From patchwork Thu Jun 29 13:54:20 2017 Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: 7bit X-Patchwork-Submitter: Christoph Hellwig X-Patchwork-Id: 782296 Return-Path: X-Original-To: patchwork-incoming@ozlabs.org Delivered-To: patchwork-incoming@ozlabs.org Received: from vger.kernel.org (vger.kernel.org [209.132.180.67]) by ozlabs.org (Postfix) with ESMTP id 3wz1PG0VB5z9s3s for ; Thu, 29 Jun 2017 23:54:42 +1000 (AEST) Authentication-Results: ozlabs.org; dkim=fail reason="signature verification failed" (2048-bit key; unprotected) header.d=infradead.org header.i=@infradead.org header.b="QzwrNNUF"; dkim-atps=neutral Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1753229AbdF2Nyk (ORCPT ); Thu, 29 Jun 2017 09:54:40 -0400 Received: from bombadil.infradead.org ([65.50.211.133]:42743 "EHLO bombadil.infradead.org" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1752956AbdF2Ny0 (ORCPT ); Thu, 29 Jun 2017 09:54:26 -0400 DKIM-Signature: v=1; a=rsa-sha256; q=dns/txt; c=relaxed/relaxed; d=infradead.org; s=bombadil.20170209; h=References:In-Reply-To:Message-Id: Date:Subject:Cc:To:From:Sender:Reply-To:MIME-Version:Content-Type: Content-Transfer-Encoding:Content-ID:Content-Description:Resent-Date: Resent-From:Resent-Sender:Resent-To:Resent-Cc:Resent-Message-ID:List-Id: List-Help:List-Unsubscribe:List-Subscribe:List-Post:List-Owner:List-Archive; bh=fQpyofeK1TgtZtQBzFHLUv5hXpuBhnqAJy31xABvd/I=; b=QzwrNNUFCtZEu6DNzOOJzzQ6X Zhv3aS/9BmjqOzM6GDqzKVkzM6DJwhXkZsFM7AyzCksvGk2bEKOEIIdr4iEH5a5gqySLXDi6Hmen5 mvNZDr5GpizZZ02iYTN5C8J71EKLVdDOXY/17QC3W9w2OiXRurijm1jGFQkJScuYfbOrbORqrnYkv rVeMPHW30eV2d9hDRkUQ5bGWKA350btKb4rEdx1Sp/LgcKUn7IioxQO/xyour9EUBtlcXMfIpjyNU CeXCpefVj28sAfq239MIIUi9Ak8kMI7eZsiOI/MvhyOE18lnh9l/v0IwR0VnmuHR4vCp1LoULLnpA tuQ6ffZVw==; Received: from ip-64-134-224-158.public.wayport.net ([64.134.224.158] helo=localhost) by bombadil.infradead.org with esmtpsa (Exim 4.87 #1 (Red Hat Linux)) id 1dQZtn-0004P2-UU; Thu, 29 Jun 2017 13:54:24 +0000 From: Christoph Hellwig To: Andreas Gruenbacher Cc: Jan Kara , linux-fsdevel@vger.kernel.org, linux-xfs@vger.kernel.org, linux-ext4@vger.kernel.org Subject: [PATCH 5/5] ext4: Switch to iomap for SEEK_HOLE / SEEK_DATA Date: Thu, 29 Jun 2017 06:54:20 -0700 Message-Id: <20170629135420.21357-6-hch@lst.de> X-Mailer: git-send-email 2.11.0 In-Reply-To: <20170629135420.21357-1-hch@lst.de> References: <20170629135420.21357-1-hch@lst.de> X-SRS-Rewrite: SMTP reverse-path rewritten from by bombadil.infradead.org. See http://www.infradead.org/rpr.html Sender: linux-ext4-owner@vger.kernel.org Precedence: bulk List-ID: X-Mailing-List: linux-ext4@vger.kernel.org Switch to the iomap_seek_hole and iomap_seek_data helpers for implementing lseek SEEK_HOLE / SEEK_DATA, and remove all the code that isn't needed any more. Note that with this patch ext4 will now always depend on the iomap code instead of only when CONFIG_DAX is enabled, and it requires adding a call into the extent status tree for iomap_begin as well to properly deal with delalloc extents. Signed-off-by: Christoph Hellwig --- fs/ext4/Kconfig | 1 + fs/ext4/ext4.h | 3 - fs/ext4/file.c | 264 +++----------------------------------------------------- fs/ext4/inode.c | 96 ++++++--------------- 4 files changed, 36 insertions(+), 328 deletions(-) diff --git a/fs/ext4/Kconfig b/fs/ext4/Kconfig index e38039fd96ff..73b850f5659c 100644 --- a/fs/ext4/Kconfig +++ b/fs/ext4/Kconfig @@ -37,6 +37,7 @@ config EXT4_FS select CRC16 select CRYPTO select CRYPTO_CRC32C + select FS_IOMAP help This is the next generation of the ext3 filesystem. diff --git a/fs/ext4/ext4.h b/fs/ext4/ext4.h index 32191548abed..eb0a1f221af3 100644 --- a/fs/ext4/ext4.h +++ b/fs/ext4/ext4.h @@ -2490,9 +2490,6 @@ extern void ext4_da_update_reserve_space(struct inode *inode, int used, int quota_claim); extern int ext4_issue_zeroout(struct inode *inode, ext4_lblk_t lblk, ext4_fsblk_t pblk, ext4_lblk_t len); -extern int ext4_get_next_extent(struct inode *inode, ext4_lblk_t lblk, - unsigned int map_len, - struct extent_status *result); /* indirect.c */ extern int ext4_ind_map_blocks(handle_t *handle, struct inode *inode, diff --git a/fs/ext4/file.c b/fs/ext4/file.c index 02ce7e7bbdf5..02bbf2ce7517 100644 --- a/fs/ext4/file.c +++ b/fs/ext4/file.c @@ -20,6 +20,7 @@ #include #include +#include #include #include #include @@ -439,253 +440,6 @@ static int ext4_file_open(struct inode * inode, struct file * filp) } /* - * Here we use ext4_map_blocks() to get a block mapping for a extent-based - * file rather than ext4_ext_walk_space() because we can introduce - * SEEK_DATA/SEEK_HOLE for block-mapped and extent-mapped file at the same - * function. When extent status tree has been fully implemented, it will - * track all extent status for a file and we can directly use it to - * retrieve the offset for SEEK_DATA/SEEK_HOLE. - */ - -/* - * When we retrieve the offset for SEEK_DATA/SEEK_HOLE, we would need to - * lookup page cache to check whether or not there has some data between - * [startoff, endoff] because, if this range contains an unwritten extent, - * we determine this extent as a data or a hole according to whether the - * page cache has data or not. - */ -static int ext4_find_unwritten_pgoff(struct inode *inode, - int whence, - ext4_lblk_t end_blk, - loff_t *offset) -{ - struct pagevec pvec; - unsigned int blkbits; - pgoff_t index; - pgoff_t end; - loff_t endoff; - loff_t startoff; - loff_t lastoff; - int found = 0; - - blkbits = inode->i_sb->s_blocksize_bits; - startoff = *offset; - lastoff = startoff; - endoff = (loff_t)end_blk << blkbits; - - index = startoff >> PAGE_SHIFT; - end = (endoff - 1) >> PAGE_SHIFT; - - pagevec_init(&pvec, 0); - do { - int i, num; - unsigned long nr_pages; - - num = min_t(pgoff_t, end - index, PAGEVEC_SIZE - 1) + 1; - nr_pages = pagevec_lookup(&pvec, inode->i_mapping, index, - (pgoff_t)num); - if (nr_pages == 0) - break; - - for (i = 0; i < nr_pages; i++) { - struct page *page = pvec.pages[i]; - struct buffer_head *bh, *head; - - /* - * If current offset is smaller than the page offset, - * there is a hole at this offset. - */ - if (whence == SEEK_HOLE && lastoff < endoff && - lastoff < page_offset(pvec.pages[i])) { - found = 1; - *offset = lastoff; - goto out; - } - - if (page->index > end) - goto out; - - lock_page(page); - - if (unlikely(page->mapping != inode->i_mapping)) { - unlock_page(page); - continue; - } - - if (!page_has_buffers(page)) { - unlock_page(page); - continue; - } - - if (page_has_buffers(page)) { - lastoff = page_offset(page); - bh = head = page_buffers(page); - do { - if (buffer_uptodate(bh) || - buffer_unwritten(bh)) { - if (whence == SEEK_DATA) - found = 1; - } else { - if (whence == SEEK_HOLE) - found = 1; - } - if (found) { - *offset = max_t(loff_t, - startoff, lastoff); - unlock_page(page); - goto out; - } - lastoff += bh->b_size; - bh = bh->b_this_page; - } while (bh != head); - } - - lastoff = page_offset(page) + PAGE_SIZE; - unlock_page(page); - } - - /* The no. of pages is less than our desired, we are done. */ - if (nr_pages < num) - break; - - index = pvec.pages[i - 1]->index + 1; - pagevec_release(&pvec); - } while (index <= end); - - if (whence == SEEK_HOLE && lastoff < endoff) { - found = 1; - *offset = lastoff; - } -out: - pagevec_release(&pvec); - return found; -} - -/* - * ext4_seek_data() retrieves the offset for SEEK_DATA. - */ -static loff_t ext4_seek_data(struct file *file, loff_t offset, loff_t maxsize) -{ - struct inode *inode = file->f_mapping->host; - struct extent_status es; - ext4_lblk_t start, last, end; - loff_t dataoff, isize; - int blkbits; - int ret; - - inode_lock(inode); - - isize = i_size_read(inode); - if (offset >= isize) { - inode_unlock(inode); - return -ENXIO; - } - - blkbits = inode->i_sb->s_blocksize_bits; - start = offset >> blkbits; - last = start; - end = isize >> blkbits; - dataoff = offset; - - do { - ret = ext4_get_next_extent(inode, last, end - last + 1, &es); - if (ret <= 0) { - /* No extent found -> no data */ - if (ret == 0) - ret = -ENXIO; - inode_unlock(inode); - return ret; - } - - last = es.es_lblk; - if (last != start) - dataoff = (loff_t)last << blkbits; - if (!ext4_es_is_unwritten(&es)) - break; - - /* - * If there is a unwritten extent at this offset, - * it will be as a data or a hole according to page - * cache that has data or not. - */ - if (ext4_find_unwritten_pgoff(inode, SEEK_DATA, - es.es_lblk + es.es_len, &dataoff)) - break; - last += es.es_len; - dataoff = (loff_t)last << blkbits; - cond_resched(); - } while (last <= end); - - inode_unlock(inode); - - if (dataoff > isize) - return -ENXIO; - - return vfs_setpos(file, dataoff, maxsize); -} - -/* - * ext4_seek_hole() retrieves the offset for SEEK_HOLE. - */ -static loff_t ext4_seek_hole(struct file *file, loff_t offset, loff_t maxsize) -{ - struct inode *inode = file->f_mapping->host; - struct extent_status es; - ext4_lblk_t start, last, end; - loff_t holeoff, isize; - int blkbits; - int ret; - - inode_lock(inode); - - isize = i_size_read(inode); - if (offset >= isize) { - inode_unlock(inode); - return -ENXIO; - } - - blkbits = inode->i_sb->s_blocksize_bits; - start = offset >> blkbits; - last = start; - end = isize >> blkbits; - holeoff = offset; - - do { - ret = ext4_get_next_extent(inode, last, end - last + 1, &es); - if (ret < 0) { - inode_unlock(inode); - return ret; - } - /* Found a hole? */ - if (ret == 0 || es.es_lblk > last) { - if (last != start) - holeoff = (loff_t)last << blkbits; - break; - } - /* - * If there is a unwritten extent at this offset, - * it will be as a data or a hole according to page - * cache that has data or not. - */ - if (ext4_es_is_unwritten(&es) && - ext4_find_unwritten_pgoff(inode, SEEK_HOLE, - last + es.es_len, &holeoff)) - break; - - last += es.es_len; - holeoff = (loff_t)last << blkbits; - cond_resched(); - } while (last <= end); - - inode_unlock(inode); - - if (holeoff > isize) - holeoff = isize; - - return vfs_setpos(file, holeoff, maxsize); -} - -/* * ext4_llseek() handles both block-mapped and extent-mapped maxbytes values * by calling generic_file_llseek_size() with the appropriate maxbytes * value for each. @@ -701,18 +455,20 @@ loff_t ext4_llseek(struct file *file, loff_t offset, int whence) maxbytes = inode->i_sb->s_maxbytes; switch (whence) { - case SEEK_SET: - case SEEK_CUR: - case SEEK_END: + default: return generic_file_llseek_size(file, offset, whence, maxbytes, i_size_read(inode)); - case SEEK_DATA: - return ext4_seek_data(file, offset, maxbytes); case SEEK_HOLE: - return ext4_seek_hole(file, offset, maxbytes); + offset = iomap_seek_hole(inode, offset, &ext4_iomap_ops); + break; + case SEEK_DATA: + offset = iomap_seek_data(inode, offset, &ext4_iomap_ops); + break; } - return -EINVAL; + if (offset < 0) + return offset; + return vfs_setpos(file, offset, maxbytes); } const struct file_operations ext4_file_operations = { diff --git a/fs/ext4/inode.c b/fs/ext4/inode.c index 5cf82d03968c..56a3b042b0ce 100644 --- a/fs/ext4/inode.c +++ b/fs/ext4/inode.c @@ -3350,7 +3350,6 @@ static int ext4_releasepage(struct page *page, gfp_t wait) return try_to_free_buffers(page); } -#ifdef CONFIG_FS_DAX static int ext4_iomap_begin(struct inode *inode, loff_t offset, loff_t length, unsigned flags, struct iomap *iomap) { @@ -3359,6 +3358,7 @@ static int ext4_iomap_begin(struct inode *inode, loff_t offset, loff_t length, unsigned long first_block = offset >> blkbits; unsigned long last_block = (offset + length - 1) >> blkbits; struct ext4_map_blocks map; + bool delalloc = false; int ret; if (WARN_ON_ONCE(ext4_has_inline_data(inode))) @@ -3369,6 +3369,27 @@ static int ext4_iomap_begin(struct inode *inode, loff_t offset, loff_t length, if (!