dm-bufio: introduce a global cache replacement
Mikulas Patocka
maintainerPATCH 5/5 dm-bufio: introduce a global cache replacement
This patch introduces a global cache replacement (instead of per-client cleanup).
If one bufio client uses the cache heavily and another client is not using it, we want to let the first client use most of the cache. The old algorithm would partition the cache equally betwen the clients and that is inoptimal.
For cache replacement, we use the clock algorithm because it doesn't require taking any lock when the buffer is accessed.
--- drivers/md/dm-bufio.c | 103 ++++++++++++++++++++++++++++++++++++++++++++++---- 1 file changed, 96 insertions(+), 7 deletions(-)
drivers/md/dm-bufio.c +96 -7
@@ -34,6 +34,7 @@#define DM_BUFIO_MEMORY_PERCENT 2#define DM_BUFIO_VMALLOC_PERCENT 25#define DM_BUFIO_WRITEBACK_RATIO 3 +#define DM_BUFIO_LOW_WATERMARK 16/* * Check buffer ages in this interval (seconds)
@@ -140,6 +141,7 @@ struct dm_buffer {unsigned char list_mode;/* LIST_* */ blk_status_t read_error;blk_status_t write_error; +unsigned accessed;unsigned hold_count;unsigned long state;unsigned long last_accessed;
@@ -198,6 +200,8 @@ static DEFINE_SPINLOCK(global_spinlock);static LIST_HEAD (global_queue); +static unsigned long global_num =0 ; +/* * Buffers are freed after this timeout */
@@ -227,6 +231,12 @@ static LIST_HEAD(dm_bufio_all_clients);*/ static DEFINE_MUTEX (dm_bufio_clients_lock); + +static struct workqueue_struct *dm_bufio_wq; +static struct delayed_work dm_bufio_cleanup_old_work; +static struct work_struct dm_bufio_replacement_work; + +#ifdef CONFIG_DM_DEBUG_BLOCK_STACK_TRACING static void buffer_record_stack (struct dm_buffer *b) {
@@ -308,10 +318,16 @@ static void adjust_total_allocated(strucif (dm_bufio_current_allocated > dm_bufio_peak_allocated) dm_bufio_peak_allocated = dm_bufio_current_allocated; + b->accessed =1 ; +if (!unlink) {list_add (&b->global_list, &global_queue); + global_num++; +if (dm_bufio_current_allocated > dm_bufio_cache_size) +queue_work (dm_bufio_wq, &dm_bufio_replacement_work); }else {list_del (&b->global_list); + global_num--; }spin_unlock (&global_spinlock);
@@ -497,6 +513,8 @@ static void __relink_lru(struct dm_buffe {struct dm_bufio_client *c = b->c; + b->accessed =1 ; +BUG_ON (!c->n_buffers[b->list_mode]); c->n_buffers[b->list_mode]--;
@@ -1814,6 +1832,78 @@ static void __evict_old_buffers(struct ddm_bufio_unlock (c); } +static void do_global_cleanup (struct work_struct *w) +{ +struct dm_bufio_client *locked_client =NULL ; +struct dm_bufio_client *current_client; +struct dm_buffer *b; +unsigned spinlock_hold_count; +unsigned long threshold = dm_bufio_cache_size - dm_bufio_cache_size /DM_BUFIO_LOW_WATERMARK ; +unsigned long loops = global_num *2 ; + +mutex_lock (&dm_bufio_clients_lock); + +reacquire_spinlock: +cond_resched (); + +spin_lock (&global_spinlock); +if (unlikely (dm_bufio_current_allocated <= threshold)) + goto exit; + + spinlock_hold_count =0 ; +get_next: +if (!loops--) + goto exit; +if (unlikely (list_empty (&global_queue))) + goto exit; + b = list_entry(global_queue.prev, struct dm_buffer, global_list); + +if (b->accessed) { + b->accessed =0 ; +list_del (&b->global_list); +list_add (&b->global_list, &global_queue); +if (likely (++spinlock_hold_count <16 )) { + goto get_next; + } +spin_unlock (&global_spinlock); + goto reacquire_spinlock; + } + + current_client = b->c; +if (unlikely (current_client != locked_client)) { +if (locked_client) +dm_bufio_unlock (locked_client); + +if (!dm_bufio_trylock (current_client)) { +spin_unlock (&global_spinlock); +dm_bufio_lock (current_client); + locked_client = current_client; + goto reacquire_spinlock; + } + + locked_client = current_client; + } + +spin_unlock (&global_spinlock); + +if (unlikely (!