001 From: Ye Bin Date: Tue, 1 Sep 2020 14:25:43 +0800 Subject: dm thin metadata: Avoid returning cmd->bm wild pointer on error Maybe __create_persistent_data_objects() caller will use PTR_ERR as a pointer, it will lead to some strange things. Signed-off-by: Ye Bin Cc: stable@vger.kernel.org Signed-off-by: Mike Snitzer --- drivers/md/dm-thin-metadata.c | 8 ++++++-- 1 file changed, 6 insertions(+), 2 deletions(-) (limited to 'drivers') diff --git a/drivers/md/dm-thin-metadata.c b/drivers/md/dm-thin-metadata.c index 76b6b323bf4b..ddb7f1f0bc48 100644 --- a/drivers/md/dm-thin-metadata.c +++ b/drivers/md/dm-thin-metadata.c @@ -739,12 +739,16 @@ static int __create_persistent_data_objects(struct dm_pool_metadata *pmd, bool f THIN_MAX_CONCURRENT_LOCKS); if (IS_ERR(pmd->bm)) { DMERR("could not create block manager"); - return PTR_ERR(pmd->bm); + r = PTR_ERR(pmd->bm); + pmd->bm = NULL; + return r; } r = __open_or_format_metadata(pmd, format_device); - if (r) + if (r) { dm_block_manager_destroy(pmd->bm); + pmd->bm = NULL; + } return r; } -- cgit From 3a653b205f29b3f9827a01a0c88bfbcb0d169494 Mon Sep 17 00:00:00 2001 From: Ye Bin Date: Tue, 1 Sep 2020 14:25:44 +0800 Subject: dm thin metadata: Fix use-after-free in dm_bm_set_read_only The following error ocurred when testing disk online/offline: [ 301.798344] device-mapper: thin: 253:5: aborting current metadata transaction [ 301.848441] device-mapper: thin: 253:5: failed to abort metadata transaction [ 301.849206] Aborting journal on device dm-26-8. [ 301.850489] EXT4-fs error (device dm-26) in __ext4_new_inode:943: Journal has aborted [ 301.851095] EXT4-fs (dm-26): Delayed block allocation failed for inode 398742 at logical offset 181 with max blocks 19 with error 30 [ 301.854476] BUG: KASAN: use-after-free in dm_bm_set_read_only+0x3a/0x40 [dm_persistent_data] Reason is: metadata_operation_failed abort_transaction dm_pool_abort_metadata __create_persistent_data_objects r = __open_or_format_metadata if (r) --> If failed will free pmd->bm but pmd->bm not set NULL dm_block_manager_destroy(pmd->bm); set_pool_mode dm_pool_metadata_read_only(pool->pmd); dm_bm_set_read_only(pmd->bm); --> use-after-free Add checks to see if pmd->bm is NULL in dm_bm_set_read_only and dm_bm_set_read_write functions. If bm is NULL it means creating the bm failed and so dm_bm_is_read_only must return true. Signed-off-by: Ye Bin Cc: stable@vger.kernel.org Signed-off-by: Mike Snitzer --- drivers/md/dm-thin-metadata.c | 2 +- drivers/md/persistent-data/dm-block-manager.c | 14 ++++++++------ 2 files changed, 9 insertions(+), 7 deletions(-) (limited to 'drivers') diff --git a/drivers/md/dm-thin-metadata.c b/drivers/md/dm-thin-metadata.c index ddb7f1f0bc48..b461836b6d26 100644 --- a/drivers/md/dm-thin-metadata.c +++ b/drivers/md/dm-thin-metadata.c @@ -958,7 +958,7 @@ int dm_pool_metadata_close(struct dm_pool_metadata *pmd) } pmd_write_lock_in_core(pmd); - if (!dm_bm_is_read_only(pmd->bm) && !pmd->fail_io) { + if (!pmd->fail_io && !dm_bm_is_read_only(pmd->bm)) { r = __commit_transaction(pmd); if (r < 0) DMWARN("%s: __commit_transaction() failed, error = %d", diff --git a/drivers/md/persistent-data/dm-block-manager.c b/drivers/md/persistent-data/dm-block-manager.c index 749ec268d957..54c089a50b15 100644 --- a/drivers/md/persistent-data/dm-block-manager.c +++ b/drivers/md/persistent-data/dm-block-manager.c @@ -493,7 +493,7 @@ int dm_bm_write_lock(struct dm_block_manager *bm, void *p; int r; - if (bm->read_only) + if (dm_bm_is_read_only(bm)) return -EPERM; p = dm_bufio_read(bm->bufio, b, (struct dm_buffer **) result); @@ -562,7 +562,7 @@ int dm_bm_write_lock_zero(struct dm_block_manager *bm, struct buffer_aux *aux; void *p; - if (bm->read_only) + if (dm_bm_is_read_only(bm)) return -EPERM; p = dm_bufio_new(bm->bufio, b, (struct dm_buffer **) result); @@ -602,7 +602,7 @@ EXPORT_SYMBOL_GPL(dm_bm_unlock); int dm_bm_flush(struct dm_block_manager *bm) { - if (bm->read_only) + if (dm_bm_is_read_only(bm)) return -EPERM; return dm_bufio_write_dirty_buffers(bm->bufio); @@ -616,19 +616,21 @@ void dm_bm_prefetch(struct dm_block_manager *bm, dm_block_t b) bool dm_bm_is_read_only(struct dm_block_manager *bm) { - return bm->read_only; + return (bm ? bm->read_only : true); } EXPORT_SYMBOL_GPL(dm_bm_is_read_only); void dm_bm_set_read_only(struct dm_block_manager *bm) { - bm->read_only = true; + if (bm) + bm->read_only = true; } EXPORT_SYMBOL_GPL(dm_bm_set_read_only); void dm_bm_set_read_write(struct dm_block_manager *bm) { - bm->read_only = false; + if (bm) + bm->read_only = false; } EXPORT_SYMBOL_GPL(dm_bm_set_read_write); -- cgit