X-Git-Url: http://git.onelab.eu/?a=blobdiff_plain;f=fs%2Fxfs%2Flinux-2.6%2Fxfs_lrw.c;h=ee788b1cb3641a355219183f9217037d04df2a94;hb=16c70f8c1b54b61c3b951b6fb220df250fe09b32;hp=9bec802f12061e8ba657a543ac30d7a1d5c381c5;hpb=6a77f38946aaee1cd85eeec6cf4229b204c15071;p=linux-2.6.git diff --git a/fs/xfs/linux-2.6/xfs_lrw.c b/fs/xfs/linux-2.6/xfs_lrw.c index 9bec802f1..ee788b1cb 100644 --- a/fs/xfs/linux-2.6/xfs_lrw.c +++ b/fs/xfs/linux-2.6/xfs_lrw.c @@ -1,65 +1,43 @@ /* - * Copyright (c) 2000-2003 Silicon Graphics, Inc. All Rights Reserved. + * Copyright (c) 2000-2003,2005 Silicon Graphics, Inc. + * All Rights Reserved. * - * This program is free software; you can redistribute it and/or modify it - * under the terms of version 2 of the GNU General Public License as + * This program is free software; you can redistribute it and/or + * modify it under the terms of the GNU General Public License as * published by the Free Software Foundation. * - * This program is distributed in the hope that it would be useful, but - * WITHOUT ANY WARRANTY; without even the implied warranty of - * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. + * This program is distributed in the hope that it would be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + * GNU General Public License for more details. * - * Further, this software is distributed without any warranty that it is - * free of the rightful claim of any third person regarding infringement - * or the like. Any license provided herein, whether implied or - * otherwise, applies only to this software file. Patent licenses, if - * any, provided herein do not apply to combinations of this program with - * other software, or any other product whatsoever. - * - * You should have received a copy of the GNU General Public License along - * with this program; if not, write the Free Software Foundation, Inc., 59 - * Temple Place - Suite 330, Boston MA 02111-1307, USA. - * - * Contact information: Silicon Graphics, Inc., 1600 Amphitheatre Pkwy, - * Mountain View, CA 94043, or: - * - * http://www.sgi.com - * - * For further information regarding this notice, see: - * - * http://oss.sgi.com/projects/GenInfo/SGIGPLNoticeExplan/ + * You should have received a copy of the GNU General Public License + * along with this program; if not, write the Free Software Foundation, + * Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA */ -/* - * fs/xfs/linux/xfs_lrw.c (Linux Read Write stuff) - * - */ - #include "xfs.h" - #include "xfs_fs.h" -#include "xfs_inum.h" +#include "xfs_bit.h" #include "xfs_log.h" +#include "xfs_inum.h" #include "xfs_trans.h" #include "xfs_sb.h" #include "xfs_ag.h" -#include "xfs_dir.h" #include "xfs_dir2.h" #include "xfs_alloc.h" #include "xfs_dmapi.h" #include "xfs_quota.h" #include "xfs_mount.h" -#include "xfs_alloc_btree.h" #include "xfs_bmap_btree.h" +#include "xfs_alloc_btree.h" #include "xfs_ialloc_btree.h" -#include "xfs_btree.h" -#include "xfs_ialloc.h" -#include "xfs_attr_sf.h" -#include "xfs_dir_sf.h" #include "xfs_dir2_sf.h" +#include "xfs_attr_sf.h" #include "xfs_dinode.h" #include "xfs_inode.h" #include "xfs_bmap.h" -#include "xfs_bit.h" +#include "xfs_btree.h" +#include "xfs_ialloc.h" #include "xfs_rtalloc.h" #include "xfs_error.h" #include "xfs_itable.h" @@ -103,7 +81,7 @@ xfs_rw_enter_trace( (void *)((unsigned long)ioflags), (void *)((unsigned long)((io->io_new_size >> 32) & 0xffffffff)), (void *)((unsigned long)(io->io_new_size & 0xffffffff)), - (void *)NULL, + (void *)((unsigned long)current_pid()), (void *)NULL, (void *)NULL, (void *)NULL, @@ -133,7 +111,7 @@ xfs_inval_cached_trace( (void *)((unsigned long)(first & 0xffffffff)), (void *)((unsigned long)((last >> 32) & 0xffffffff)), (void *)((unsigned long)(last & 0xffffffff)), - (void *)NULL, + (void *)((unsigned long)current_pid()), (void *)NULL, (void *)NULL, (void *)NULL, @@ -209,30 +187,6 @@ unlock: return (-status); } -/* - * xfs_inval_cached_pages - * - * This routine is responsible for keeping direct I/O and buffered I/O - * somewhat coherent. From here we make sure that we're at least - * temporarily holding the inode I/O lock exclusively and then call - * the page cache to flush and invalidate any cached pages. If there - * are no cached pages this routine will be very quick. - */ -void -xfs_inval_cached_pages( - vnode_t *vp, - xfs_iocore_t *io, - xfs_off_t offset, - int write, - int relock) -{ - if (VN_CACHED(vp)) { - xfs_inval_cached_trace(io, offset, -1, ctooff(offtoct(offset)), -1); - VOP_FLUSHINVAL_PAGES(vp, ctooff(offtoct(offset)), -1, FI_REMAPF_LOCKED); - } - -} - ssize_t /* bytes read, or (-) error */ xfs_read( bhv_desc_t *bdp, @@ -250,7 +204,7 @@ xfs_read( xfs_fsize_t n; xfs_inode_t *ip; xfs_mount_t *mp; - vnode_t *vp; + bhv_vnode_t *vp; unsigned long seg; ip = XFS_BHVTOI(bdp); @@ -277,8 +231,8 @@ xfs_read( xfs_buftarg_t *target = (ip->i_d.di_flags & XFS_DIFLAG_REALTIME) ? mp->m_rtdev_targp : mp->m_ddev_targp; - if ((*offset & target->pbr_smask) || - (size & target->pbr_smask)) { + if ((*offset & target->bt_smask) || + (size & target->bt_smask)) { if (*offset == ip->i_d.di_size) { return (0); } @@ -293,27 +247,36 @@ xfs_read( if (n < size) size = n; - if (XFS_FORCED_SHUTDOWN(mp)) { + if (XFS_FORCED_SHUTDOWN(mp)) return -EIO; - } if (unlikely(ioflags & IO_ISDIRECT)) - down(&inode->i_sem); + mutex_lock(&inode->i_mutex); xfs_ilock(ip, XFS_IOLOCK_SHARED); if (DM_EVENT_ENABLED(vp->v_vfsp, ip, DM_EVENT_READ) && !