__try_evict_buffer (b,GFP_KERNEL ))) { +spin_lock (&global_spinlock); +list_del (&b->global_list); +list_add (&b->global_list, &global_queue); +spin_unlock (&global_spinlock); + } + + goto reacquire_spinlock; + +exit: +spin_unlock (&global_spinlock); + +if (locked_client) +dm_bufio_unlock (locked_client); + +mutex_unlock (&dm_bufio_clients_lock); +} +static void cleanup_old_buffers (void ) {unsigned long max_age_hz =get_max_age_hz ();
@@ -1829,14 +1919,11 @@ static void cleanup_old_buffers(void)mutex_unlock (&dm_bufio_clients_lock); } -static struct workqueue_struct *dm_bufio_wq; -static struct delayed_work dm_bufio_work; -static void work_fn (struct work_struct *w) {cleanup_old_buffers (); -queue_delayed_work (dm_bufio_wq, &dm_bufio_work, +queue_delayed_work (dm_bufio_wq, &dm_bufio_cleanup_old_work,DM_BUFIO_WORK_TIMER_SECS *HZ ); }
@@ -1878,8 +1965,9 @@ static int __init dm_bufio_init(void)if (!dm_bufio_wq)return -ENOMEM ; -INIT_DELAYED_WORK (&dm_bufio_work, work_fn); -queue_delayed_work (dm_bufio_wq, &dm_bufio_work, +INIT_DELAYED_WORK (&dm_bufio_cleanup_old_work, work_fn); +INIT_WORK (&dm_bufio_replacement_work, do_global_cleanup); +queue_delayed_work (dm_bufio_wq, &dm_bufio_cleanup_old_work,DM_BUFIO_WORK_TIMER_SECS *HZ );return 0 ;
@@ -1892,7 +1980,8 @@ static void __exit dm_bufio_exit(void) {int bug =0 ; -cancel_delayed_work_sync (&dm_bufio_work); +cancel_delayed_work_sync (&dm_bufio_cleanup_old_work); +flush_workqueue (dm_bufio_wq);destroy_workqueue (dm_bufio_wq);if (dm_bufio_client_count) {
Mikulas,
please use list_move instead of list_del/list_add pairs.
Heinz
Mikulas Patocka
maintainerPATCH v2 5/5 dm-bufio: introduce a global cache replacement
Quoted 5 lines · Heinz Mauelshagen
Mikulas,
please use
list_moveinstead of list_del/list_add pairs.Heinz
OK. Here I resend it.
From: Mikulas Patocka <[email protected]>
This patch introduces a global cache replacement (instead of per-client cleanup).
If one bufio client uses the cache heavily and another client is not using it, we want to let the first client use most of the cache. The old algorithm would partition the cache equally betwen the clients and that is inoptimal.
For cache replacement, we use the clock algorithm because it doesn't require taking any lock when the buffer is accessed.
--- drivers/md/dm-bufio.c | 101 ++++++++++++++++++++++++++++++++++++++++++++++---- 1 file changed, 94 insertions(+), 7 deletions(-)
drivers/md/dm-bufio.c +94 -7
@@ -34,6 +34,7 @@#define DM_BUFIO_MEMORY_PERCENT 2#define DM_BUFIO_VMALLOC_PERCENT 25#define DM_BUFIO_WRITEBACK_RATIO 3 +#define DM_BUFIO_LOW_WATERMARK_RATIO 16/* * Check buffer ages in this interval (seconds)
@@ -140,6 +141,7 @@ struct dm_buffer {unsigned char list_mode;/* LIST_* */ blk_status_t read_error;blk_status_t write_error; +unsigned accessed;unsigned hold_count;unsigned long state;unsigned long last_accessed;
@@ -198,6 +200,8 @@ static DEFINE_SPINLOCK(global_spinlock);static LIST_HEAD (global_queue); +static unsigned long global_num =0 ; +/* * Buffers are freed after this timeout */
@@ -227,6 +231,12 @@ static LIST_HEAD(dm_bufio_all_clients);*/ static DEFINE_MUTEX (dm_bufio_clients_lock); + +static struct workqueue_struct *dm_bufio_wq; +static struct delayed_work dm_bufio_cleanup_old_work; +static struct work_struct dm_bufio_replacement_work; + +#ifdef CONFIG_DM_DEBUG_BLOCK_STACK_TRACING static void buffer_record_stack (struct dm_buffer *b) {
