From a0da84f35b25875870270d16b6eccda4884d61a7 Mon Sep 17 00:00:00 2001 From: Neil Brown Date: Sat, 28 Jun 2008 08:31:22 +1000 Subject: Improve setting of "events_cleared" for write-intent bitmaps. When an array is degraded, bits in the write-intent bitmap are not cleared, so that if the missing device is re-added, it can be synced by only updated those parts of the device that have changed since it was removed. The enable this a 'events_cleared' value is stored. It is the event counter for the array the last time that any bits were cleared. Sometimes - if a device disappears from an array while it is 'clean' - the events_cleared value gets updated incorrectly (there are subtle ordering issues between updateing events in the main metadata and the bitmap metadata) resulting in the missing device appearing to require a full resync when it is re-added. With this patch, we update events_cleared precisely when we are about to clear a bit in the bitmap. We record events_cleared when we clear the bit internally, and copy that to the superblock which is written out before the bit on storage. This makes it more "obviously correct". We also need to update events_cleared when the event_count is going backwards (as happens on a dirty->clean transition of a non-degraded array). Thanks to Mike Snitzer for identifying this problem and testing early "fixes". Cc: "Mike Snitzer" Signed-off-by: Neil Brown --- include/linux/raid/bitmap.h | 1 + 1 file changed, 1 insertion(+) (limited to 'include') diff --git a/include/linux/raid/bitmap.h b/include/linux/raid/bitmap.h index 78bfdea24a8..e98900671ca 100644 --- a/include/linux/raid/bitmap.h +++ b/include/linux/raid/bitmap.h @@ -221,6 +221,7 @@ struct bitmap { unsigned long syncchunk; __u64 events_cleared; + int need_sync; /* bitmap spinlock */ spinlock_t lock; -- cgit v1.2.3 From 5e96ee65c8bd629ce093da67a066d3946468298a Mon Sep 17 00:00:00 2001 From: Neil Brown Date: Sat, 28 Jun 2008 08:31:24 +1000 Subject: Allow setting start point for requested check/repair This makes it possible to just resync a small part of an array. e.g. if a drive reports that it has questionable sectors, a 'repair' of just the region covering those sectors will cause them to be read and, if there is an error, re-written with correct data. Signed-off-by: Neil Brown --- include/linux/raid/md_k.h | 2 ++ 1 file changed, 2 insertions(+) (limited to 'include') diff --git a/include/linux/raid/md_k.h b/include/linux/raid/md_k.h index 3dea9f545c8..780e0613e6d 100644 --- a/include/linux/raid/md_k.h +++ b/include/linux/raid/md_k.h @@ -227,6 +227,8 @@ struct mddev_s atomic_t recovery_active; /* blocks scheduled, but not written */ wait_queue_head_t recovery_wait; sector_t recovery_cp; + sector_t resync_min; /* user requested sync + * starts here */ sector_t resync_max; /* resync should pause * when it gets here */ -- cgit v1.2.3 From 72a23c211e4587859d5bf61ac4962d76e593fb02 Mon Sep 17 00:00:00 2001 From: Neil Brown Date: Sat, 28 Jun 2008 08:31:41 +1000 Subject: Make sure all changes to md/sync_action are notified. When the 'resync' thread starts or stops, when we explicitly set sync_action, or when we determine that there is definitely nothing to do, we notify sync_action. To stop "sync_action" from occasionally showing the wrong value, we introduce a new flags - MD_RECOVERY_RECOVER - to say that a recovery is probably needed or happening, and we make sure that we set MD_RECOVERY_RUNNING before