Spinlocks don't have a buslocked unlock and are faster. On a P4, time to write a 4M file with 4M one-byte-write()s: Before: 0.72s user 5.47s system 99% cpu 6.227 total 0.76s user 5.40s system 100% cpu 6.154 total 0.77s user 5.38s system 100% cpu 6.146 total After: 1.09s user 4.92s system 99% cpu 6.014 total 0.74s user 5.28s system 99% cpu 6.023 total 1.03s user 4.97s system 100% cpu 5.991 total fs/fs-writeback.c | 4 ++-- fs/inode.c | 2 +- fs/mpage.c | 8 ++++---- include/linux/fs.h | 2 +- mm/filemap.c | 42 +++++++++++++++++++++--------------------- mm/page-writeback.c | 14 +++++++------- mm/readahead.c | 8 ++++---- mm/swap_state.c | 22 +++++++++++----------- mm/swapfile.c | 8 ++++---- mm/truncate.c | 6 +++--- mm/vmscan.c | 14 +++++++------- 11 files changed, 65 insertions(+), 65 deletions(-) diff -puN fs/fs-writeback.c~page-lock-is-spin_lock fs/fs-writeback.c --- 25/fs/fs-writeback.c~page-lock-is-spin_lock 2003-04-08 03:16:30.000000000 -0700 +++ 25-akpm/fs/fs-writeback.c 2003-04-08 03:16:30.000000000 -0700 @@ -149,10 +149,10 @@ __sync_single_inode(struct inode *inode, * read speculatively by this cpu before &= ~I_DIRTY -- mikulas */ - write_lock(&mapping->page_lock); + spin_lock(&mapping->page_lock); if (wait || !wbc->for_kupdate || list_empty(&mapping->io_pages)) list_splice_init(&mapping->dirty_pages, &mapping->io_pages); - write_unlock(&mapping->page_lock); + spin_unlock(&mapping->page_lock); spin_unlock(&inode_lock); do_writepages(mapping, wbc); diff -puN fs/inode.c~page-lock-is-spin_lock fs/inode.c --- 25/fs/inode.c~page-lock-is-spin_lock 2003-04-08 03:16:30.000000000 -0700 +++ 25-akpm/fs/inode.c 2003-04-08 03:16:30.000000000 -0700 @@ -181,7 +181,7 @@ void inode_init_once(struct inode *inode INIT_LIST_HEAD(&inode->i_devices); sema_init(&inode->i_sem, 1); INIT_RADIX_TREE(&inode->i_data.page_tree, GFP_ATOMIC); - rwlock_init(&inode->i_data.page_lock); + spin_lock_init(&inode->i_data.page_lock); init_MUTEX(&inode->i_data.i_shared_sem); INIT_LIST_HEAD(&inode->i_data.private_list); spin_lock_init(&inode->i_data.private_lock); diff -puN fs/mpage.c~page-lock-is-spin_lock fs/mpage.c --- 25/fs/mpage.c~page-lock-is-spin_lock 2003-04-08 03:16:30.000000000 -0700 +++ 25-akpm/fs/mpage.c 2003-04-08 03:16:30.000000000 -0700 @@ -627,7 +627,7 @@ mpage_writepages(struct address_space *m writepage = mapping->a_ops->writepage; pagevec_init(&pvec, 0); - write_lock(&mapping->page_lock); + spin_lock(&mapping->page_lock); while (!list_empty(&mapping->io_pages) && !done) { struct page *page = list_entry(mapping->io_pages.prev, struct page, list); @@ -647,7 +647,7 @@ mpage_writepages(struct address_space *m list_add(&page->list, &mapping->locked_pages); page_cache_get(page); - write_unlock(&mapping->page_lock); + spin_unlock(&mapping->page_lock); /* * At this point we hold neither mapping->page_lock nor @@ -679,12 +679,12 @@ mpage_writepages(struct address_space *m unlock_page(page); } page_cache_release(page); - write_lock(&mapping->page_lock); + spin_lock(&mapping->page_lock); } /* * Leave any remaining dirty pages on ->io_pages */ - write_unlock(&mapping->page_lock); + spin_unlock(&mapping->page_lock); if (bio) mpage_bio_submit(WRITE, bio); return ret; diff -puN include/linux/fs.h~page-lock-is-spin_lock include/linux/fs.h --- 25/include/linux/fs.h~page-lock-is-spin_lock 2003-04-08 03:16:30.000000000 -0700 +++ 25-akpm/include/linux/fs.h 2003-04-08 03:16:30.000000000 -0700 @@ -313,7 +313,7 @@ struct backing_dev_info; struct address_space { struct inode *host; /* owner: inode, block_device */ struct radix_tree_root page_tree; /* radix tree of all pages */ - rwlock_t page_lock; /* and rwlock protecting it */ + spinlock_t page_lock; /* and rwlock protecting it */ struct list_head clean_pages; /* list of clean pages */ struct list_head dirty_pages; /* list of dirty pages */ struct list_head locked_pages; /* list of locked pages */ diff -puN mm/filemap.c~page-lock-is-spin_lock mm/filemap.c --- 25/mm/filemap.c~page-lock-is-spin_lock 2003-04-08 03:16:30.000000000 -0700 +++ 25-akpm/mm/filemap.c 2003-04-08 03:16:30.000000000 -0700 @@ -99,9 +99,9 @@ void remove_from_page_cache(struct page if (unlikely(!PageLocked(page))) PAGE_BUG(page); - write_lock(&mapping->page_lock); + spin_lock(&mapping->page_lock); __remove_from_page_cache(page); - write_unlock(&mapping->page_lock); + spin_unlock(&mapping->page_lock); } static inline int sync_page(struct page *page) @@ -133,9 +133,9 @@ static int __filemap_fdatawrite(struct a if (mapping->backing_dev_info->memory_backed) return 0; - write_lock(&mapping->page_lock); + spin_lock(&mapping->page_lock); list_splice_init(&mapping->dirty_pages, &mapping->io_pages); - write_unlock(&mapping->page_lock); + spin_unlock(&mapping->page_lock); ret = do_writepages(mapping, &wbc); return ret; } @@ -166,7 +166,7 @@ int filemap_fdatawait(struct address_spa restart: progress = 0; - write_lock(&mapping->page_lock); + spin_lock(&mapping->page_lock); while (!list_empty(&mapping->locked_pages)) { struct page *page; @@ -180,7 +180,7 @@ restart: if (!PageWriteback(page)) { if (++progress > 32) { if (need_resched()) { - write_unlock(&mapping->page_lock); + spin_unlock(&mapping->page_lock); __cond_resched(); goto restart; } @@ -190,16 +190,16 @@ restart: progress = 0; page_cache_get(page); - write_unlock(&mapping->page_lock); + spin_unlock(&mapping->page_lock); wait_on_page_writeback(page); if (PageError(page)) ret = -EIO; page_cache_release(page); - write_lock(&mapping->page_lock); + spin_lock(&mapping->page_lock); } - write_unlock(&mapping->page_lock); + spin_unlock(&mapping->page_lock); return ret; } @@ -227,7 +227,7 @@ int add_to_page_cache(struct page *page, if (error == 0) { page_cache_get(page); - write_lock(&mapping->page_lock); + spin_lock(&mapping->page_lock); error = radix_tree_insert(&mapping->page_tree, offset, page); if (!error) { SetPageLocked(page); @@ -235,7 +235,7 @@ int add_to_page_cache(struct page *page, } else { page_cache_release(page); } - write_unlock(&mapping->page_lock); + spin_unlock(&mapping->page_lock); radix_tree_preload_end(); } return error; @@ -364,11 +364,11 @@ struct page * find_get_page(struct addre * We scan the hash list read-only. Addition to and removal from * the hash-list needs a held write-lock. */ - read_lock(&mapping->page_lock); + spin_lock(&mapping->page_lock); page = radix_tree_lookup(&mapping->page_tree, offset); if (page) page_cache_get(page); - read_unlock(&mapping->page_lock); + spin_unlock(&mapping->page_lock); return page; } @@ -379,11 +379,11 @@ struct page *find_trylock_page(struct ad { struct page *page; - read_lock(&mapping->page_lock); + spin_lock(&mapping->page_lock); page = radix_tree_lookup(&mapping->page_tree, offset); if (page && TestSetPageLocked(page)) page = NULL; - read_unlock(&mapping->page_lock); + spin_unlock(&mapping->page_lock); return page; } @@ -403,15 +403,15 @@ struct page *find_lock_page(struct addre { struct page *page; - read_lock(&mapping->page_lock); + spin_lock(&mapping->page_lock); repeat: page = radix_tree_lookup(&mapping->page_tree, offset); if (page) { page_cache_get(page); if (TestSetPageLocked(page)) { - read_unlock(&mapping->page_lock); + spin_unlock(&mapping->page_lock); lock_page(page); - read_lock(&mapping->page_lock); + spin_lock(&mapping->page_lock); /* Has the page been truncated while we slept? */ if (page->mapping != mapping || page->index != offset) { @@ -421,7 +421,7 @@ repeat: } } } - read_unlock(&mapping->page_lock); + spin_unlock(&mapping->page_lock); return page; } @@ -491,12 +491,12 @@ unsigned int find_get_pages(struct addre unsigned int i; unsigned int ret; - read_lock(&mapping->page_lock); + spin_lock(&mapping->page_lock); ret = radix_tree_gang_lookup(&mapping->page_tree, (void **)pages, start, nr_pages); for (i = 0; i < ret; i++) page_cache_get(pages[i]); - read_unlock(&mapping->page_lock); + spin_unlock(&mapping->page_lock); return ret; } diff -puN mm/page-writeback.c~page-lock-is-spin_lock mm/page-writeback.c --- 25/mm/page-writeback.c~page-lock-is-spin_lock 2003-04-08 03:16:30.000000000 -0700 +++ 25-akpm/mm/page-writeback.c 2003-04-08 03:16:30.000000000 -0700 @@ -425,12 +425,12 @@ int write_one_page(struct page *page, in if (wait && PageWriteback(page)) wait_on_page_writeback(page); - write_lock(&mapping->page_lock); + spin_lock(&mapping->page_lock); list_del(&page->list); if (test_clear_page_dirty(page)) { list_add(&page->list, &mapping->locked_pages); page_cache_get(page); - write_unlock(&mapping->page_lock); + spin_unlock(&mapping->page_lock); ret = mapping->a_ops->writepage(page, &wbc); if (ret == 0 && wait) { wait_on_page_writeback(page); @@ -440,7 +440,7 @@ int write_one_page(struct page *page, in page_cache_release(page); } else { list_add(&page->list, &mapping->clean_pages); - write_unlock(&mapping->page_lock); + spin_unlock(&mapping->page_lock); unlock_page(page); } return ret; @@ -513,14 +513,14 @@ int __set_page_dirty_buffers(struct page spin_unlock(&mapping->private_lock); if (!TestSetPageDirty(page)) { - write_lock(&mapping->page_lock); + spin_lock(&mapping->page_lock); if (page->mapping) { /* Race with truncate? */ if (!mapping->backing_dev_info->memory_backed) inc_page_state(nr_dirty); list_del(&page->list); list_add(&page->list, &mapping->dirty_pages); } - write_unlock(&mapping->page_lock); + spin_unlock(&mapping->page_lock); __mark_inode_dirty(mapping->host, I_DIRTY_PAGES); } @@ -550,7 +550,7 @@ int __set_page_dirty_nobuffers(struct pa struct address_space *mapping = page->mapping; if (mapping) { - write_lock(&mapping->page_lock); + spin_lock(&mapping->page_lock); if (page->mapping) { /* Race with truncate? */ BUG_ON(page->mapping != mapping); if (!mapping->backing_dev_info->memory_backed) @@ -558,7 +558,7 @@ int __set_page_dirty_nobuffers(struct pa list_del(&page->list); list_add(&page->list, &mapping->dirty_pages); } - write_unlock(&mapping->page_lock); + spin_unlock(&mapping->page_lock); __mark_inode_dirty(mapping->host, I_DIRTY_PAGES); } } diff -puN mm/readahead.c~page-lock-is-spin_lock mm/readahead.c --- 25/mm/readahead.c~page-lock-is-spin_lock 2003-04-08 03:16:30.000000000 -0700 +++ 25-akpm/mm/readahead.c 