@@ -308,10 +318,16 @@ static void adjust_total_allocated(strucif (dm_bufio_current_allocated > dm_bufio_peak_allocated) dm_bufio_peak_allocated = dm_bufio_current_allocated; + b->accessed =1 ; +if (!unlink) {list_add (&b->global_list, &global_queue); + global_num++; +if (dm_bufio_current_allocated > dm_bufio_cache_size) +queue_work (dm_bufio_wq, &dm_bufio_replacement_work); }else {list_del (&b->global_list); + global_num--; }spin_unlock (&global_spinlock);
@@ -497,6 +513,8 @@ static void __relink_lru(struct dm_buffe {struct dm_bufio_client *c = b->c; + b->accessed =1 ; +BUG_ON (!c->n_buffers[b->list_mode]); c->n_buffers[b->list_mode]--;
@@ -1814,6 +1832,76 @@ static void __evict_old_buffers(struct ddm_bufio_unlock (c); } +static void do_global_cleanup (struct work_struct *w) +{ +struct dm_bufio_client *locked_client =NULL ; +struct dm_bufio_client *current_client; +struct dm_buffer *b; +unsigned spinlock_hold_count; +unsigned long threshold = dm_bufio_cache_size - dm_bufio_cache_size /DM_BUFIO_LOW_WATERMARK_RATIO ; +unsigned long loops = global_num *2 ; + +mutex_lock (&dm_bufio_clients_lock); + +reacquire_spinlock: +cond_resched (); + +spin_lock (&global_spinlock); +if (unlikely (dm_bufio_current_allocated <= threshold)) + goto exit; + + spinlock_hold_count =0 ; +get_next: +if (!loops--) + goto exit; +if (unlikely (list_empty (&global_queue))) + goto exit; + b = list_entry(global_queue.prev, struct dm_buffer, global_list); + +if (b->accessed) { + b->accessed =0 ; +list_move (&b->global_list, &global_queue); +if (likely (++spinlock_hold_count <16 )) { + goto get_next; + } +spin_unlock (&global_spinlock); + goto reacquire_spinlock; + } + + current_client = b->c; +if (unlikely (current_client != locked_client)) { +if (locked_client) +dm_bufio_unlock (locked_client); + +if (!dm_bufio_trylock (current_client)) { +spin_unlock (&global_spinlock); +dm_bufio_lock (current_client); + locked_client = current_client; + goto reacquire_spinlock; + } + + locked_client = current_client; + } + +spin_unlock (&global_spinlock); + +if (unlikely (!__try_evict_buffer (b,GFP_KERNEL ))) { +spin_lock (&global_spinlock); +list_move (&b->global_list, &global_queue); +spin_unlock (&global_spinlock); + } + + goto reacquire_spinlock; + +exit: +spin_unlock (&global_spinlock); + +if (locked_client) +dm_bufio_unlock (locked_client); + +mutex_unlock (&dm_bufio_clients_lock); +} +static void cleanup_old_buffers (void ) {unsigned long max_age_hz =get_max_age_hz ();
@@ -1829,14 +1917,11 @@ static void cleanup_old_buffers(void)mutex_unlock (&dm_bufio_clients_lock); } -static struct workqueue_struct *dm_bufio_wq; -static struct delayed_work dm_bufio_work; -static void work_fn (struct work_struct *w) {cleanup_old_buffers (); -queue_delayed_work (dm_bufio_wq, &dm_bufio_work, +queue_delayed_work (dm_bufio_wq, &dm_bufio_cleanup_old_work,DM_BUFIO_WORK_TIMER_SECS *HZ ); }
@@ -1878,8 +1963,9 @@ static int __init dm_bufio_init(void)if (!dm_bufio_wq)return -ENOMEM ; -INIT_DELAYED_WORK (&dm_bufio_work, work_fn); -queue_delayed_work (dm_bufio_wq, &dm_bufio_work, +INIT_DELAYED_WORK (&dm_bufio_cleanup_old_work, work_fn); +INIT_WORK (&dm_bufio_replacement_work, do_global_cleanup); +queue_delayed_work (dm_bufio_wq, &dm_bufio_cleanup_old_work,DM_BUFIO_WORK_TIMER_SECS *HZ );return 0 ;
@@ -1892,7 +1978,8 @@ static void __exit dm_bufio_exit(void) {int bug =0 ; -cancel_delayed_work_sync (&dm_bufio_work); +cancel_delayed_work_sync (&dm_bufio_cleanup_old_work); +flush_workqueue (dm_bufio_wq);destroy_workqueue (dm_bufio_wq);if (dm_bufio_client_count) {
Quoted 24 lines · Mikulas Patocka
Quoted 5 lines · Heinz Mauelshagen
Mikulas,
please use
list_moveinstead of list_del/list_add pairs.Heinz
OK. Here I resend it.