clearing MD_RECOVERY_NEEDED. Signed-off-by: Neil Brown --- include/linux/raid/md_k.h | 2 ++ 1 file changed, 2 insertions(+) (limited to 'include') diff --git a/include/linux/raid/md_k.h b/include/linux/raid/md_k.h index 780e0613e6d..62aa9c9a6dd 100644 --- a/include/linux/raid/md_k.h +++ b/include/linux/raid/md_k.h @@ -188,6 +188,7 @@ struct mddev_s * NEEDED: we might need to start a resync/recover * RUNNING: a thread is running, or about to be started * SYNC: actually doing a resync, not a recovery + * RECOVER: doing recovery, or need to try it. * INTR: resync needs to be aborted for some reason * DONE: thread is done and is waiting to be reaped * REQUEST: user-space has requested a sync (used with SYNC) @@ -198,6 +199,7 @@ struct mddev_s */ #define MD_RECOVERY_RUNNING 0 #define MD_RECOVERY_SYNC 1 +#define MD_RECOVERY_RECOVER 2 #define MD_RECOVERY_INTR 3 #define MD_RECOVERY_DONE 4 #define MD_RECOVERY_NEEDED 5 -- cgit v1.2.3 From 526647320e696f434647f38421a6ecf65b859c43 Mon Sep 17 00:00:00 2001 From: Neil Brown Date: Sat, 28 Jun 2008 08:31:44 +1000 Subject: Make sure all changes to md/dev-XX/state are notified The important state change happens during an interrupt in md_error. So just set a flag there and call sysfs_notify later in process context. Signed-off-by: Neil Brown --- include/linux/raid/md_k.h | 3 +++ 1 file changed, 3 insertions(+) (limited to 'include') diff --git a/include/linux/raid/md_k.h b/include/linux/raid/md_k.h index 62aa9c9a6dd..df30c439587 100644 --- a/include/linux/raid/md_k.h +++ b/include/linux/raid/md_k.h @@ -87,6 +87,9 @@ struct mdk_rdev_s #define Blocked 8 /* An error occured on an externally * managed array, don't allow writes * until it is cleared */ +#define StateChanged 9 /* Faulty or Blocked has changed during + * interrupt, so it needs to be + * notified by the thread */ wait_queue_head_t blocked_wait; int desc_nr; /* descriptor index in the superblock */ -- cgit v1.2.3 From b203886edbcaac3ca427cf4dbcb50b18bdb346fd Mon Sep 17 00:00:00 2001 From: Dan Williams Date: Sat, 28 Jun 2008 08:31:50 +1000 Subject: md: kill STRIPE_OP_MOD_DMA in raid5 offload From: Dan Williams This micro-optimization allowed the raid code to skip a re-read of the parity block after checking parity. It took advantage of the fact that xor-offload-engines have their own internal result buffer and can check parity without writing to memory. Remove it for the following reasons: 1/ It is a layering violation for MD to need to manage the DMA and non-DMA paths within async_xor_zero_sum 2/ Bad precedent to toggle the 'ops' flags outside the lock 3/ Hard to realize a performance gain as reads will not need an updated parity block and writes will dirty it anyways. Signed-off-by: Dan Williams Signed-off-by: Neil Brown --- include/linux/raid/raid5.h | 2 -- 1 file changed, 2 deletions(-) (limited to 'include') diff --git a/include/linux/raid/raid5.h b/include/linux/raid/raid5.h index f0827d31ae6..4ecae31a3dc 100644 --- a/include/linux/raid/raid5.h +++ b/include/linux/raid/raid5.h @@ -267,10 +267,8 @@ struct r6_state { /* modifiers to the base operations * STRIPE_OP_MOD_REPAIR_PD - compute the parity block and write it back - * STRIPE_OP_MOD_DMA_CHECK - parity is not corrupted by the check */ #define STRIPE_OP_MOD_REPAIR_PD 7 -#define STRIPE_OP_MOD_DMA_CHECK 8 /* * Plugging: -- cgit v1.2.3 From 2b7497f0e0a0b9cf21d822e427d5399b2056501a Mon Sep 17 00:00:00 2001 From: Dan Williams Date: Sat, 28 Jun 2008 08:31:52 +1000 Subject: md: kill STRIPE_OP_IO flag From: Dan Williams The R5_Want{Read,Write} flags already gate i/o. So, this flag is superfluous and we can unconditionally call ops_run_io(). Signed-off-by: Dan Williams Signed-off-by: Neil Brown --- include/linux/raid/raid5.h | 1 - 1 file changed, 1 deletion(-) (limited to 'include') diff --git a/include/linux/raid/raid5.h b/include/linux/raid/raid5.h index 4ecae31a3dc..1301195abf4 100644 --- a/include/linux/raid/raid5.h +++ b/include/linux/raid/raid5.h @@ -263,7 +263,6 @@ struct r6_state { #define STRIPE_OP_BIODRAIN 3 #define STRIPE_OP_POSTXOR 4 #define STRIPE_OP_CHECK 5 -#define STRIPE_OP_IO 6 /* modifiers to the base operations * STRIPE_OP_MOD_REPAIR_PD - compute the parity block and write it back -- cgit v1.2.3 From ecc65c9b3f9b9d740a5deade3d85b39be56401b6 Mon Sep 17 00:00:00 2001 From: Dan Williams Date: Sat, 28 Jun 2008 08:31:57 +1000 Subject: md: replace STRIPE_OP_CHECK with 'check_states' From: Dan Williams The STRIPE_OP_* flags record the state of stripe operations which are performed outside the stripe lock. Their use in indicating which operations need to be run is straightforward; however, interpolating what the next state of the stripe should be based on a given combination of these flags is not straightforward, and has led to bugs. An easier to read implementation with minimal degrees of freedom is needed. Towards this goal, this patch introduces explicit states to replace what was previously interpolated from the STRIPE_OP_* flags. For now this only converts the handle_parity_checks5 path, removing a user of the ops.{pending,ack,complete,count} fields of struct stripe_operations. This conversion also found a remaining issue with the current code. There is a small window for a drive to fail between when we schedule a repair and when the parity calculation for that repair completes. When this happens we will writeback to 'failed_num' when we really want to write back to 'pd_idx'. Signed-off-by: Dan Williams Signed-off-by: Neil Brown --- include/linux/raid/raid5.h | 46 ++++++++++++++++++++++++++++++++++++++++------ 1 file changed, 40 insertions(+), 6 deletions(-) (limited to 'include') diff --git a/include/linux/raid/raid5.h b/include/linux/raid/raid5.h index 1301195abf4..2c96d5fd54b 100644 --- a/include/linux/raid/raid5.h +++ b/include/linux/raid/raid5.h @@ -158,6 +158,41 @@ * the compute block completes. */ +/* + * Operations state - intermediate states that are visible outside of sh->lock + * In general _idle indicates nothing is running, _run indicates a data + * processing operation is active, and _result means the data processing result + * is stable and can be acted upon. For simple operations like biofill and + * compute that only have an _idle and _run state they are indicated with + * sh->state flags (STRIPE_BIOFILL_RUN and STRIPE_COMPUTE_RUN) + */ +/** + * enum check_states - handles syncing / repairing a stripe + * @check_state_idle - check operations are quiesced + * @check_state_run - check operation is running + * @check_state_result - set outside lock when check result is valid + * @check_state_compute_run - check failed and we are repairing + * @check_state_compute_result - set outside lock when compute result is valid + */ +enum check_states { + check_state_idle = 0, + check_state_run, /* parity check */ + check_state_check_result, + check_state_compute_run, /* parity repair */ + check_state_compute_result, +}; + +/** + * enum reconstruct_states - handles writing or expanding a stripe + */ +enum reconstruct_states { + reconstruct_state_idle = 0, + reconstruct_state_drain_run, /* write */ + reconstruct_state_run, /* expand */ + reconstruct_state_drain_result, + reconstruct_state_result, +}; + struct stripe_head { struct hlist_node hash; struct list_head lru; /* inactive_list or handle_list */ @@ -169,6 +204,7 @@ struct stripe_head { spinlock_t lock; int bm_seq; /* sequence number for bitmap flushes */ int disks; /* disks in stripe */ + enum check_states check_state; /* stripe_operations * @pending - pending ops flags (set for request->issue->complete) * @ack - submitted ops flags (set for issue->complete) @@ -202,6 +238,7 @@ struct stripe_head_state { int locked, uptodate, to_read, to_write, failed, written; int to_fill, compute, req_compute, non_overwrite; int failed_num; + unsigned long ops_request; }; /* r6_state - extra state data only relevant to r6 */ @@ -254,8 +291,10 @@ struct r6_state { #define STRIPE_EXPAND_READY 11 #define STRIPE_IO_STARTED 12 /* do not count towards 'bypass_count' */ #define STRIPE_FULL_WRITE 13 /* all blocks are set to be overwritten */ +#define STRIPE_BIOFILL_RUN 14 +#define STRIPE_COMPUTE_RUN 15 /* - * Operations flags (in issue order) + * Operation request flags */ #define STRIPE_OP_BIOFILL 0 #define STRIPE_OP_COMPUTE_BLK 1 @@ -264,11 +303,6 @@ struct r6_state { #define STRIPE_OP_POSTXOR 4 #define STRIPE_OP_CHECK 5 -/* modifiers to the base operations - * STRIPE_OP_MOD_REPAIR_PD - compute the parity block and write it back - */ -#define STRIPE_OP_MOD_REPAIR_PD 7 - /* * Plugging: * -- cgit v1.2.3 From 600aa10993012ff2dd5617720dac081e4f992017 Mon Sep 17 00:00:00 2001 From: Dan Williams Date: Sat, 28 Jun 2008 08:32:05 +1000 Subject: md: replace STRIPE_OP_{BIODRAIN,PREXOR,POSTXOR} with 'reconstruct_states' From: Dan Williams Track the state of reconstruct operations (recalculating the parity block usually due to incoming writes, or as part of array expansion) Reduces the scope of the STRIPE_OP_{BIODRAIN,PREXOR,POSTXOR} flags to only tracking whether a reconstruct operation has been requested via the ops_request field of struct stripe_head_state. This is the final step in the removal of ops.{pending,ack,complete,count}, i.e. the STRIPE_OP_{BIODRAIN,PREXOR,POSTXOR} flags only request an operation and do not track the state of the operation. Signed-off-by: Dan Williams Signed-off-by: Neil Brown --- include/linux/raid/raid5.h | 9 +-------- 1 file changed, 1 insertion(+), 8 deletions(-) (limited to 'include') diff --git a/include/linux/raid/raid5.h b/include/linux/raid/raid5.h index 2c96d5fd54b..5f3e674b87d 100644 --- a/include/linux/raid/raid5.h +++ b/include/linux/raid/raid5.h @@ -205,19 +205,12 @@ struct stripe_head { int bm_seq; /* sequence number for bitmap flushes */ int disks; /* disks in stripe */ enum check_states check_state; + enum reconstruct_states reconstruct_state; /* stripe_operations - * @pending - pending ops flags (set for request->issue->complete) - * @ack - submitted ops flags (set for issue->complete) - * @complete - completed ops flags (set for complete) * @target - STRIPE_OP_COMPUTE_BLK target - * @count - raid5_runs_ops is set to run when this is non-zero */ struct stripe_operations { - unsigned long pending; - unsigned long ack; - unsigned long complete; int target; - int count; u32 zero_sum_result; } ops; struct r5dev { -- cgit v1.2.3 From d8ee0728b5b30d7a6f62c399a95e953616d31f23 Mon Sep 17 00:00:00 2001 From: Dan Williams Date: Sat, 28 Jun 2008 08:32:06 +1000 Subject: md: replace R5_WantPrexor with R5_WantDrain, add 'prexor' reconstruct_states From: Dan Williams Currently ops_run_biodrain and other locations have extra logic to determine which blocks are processed in the prexor and non-prexor cases. This can be eliminated if handle_write_operations5 flags the blocks to be processed in all cases via R5_Wantdrain. The presence of the prexor operation is tracked in sh->reconstruct_state. Signed-off-by: Dan Williams Signed-off-by: Neil Brown --- include/linux/raid/raid5.h | 6 +++--- 1 file changed, 3 insertions(+), 3 deletions(-) (limited to 'include') diff --git a/include/linux/raid/raid5.h b/include/linux/raid/raid5.h index 5f3e674b87d..3b267279245 100644 --- a/include/linux/raid/raid5.h +++ b/include/linux/raid/raid5.h @@ -187,8 +187,10 @@ enum check_states { */ enum reconstruct_states { reconstruct_state_idle = 0, + reconstruct_state_prexor_drain_run, /* prexor-write */ reconstruct_state_drain_run, /* write */ reconstruct_state_run, /* expand */ + reconstruct_state_prexor_drain_result, reconstruct_state_drain_result, reconstruct_state_result, }; @@ -258,9 +260,7 @@ struct r6_state { #define R5_Wantfill 12 /* dev->toread contains a bio that needs * filling */ -#define R5_Wantprexor 13 /* distinguish blocks ready for rmw from - * other "towrites" - */ +#define R5_Wantdrain 13 /* dev->towrite needs to be drained */ /* * Write method */ -- cgit v1.2.3 From b5470dc5fc18a8ff6517c3bb538d1479e58ecb02 Mon Sep 17 00:00:00 2001 From: Dan Williams Date: Fri, 27 Jun 2008 21:44:04 -0700 Subject: md: resolve external metadata handling deadlock in md_allow_write md_allow_write() marks the metadata dirty while holding mddev->lock and then waits for the write to complete. For externally managed metadata this causes a deadlock as userspace needs to take the lock to communicate that the metadata update has completed. Change md_allow_write() in the 'external' case to start the 'mark active' operation and then return -EAGAIN. The expected side effects while waiting for userspace to write 'active' to 'array_state' are holding off reshape (code currently handles -ENOMEM), cause some 'stripe_cache_size' change requests to fail, cause some GET_BITMAP_FILE ioctl requests to fall back to GFP_NOIO, and cause updates to 'raid_disks' to fail. Except for 'stripe_cache_size' changes these failures can be mitigated by coordinating with mdmon. md_write_start() still prevents writes from occurring until the metadata handler has had a chance to take action as it unconditionally waits for MD_CHANGE_CLEAN to be cleared. [neilb@suse.de: return -EAGAIN, try GFP_NOIO] Signed-off-by: Dan Williams --- include/linux/raid/md.h | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'include') diff --git a/include/linux/raid/md.h b/include/linux/raid/md.h index b7386ae9d28..dc0e3fcb9f2 100644 --- a/include/linux/raid/md.h +++ b/include/linux/raid/md.h @@ -95,7 +95,7 @@ extern int sync_page_io(struct block_device *bdev, sector_t sector, int size, struct page *page, int rw); extern void md_do_sync(mddev_t *mddev); extern void md_new_event(mddev_t *mddev); -extern void md_allow_write(mddev_t *mddev); +extern int md_allow_write(mddev_t *mddev); extern void md_wait_for_blocked_rdev(mdk_rdev_t *rdev, mddev_t *mddev); #endif /* CONFIG_MD */ -- cgit v1.2.3 