(ioflags & IO_INVIS)) { - vrwlock_t locktype = VRWLOCK_READ; + bhv_vrwlock_t locktype = VRWLOCK_READ; + int dmflags = FILP_DELAY_FLAG(file) | DM_SEM_FLAG_RD(ioflags); ret = -XFS_SEND_DATA(mp, DM_EVENT_READ, BHV_TO_VNODE(bdp), *offset, size, - FILP_DELAY_FLAG(file), &locktype); + dmflags, &locktype); if (ret) { xfs_iunlock(ip, XFS_IOLOCK_SHARED); - goto unlock_isem; + if (unlikely(ioflags & IO_ISDIRECT)) + mutex_unlock(&inode->i_mutex); + return ret; } } + if (unlikely((ioflags & IO_ISDIRECT) && VN_CACHED(vp))) + bhv_vop_flushinval_pages(vp, ctooff(offtoct(*offset)), + -1, FI_REMAPF_LOCKED); + + if (unlikely(ioflags & IO_ISDIRECT)) + mutex_unlock(&inode->i_mutex); + xfs_rw_enter_trace(XFS_READ_ENTER, &ip->i_iocore, (void *)iovp, segs, *offset, ioflags); ret = __generic_file_aio_read(iocb, iovp, segs, offset); @@ -323,13 +286,6 @@ xfs_read( XFS_STATS_ADD(xs_read_bytes, ret); xfs_iunlock(ip, XFS_IOLOCK_SHARED); - - if (likely(!(ioflags & IO_INVIS))) - xfs_ichgtime(ip, XFS_ICHGTIME_ACC); - -unlock_isem: - if (unlikely(ioflags & IO_ISDIRECT)) - up(&inode->i_sem); return ret; } @@ -344,36 +300,23 @@ xfs_sendfile( void *target, cred_t *credp) { + xfs_inode_t *ip = XFS_BHVTOI(bdp); + xfs_mount_t *mp = ip->i_mount; ssize_t ret; - xfs_fsize_t n; - xfs_inode_t *ip; - xfs_mount_t *mp; - vnode_t *vp; - - ip = XFS_BHVTOI(bdp); - vp = BHV_TO_VNODE(bdp); - mp = ip->i_mount; XFS_STATS_INC(xs_read_calls); - - n = XFS_MAXIOFFSET(mp) - *offset; - if ((n <= 0) || (count == 0)) - return 0; - - if (n < count) - count = n; - - if (XFS_FORCED_SHUTDOWN(ip->i_mount)) + if (XFS_FORCED_SHUTDOWN(mp)) return -EIO; xfs_ilock(ip, XFS_IOLOCK_SHARED); - if (DM_EVENT_ENABLED(vp->v_vfsp, ip, DM_EVENT_READ) && + if (DM_EVENT_ENABLED(BHV_TO_VNODE(bdp)->v_vfsp, ip, DM_EVENT_READ) && (!(ioflags & IO_INVIS))) { - vrwlock_t locktype = VRWLOCK_READ; + bhv_vrwlock_t locktype = VRWLOCK_READ; int error; - error = XFS_SEND_DATA(mp, DM_EVENT_READ, BHV_TO_VNODE(bdp), *offset, count, + error = XFS_SEND_DATA(mp, DM_EVENT_READ, BHV_TO_VNODE(bdp), + *offset, count, FILP_DELAY_FLAG(filp), &locktype); if (error) { xfs_iunlock(ip, XFS_IOLOCK_SHARED); @@ -383,15 +326,114 @@ xfs_sendfile( xfs_rw_enter_trace(XFS_SENDFILE_ENTER, &ip->i_iocore, (void *)(unsigned long)target, count, *offset, ioflags); ret = generic_file_sendfile(filp, offset, count, actor, target); + if (ret > 0) + XFS_STATS_ADD(xs_read_bytes, ret); xfs_iunlock(ip, XFS_IOLOCK_SHARED); + return ret; +} + +ssize_t +xfs_splice_read( + bhv_desc_t *bdp, + struct file *infilp, + loff_t *ppos, + struct pipe_inode_info *pipe, + size_t count, + int flags, + int ioflags, + cred_t *credp) +{ + xfs_inode_t *ip = XFS_BHVTOI(bdp); + xfs_mount_t *mp = ip->i_mount; + ssize_t ret; + + XFS_STATS_INC(xs_read_calls); + if (XFS_FORCED_SHUTDOWN(ip->i_mount)) + return -EIO; + + xfs_ilock(ip, XFS_IOLOCK_SHARED); + if (DM_EVENT_ENABLED(BHV_TO_VNODE(bdp)->v_vfsp, ip, DM_EVENT_READ) && + (!