From a9cc799eca0c798ab5dd8648564fc2025bdd9bd2 Mon Sep 17 00:00:00 2001 From: Eric Sandeen Date: Wed, 3 Feb 2010 17:50:13 +0000 Subject: xfs: increase readdir buffer size While doing some testing of readdir perf a while back, I noticed that the buffer size we're using internally is smaller than what glibc gives us by default. Upping this size helped a bit, and seems safe. glibc's __alloc_dir() does: const size_t default_allocation = (4 * BUFSIZ < sizeof (struct dirent64) ? sizeof (struct dirent64) : 4 * BUFSIZ); const size_t small_allocation = (BUFSIZ < sizeof (struct dirent64) ? sizeof (struct dirent64) : BUFSIZ); size_t allocation = default_allocation; #ifdef _STATBUF_ST_BLKSIZE if (statp != NULL && default_allocation < statp->st_blksize) allocation = statp->st_blksize; #endif and #define _G_BUFSIZ 8192 #define _IO_BUFSIZ _G_BUFSIZ # define BUFSIZ _IO_BUFSIZ so the default buffer is 4 * 8192 = 32768 (except in the unlikely case of blocks > 32k....) Signed-off-by: Eric Sandeen Reviewed-by: Christoph Hellwig Signed-off-by: Alex Elder --- fs/xfs/linux-2.6/xfs_file.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) (limited to 'fs/xfs/linux-2.6/xfs_file.c') diff --git a/fs/xfs/linux-2.6/xfs_file.c b/fs/xfs/linux-2.6/xfs_file.c index e4caeb28ce2e..3805ada98747 100644 --- a/fs/xfs/linux-2.6/xfs_file.c +++ b/fs/xfs/linux-2.6/xfs_file.c @@ -203,9 +203,9 @@ xfs_file_readdir( * * Try to give it an estimate that's good enough, maybe at some * point we can change the ->readdir prototype to include the - * buffer size. + * buffer size. For now we use the current glibc buffer size. */ - bufsize = (size_t)min_t(loff_t, PAGE_SIZE, ip->i_d.di_size); + bufsize = (size_t)min_t(loff_t, 32768, ip->i_d.di_size); error = xfs_readdir(ip, dirent, bufsize, (xfs_off_t *)&filp->f_pos, filldir); -- cgit v1.2.3 From dda35b8f84d209784041bbad47f9e195a08a7527 Mon Sep 17 00:00:00 2001 From: Christoph Hellwig Date: Mon, 15 Feb 2010 09:44:46 +0000 Subject: xfs: merge xfs_lrw.c into xfs_file.c Currently the code to implement the file operations is split over two small files. Merge the content of xfs_lrw.c into xfs_file.c to have it in one place. Note that I haven't done various cleanups that are possible after this yet, they will follow in the next patch. Also the function xfs_dev_is_read_only which was in xfs_lrw.c before really doesn't fit in here at all and was moved to xfs_mount.c. Signed-off-by: Christoph Hellwig Reviewed-by: Dave Chinner Signed-off-by: Alex Elder --- fs/xfs/linux-2.6/xfs_file.c | 723 ++++++++++++++++++++++++++++++++++++++++++++ 1 file changed, 723 insertions(+) (limited to 'fs/xfs/linux-2.6/xfs_file.c') diff --git a/fs/xfs/linux-2.6/xfs_file.c b/fs/xfs/linux-2.6/xfs_file.c index 3805ada98747..51fc510828a4 100644 --- a/fs/xfs/linux-2.6/xfs_file.c +++ b/fs/xfs/linux-2.6/xfs_file.c @@ -16,6 +16,7 @@ * Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA */ #include "xfs.h" +#include "xfs_fs.h" #include "xfs_bit.h" #include "xfs_log.h" #include "xfs_inum.h" @@ -34,16 +35,738 @@ #include "xfs_dir2_sf.h" #include "xfs_dinode.h" #include "xfs_inode.h" +#include "xfs_bmap.h" #include "xfs_error.h" #include "xfs_rw.h" #include "xfs_vnodeops.h" #include "xfs_da_btree.h" #include "xfs_ioctl.h" +#include "xfs_trace.h" #include static const struct vm_operations_struct xfs_file_vm_ops; +/* + * xfs_iozero + * + * xfs_iozero clears the specified range of buffer supplied, + * and marks all the affected blocks as valid and modified. If + * an affected block is not allocated, it will be allocated. If + * an affected block is not completely overwritten, and is not + * valid before the operation, it will be read from disk before + * being partially zeroed. + */ +STATIC int +xfs_iozero( + struct xfs_inode *ip, /* inode */ + loff_t pos, /* offset in file */ + size_t count) /* size of data to zero */ +{ + struct page *page; + struct address_space *mapping; + int status; + + mapping = VFS_I(ip)->i_mapping; + do { + unsigned offset, bytes; + void *fsdata; + + offset = (pos & (PAGE_CACHE_SIZE -1)); /* Within page */ + bytes = PAGE_CACHE_SIZE - offset; + if (bytes > count) + bytes = count; + + status = pagecache_write_begin(NULL, mapping, pos, bytes, + AOP_FLAG_UNINTERRUPTIBLE, + &page, &fsdata); + if (status) + break; + + zero_user(page, offset, bytes); + + status = pagecache_write_end(NULL, mapping, pos, bytes, bytes, + page, fsdata); + WARN_ON(status <= 0); /* can't return less than zero! */ + pos += bytes; + count -= bytes; + status = 0; + } while (count); + + return (-status); +} + +ssize_t /* bytes read, or (-) error */ +xfs_read( + xfs_inode_t *ip, + struct kiocb *iocb, + const struct iovec *iovp, + unsigned int segs, + loff_t *offset, + int ioflags) +{ + struct file *file = iocb->ki_filp; + struct inode *inode = file->f_mapping->host; + xfs_mount_t *mp = ip->i_mount; + size_t size = 0; + ssize_t ret = 0; + xfs_fsize_t n; + unsigned long seg; + + + XFS_STATS_INC(xs_read_calls); + + /* START copy & waste from filemap.c */ + for (seg = 0; seg < segs; seg++) { + const struct iovec *iv = &iovp[seg]; + + /* + * If any segment has a negative length, or the cumulative + * length ever wraps negative then return -EINVAL. + */ + size += iv->iov_len; + if (unlikely((ssize_t)(size|iv->iov_len) < 0)) + return XFS_ERROR(-EINVAL); + } + /* END copy & waste from filemap.c */ + + if (unlikely(ioflags & IO_ISDIRECT)) { + xfs_buftarg_t *target = + XFS_IS_REALTIME_INODE(ip) ? + mp->m_rtdev_targp : mp->m_ddev_targp; + if ((*offset & target->bt_smask) || + (size & target->bt_smask)) { + if (*offset == ip->i_size) { + return (0); + } + return -XFS_ERROR(EINVAL); + } + } + + n = XFS_MAXIOFFSET(mp) - *offset; + if ((n <= 0) || (size == 0)) + return 0; + + if (n < size) + size = n; + + if (XFS_FORCED_SHUTDOWN(mp)) + return -EIO; + + if (unlikely(ioflags & IO_ISDIRECT)) + mutex_lock(&inode->i_mutex); + xfs_ilock(ip, XFS_IOLOCK_SHARED); + + if (DM_EVENT_ENABLED(ip, DM_EVENT_READ) && !(ioflags & IO_INVIS)) { + int dmflags = FILP_DELAY_FLAG(file) | DM_SEM_FLAG_RD(ioflags); + int iolock = XFS_IOLOCK_SHARED; + + ret = -XFS_SEND_DATA(mp, DM_EVENT_READ, ip, *offset, size, + dmflags, &iolock); + if (ret) { + xfs_iunlock(ip, XFS_IOLOCK_SHARED); + if (unlikely(ioflags & IO_ISDIRECT)) + mutex_unlock(&inode->i_mutex); + return ret; + } + } + + if (unlikely(ioflags & IO_ISDIRECT)) { + if (inode->i_mapping->nrpages) + ret = -xfs_flushinval_pages(ip, (*offset & PAGE_CACHE_MASK), + -1, FI_REMAPF_LOCKED); + mutex_unlock(&inode->i_mutex); + if (ret) { + xfs_iunlock(ip, XFS_IOLOCK_SHARED); + return ret; + } + } + + trace_xfs_file_read(ip, size, *offset, ioflags); + + iocb->ki_pos = *offset; + ret = generic_file_aio_read(iocb, iovp, segs, *offset); + if (ret > 0) + XFS_STATS_ADD(xs_read_bytes, ret); + + xfs_iunlock(ip, XFS_IOLOCK_SHARED); + return ret; +} + +ssize_t +xfs_splice_read( + xfs_inode_t *ip, + struct file *infilp, + loff_t *ppos, + struct pipe_inode_info *pipe, + size_t count, + int flags, + int ioflags) +{ + 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(ip, DM_EVENT_READ) && !(ioflags & IO_INVIS)) { + int iolock = XFS_IOLOCK_SHARED; + int error; + + error = XFS_SEND_DATA(mp, DM_EVENT_READ, ip, *ppos, count, + FILP_DELAY_FLAG(infilp), &iolock); + if (error) { + xfs_iunlock(ip, XFS_IOLOCK_SHARED); + return -error; + } + } + + trace_xfs_file_splice_read(ip, count, *ppos, ioflags); + + ret = generic_file_splice_read(infilp, ppos, pipe, count, flags); + if (ret > 0) + XFS_STATS_ADD(xs_read_bytes, ret); + + xfs_iunlock(ip, XFS_IOLOCK_SHARED); + return ret; +} + +ssize_t +xfs_splice_write( + xfs_inode_t *ip, + struct pipe_inode_info *pipe, + struct file *outfilp, + loff_t *ppos, + size_t count, + int flags, + int ioflags) +{ + xfs_mount_t *mp = ip->i_mount; + ssize_t ret; + struct inode *inode = outfilp->f_mapping->host; + xfs_fsize_t isize, new_size; + + 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(ip, DM_EVENT_WRITE) && !(ioflags & IO_INVIS)) { + int iolock = XFS_IOLOCK_EXCL; + int error; + + error = XFS_SEND_DATA(mp, DM_EVENT_WRITE, ip, *ppos, count, + FILP_DELAY_FLAG(outfilp), &iolock); + if (error) { + xfs_iunlock(ip, XFS_IOLOCK_EXCL); + return -error; + } + } + + new_size = *ppos + count; + + xfs_ilock(ip, XFS_ILOCK_EXCL); + if (new_size > ip->i_size) + ip->i_new_size = new_size; + xfs_iunlock(ip, XFS_ILOCK_EXCL); + + trace_xfs_file_splice_write(ip, 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_size) { + xfs_ilock(ip, XFS_ILOCK_EXCL); + if (*ppos > ip->i_size) + ip->i_size = *ppos; + xfs_iunlock(ip, XFS_ILOCK_EXCL); + } + + if (ip->i_new_size) { + xfs_ilock(ip, XFS_ILOCK_EXCL); + ip->i_new_size = 0; + if (ip->i_d.di_size > ip->i_size) + ip->i_d.di_size = ip->i_size; + xfs_iunlock(ip, XFS_ILOCK_EXCL); + } + xfs_iunlock(ip, XFS_IOLOCK_EXCL); + return ret; +} + +/* + * This routine is called to handle zeroing any space in the last + * block of the file that is beyond the EOF. We do this since the + * size is being increased without writing anything to that block + * and we don't want anyone to read the garbage on the disk. + */ +STATIC int /* error (positive) */ +xfs_zero_last_block( + xfs_inode_t *ip, + xfs_fsize_t offset, + xfs_fsize_t isize) +{ + xfs_fileoff_t last_fsb; + xfs_mount_t *mp = ip->i_mount; + int nimaps; + int zero_offset; + int zero_len; + int error = 0; + xfs_bmbt_irec_t imap; + + ASSERT(xfs_isilocked(ip, XFS_ILOCK_EXCL)); + + 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. + */ + return 0; + } + + last_fsb = XFS_B_TO_FSBT(mp, isize); + nimaps = 1; + error = xfs_bmapi(NULL, ip, last_fsb, 1, 0, NULL, 0, &imap, + &nimaps, NULL, NULL); + if (error) { + return error; + } + ASSERT(nimaps > 0); + /* + * If the block underlying isize is just a hole, then there + * is nothing to zero. + */ + if (imap.br_startblock == HOLESTARTBLOCK) { + return 0; + } + /* + * Zero the part of the last block beyond the EOF, and write it + * out sync. We need to drop the ilock while we do this so we + * don't deadlock when the buffer cache calls back to us. + */ + xfs_iunlock(ip, XFS_ILOCK_EXCL); + + zero_len = mp->m_sb.sb_blocksize - zero_offset; + if (isize + zero_len > offset) + zero_len = offset - isize; + error = xfs_iozero(ip, isize, zero_len); + + xfs_ilock(ip, XFS_ILOCK_EXCL); + ASSERT(error >= 0); + return error; +} + +/* + * Zero any on disk space between the current EOF and the new, + * larger EOF. This handles the normal case of zeroing the remainder + * of the last block in the file and the unusual case of zeroing blocks + * out