From 0f420358e3a2abc028320ace7783e2e38cae77bf Mon Sep 17 00:00:00 2001 From: Andre Noll Date: Fri, 11 Jul 2008 22:02:23 +1000 Subject: md: Turn rdev->sb_offset into a sector-based quantity. Rename it to sb_start to make sure all users have been converted. Signed-off-by: Andre Noll Signed-off-by: Neil Brown --- include/linux/raid/md_k.h | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'include') diff --git a/include/linux/raid/md_k.h b/include/linux/raid/md_k.h index df30c439587..e37aaa41efc 100644 --- a/include/linux/raid/md_k.h +++ b/include/linux/raid/md_k.h @@ -59,7 +59,7 @@ struct mdk_rdev_s int sb_loaded; __u64 sb_events; sector_t data_offset; /* start of data in array */ - sector_t sb_offset; + sector_t sb_start; /* offset of the super block (in 512byte sectors) */ int sb_size; /* bytes in the superblock */ int preferred_minor; /* autorun support */ -- cgit v1.2.3 From 7e93a89251d4ed7bd4475db62616ccd03ddfd01a Mon Sep 17 00:00:00 2001 From: Andre Noll Date: Fri, 11 Jul 2008 22:02:23 +1000 Subject: md: Remove some unused macros. Signed-off-by: Andre Noll Signed-off-by: Neil Brown --- include/linux/raid/md_p.h | 3 --- 1 file changed, 3 deletions(-) (limited to 'include') diff --git a/include/linux/raid/md_p.h b/include/linux/raid/md_p.h index 3f2cd98c508..8b4de4a41ff 100644 --- a/include/linux/raid/md_p.h +++ b/include/linux/raid/md_p.h @@ -43,14 +43,11 @@ */ #define MD_RESERVED_BYTES (64 * 1024) #define MD_RESERVED_SECTORS (MD_RESERVED_BYTES / 512) -#define MD_RESERVED_BLOCKS (MD_RESERVED_BYTES / BLOCK_SIZE) #define MD_NEW_SIZE_SECTORS(x) ((x & ~(MD_RESERVED_SECTORS - 1)) - MD_RESERVED_SECTORS) -#define MD_NEW_SIZE_BLOCKS(x) ((x & ~(MD_RESERVED_BLOCKS - 1)) - MD_RESERVED_BLOCKS) #define MD_SB_BYTES 4096 #define MD_SB_WORDS (MD_SB_BYTES / 4) -#define MD_SB_BLOCKS (MD_SB_BYTES / BLOCK_SIZE) #define MD_SB_SECTORS (MD_SB_BYTES / 512) /* -- cgit v1.2.3 From f233ea5c9e0d8b95e4283bf6a3436b88f6fd3586 Mon Sep 17 00:00:00 2001 From: Andre Noll Date: Mon, 21 Jul 2008 17:05:22 +1000 Subject: md: Make mddev->array_size sector-based. This patch renames the array_size field of struct mddev_s to array_sectors and converts all instances to use units of 512 byte sectors instead of 1k blocks. Signed-off-by: Andre Noll Signed-off-by: NeilBrown --- include/linux/raid/md_k.h | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'include') diff --git a/include/linux/raid/md_k.h b/include/linux/raid/md_k.h index e37aaa41efc..6f72b47ae41 100644 --- a/include/linux/raid/md_k.h +++ b/include/linux/raid/md_k.h @@ -150,7 +150,7 @@ struct mddev_s int raid_disks; int max_disks; sector_t size; /* used size of component devices */ - sector_t array_size; /* exported array size */ + sector_t array_sectors; /* exported array size */ __u64 events; char uuid[16]; -- cgit v1.2.3 From d6e2215052810678bc9782fd980b52706fc71f50 Mon Sep 17 00:00:00 2001 From: Andre Noll Date: Mon, 21 Jul 2008 17:05:25 +1000 Subject: md: linear: Make array_size sector-based and rename it to array_sectors. Signed-off-by: Andre Noll Signed-off-by: NeilBrown --- include/linux/raid/linear.h | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'include') diff --git a/include/linux/raid/linear.h b/include/linux/raid/linear.h index ba15469daf1..7e375111d00 100644 --- a/include/linux/raid/linear.h +++ b/include/linux/raid/linear.h @@ -16,7 +16,7 @@ struct linear_private_data struct linear_private_data *prev; /* earlier version */ dev_info_t **hash_table; sector_t hash_spacing; - sector_t array_size; + sector_t array_sectors; int preshift; /* shift before dividing by hash_spacing */ dev_info_t disks[0]; }; -- cgit v1.2.3 From f2ea68cf42aafdd93393b6b8b20fc3c2b5f4390c Mon Sep 17 00:00:00 2001 From: NeilBrown Date: Mon, 21 Jul 2008 17:05:25 +1000 Subject: md: only count actual openers as access which prevent a 'stop' Open isn't the only thing that increments ->active. e.g. reading /proc/mdstat will increment it briefly. So to avoid false positives in testing for concurrent access, introduce a new counter that counts just the number of times the md device it open. Signed-off-by: NeilBrown --- include/linux/raid/md_k.h | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-) (limited to 'include') diff --git a/include/linux/raid/md_k.h b/include/linux/raid/md_k.h index 6f72b47ae41..4bef4791d80 100644 --- a/include/linux/raid/md_k.h +++ b/include/linux/raid/md_k.h @@ -215,7 +215,8 @@ struct mddev_s int in_sync; /* know to not need resync */ struct mutex reconfig_mutex; - atomic_t active; + atomic_t active; /* general refcount */ + atomic_t openers; /* number of active opens */ int changed; /* true if we might need to reread partition info */ int degraded; /* whether md should consider -- cgit v1.2.3 From 4b80991c6cb9efa607bc4fd6f3ecdf5511c31bb0 Mon Sep 17 00:00:00 2001 From: NeilBrown Date: Mon, 21 Jul 2008 17:05:25 +1000 Subject: md: Protect access to mddev->disks list using RCU All modifications and most access to the mddev->disks list are made under the reconfig_mutex lock. However there are three places where the list is walked without any locking. If a reconfig happens at this time, havoc (and oops) can ensue. So use RCU to protect these accesses: - wrap them in rcu_read_{,un}lock() - use list_for_each_entry_rcu - add to the list with list_add_rcu - delete from the list with list_del_rcu - delay the 'free' with call_rcu rather than schedule_work Note that export_rdev did a list_del_init on this list. In almost all cases the entry was not in the list anymore so it was a no-op and so safe. It is no longer safe as after list_del_rcu we may not touch the list_head. An audit shows that export_rdev is called: - after unbind_rdev_from_array, in which case the delete has already been done, - after bind_rdev_to_array fails, in which case the delete isn't needed. - before the device has been put on a list at all (e.g. in add_new_disk where reading the superblock fails). - and in autorun devices after a failure when the device is on a different list. So remove the list_del_init call from export_rdev, and add it back immediately before the called to export_rdev for that last case. Note also that ->same_set is sometimes used for lists other than mddev->list (e.g. candidates). In these cases rcu is not needed. Signed-off-by: NeilBrown --- include/linux/raid/md_k.h | 3 +++ 1 file changed, 3 insertions(+) (limited to 'include') diff --git a/include/linux/raid/md_k.h b/include/linux/raid/md_k.h index 4bef4791d80..9f2549ac0e2 100644 --- a/include/linux/raid/md_k.h +++ b/include/linux/raid/md_k.h @@ -339,6 +339,9 @@ static inline char * mdname (mddev_t * mddev) #define rdev_for_each(rdev, tmp, mddev) \ rdev_for_each_list(rdev, tmp, (mddev)->disks) +#define rdev_for_each_rcu(rdev, mddev) \ + list_for_each_entry_rcu(rdev, &((mddev)->disks), same_set) + typedef struct mdk_thread_s { void (*run) (mddev_t *mddev); mddev_t *mddev; -- cgit v1.2.3