(ioflags & IO_INVIS))) { + bhv_vrwlock_t locktype = VRWLOCK_READ; + int error; + + error = XFS_SEND_DATA(mp, DM_EVENT_READ, BHV_TO_VNODE(bdp), + *ppos, count, + FILP_DELAY_FLAG(infilp), &locktype); + if (error) { + xfs_iunlock(ip, XFS_IOLOCK_SHARED); + return -error; + } + } + xfs_rw_enter_trace(XFS_SPLICE_READ_ENTER, &ip->i_iocore, + pipe, count, *ppos, ioflags); + ret = generic_file_splice_read(infilp, ppos, pipe, count, flags); if (ret > 0) XFS_STATS_ADD(xs_read_bytes, ret); - if (likely(!(ioflags & IO_INVIS))) - xfs_ichgtime(ip, XFS_ICHGTIME_ACC); + xfs_iunlock(ip, XFS_IOLOCK_SHARED); + return ret; +} + +ssize_t +xfs_splice_write( + bhv_desc_t *bdp, + struct pipe_inode_info *pipe, + struct file *outfilp, + loff_t *ppos, + size_t count, + int flags, + int ioflags, + cred_t *credp) +{ + xfs_inode_t *ip = XFS_BHVTOI(bdp); + xfs_mount_t *mp = ip->i_mount; + ssize_t ret; + struct inode *inode = outfilp->f_mapping->host; + xfs_fsize_t isize; + XFS_STATS_INC(xs_write_calls); + if (XFS_FORCED_SHUTDOWN(ip->i_mount)) + return -EIO; + + xfs_ilock(ip, XFS_IOLOCK_EXCL); + + if (DM_EVENT_ENABLED(BHV_TO_VNODE(bdp)->v_vfsp, ip, DM_EVENT_WRITE) && + (!(ioflags & IO_INVIS))) { + bhv_vrwlock_t locktype = VRWLOCK_WRITE; + int error; + + error = XFS_SEND_DATA(mp, DM_EVENT_WRITE, BHV_TO_VNODE(bdp), + *ppos, count, + FILP_DELAY_FLAG(outfilp), &locktype); + if (error) { + xfs_iunlock(ip, XFS_IOLOCK_EXCL); + return -error; + } + } + xfs_rw_enter_trace(XFS_SPLICE_WRITE_ENTER, &ip->i_iocore, + pipe, count, *ppos, ioflags); + ret = generic_file_splice_write(pipe, outfilp, ppos, count, flags); + if (ret > 0) + XFS_STATS_ADD(xs_write_bytes, ret); + + isize = i_size_read(inode); + if (unlikely(ret < 0 && ret != -EFAULT && *ppos > isize)) + *ppos = isize; + + if (*ppos > ip->i_d.di_size) { + xfs_ilock(ip, XFS_ILOCK_EXCL); + if (*ppos > ip->i_d.di_size) { + ip->i_d.di_size = *ppos; + i_size_write(inode, *ppos); + ip->i_update_core = 1; + ip->i_update_size = 1; + } + xfs_iunlock(ip, XFS_ILOCK_EXCL); + } + xfs_iunlock(ip, XFS_IOLOCK_EXCL); return ret; } @@ -405,28 +447,22 @@ STATIC int /* error (positive) */ xfs_zero_last_block( struct inode *ip, xfs_iocore_t *io, - xfs_off_t offset, xfs_fsize_t isize, xfs_fsize_t end_size) { xfs_fileoff_t last_fsb; - xfs_mount_t *mp; + xfs_mount_t *mp = io->io_mount; int nimaps; int zero_offset; int zero_len; - int isize_fsb_offset; int error = 0; xfs_bmbt_irec_t imap; loff_t loff; - size_t lsize; ASSERT(ismrlocked(io->io_lock, MR_UPDATE) != 0); - ASSERT(offset > isize); - mp = io->io_mount; - - isize_fsb_offset = XFS_B_FSB_OFFSET(mp, isize); - if (isize_fsb_offset == 0) { + zero_offset = XFS_B_FSB_OFFSET(mp, isize); + if (zero_offset == 0) { /* * There are no extra bytes in the last block on disk to * zero, so return. @@ -437,7 +473,7 @@ xfs_zero_last_block( last_fsb = XFS_B_TO_FSBT(mp, isize); nimaps = 1; error = XFS_BMAPI(mp, NULL, io, last_fsb, 1, 0, NULL, 0, &imap, - &nimaps, NULL); + &nimaps, NULL, NULL); if (error) { return error; } @@ -455,12 +491,9 @@ xfs_zero_last_block( * don't deadlock when the buffer cache calls back to us. */ XFS_IUNLOCK(mp, io, XFS_ILOCK_EXCL| XFS_EXTSIZE_RD); - loff = XFS_FSB_TO_B(mp, last_fsb); - lsize = XFS_FSB_TO_B(mp, 1); - - zero_offset = isize_fsb_offset; - zero_len = mp->m_sb.sb_blocksize - isize_fsb_offset; + loff = XFS_FSB_TO_B(mp, last_fsb); + zero_len = mp->m_sb.sb_blocksize - zero_offset; error = xfs_iozero(ip, loff + zero_offset, zero_len, end_size); XFS_ILOCK(mp, io, XFS_ILOCK_EXCL|XFS_EXTSIZE_RD); @@ -481,37 +514,31 @@ xfs_zero_last_block( int /* error (positive) */ xfs_zero_eof( - vnode_t *vp, + bhv_vnode_t *vp, xfs_iocore_t *io, xfs_off_t offset, /* starting I/O offset */ xfs_fsize_t isize, /* current inode size */ xfs_fsize_t end_size) /* terminal inode size */ { - struct inode *ip = LINVFS_GET_IP(vp); + struct inode *ip = vn_to_inode(vp); xfs_fileoff_t start_zero_fsb; xfs_fileoff_t end_zero_fsb; - xfs_fileoff_t prev_zero_fsb; xfs_fileoff_t zero_count_fsb; xfs_fileoff_t last_fsb; - xfs_extlen_t buf_len_fsb; - xfs_extlen_t prev_zero_count; - xfs_mount_t *mp; + xfs_mount_t *mp = io->io_mount; int nimaps; int error = 0; xfs_bmbt_irec_t imap; - loff_t loff; - size_t lsize; ASSERT(ismrlocked(io->io_lock, MR_UPDATE)); ASSERT(ismrlocked(io->io_iolock, MR_UPDATE)); - - mp = io->io_mount; + ASSERT(offset > isize); /* * First handle zeroing the block on which isize resides. * We only zero a part of that block so it is handled specially. */ - error = xfs_zero_last_block(ip, io, offset, isize, end_size); + error = xfs_zero_last_block(ip, io, isize, end_size); if (error) { ASSERT(ismrlocked(io->io_lock, MR_UPDATE)); ASSERT(ismrlocked(io->io_iolock, MR_UPDATE)); @@ -539,13 +566,11 @@ xfs_zero_eof( } ASSERT(start_zero_fsb <= end_zero_fsb); - prev_zero_fsb = NULLFILEOFF; - prev_zero_count = 0; while (start_zero_fsb <= end_zero_fsb) { nimaps = 1; zero_count_fsb = end_zero_fsb - start_zero_fsb + 1; error = XFS_BMAPI(mp, NULL, io, start_zero_fsb, zero_count_fsb, - 0, NULL, 0, &imap, &nimaps, NULL); + 0, NULL, 0, &imap, &nimaps, NULL, NULL); if (error) { ASSERT(ismrlocked(io->io_lock, MR_UPDATE)); ASSERT(ismrlocked(io->io_iolock, MR_UPDATE)); @@ -562,42 +587,27 @@ xfs_zero_eof( * that sits on a hole and sets the page as P_HOLE * and calls remapf if it is a mapped file. */ - prev_zero_fsb = NULLFILEOFF; - prev_zero_count = 0; - start_zero_fsb = imap.br_startoff + - imap.br_blockcount; + start_zero_fsb = imap.br_startoff + imap.br_blockcount; ASSERT(start_zero_fsb <= (end_zero_fsb + 1)); continue; } /* - * There are blocks in the range requested. - * Zero them a single write at a time. We actually - * don't zero the entire range returned if it is - * too big and simply loop around to get the rest. - * That is not the most efficient thing to do, but it - * is simple and this path should not be exercised often. - */ - buf_len_fsb = XFS_FILBLKS_MIN(imap.br_blockcount, - mp->m_writeio_blocks << 8); - /* + * There are blocks we need to zero. * Drop the inode lock while we're doing the I/O. * We'll still have the iolock to protect us. */ XFS_IUNLOCK(mp, io, XFS_ILOCK_EXCL|XFS_EXTSIZE_RD); - loff = XFS_FSB_TO_B(mp, start_zero_fsb); - lsize = XFS_FSB_TO_B(mp, buf_len_fsb); - - error = xfs_iozero(ip, loff, lsize, end_size); - + error = xfs_iozero(ip, + XFS_FSB_TO_B(mp, start_zero_fsb), + XFS_FSB_TO_B(mp, imap.br_blockcount), + end_size); if (error) { goto out_lock; } - prev_zero_fsb = start_zero_fsb; - prev_zero_count = buf_len_fsb; - start_zero_fsb = imap.br_startoff + buf_len_fsb; + start_zero_fsb = imap.br_startoff + imap.br_blockcount; ASSERT(start_zero_fsb <= (end_zero_fsb + 1)); XFS_ILOCK(mp, io, XFS_ILOCK_EXCL|XFS_EXTSIZE_RD); @@ -631,14 +641,14 @@ xfs_write( ssize_t ret = 0, error = 0; xfs_fsize_t isize, new_size; xfs_iocore_t *io; - vnode_t *vp; + bhv_vnode_t *vp; unsigned long seg; int iolock; int eventsent = 0; - vrwlock_t locktype; + bhv_vrwlock_t locktype; size_t ocount = 0, count; loff_t pos; - int need_isem = 1, need_flush = 0; + int need_i_mutex = 1, need_flush = 0; XFS_STATS_INC(xs_write_calls); @@ -673,32 +683,32 @@ xfs_write( io = &xip->i_iocore; mp = io->io_mount; + vfs_wait_for_freeze(vp->v_vfsp, SB_FREEZE_WRITE); + if (XFS_FORCED_SHUTDOWN(mp)) return -EIO; - fs_check_frozen(vp->v_vfsp, SB_FREEZE_WRITE); - if (ioflags & IO_ISDIRECT) { xfs_buftarg_t *target = (xip->i_d.di_flags & XFS_DIFLAG_REALTIME) ? mp->m_rtdev_targp : mp->m_ddev_targp; - if ((pos & target->pbr_smask) || (count & target->pbr_smask)) + if ((pos & target->bt_smask) || (count & target->bt_smask)) return XFS_ERROR(-EINVAL); if (!VN_CACHED(vp) && pos < i_size_read(inode)) - need_isem = 0; + need_i_mutex = 0; if (VN_CACHED(vp)) need_flush = 1; } relock: - if (need_isem) { + if (need_i_mutex) { iolock = XFS_IOLOCK_EXCL; locktype = VRWLOCK_WRITE; - down(&inode->i_sem); + mutex_lock(&inode->i_mutex); } else { iolock = XFS_IOLOCK_SHARED; locktype = VRWLOCK_WRITE_DIRECT; @@ -716,7 +726,7 @@ start: S_ISBLK(inode->i_mode)); if (error) { xfs_iunlock(xip, XFS_ILOCK_EXCL|iolock); - goto out_unlock_isem; + goto out_unlock_mutex; } new_size = pos + count; @@ -728,8 +738,8 @@ start: loff_t savedsize = pos; int dmflags = FILP_DELAY_FLAG(file); - if (need_isem) - dmflags |= DM_FLAGS_ISEM; + if (need_i_mutex) + dmflags |= DM_FLAGS_IMUX; xfs_iunlock(xip, XFS_ILOCK_EXCL); error = XFS_SEND_DATA(xip->i_mount, DM_EVENT_WRITE, vp, @@ -737,13 +747,13 @@ start: dmflags, &locktype); if (error) { xfs_iunlock(xip, iolock); - goto out_unlock_isem; + goto out_unlock_mutex; } xfs_ilock(xip, XFS_ILOCK_EXCL); eventsent = 1; /* - * The iolock was dropped and reaquired in XFS_SEND_DATA + * The iolock was dropped and reacquired in XFS_SEND_DATA * so we have to recheck the size when appending. * We will only "goto start;" once, since having sent the * event prevents another call to XFS_SEND_DATA, which is @@ -755,15 +765,10 @@ start: } } - /* - * On Linux, generic_file_write updates the times even if - * no data is copied in so long as the write had a size. - * - * We must update xfs' times since revalidate will overcopy xfs. - */ - if (!(ioflags & IO_INVIS)) { - xfs_ichgtime(xip, XFS_ICHGTIME_MOD | XFS_ICHGTIME_CHG); - inode_update_time(inode, 1); + if (likely(!(ioflags & IO_INVIS))) { + file_update_time(file); + xfs_ichgtime_fast(xip, inode, + XFS_ICHGTIME_MOD | XFS_ICHGTIME_CHG); } /* @@ -780,7 +785,7 @@ start: isize, pos + count); if (error) { xfs_iunlock(xip, XFS_ILOCK_EXCL|iolock); - goto out_unlock_isem; + goto out_unlock_mutex; } } xfs_iunlock(xip, XFS_ILOCK_EXCL); @@ -801,7 +806,7 @@ start: error = -remove_suid(file->f_dentry); if (unlikely(error)) { xfs_iunlock(xip, iolock); - goto out_unlock_isem; + goto out_unlock_mutex; } } @@ -813,18 +818,18 @@ retry: if (need_flush) { xfs_inval_cached_trace(io, pos, -1, ctooff(offtoct(pos)), -1); - VOP_FLUSHINVAL_PAGES(vp, ctooff(offtoct(pos)), + bhv_vop_flushinval_pages(vp, ctooff(offtoct(pos)), -1, FI_REMAPF_LOCKED); } - if (need_isem) { + if (need_i_mutex) { /* demote the lock now the cached pages are gone */ XFS_ILOCK_DEMOTE(mp, io, XFS_IOLOCK_EXCL); - up(&inode->i_sem); + mutex_unlock(&inode->i_mutex); iolock = XFS_IOLOCK_SHARED; locktype = VRWLOCK_WRITE_DIRECT; - need_isem = 0; + need_i_mutex = 0; } xfs_rw_enter_trace(XFS_DIOWR_ENTER, io, (void *)iovp, segs, @@ -842,7 +847,7 @@ retry: pos += ret; count -= ret; - need_isem = 1; + need_i_mutex = 1; ioflags &= ~IO_ISDIRECT; xfs_iunlock(xip, iolock); goto relock; @@ -864,16 +869,25 @@ retry: !(ioflags & IO_INVIS)) { xfs_rwunlock(bdp, locktype); + if (need_i_mutex) + mutex_unlock(&inode->i_mutex); error = XFS_SEND_NAMESP(xip->i_mount, DM_EVENT_NOSPACE, vp, DM_RIGHT_NULL, vp, DM_RIGHT_NULL, NULL, NULL, 0, 0, 0); /* Delay flag intentionally unused */ if (error) - goto out_unlock_isem; + goto out_nounlocks; + if (need_i_mutex) + mutex_lock(&inode->i_mutex); xfs_rwlock(bdp, locktype); pos = xip->i_d.di_size; + ret = 0; goto retry; } + isize = i_size_read(inode); + if (unlikely(ret < 0 && ret != -EFAULT && *offset > isize)) + *offset = isize; + if (*offset > xip->i_d.di_size) { xfs_ilock(xip, XFS_ILOCK_EXCL); if (*offset > xip->i_d.di_size) { @@ -893,83 +907,13 @@ retry: /* Handle various SYNC-type writes */ if ((file->f_flags & O_SYNC) || IS_SYNC(inode)) { - /* - * If we're treating this as O_DSYNC and we have not updated the - * size, force the log. - */ - if (!(mp->m_flags & XFS_MOUNT_OSYNCISOSYNC) && - !