2003-04-08 03:16:30.000000000 -0700 @@ -217,7 +217,7 @@ __do_page_cache_readahead(struct address /* * Preallocate as many pages as we will need. */ - read_lock(&mapping->page_lock); + spin_lock(&mapping->page_lock); for (page_idx = 0; page_idx < nr_to_read; page_idx++) { unsigned long page_offset = offset + page_idx; @@ -228,16 +228,16 @@ __do_page_cache_readahead(struct address if (page) continue; - read_unlock(&mapping->page_lock); + spin_unlock(&mapping->page_lock); page = page_cache_alloc_cold(mapping); - read_lock(&mapping->page_lock); + spin_lock(&mapping->page_lock); if (!page) break; page->index = page_offset; list_add(&page->list, &page_pool); ret++; } - read_unlock(&mapping->page_lock); + spin_unlock(&mapping->page_lock); /* * Now start the IO. We ignore I/O errors - if the page is not diff -puN mm/swapfile.c~page-lock-is-spin_lock mm/swapfile.c --- 25/mm/swapfile.c~page-lock-is-spin_lock 2003-04-08 03:16:30.000000000 -0700 +++ 25-akpm/mm/swapfile.c 2003-04-08 03:16:30.000000000 -0700 @@ -248,10 +248,10 @@ static int exclusive_swap_page(struct pa /* Is the only swap cache user the cache itself? */ if (p->swap_map[swp_offset(entry)] == 1) { /* Recheck the page count with the pagecache lock held.. */ - read_lock(&swapper_space.page_lock); + spin_lock(&swapper_space.page_lock); if (page_count(page) - !!PagePrivate(page) == 2) retval = 1; - read_unlock(&swapper_space.page_lock); + spin_unlock(&swapper_space.page_lock); } swap_info_put(p); } @@ -319,13 +319,13 @@ int remove_exclusive_swap_page(struct pa retval = 0; if (p->swap_map[swp_offset(entry)] == 1) { /* Recheck the page count with the pagecache lock held.. */ - write_lock(&swapper_space.page_lock); + spin_lock(&swapper_space.page_lock); if ((page_count(page) == 2) && !PageWriteback(page)) { __delete_from_swap_cache(page); SetPageDirty(page); retval = 1; } - write_unlock(&swapper_space.page_lock); + spin_unlock(&swapper_space.page_lock); } swap_info_put(p); diff -puN mm/swap_state.c~page-lock-is-spin_lock mm/swap_state.c --- 25/mm/swap_state.c~page-lock-is-spin_lock 2003-04-08 03:16:30.000000000 -0700 +++ 25-akpm/mm/swap_state.c 2003-04-08 03:16:30.000000000 -0700 @@ -34,7 +34,7 @@ extern struct address_space_operations s struct address_space swapper_space = { .page_tree = RADIX_TREE_INIT(GFP_ATOMIC), - .page_lock = RW_LOCK_UNLOCKED, + .page_lock = SPIN_LOCK_UNLOCKED, .clean_pages = LIST_HEAD_INIT(swapper_space.clean_pages), .dirty_pages = LIST_HEAD_INIT(swapper_space.dirty_pages), .io_pages = LIST_HEAD_INIT(swapper_space.io_pages), @@ -191,9 +191,9 @@ void delete_from_swap_cache(struct page entry.val = page->index; - write_lock(&swapper_space.page_lock); + spin_lock(&swapper_space.page_lock); __delete_from_swap_cache(page); - write_unlock(&swapper_space.page_lock); + spin_unlock(&swapper_space.page_lock); swap_free(entry); page_cache_release(page); @@ -204,8 +204,8 @@ int move_to_swap_cache(struct page *page struct address_space *mapping = page->mapping; int err; - write_lock(&swapper_space.page_lock); - write_lock(&mapping->page_lock); + spin_lock(&swapper_space.page_lock); + spin_lock(&mapping->page_lock); err = radix_tree_insert(&swapper_space.page_tree, entry.val, page); if (!err) { @@ -213,8 +213,8 @@ int move_to_swap_cache(struct page *page ___add_to_page_cache(page, &swapper_space, entry.val); } - write_unlock(&mapping->page_lock); - write_unlock(&swapper_space.page_lock); + spin_unlock(&mapping->page_lock); + spin_unlock(&swapper_space.page_lock); if (!err) { if (!swap_duplicate(entry)) @@ -240,8 +240,8 @@ int move_from_swap_cache(struct page *pa entry.val = page->index; - write_lock(&swapper_space.page_lock); - write_lock(&mapping->page_lock); + spin_lock(&swapper_space.page_lock); + spin_lock(&mapping->page_lock); err = radix_tree_insert(&mapping->page_tree, index, page); if (!err) { @@ -249,8 +249,8 @@ int move_from_swap_cache(struct page *pa ___add_to_page_cache(page, mapping, index); } - write_unlock(&mapping->page_lock); - write_unlock(&swapper_space.page_lock); + spin_unlock(&mapping->page_lock); + spin_unlock(&swapper_space.page_lock); if (!err) { swap_free(entry); diff -puN mm/truncate.c~page-lock-is-spin_lock mm/truncate.c --- 25/mm/truncate.c~page-lock-is-spin_lock 2003-04-08 03:16:30.000000000 -0700 +++ 25-akpm/mm/truncate.c 2003-04-08 03:16:30.000000000 -0700 @@ -73,13 +73,13 @@ invalidate_complete_page(struct address_ if (PagePrivate(page) && !try_to_release_page(page, 0)) return 0; - write_lock(&mapping->page_lock); + spin_lock(&mapping->page_lock); if (PageDirty(page)) { - write_unlock(&mapping->page_lock); + spin_unlock(&mapping->page_lock); return 0; } __remove_from_page_cache(page); - write_unlock(&mapping->page_lock); + spin_unlock(&mapping->page_lock); ClearPageUptodate(page); page_cache_release(page); /* pagecache ref */ return 1; diff -puN mm/vmscan.c~page-lock-is-spin_lock mm/vmscan.c --- 25/mm/vmscan.c~page-lock-is-spin_lock 2003-04-08 03:16:30.000000000 -0700 +++ 25-akpm/mm/vmscan.c 2003-04-08 03:16:30.000000000 -0700 @@ -325,7 +325,7 @@ shrink_list(struct list_head *page_list, goto keep_locked; if (!may_write_to_queue(mapping->backing_dev_info)) goto keep_locked; - write_lock(&mapping->page_lock); + spin_lock(&mapping->page_lock); if (test_clear_page_dirty(page)) { int res; struct writeback_control wbc = { @@ -336,7 +336,7 @@ shrink_list(struct list_head *page_list, }; list_move(&page->list, &mapping->locked_pages); - write_unlock(&mapping->page_lock); + spin_unlock(&mapping->page_lock); SetPageReclaim(page); res = mapping->a_ops->writepage(page, &wbc); @@ -351,7 +351,7 @@ shrink_list(struct list_head *page_list, } goto keep; } - write_unlock(&mapping->page_lock); + spin_unlock(&mapping->page_lock); } /* @@ -385,7 +385,7 @@ shrink_list(struct list_head *page_list, if (!mapping) goto keep_locked; /* truncate got there first */ - write_lock(&mapping->page_lock); + spin_lock(&mapping->page_lock); /* * The non-racy check for busy page. It is critical to check @@ -393,7 +393,7 @@ shrink_list(struct list_head *page_list, * not in use by anybody. (pagecache + us == 2) */ if (page_count(page) != 2 || PageDirty(page)) { - write_unlock(&mapping->page_lock); + spin_unlock(&mapping->page_lock); goto keep_locked; } @@ -401,7 +401,7 @@ shrink_list(struct list_head *page_list, if (PageSwapCache(page)) { swp_entry_t swap = { .val = page->index }; __delete_from_swap_cache(page); - write_unlock(&mapping->page_lock); + spin_unlock(&mapping->page_lock); swap_free(swap); __put_page(page); /* The pagecache ref */ goto free_it; @@ -409,7 +409,7 @@ shrink_list(struct list_head *page_list, #endif /* CONFIG_SWAP */ __remove_from_page_cache(page); - write_unlock(&mapping->page_lock); + spin_unlock(&mapping->page_lock); __put_page(page); free_it: _