From: Mikulas Patocka <[email protected]>
This patch introduces a global cache replacement (instead of per-client cleanup).
If one bufio client uses the cache heavily and another client is not using it, we want to let the first client use most of the cache. The old algorithm would partition the cache equally betwen the clients and that is inoptimal.
Quoted 4 lines
For cache replacement, we use the clock algorithm because it doesn't require taking any lock when the buffer is accessed.
Signed-off-by Mikulas Patocka <[email protected]>
I'd like to fold in this cleanup if you're OK with it.
Rather use a main control structure for the loop rather than gotos.
You OK with this?
drivers/md/dm-bufio.c +38 -41
@@ -230,7 +230,6 @@ static LIST_HEAD(dm_bufio_all_clients);*/ static DEFINE_MUTEX (dm_bufio_clients_lock); -static struct workqueue_struct *dm_bufio_wq;static struct delayed_work dm_bufio_cleanup_old_work;static struct work_struct dm_bufio_replacement_work;
@@ -1827,62 +1826,60 @@ static void do_global_cleanup(struct work_struct *w)struct dm_bufio_client *current_client;struct dm_buffer *b;unsigned spinlock_hold_count; -unsigned long threshold = dm_bufio_cache_size - dm_bufio_cache_size /DM_BUFIO_LOW_WATERMARK_RATIO ; +unsigned long threshold = dm_bufio_cache_size - + dm_bufio_cache_size /DM_BUFIO_LOW_WATERMARK_RATIO ;unsigned long loops = global_num *2 ;mutex_lock (&dm_bufio_clients_lock); -reacquire_spinlock: -cond_resched (); +while (1 ) { +cond_resched (); -spin_lock (&global_spinlock); -if (unlikely (dm_bufio_current_allocated <= threshold)) - goto exit; +spin_lock (&global_spinlock); +if (unlikely (dm_bufio_current_allocated <= threshold)) +break ; - spinlock_hold_count =0 ; + spinlock_hold_count =0 ; get_next: -if (!loops--) - goto exit; -if (unlikely (list_empty (&global_queue))) - goto exit; - b = list_entry(global_queue.prev, struct dm_buffer, global_list); - -if (b->accessed) { - b->accessed =0 ; -list_move (&b->global_list, &global_queue); -if (likely (++spinlock_hold_count <16 )) { - goto get_next; - } -spin_unlock (&global_spinlock); - goto reacquire_spinlock; - } - - current_client = b->c; -if (unlikely (current_client != locked_client)) { -if (locked_client) -dm_bufio_unlock (locked_client); +if (!loops--) +break ; +if (unlikely (list_empty (&global_queue))) +break ; + b = list_entry(global_queue.prev, struct dm_buffer, global_list); -if (!dm_bufio_trylock (current_client)) { +if (b->accessed) { + b->accessed =0 ; +list_move (&b->global_list, &global_queue); +if (likely (++spinlock_hold_count <16 )) + goto get_next;spin_unlock (&global_spinlock); -dm_bufio_lock (current_client); - locked_client = current_client; - goto reacquire_spinlock; +continue ; } - locked_client = current_client; - } + current_client = b->c; +if (unlikely (current_client != locked_client)) { +if (locked_client) +dm_bufio_unlock (locked_client); -spin_unlock (&global_spinlock); +if (!dm_bufio_trylock (current_client)) { +spin_unlock (&global_spinlock); +dm_bufio_lock (current_client); + locked_client = current_client; +continue ; + } + + locked_client = current_client; + } -if (unlikely (!__try_evict_buffer (b,GFP_KERNEL ))) { -spin_lock (&global_spinlock); -list_move (&b->global_list, &global_queue);spin_unlock (&global_spinlock); - } - goto reacquire_spinlock; +if (unlikely (!__try_evict_buffer (b,GFP_KERNEL ))) { +spin_lock (&global_spinlock); +list_move (&b->global_list, &global_queue); +spin_unlock (&global_spinlock); + } + } -exit:spin_unlock (&global_spinlock);if (locked_client)
Mikulas Patocka
maintainerQuoted 30 lines · Mike Snitzer
Quoted 27 lines · Mikulas Patocka
Quoted 5 lines · Heinz Mauelshagen
Mikulas,
please use
list_moveinstead of list_del/list_add pairs.Heinz
OK. Here I resend it.
