From 35a82d1a53e1a9ad54efafcc940f9335beaed5c3 Mon Sep 17 00:00:00 2001 From: Nick Piggin Date: Thu, 23 Jun 2005 00:09:06 -0700 Subject: [PATCH] optimise loop driver a bit Looks like locking can be optimised quite a lot. Increase lock widths slightly so lo_lock is taken fewer times per request. Also it was quite trivial to cover lo_pending with that lock, and remove the atomic requirement. This also makes memory ordering explicitly correct, which is nice (not that I particularly saw any mem ordering bugs). Test was reading 4 250MB files in parallel on ext2-on-tmpfs filesystem (1K block size, 4K page size). System is 2 socket Xeon with HT (4 thread). intel:/home/npiggin# umount /dev/loop0 ; mount /dev/loop0 /mnt/loop ; /usr/bin/time ./mtloop.sh Before: 0.24user 5.51system 0:02.84elapsed 202%CPU (0avgtext+0avgdata 0maxresident)k 0.19user 5.52system 0:02.88elapsed 198%CPU (0avgtext+0avgdata 0maxresident)k 0.19user 5.57system 0:02.89elapsed 198%CPU (0avgtext+0avgdata 0maxresident)k 0.22user 5.51system 0:02.90elapsed 197%CPU (0avgtext+0avgdata 0maxresident)k 0.19user 5.44system 0:02.91elapsed 193%CPU (0avgtext+0avgdata 0maxresident)k After: 0.07user 2.34system 0:01.68elapsed 143%CPU (0avgtext+0avgdata 0maxresident)k 0.06user 2.37system 0:01.68elapsed 144%CPU (0avgtext+0avgdata 0maxresident)k 0.06user 2.39system 0:01.68elapsed 145%CPU (0avgtext+0avgdata 0maxresident)k 0.06user 2.36system 0:01.68elapsed 144%CPU (0avgtext+0avgdata 0maxresident)k 0.06user 2.42system 0:01.68elapsed 147%CPU (0avgtext+0avgdata 0maxresident)k Signed-off-by: Nick Piggin Signed-off-by: Andrew Morton Signed-off-by: Linus Torvalds --- drivers/block/loop.c | 81 ++++++++++++++++++++++++---------------------------- 1 file changed, 38 insertions(+), 43 deletions(-) (limited to 'drivers') diff --git a/drivers/block/loop.c b/drivers/block/loop.c index 6f011d0d8e9..b35e08876dd 100644 --- a/drivers/block/loop.c +++ b/drivers/block/loop.c @@ -472,17 +472,11 @@ static int do_bio_filebacked(struct loop_device *lo, struct bio *bio) */ static void loop_add_bio(struct loop_device *lo, struct bio *bio) { - unsigned long flags; - - spin_lock_irqsave(&lo->lo_lock, flags); if (lo->lo_biotail) { lo->lo_biotail->bi_next = bio; lo->lo_biotail = bio; } else lo->lo_bio = lo->lo_biotail = bio; - spin_unlock_irqrestore(&lo->lo_lock, flags); - - up(&lo->lo_bh_mutex); } /* @@ -492,14 +486,12 @@ static struct bio *loop_get_bio(struct loop_device *lo) { struct bio *bio; - spin_lock_irq(&lo->lo_lock); if ((bio = lo->lo_bio)) { if (bio == lo->lo_biotail) lo->lo_biotail = NULL; lo->lo_bio = bio->bi_next; bio->bi_next = NULL; } - spin_unlock_irq(&lo->lo_lock); return bio; } @@ -509,35 +501,28 @@ static int loop_make_request(request_queue_t *q, struct bio *old_bio) struct loop_device *lo = q->queuedata; int rw = bio_rw(old_bio); - if (!lo) - goto out; + if (rw == READA) + rw = READ; + + BUG_ON(!lo || (rw != READ && rw != WRITE)); spin_lock_irq(&lo->lo_lock); if (lo->lo_state != Lo_bound) - goto inactive; - atomic_inc(&lo->lo_pending); - spin_unlock_irq(&lo->lo_lock); - - if (rw == WRITE) { - if (lo->lo_flags & LO_FLAGS_READ_ONLY) - goto err; - } else if (rw == READA) { - rw = READ; - } else if (rw != READ) { - printk(KERN_ERR "loop: unknown command (%x)\n", rw); - goto err; - } + goto out; + if (unlikely(rw == WRITE && (lo->lo_flags & LO_FLAGS_READ_ONLY))) + goto out; + lo->lo_pending++; loop_add_bio(lo, old_bio); + spin_unlock_irq(&lo->lo_lock); + up(&lo->lo_bh_mutex); return 0; -err: - if (atomic_dec_and_test(&lo->lo_pending)) - up(&lo->lo_bh_mutex); + out: + if (lo->lo_pending == 0) + up(&lo->lo_bh_mutex); + spin_unlock_irq(&lo->lo_lock); bio_io_error(old_bio, old_bio->bi_size); return 0; -inactive: - spin_unlock_irq(&lo->lo_lock); - goto out; } /* @@ -560,13 +545,11 @@ static void do_loop_switch(struct loop_device *, struct switch_request *); static inline void loop_handle_bio(struct loop_device *lo, struct bio *bio) { - int ret; - if (unlikely(!bio->bi_bdev)) { do_loop_switch(lo, bio->bi_private); bio_put(bio); } else { - ret = do_bio_filebacked(lo, bio); + int ret = do_bio_filebacked(lo, bio); bio_endio(bio, bio->bi_size, ret); } } @@ -594,7 +577,7 @@ static int loop_thread(void *data) set_user_nice(current, -20); lo->lo_state = Lo_bound; - atomic_inc(&lo->lo_pending); + lo->lo_pending = 1; /* * up sem, we are running @@ -602,26 +585,37 @@ static int loop_thread(void *data) up(&lo->lo_sem); for (;;) { - down_interruptible(&lo->lo_bh_mutex); + int pending; + /* - * could be upped because of tear-down, not because of - * pending work + * interruptible just to not contribute to load avg */ - if (!atomic_read(&lo->lo_pending)) + if (down_interruptible(&lo->lo_bh_mutex)) + continue; + + spin_lock_irq(&lo->lo_lock); + + /* + * could be upped because of tear-down, not pending work + */ + if (unlikely(!lo->lo_pending)) { + spin_unlock_irq(&lo->lo_lock); break; + } bio = loop_get_bio(lo); - if (!bio) { - printk("loop: missing bio\n"); - continue; - } + lo->lo_pending--; + pending = lo->lo_pending; + spin_unlock_irq(&lo->lo_lock); + + BUG_ON(!bio); loop_handle_bio(lo, bio); /* * upped both for pending work and tear-down, lo_pending * will hit zero then */ - if (atomic_dec_and_test(&lo->lo_pending)) + if (unlikely(!pending)) break; } @@ -900,7 +894,8 @@ static int loop_clr_fd(struct loop_device *lo, struct block_device *bdev) spin_lock_irq(&lo->lo_lock); lo->lo_state = Lo_rundown; - if (atomic_dec_and_test(&lo->lo_pending)) + lo->lo_pending--; + if (!lo->lo_pending) up(&lo->lo_bh_mutex); spin_unlock_irq(&lo->lo_lock); -- cgit v1.2.3