md: Introduction - assorted fixes
Subsystems (2)
PATCH 1/1 md: Introduction - assorted fixes
The following 5 patches are assorted fixes for md in 2.6.16-whatever
Patch 4 isn't exactly a 'fix'. It adds functionality to raid6 to match raid5.
All should be suitable for 2.6.16-rc2
Thanks,
NeilBrown
[PATCH 001 of 5] md: Fix device-size updates in md.
[PATCH 002 of 5] md: Make sure array geometry changes persist with version-1 superblocks. [PATCH 003 of 5] md: Don't remove bitmap from md array when switching to read-only [PATCH 004 of 5] md: Add sysfs access to raid6 stripe cache size
[PATCH 005 of 5] md: Make sure QUEUE_FLAG_CLUSTER is set properly for md.
PATCH 1/1 md: Fix device-size updates in md.
As 'array_size' is a 'sector_t', it may overflow inappropriately when shifted 10 bits. So We should cast it to a loff_t first.
There are two places with this problem, but the second (in update_raid_disks) isn't needed so just remove it:
The only personality that handles ->reshape currently is raid1, and it doesn't change the size of the array.
When added for raid5/6, reshape again won't change the size of the array, at least not straight away.
This code might be need for reshaping 'linear' but linear->shape, if implemented, should probably do the i_size_write itself.
### Diffstat output
./drivers/md/md.c | 13 +------------ 1 file changed, 1 insertion(+), 12 deletions(-)
drivers/md/md.c renamed from drivers/md/md.c~current~ +1 -12
@@ -3466,7 +3466,7 @@ static int update_size(mddev_t *mddev, u bdev =bdget_disk (mddev->gendisk,0 );if (bdev) {mutex_lock (&bdev->bd_inode->i_mutex); -i_size_write (bdev->bd_inode, mddev->array_size <<10 ); +i_size_write (bdev->bd_inode, (loff_t )mddev->array_size <<10 );mutex_unlock (&bdev->bd_inode->i_mutex);bdput (bdev); }
@@ -3486,17 +3486,6 @@ static int update_raid_disks(mddev_t *mdif (mddev->sync_thread)return -EBUSY ; rv = mddev->pers->reshape (mddev, raid_disks); -if (!rv) { -struct block_device *bdev; - - bdev =bdget_disk (mddev->gendisk,0 ); -if (bdev) { -mutex_lock (&bdev->bd_inode->i_mutex); -i_size_write (bdev->bd_inode, mddev->array_size <<10 ); -mutex_unlock (&bdev->bd_inode->i_mutex); -bdput (bdev); - } - }return rv; }
PATCH 1/1 md: Make sure array geometry changes persist with version-1 superblocks.
In reply to NeilBrownsuper_1_sync only updates fields in the superblock that might have changed.
'raid_disks' and 'size' could have changed, but this information doesn't get updated.... until this patch.
### Diffstat output
./drivers/md/md.c | 3 +++ 1 file changed, 3 insertions(+)
drivers/md/md.c renamed from drivers/md/md.c~current~ +3 -0
@@ -1162,6 +1162,9 @@ static void super_1_sync(mddev_t *mddev, sb->cnt_corrected_read =atomic_read (&rdev->corrected_errors); + sb->raid_disks =cpu_to_le32 (mddev->raid_disks); + sb->size =cpu_to_le64 (mddev->size); +if (mddev->bitmap && mddev->bitmap_file ==NULL ) { sb->bitmap_offset =cpu_to_le32 ((__u32 )mddev->bitmap_offset); sb->feature_map =cpu_to_le32 (MD_FEATURE_BITMAP_OFFSET );
PATCH 1/1 md: Don't remove bitmap from md array when switching to read-only
In reply to NeilBrownWhile a read-only array doesn't not really need a bitmap, we should not remove the bitmap when switching an array to read-only because
a/ There is no code to re-add the bitmap which switching to read-write, b/ There is insufficient locking - the bitmap could be accessed while it is
being removed.
### Diffstat output
./drivers/md/md.c | 16 ++++++++-------- 1 file changed, 8 insertions(+), 8 deletions(-)
drivers/md/md.c renamed from drivers/md/md.c~current~ +8 -8
@@ -2690,14 +2690,6 @@ static int do_md_stop(mddev_t * mddev, iset_disk_ro (disk,1 ); } -bitmap_destroy (mddev); -if (mddev->bitmap_file) { -atomic_set (&mddev->bitmap_file->f_dentry->d_inode->i_writecount,1 ); -fput (mddev->bitmap_file); - mddev->bitmap_file =NULL ; - } - mddev->bitmap_offset =0 ; -/* * Free resources if final stop */
@@ -2707,6 +2699,14 @@ static int do_md_stop(mddev_t * mddev, istruct gendisk *disk;printk (KERN_INFO "md: %s stopped.\n" ,mdname (mddev)); +bitmap_destroy (mddev); +if (mddev->bitmap_file) { +atomic_set (&mddev->bitmap_file->f_dentry->d_inode->i_writecount,1 ); +fput (mddev->bitmap_file); + mddev->bitmap_file =NULL ; + } + mddev->bitmap_offset =0 ; + ITERATE_RDEV(mddev,rdev,tmp)if (rdev->raid_disk >=0 ) {char nm[20 ];
PATCH 1/1 md: Add sysfs access to raid6 stripe cache size
In reply to NeilBrown.. just as we already have for raid5.
### Diffstat output
./drivers/md/raid6main.c | 149 ++++++++++++++++++++++++++++++++++++----------- 1 file changed, 115 insertions(+), 34 deletions(-)
drivers/md/raid6main.c renamed from drivers/md/raid6main.c~current~ +115 -34
@@ -115,7 +115,7 @@ static void __release_stripe(raid6_conf_list_add_tail (&sh->lru, &conf->inactive_list);atomic_dec (&conf->active_stripes);if (!conf->inactive_blocked || -atomic_read (&conf->active_stripes) < (NR_STRIPES *3 /4 )) +atomic_read (&conf->active_stripes) < (conf->max_nr_stripes*3 /4 ))wake_up (&conf->wait_for_stripe); } }
@@ -273,7 +273,8 @@ static struct stripe_head *get_active_st conf->inactive_blocked =1 ;wait_event_lock_irq (conf->wait_for_stripe, !list_empty (&conf->inactive_list) && - (atomic_read (&conf->active_stripes) < (NR_STRIPES *3 /4 ) + (atomic_read (&conf->active_stripes) + < (conf->max_nr_stripes *3 /4 ) || !conf->inactive_blocked), conf->device_lock, unplug_slaves(conf->mddev);
@@ -302,9 +303,31 @@ static struct stripe_head *get_active_streturn sh; } -static int grow_stripes (raid6_conf_t *conf,int num) +static int grow_one_stripe (raid6_conf_t *conf) {struct stripe_head *sh; + sh =kmem_cache_alloc (conf->slab_cache,GFP_KERNEL ); +if (!sh) +return 0 ; +memset (sh,0 ,sizeof (*sh) + (conf->raid_disks-1 )*sizeof (struct r5dev )); + sh->raid_conf = conf; +spin_lock_init (&sh->lock); + +if (grow_buffers (sh, conf->raid_disks)) { +shrink_buffers (sh, conf->raid_disks); +kmem_cache_free (conf->slab_cache, sh); +return 0 ; + } +/* we just created an active stripe so... */ +atomic_set (&sh->count,1 ); +atomic_inc (&conf->active_stripes); +INIT_LIST_HEAD (&sh->lru); +release_stripe (sh); +return 1 ; +} + +static int grow_stripes (raid6_conf_t *conf,int num) +{kmem_cache_t *sc;int devs = conf->raid_disks;
@@ -316,44 +339,33 @@ static int grow_stripes(raid6_conf_t *coif (!sc)return 1 ; conf->slab_cache = sc; -while (num--) { - sh =kmem_cache_alloc (sc,GFP_KERNEL ); -if (!sh) +while (num--) +if (!grow_one_stripe (conf))return 1 ; -memset (sh,0 ,sizeof (*sh) + (devs-1 )*sizeof (struct r5dev )); - sh->raid_conf = conf; -spin_lock_init (&sh->lock); - -if (grow_buffers (sh, conf->raid_disks)) { -shrink_buffers (sh, conf->raid_disks); -kmem_cache_free (sc, sh); -return 1 ; - } -/* we just created an active stripe so... */ -atomic_set (&sh->count,1 ); -atomic_inc (&conf->active_stripes); -INIT_LIST_HEAD (&sh->lru); -release_stripe (sh); - }return 0 ; } -static void shrink_stripes (raid6_conf_t *conf) +static int drop_one_stripe (raid6_conf_t *conf) {struct stripe_head *sh; +spin_lock_irq (&conf->device_lock); + sh =get_free_stripe (conf); +spin_unlock_irq (&conf->device_lock); +if (!sh) +return 0 ; +if (atomic_read (&sh->count)) +BUG (); +shrink_buffers (sh, conf->raid_disks); +kmem_cache_free (conf->slab_cache, sh); +atomic_dec (&conf->active_stripes); +return 1 ; +} + +static void shrink_stripes (raid6_conf_t *conf) +{ +while (drop_one_stripe (conf)) + ; -while (1 ) { -spin_lock_irq (&conf->device_lock); - sh =get_free_stripe (conf); -spin_unlock_irq (&conf->device_lock); -if (!sh) -break ; -if (atomic_read (&sh->count)) -BUG (); -shrink_buffers (sh, conf->raid_disks); -kmem_cache_free (conf->slab_cache, sh); -atomic_dec (&conf->active_stripes); - }kmem_cache_destroy (conf->slab_cache); conf->slab_cache =NULL ; }
@@ -1912,6 +1924,74 @@ static void raid6d (mddev_t *mddev)PRINTK ("--- raid6d inactive\n" ); } +static ssize_t +raid6_show_stripe_cache_size (mddev_t *mddev,char *page) +{ +raid6_conf_t *conf =mddev_to_conf (mddev); +if (conf) +return sprintf (page,"%d\n" , conf->max_nr_stripes); +else +return 0 ; +} + +static ssize_t +raid6_store_stripe_cache_size (mddev_t *mddev,const char *page,size_t len) +{ +raid6_conf_t *conf =mddev_to_conf (mddev); +char *end; +int new; +if (len >=PAGE_SIZE ) +return -EINVAL ; +if (!conf) +return -ENODEV ; + + new =simple_strtoul (page, &end,10 ); +if (!*page || (*end && *end !='\n' ) ) +return -EINVAL ; +if (new <=16 || new >32768 ) +return -EINVAL ; +while (new < conf->max_nr_stripes) { +if (drop_one_stripe (conf)) + conf->max_nr_stripes--; +else +break ; + } +while (new > conf->max_nr_stripes) { +if (grow_one_stripe (conf)) + conf->max_nr_stripes++; +else break ; + } +return len; +} + +static struct md_sysfs_entry +raid6_stripecache_size =__ATTR (stripe_cache_size,S_IRUGO |S_IWUSR , + raid6_show_stripe_cache_size, + raid6_store_stripe_cache_size); + +static ssize_t +stripe_cache_active_show (mddev_t *mddev,char *page) +{ +raid6_conf_t *conf =mddev_to_conf (mddev); +if (conf) +return sprintf (page,"%d\n" ,atomic_read (&conf->active_stripes)); +else +return 0 ; +} + +static struct md_sysfs_entry +raid6_stripecache_active =__ATTR_RO (stripe_cache_active); + +static struct attribute *raid6_attrs[] = { + &raid6_stripecache_size.attr, + &raid6_stripecache_active.attr, +NULL , +}; +static struct attribute_group raid6_attrs_group = { + .name =NULL , + .attrs = raid6_attrs, +}; +static int run (mddev_t *mddev) {raid6_conf_t *conf;
@@ -2095,6 +2175,7 @@ static int stop (mddev_t *mddev)shrink_stripes (conf);kfree (conf->stripe_hashtbl);blk_sync_queue (mddev->queue);/* the unplug fn references 'conf'*/ +sysfs_remove_group (&mddev->kobj, &raid6_attrs_group);kfree (conf); mddev->private =NULL ;return 0 ;
PATCH 1/1 md: Make sure QUEUE_FLAG_CLUSTER is set properly for md.
In reply to NeilBrownThis flag should be set for a virtual device iff it is set for all underlying devices.
### Diffstat output
./block/ll_rw_blk.c | 2 ++ ./drivers/md/md.c | 1 + 2 files changed, 3 insertions(+)
block/ll_rw_blk.c renamed from block/ll_rw_blk.c~current~ +2 -0
@@ -778,6 +778,8 @@ void blk_queue_stack_limits(request_queu t->max_hw_segments =min (t->max_hw_segments,b->max_hw_segments); t->max_segment_size =min (t->max_segment_size,b->max_segment_size); t->hardsect_size =max (t->hardsect_size,b->hardsect_size); +if (!test_bit (QUEUE_FLAG_CLUSTER , &b->queue_flags)) +clear_bit (QUEUE_FLAG_CLUSTER , &t->queue_flags); }EXPORT_SYMBOL (blk_queue_stack_limits);
drivers/md/md.c renamed from drivers/md/md.c~current~ +1 -0
@@ -264,6 +264,7 @@ static mddev_t * mddev_find(dev_t unit)kfree (new);return NULL ; } +set_bit (QUEUE_FLAG_CLUSTER , &new->queue->queue_flags);blk_queue_make_request (new->queue, md_fail_request);