From: Mikulas Patocka <[email protected]>
This patch introduces a global cache replacement (instead of per-client cleanup).
If one bufio client uses the cache heavily and another client is not using it, we want to let the first client use most of the cache. The old algorithm would partition the cache equally betwen the clients and that is inoptimal.
For cache replacement, we use the clock algorithm because it doesn't require taking any lock when the buffer is accessed.
Quoted 7 lines
Quoted 1 line
Signed-off-by Mikulas Patocka <[email protected]>I'd like to fold in this cleanup if you're OK with it.
Rather use a main control structure for the loop rather than gotos.
You OK with this?
Yes - you can replace gotos with the loop.
Mikulas
Quoted 114 lines
drivers/md/dm-bufio.c +38
-41@@ -230,7 +230,6 @@ static LIST_HEAD(dm_bufio_all_clients);*/ static DEFINE_MUTEX (dm_bufio_clients_lock); -static struct workqueue_struct *dm_bufio_wq;static struct delayed_work dm_bufio_cleanup_old_work;static struct work_struct dm_bufio_replacement_work;@@ -1827,62 +1826,60 @@ static void do_global_cleanup(struct work_struct *w)struct dm_bufio_client *current_client;struct dm_buffer *b;unsigned spinlock_hold_count; -unsigned long threshold = dm_bufio_cache_size - dm_bufio_cache_size /DM_BUFIO_LOW_WATERMARK_RATIO ; +unsigned long threshold = dm_bufio_cache_size - + dm_bufio_cache_size /DM_BUFIO_LOW_WATERMARK_RATIO ;unsigned long loops = global_num *2 ;mutex_lock (&dm_bufio_clients_lock); -reacquire_spinlock: -cond_resched (); +while (1 ) { +cond_resched (); -spin_lock (&global_spinlock); -if (unlikely (dm_bufio_current_allocated <= threshold)) - goto exit; +spin_lock (&global_spinlock); +if (unlikely (dm_bufio_current_allocated <= threshold)) +break ; - spinlock_hold_count =0 ; + spinlock_hold_count =0 ; get_next: -if (!loops--) - goto exit; -if (unlikely (list_empty (&global_queue))) - goto exit; - b = list_entry(global_queue.prev, struct dm_buffer, global_list); - -if (b->accessed) { - b->accessed =0 ; -list_move (&b->global_list, &global_queue); -if (likely (++spinlock_hold_count <16 )) { - goto get_next; - } -spin_unlock (&global_spinlock); - goto reacquire_spinlock; - } - - current_client = b->c; -if (unlikely (current_client != locked_client)) { -if (locked_client) -dm_bufio_unlock (locked_client); +if (!loops--) +break ; +if (unlikely (list_empty (&global_queue))) +break ; + b = list_entry(global_queue.prev, struct dm_buffer, global_list); -if (!dm_bufio_trylock (current_client)) { +if (b->accessed) { + b->accessed =0 ; +list_move (&b->global_list, &global_queue); +if (likely (++spinlock_hold_count <16 )) + goto get_next;spin_unlock (&global_spinlock); -dm_bufio_lock (current_client); - locked_client = current_client; - goto reacquire_spinlock; +continue ; } - locked_client = current_client; - } + current_client = b->c; +if (unlikely (current_client != locked_client)) { +if (locked_client) +dm_bufio_unlock (locked_client); -spin_unlock (&global_spinlock); +if (!dm_bufio_trylock (current_client)) { +spin_unlock (&global_spinlock); +dm_bufio_lock (current_client); + locked_client = current_client; +continue ; + } + + locked_client = current_client; + } -if (unlikely (!__try_evict_buffer (b,GFP_KERNEL ))) { -spin_lock (&global_spinlock); -list_move (&b->global_list, &global_queue);spin_unlock (&global_spinlock); - } - goto reacquire_spinlock; +if (unlikely (!__try_evict_buffer (b,GFP_KERNEL ))) { +spin_lock (&global_spinlock); +list_move (&b->global_list, &global_queue); +spin_unlock (&global_spinlock); + } + } -exit:spin_unlock (&global_spinlock);if (locked_client)