(xip->i_update_size)) { - xfs_inode_log_item_t *iip = xip->i_itemp; - - /* - * If an allocation transaction occurred - * without extending the size, then we have to force - * the log up the proper point to ensure that the - * allocation is permanent. We can't count on - * the fact that buffered writes lock out direct I/O - * writes - the direct I/O write could have extended - * the size nontransactionally, then finished before - * we started. xfs_write_file will think that the file - * didn't grow but the update isn't safe unless the - * size change is logged. - * - * Force the log if we've committed a transaction - * against the inode or if someone else has and - * the commit record hasn't gone to disk (e.g. - * the inode is pinned). This guarantees that - * all changes affecting the inode are permanent - * when we return. - */ - if (iip && iip->ili_last_lsn) { - xfs_log_force(mp, iip->ili_last_lsn, - XFS_LOG_FORCE | XFS_LOG_SYNC); - } else if (xfs_ipincount(xip) > 0) { - xfs_log_force(mp, (xfs_lsn_t)0, - XFS_LOG_FORCE | XFS_LOG_SYNC); - } - - } else { - xfs_trans_t *tp; - - /* - * O_SYNC or O_DSYNC _with_ a size update are handled - * the same way. - * - * If the write was synchronous then we need to make - * sure that the inode modification time is permanent. - * We'll have updated the timestamp above, so here - * we use a synchronous transaction to log the inode. - * It's not fast, but it's necessary. - * - * If this a dsync write and the size got changed - * non-transactionally, then we need to ensure that - * the size change gets logged in a synchronous - * transaction. - */ + error = xfs_write_sync_logforce(mp, xip); + if (error) + goto out_unlock_internal; - tp = xfs_trans_alloc(mp, XFS_TRANS_WRITE_SYNC); - if ((error = xfs_trans_reserve(tp, 0, - XFS_SWRITE_LOG_RES(mp), - 0, 0, 0))) { - /* Transaction reserve failed */ - xfs_trans_cancel(tp, 0); - } else { - /* Transaction reserve successful */ - xfs_ilock(xip, XFS_ILOCK_EXCL); - xfs_trans_ijoin(tp, xip, XFS_ILOCK_EXCL); - xfs_trans_ihold(tp, xip); - xfs_trans_log_inode(tp, xip, XFS_ILOG_CORE); - xfs_trans_set_sync(tp); - error = xfs_trans_commit(tp, 0, NULL); - xfs_iunlock(xip, XFS_ILOCK_EXCL); - } - if (error) - goto out_unlock_internal; - } - xfs_rwunlock(bdp, locktype); - if (need_isem) - up(&inode->i_sem); + if (need_i_mutex) + mutex_unlock(&inode->i_mutex); error = sync_page_range(inode, mapping, pos, ret); if (!error) @@ -979,9 +923,10 @@ retry: out_unlock_internal: xfs_rwunlock(bdp, locktype); - out_unlock_isem: - if (need_isem) - up(&inode->i_sem); + out_unlock_mutex: + if (need_i_mutex) + mutex_unlock(&inode->i_mutex); + out_nounlocks: return -error; } @@ -998,7 +943,7 @@ xfs_bdstrat_cb(struct xfs_buf *bp) mp = XFS_BUF_FSPRIVATE3(bp, xfs_mount_t *); if (!XFS_FORCED_SHUTDOWN(mp)) { - pagebuf_iorequest(bp); + xfs_buf_iorequest(bp); return 0; } else { xfs_buftrace("XFS__BDSTRAT IOERROR", bp); @@ -1051,7 +996,7 @@ xfsbdstrat( * if (XFS_BUF_IS_GRIO(bp)) { */ - pagebuf_iorequest(bp); + xfs_buf_iorequest(bp); return 0; }