Commit 27f3c465 authored by Andrew Morton's avatar Andrew Morton Committed by Linus Torvalds

[PATCH] AS: no trinary states

From: Nick Piggin <piggin@cyberone.com.au>

Gets rid of the silly (and undocumented) trinary stateness of
ad->changed_batch.

Kernel style flags field could now save 20 bytes per asd (request queue) and
4 per arq.  But I figure that change can wait until I do autotuning...
parent 4f0c0d09
...@@ -101,7 +101,8 @@ struct as_data { ...@@ -101,7 +101,8 @@ struct as_data {
struct list_head *hash; /* request hash */ struct list_head *hash; /* request hash */
unsigned long current_batch_expires; unsigned long current_batch_expires;
unsigned long last_check_fifo[2]; unsigned long last_check_fifo[2];
int changed_batch; int changed_batch; /* 1: waiting for old batch to end */
int new_batch; /* 1: waiting on first read complete */
int batch_data_dir; /* current batch REQ_SYNC / REQ_ASYNC */ int batch_data_dir; /* current batch REQ_SYNC / REQ_ASYNC */
int write_batch_count; /* max # of reqs in a write batch */ int write_batch_count; /* max # of reqs in a write batch */
int current_write_count; /* how many requests left this batch */ int current_write_count; /* how many requests left this batch */
...@@ -910,8 +911,13 @@ static void as_completed_request(request_queue_t *q, struct request *rq) ...@@ -910,8 +911,13 @@ static void as_completed_request(request_queue_t *q, struct request *rq)
return; return;
if (ad->changed_batch && ad->nr_dispatched == 1) { if (ad->changed_batch && ad->nr_dispatched == 1) {
WARN_ON(ad->batch_data_dir == arq->is_sync);
kblockd_schedule_work(&ad->antic_work); kblockd_schedule_work(&ad->antic_work);
ad->changed_batch = 2; ad->changed_batch = 0;
if (ad->batch_data_dir == REQ_SYNC)
ad->new_batch = 1;
} }
ad->nr_dispatched--; ad->nr_dispatched--;
...@@ -920,12 +926,12 @@ static void as_completed_request(request_queue_t *q, struct request *rq) ...@@ -920,12 +926,12 @@ static void as_completed_request(request_queue_t *q, struct request *rq)
* actually serviced. This should help devices with big TCQ windows * actually serviced. This should help devices with big TCQ windows
* and writeback caches * and writeback caches
*/ */
if (ad->batch_data_dir == REQ_SYNC && ad->changed_batch if (ad->new_batch && ad->batch_data_dir == arq->is_sync) {
&& ad->batch_data_dir == arq->is_sync) { WARN_ON(ad->batch_data_dir != REQ_SYNC);
update_write_batch(ad); update_write_batch(ad);
ad->current_batch_expires = jiffies + ad->current_batch_expires = jiffies +
ad->batch_expire[REQ_SYNC]; ad->batch_expire[REQ_SYNC];
ad->changed_batch = 0; ad->new_batch = 0;
} }
if (!arq->io_context) if (!arq->io_context)
...@@ -1070,7 +1076,7 @@ static int as_fifo_expired(struct as_data *ad, int adir) ...@@ -1070,7 +1076,7 @@ static int as_fifo_expired(struct as_data *ad, int adir)
*/ */
static inline int as_batch_expired(struct as_data *ad) static inline int as_batch_expired(struct as_data *ad)
{ {
if (ad->changed_batch) if (ad->changed_batch || ad->new_batch)
return 0; return 0;
if (ad->batch_data_dir == REQ_SYNC) if (ad->batch_data_dir == REQ_SYNC)
...@@ -1150,7 +1156,7 @@ static int as_dispatch_request(struct as_data *ad) ...@@ -1150,7 +1156,7 @@ static int as_dispatch_request(struct as_data *ad)
if (!(reads || writes) if (!(reads || writes)
|| ad->antic_status == ANTIC_WAIT_REQ || ad->antic_status == ANTIC_WAIT_REQ
|| ad->antic_status == ANTIC_WAIT_NEXT || ad->antic_status == ANTIC_WAIT_NEXT
|| ad->changed_batch == 1) || ad->changed_batch)
return 0; return 0;
if (!(reads && writes && as_batch_expired(ad)) ) { if (!(reads && writes && as_batch_expired(ad)) ) {
...@@ -1192,8 +1198,10 @@ static int as_dispatch_request(struct as_data *ad) ...@@ -1192,8 +1198,10 @@ static int as_dispatch_request(struct as_data *ad)
*/ */
goto dispatch_writes; goto dispatch_writes;
if (ad->batch_data_dir == REQ_ASYNC) if (ad->batch_data_dir == REQ_ASYNC) {
WARN_ON(ad->new_batch);
ad->changed_batch = 1; ad->changed_batch = 1;
}
ad->batch_data_dir = REQ_SYNC; ad->batch_data_dir = REQ_SYNC;
arq = list_entry_fifo(ad->fifo_list[ad->batch_data_dir].next); arq = list_entry_fifo(ad->fifo_list[ad->batch_data_dir].next);
ad->last_check_fifo[ad->batch_data_dir] = jiffies; ad->last_check_fifo[ad->batch_data_dir] = jiffies;
...@@ -1208,8 +1216,16 @@ static int as_dispatch_request(struct as_data *ad) ...@@ -1208,8 +1216,16 @@ static int as_dispatch_request(struct as_data *ad)
dispatch_writes: dispatch_writes:
BUG_ON(RB_EMPTY(&ad->sort_list[REQ_ASYNC])); BUG_ON(RB_EMPTY(&ad->sort_list[REQ_ASYNC]));
if (ad->batch_data_dir == REQ_SYNC) if (ad->batch_data_dir == REQ_SYNC) {
ad->changed_batch = 1; ad->changed_batch = 1;
/*
* new_batch might be 1 when the queue runs out of
* reads. A subsequent submission of a write might
* cause a change of batch before the read is finished.
*/
ad->new_batch = 0;
}
ad->batch_data_dir = REQ_ASYNC; ad->batch_data_dir = REQ_ASYNC;
ad->current_write_count = ad->write_batch_count; ad->current_write_count = ad->write_batch_count;
ad->write_batch_idled = 0; ad->write_batch_idled = 0;
...@@ -1232,14 +1248,19 @@ static int as_dispatch_request(struct as_data *ad) ...@@ -1232,14 +1248,19 @@ static int as_dispatch_request(struct as_data *ad)
} }
if (ad->changed_batch) { if (ad->changed_batch) {
if (ad->changed_batch == 1 && ad->nr_dispatched) WARN_ON(ad->new_batch);
if (ad->nr_dispatched)
return 0; return 0;
if (ad->batch_data_dir == REQ_ASYNC) {
if (ad->batch_data_dir == REQ_ASYNC)
ad->current_batch_expires = jiffies + ad->current_batch_expires = jiffies +
ad->batch_expire[REQ_ASYNC]; ad->batch_expire[REQ_ASYNC];
ad->changed_batch = 0; else
} else ad->new_batch = 1;
ad->changed_batch = 2;
ad->changed_batch = 0;
arq->request->flags |= REQ_SOFTBARRIER; arq->request->flags |= REQ_SOFTBARRIER;
} }
......
Markdown is supported
0%
or
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment