From patchwork Wed Oct 7 05:00:25 2015 Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: 7bit X-Patchwork-Submitter: "Darrick J. Wong" X-Patchwork-Id: 7342331 Return-Path: X-Original-To: patchwork-linux-fsdevel@patchwork.kernel.org Delivered-To: patchwork-parsemail@patchwork2.web.kernel.org Received: from mail.kernel.org (mail.kernel.org [198.145.29.136]) by patchwork2.web.kernel.org (Postfix) with ESMTP id E33ABBEEA4 for ; Wed, 7 Oct 2015 05:00:34 +0000 (UTC) Received: from mail.kernel.org (localhost [127.0.0.1]) by mail.kernel.org (Postfix) with ESMTP id 26FCE20686 for ; Wed, 7 Oct 2015 05:00:33 +0000 (UTC) Received: from vger.kernel.org (vger.kernel.org [209.132.180.67]) by mail.kernel.org (Postfix) with ESMTP id D76BA2066D for ; Wed, 7 Oct 2015 05:00:31 +0000 (UTC) Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1753294AbbJGFAa (ORCPT ); Wed, 7 Oct 2015 01:00:30 -0400 Received: from userp1040.oracle.com ([156.151.31.81]:45089 "EHLO userp1040.oracle.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1753271AbbJGFA3 (ORCPT ); Wed, 7 Oct 2015 01:00:29 -0400 Received: from aserv0021.oracle.com (aserv0021.oracle.com [141.146.126.233]) by userp1040.oracle.com (Sentrion-MTA-4.3.2/Sentrion-MTA-4.3.2) with ESMTP id t9750Qtu014113 (version=TLSv1 cipher=DHE-RSA-AES256-SHA bits=256 verify=OK); Wed, 7 Oct 2015 05:00:27 GMT Received: from aserv0122.oracle.com (aserv0122.oracle.com [141.146.126.236]) by aserv0021.oracle.com (8.13.8/8.13.8) with ESMTP id t9750Q5T000809 (version=TLSv1/SSLv3 cipher=DHE-RSA-AES256-SHA bits=256 verify=FAIL); Wed, 7 Oct 2015 05:00:26 GMT Received: from abhmp0008.oracle.com (abhmp0008.oracle.com [141.146.116.14]) by aserv0122.oracle.com (8.13.8/8.13.8) with ESMTP id t9750QXa006978; Wed, 7 Oct 2015 05:00:26 GMT Received: from localhost (/24.21.154.84) by default (Oracle Beehive Gateway v4.0) with ESMTP ; Tue, 06 Oct 2015 22:00:26 -0700 Subject: [PATCH 48/58] xfs: copy-on-write reflinked blocks when zeroing ranges of blocks From: "Darrick J. Wong" To: david@fromorbit.com, darrick.wong@oracle.com Cc: linux-fsdevel@vger.kernel.org, xfs@oss.sgi.com Date: Tue, 06 Oct 2015 22:00:25 -0700 Message-ID: <20151007050025.30457.19562.stgit@birch.djwong.org> In-Reply-To: <20151007045443.30457.47038.stgit@birch.djwong.org> References: <20151007045443.30457.47038.stgit@birch.djwong.org> User-Agent: StGit/0.17.1-dirty MIME-Version: 1.0 X-Source-IP: aserv0021.oracle.com [141.146.126.233] Sender: linux-fsdevel-owner@vger.kernel.org Precedence: bulk List-ID: X-Mailing-List: linux-fsdevel@vger.kernel.org X-Spam-Status: No, score=-6.9 required=5.0 tests=BAYES_00, RCVD_IN_DNSWL_HI, T_RP_MATCHES_RCVD, UNPARSEABLE_RELAY autolearn=unavailable version=3.3.1 X-Spam-Checker-Version: SpamAssassin 3.3.1 (2010-03-16) on mail.kernel.org X-Virus-Scanned: ClamAV using ClamSMTP When we're writing zeroes to a reflinked block (such as when we're punching a reflinked range), we need to fork the the block and write to that, otherwise we can corrupt the other reflinks. Signed-off-by: Darrick J. Wong --- fs/xfs/xfs_bmap_util.c | 25 +++++++- fs/xfs/xfs_reflink.c | 154 ++++++++++++++++++++++++++++++++++++++++++++++++ fs/xfs/xfs_reflink.h | 6 ++ 3 files changed, 183 insertions(+), 2 deletions(-) -- To unsubscribe from this list: send the line "unsubscribe linux-fsdevel" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html diff --git a/fs/xfs/xfs_bmap_util.c b/fs/xfs/xfs_bmap_util.c index 245a34a..b054b28 100644 --- a/fs/xfs/xfs_bmap_util.c +++ b/fs/xfs/xfs_bmap_util.c @@ -41,6 +41,7 @@ #include "xfs_icache.h" #include "xfs_log.h" #include "xfs_rmap_btree.h" +#include "xfs_reflink.h" /* Kernel only BMAP related definitions and functions */ @@ -1095,7 +1096,9 @@ xfs_zero_remaining_bytes( xfs_buf_t *bp; xfs_mount_t *mp = ip->i_mount; int nimap; - int error = 0; + int error = 0, err2; + bool should_fork; + struct xfs_trans *tp; /* * Avoid doing I/O beyond eof - it's not necessary @@ -1136,8 +1139,14 @@ xfs_zero_remaining_bytes( if (lastoffset > endoff) lastoffset = endoff; + /* Do we need to CoW this block? */ + error = xfs_reflink_should_fork_block(ip, &imap, offset, + &should_fork); + if (error) + return error; + /* DAX can just zero the backing device directly */ - if (IS_DAX(VFS_I(ip))) { + if (IS_DAX(VFS_I(ip)) && !should_fork) { error = dax_zero_page_range(VFS_I(ip), offset, lastoffset - offset + 1, xfs_get_blocks_direct); @@ -1158,10 +1167,22 @@ xfs_zero_remaining_bytes( (offset - XFS_FSB_TO_B(mp, imap.br_startoff)), 0, lastoffset - offset + 1); + tp = NULL; + if (should_fork) { + error = xfs_reflink_fork_buf(ip, bp, offset_fsb, &tp); + if (error) + return error; + } + error = xfs_bwrite(bp); + + err2 = xfs_reflink_finish_fork_buf(ip, bp, offset_fsb, tp, + error, imap.br_startblock); xfs_buf_relse(bp); if (error) return error; + if (err2) + return err2; } return error; } diff --git a/fs/xfs/xfs_reflink.c b/fs/xfs/xfs_reflink.c index 226e23f..f5eed2f 100644 --- a/fs/xfs/xfs_reflink.c +++ b/fs/xfs/xfs_reflink.c @@ -788,3 +788,157 @@ xfs_reflink_add_ioend( { list_add_tail(&eio->rlei_list, &ioend->io_reflink_endio_list); } + +/** + * xfs_reflink_fork_buf() - start a transaction to fork a buffer (if needed) + * + * @mp: XFS mount point + * @ip: XFS inode + * @bp: the buffer that we might need to fork + * @fileoff: file offset of the buffer + * @ptp: pointer to an XFS transaction + */ +int +xfs_reflink_fork_buf( + struct xfs_inode *ip, + struct xfs_buf *bp, + xfs_fileoff_t fileoff, + struct xfs_trans **ptp) +{ + struct xfs_mount *mp = ip->i_mount; + struct xfs_trans *tp; + xfs_fsblock_t fsbno; + xfs_fsblock_t new_fsbno; + xfs_agnumber_t agno; + xfs_agblock_t agbno; + uint resblks; + int error; + + fsbno = XFS_DADDR_TO_FSB(mp, XFS_BUF_ADDR(bp)); + agno = XFS_FSB_TO_AGNO(mp, fsbno); + agbno = XFS_FSB_TO_AGBNO(mp, fsbno); + CHECK_AG_NUMBER(mp, agno); + CHECK_AG_EXTENT(mp, agno, 1); + + /* + * Get ready to remap the thing... + */ + resblks = XFS_DIOSTRAT_SPACE_RES(mp, 3); + tp = xfs_trans_alloc(mp, XFS_TRANS_STRAT_WRITE); + error = xfs_trans_reserve(tp, &M_RES(mp)->tr_write, resblks, 0); + + /* + * check for running out of space + */ + if (error) { + /* + * Free the transaction structure. + */ + ASSERT(error == -ENOSPC || XFS_FORCED_SHUTDOWN(mp)); + goto out_cancel; + } + error = xfs_trans_reserve_quota(tp, mp, + ip->i_udquot, ip->i_gdquot, ip->i_pdquot, + resblks, 0, XFS_QMOPT_RES_REGBLKS); + if (error) + goto out_cancel; + + xfs_ilock(ip, XFS_ILOCK_EXCL); + xfs_trans_ijoin(tp, ip, XFS_ILOCK_EXCL); + + /* fork block, remap buffer */ + error = fork_one_block(mp, tp, ip, fsbno, &new_fsbno, fileoff); + if (error) + goto out_cancel; + + trace_xfs_reflink_fork_buf(ip, fileoff, fsbno, 1, new_fsbno); + + XFS_BUF_SET_ADDR(bp, XFS_FSB_TO_DADDR(mp, new_fsbno)); + *ptp = tp; + return error; + +out_cancel: + xfs_trans_cancel(tp); + trace_xfs_reflink_fork_buf_error(ip, error, _RET_IP_); + return error; +} + +/** + * xfs_reflink_finish_fork_buf() - finish forking a file buffer + * + * @ip: XFS inode + * @bp: the buffer that was forked + * @fileoff: file offset of the buffer + * @tp: transaction that was returned from xfs_reflink_fork_buf() + * @write_error: status code from writing the block + */ +int +xfs_reflink_finish_fork_buf( + struct xfs_inode *ip, + struct xfs_buf *bp, + xfs_fileoff_t fileoff, + struct xfs_trans *tp, + int write_error, + xfs_fsblock_t old_fsbno) +{ + struct xfs_mount *mp = ip->i_mount; + struct xfs_bmap_free free_list; + xfs_fsblock_t firstfsb; + xfs_fsblock_t fsbno; + struct xfs_bmbt_irec imaps[1]; + xfs_agnumber_t agno; + int nimaps = 1; + int done; + int error = write_error; + int committed; + struct xfs_owner_info oinfo; + + if (tp == NULL) + return 0; + + fsbno = XFS_DADDR_TO_FSB(mp, XFS_BUF_ADDR(bp)); + agno = XFS_FSB_TO_AGNO(mp, fsbno); + XFS_RMAP_INO_OWNER(&oinfo, ip->i_ino, XFS_DATA_FORK, fileoff); + if (write_error != 0) + goto out_write_error; + + trace_xfs_reflink_fork_buf(ip, fileoff, old_fsbno, 1, fsbno); + /* + * Remap the old blocks. + */ + xfs_bmap_init(&free_list, &firstfsb); + error = xfs_bunmapi(tp, ip, fileoff, 1, 0, 1, &firstfsb, &free_list, + &done); + if (error) + goto out_free; + ASSERT(done == 1); + + error = xfs_bmapi_write(tp, ip, fileoff, 1, XFS_BMAPI_REMAP, &fsbno, + 1, &imaps[0], &nimaps, &free_list); + if (error) + goto out_free; + + /* + * complete the transaction + */ + error = xfs_bmap_finish(&tp, &free_list, &committed); + if (error) + goto out_cancel; + + error = xfs_trans_commit(tp); + if (error) + goto out_error; + + return error; +out_free: + xfs_bmap_finish(&tp, &free_list, &committed); +out_write_error: + done = xfs_free_extent(tp, fsbno, 1, &oinfo); + if (error == 0) + error = done; +out_cancel: + xfs_trans_cancel(tp); +out_error: + trace_xfs_reflink_finish_fork_buf_error(ip, error, _RET_IP_); + return error; +} diff --git a/fs/xfs/xfs_reflink.h b/fs/xfs/xfs_reflink.h index b3e12d2..ce00cf6 100644 --- a/fs/xfs/xfs_reflink.h +++ b/fs/xfs/xfs_reflink.h @@ -38,4 +38,10 @@ extern void xfs_reflink_add_ioend(struct xfs_ioend *ioend, extern int xfs_reflink_should_fork_block(struct xfs_inode *ip, struct xfs_bmbt_irec *imap, xfs_off_t offset, bool *type); +extern int xfs_reflink_fork_buf(struct xfs_inode *ip, struct xfs_buf *bp, + xfs_fileoff_t fileoff, struct xfs_trans **ptp); +extern int xfs_reflink_finish_fork_buf(struct xfs_inode *ip, struct xfs_buf *bp, + xfs_fileoff_t fileoff, struct xfs_trans *tp, int write_error, + xfs_fsblock_t old_fsbno); + #endif /* __XFS_REFLINK_H */