(flags & IOMAP_WRITE)) { ret = ext4_map_blocks(NULL, inode, &map, 0); + if (!ret) { + struct extent_status es = {}; + + ext4_es_find_delayed_extent_range(inode, map.m_lblk, + map.m_lblk + map.m_len - 1, &es); + /* Is delalloc data before next block in extent tree? */ + if (es.es_len && es.es_lblk < map.m_lblk + map.m_len) { + ext4_lblk_t offset = 0; + + if (es.es_lblk < map.m_lblk) + offset = map.m_lblk - es.es_lblk; + map.m_lblk = es.es_lblk + offset; + map.m_pblk = ext4_es_pblock(&es) + offset; + map.m_len = es.es_len - offset; + if (ext4_es_status(&es) & EXTENT_STATUS_UNWRITTEN) + map.m_flags |= EXT4_MAP_UNWRITTEN; + if (ext4_es_status(&es) & EXTENT_STATUS_DELAYED) + delalloc = true; + ret = 1; + } + } } else { int dio_credits; handle_t *handle; @@ -3436,7 +3457,9 @@ static int ext4_iomap_begin(struct inode *inode, loff_t offset, loff_t length, iomap->blkno = IOMAP_NULL_BLOCK; iomap->length = (u64)map.m_len << blkbits; } else { - if (map.m_flags & EXT4_MAP_MAPPED) { + if (delalloc) { + iomap->type = IOMAP_DELALLOC; + } else if (map.m_flags & EXT4_MAP_MAPPED) { iomap->type = IOMAP_MAPPED; } else if (map.m_flags & EXT4_MAP_UNWRITTEN) { iomap->type = IOMAP_UNWRITTEN; @@ -3511,8 +3534,6 @@ const struct iomap_ops ext4_iomap_ops = { .iomap_end = ext4_iomap_end, }; -#endif - static int ext4_end_io_dio(struct kiocb *iocb, loff_t offset, ssize_t size, void *private) { @@ -5994,70 +6015,3 @@ int ext4_filemap_fault(struct vm_fault *vmf) return err; } - -/* - * Find the first extent at or after @lblk in an inode that is not a hole. - * Search for @map_len blocks at most. The extent is returned in @result. - * - * The function returns 1 if we found an extent. The function returns 0 in - * case there is no extent at or after @lblk and in that case also sets - * @result->es_len to 0. In case of error, the error code is returned. - */ -int ext4_get_next_extent(struct inode *inode, ext4_lblk_t lblk, - unsigned int map_len, struct extent_status *result) -{ - struct ext4_map_blocks map; - struct extent_status es = {}; - int ret; - - map.m_lblk = lblk; - map.m_len = map_len; - - /* - * For non-extent based files this loop may iterate several times since - * we do not determine full hole size. - */ - while (map.m_len > 0) { - ret = ext4_map_blocks(NULL, inode, &map, 0); - if (ret < 0) - return ret; - /* There's extent covering m_lblk? Just return it. */ - if (ret > 0) { - int status; - - ext4_es_store_pblock(result, map.m_pblk); - result->es_lblk = map.m_lblk; - result->es_len = map.m_len; - if (map.m_flags & EXT4_MAP_UNWRITTEN) - status = EXTENT_STATUS_UNWRITTEN; - else - status = EXTENT_STATUS_WRITTEN; - ext4_es_store_status(result, status); - return 1; - } - ext4_es_find_delayed_extent_range(inode, map.m_lblk, - map.m_lblk + map.m_len - 1, - &es); - /* Is delalloc data before next block in extent tree? */ - if (es.es_len && es.es_lblk < map.m_lblk + map.m_len) { - ext4_lblk_t offset = 0; - - if (es.es_lblk < lblk) - offset = lblk - es.es_lblk; - result->es_lblk = es.es_lblk + offset; - ext4_es_store_pblock(result, - ext4_es_pblock(&es) + offset); - result->es_len = es.es_len - offset; - ext4_es_store_status(result, ext4_es_status(&es)); - - return 1; - } - /* There's a hole at m_lblk, advance us after it */ - map.m_lblk += map.m_len; - map_len -= map.m_len; - map.m_len = map_len; - cond_resched(); - } - result->es_len = 0; - return 0; -}