beyond the size of the file. This second case only happens + * with fixed size extents and when the system crashes before the inode + * size was updated but after blocks were allocated. If fill is set, + * then any holes in the range are filled and zeroed. If not, the holes + * are left alone as holes. + */ + +int /* error (positive) */ +xfs_zero_eof( + xfs_inode_t *ip, + xfs_off_t offset, /* starting I/O offset */ + xfs_fsize_t isize) /* current inode size */ +{ + xfs_mount_t *mp = ip->i_mount; + xfs_fileoff_t start_zero_fsb; + xfs_fileoff_t end_zero_fsb; + xfs_fileoff_t zero_count_fsb; + xfs_fileoff_t last_fsb; + xfs_fileoff_t zero_off; + xfs_fsize_t zero_len; + int nimaps; + int error = 0; + xfs_bmbt_irec_t imap; + + ASSERT(xfs_isilocked(ip, XFS_ILOCK_EXCL|XFS_IOLOCK_EXCL)); + 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, offset, isize); + if (error) { + ASSERT(xfs_isilocked(ip, XFS_ILOCK_EXCL|XFS_IOLOCK_EXCL)); + return error; + } + + /* + * Calculate the range between the new size and the old + * where blocks needing to be zeroed may exist. To get the + * block where the last byte in the file currently resides, + * we need to subtract one from the size and truncate back + * to a block boundary. We subtract 1 in case the size is + * exactly on a block boundary. + */ + last_fsb = isize ? XFS_B_TO_FSBT(mp, isize - 1) : (xfs_fileoff_t)-1; + start_zero_fsb = XFS_B_TO_FSB(mp, (xfs_ufsize_t)isize); + end_zero_fsb = XFS_B_TO_FSBT(mp, offset - 1); + ASSERT((xfs_sfiloff_t)last_fsb < (xfs_sfiloff_t)start_zero_fsb); + if (last_fsb == end_zero_fsb) { + /* + * The size was only incremented on its last block. + * We took care of that above, so just return. + */ + return 0; + } + + ASSERT(start_zero_fsb <= end_zero_fsb); + while (start_zero_fsb <= end_zero_fsb) { + nimaps = 1; + zero_count_fsb = end_zero_fsb - start_zero_fsb + 1; + error = xfs_bmapi(NULL, ip, start_zero_fsb, zero_count_fsb, + 0, NULL, 0, &imap, &nimaps, NULL, NULL); + if (error) { + ASSERT(xfs_isilocked(ip, XFS_ILOCK_EXCL|XFS_IOLOCK_EXCL)); + return error; + } + ASSERT(nimaps > 0); + + if (imap.br_state == XFS_EXT_UNWRITTEN || + imap.br_startblock == HOLESTARTBLOCK) { + /* + * This loop handles initializing pages that were + * partially initialized by the code below this + * loop. It basically zeroes the part of the page + * that sits on a hole and sets the page as P_HOLE + * and calls remapf if it is a mapped file. + */ + start_zero_fsb = imap.br_startoff + imap.br_blockcount; + ASSERT(start_zero_fsb <= (end_zero_fsb + 1)); + continue; + } + + /* + * 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(ip, XFS_ILOCK_EXCL); + + zero_off = XFS_FSB_TO_B(mp, start_zero_fsb); + zero_len = XFS_FSB_TO_B(mp, imap.br_blockcount); + + if ((zero_off + zero_len) > offset) + zero_len = offset - zero_off; + + error = xfs_iozero(ip, zero_off, zero_len); + if (error) { + goto out_lock; + } + + start_zero_fsb = imap.br_startoff + imap.br_blockcount; + ASSERT(start_zero_fsb <= (end_zero_fsb + 1)); + + xfs_ilock(ip, XFS_ILOCK_EXCL); + } + + return 0; + +out_lock: + xfs_ilock(ip, XFS_ILOCK_EXCL); + ASSERT(error >= 0); + return error; +} + +ssize_t /* bytes written, or (-) error */ +xfs_write( + struct xfs_inode *xip, + struct kiocb *iocb, + const struct iovec *iovp, + unsigned int nsegs, + loff_t *offset, + int ioflags) +{ + struct file *file = iocb->ki_filp; + struct address_space *mapping = file->f_mapping; + struct inode *inode = mapping->host; + unsigned long segs = nsegs; + xfs_mount_t *mp; + ssize_t ret = 0, error = 0; + xfs_fsize_t isize, new_size; + int iolock; + int eventsent = 0; + size_t ocount = 0, count; + loff_t pos; + int need_i_mutex; + + XFS_STATS_INC(xs_write_calls); + + error = generic_segment_checks(iovp, &segs, &ocount, VERIFY_READ); + if (error) + return error; + + count = ocount; + pos = *offset; + + if (count == 0) + return 0; + + mp = xip->i_mount; + + xfs_wait_for_freeze(mp, SB_FREEZE_WRITE); + + if (XFS_FORCED_SHUTDOWN(mp)) + return -EIO; + +relock: + if (ioflags & IO_ISDIRECT) { + iolock = XFS_IOLOCK_SHARED; + need_i_mutex = 0; + } else { + iolock = XFS_IOLOCK_EXCL; + need_i_mutex = 1; + mutex_lock(&inode->i_mutex); + } + + xfs_ilock(xip, XFS_ILOCK_EXCL|iolock); + +start: + error = -generic_write_checks(file, &pos, &count, + S_ISBLK(inode->i_mode)); + if (error) { + xfs_iunlock(xip, XFS_ILOCK_EXCL|iolock); + goto out_unlock_mutex; + } + + if ((DM_EVENT_ENABLED(xip, DM_EVENT_WRITE) && + !(ioflags & IO_INVIS) && !eventsent)) { + int dmflags = FILP_DELAY_FLAG(file); + + if (need_i_mutex) + dmflags |= DM_FLAGS_IMUX; + + xfs_iunlock(xip, XFS_ILOCK_EXCL); + error = XFS_SEND_DATA(xip->i_mount, DM_EVENT_WRITE, xip, + pos, count, dmflags, &iolock); + if (error) { + goto out_unlock_internal; + } + xfs_ilock(xip, XFS_ILOCK_EXCL); + eventsent = 1; + + /* + * 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 + * what allows the size to change in the first place. + */ + if ((file->f_flags & O_APPEND) && pos != xip->i_size) + goto start; + } + + if (ioflags & IO_ISDIRECT) { + xfs_buftarg_t *target = + XFS_IS_REALTIME_INODE(xip) ? + mp->m_rtdev_targp : mp->m_ddev_targp; + + if ((pos & target->bt_smask) || (count & target->bt_smask)) { + xfs_iunlock(xip, XFS_ILOCK_EXCL|iolock); + return XFS_ERROR(-EINVAL); + } + + if (!need_i_mutex && (mapping->nrpages || pos > xip->i_size)) { + xfs_iunlock(xip, XFS_ILOCK_EXCL|iolock); + iolock = XFS_IOLOCK_EXCL; + need_i_mutex = 1; + mutex_lock(&inode->i_mutex); + xfs_ilock(xip, XFS_ILOCK_EXCL|iolock); + goto start; + } + } + + new_size = pos + count; + if (new_size > xip->i_size) + xip->i_new_size = new_size; + + if (likely(!(ioflags & IO_INVIS))) + file_update_time(file); + + /* + * If the offset is beyond the size of the file, we have a couple + * of things to do. First, if there is already space allocated + * we need to either create holes or zero the disk or ... + * + * If there is a page where the previous size lands, we need + * to zero it out up to the new size. + */ + + if (pos > xip->i_size) { + error = xfs_zero_eof(xip, pos, xip->i_size); + if (error) { + xfs_iunlock(xip, XFS_ILOCK_EXCL); + goto out_unlock_internal; + } + } + xfs_iunlock(xip, XFS_ILOCK_EXCL); + + /* + * If we're writing the file then make sure to clear the + * setuid and setgid bits if the process is not being run + * by root. This keeps people from modifying setuid and + * setgid binaries. + */ + error = -file_remove_suid(file); + if (unlikely(error)) + goto out_unlock_internal; + + /* We can write back this queue in page reclaim */ + current->backing_dev_info = mapping->backing_dev_info; + + if ((ioflags & IO_ISDIRECT)) { + if (mapping->nrpages) { + WARN_ON(need_i_mutex == 0); + error = xfs_flushinval_pages(xip, + (pos & PAGE_CACHE_MASK), + -1, FI_REMAPF_LOCKED); + if (error) + goto out_unlock_internal; + } + + if (need_i_mutex) { + /* demote the lock now the cached pages are gone */ + xfs_ilock_demote(xip, XFS_IOLOCK_EXCL); + mutex_unlock(&inode->i_mutex); + + iolock = XFS_IOLOCK_SHARED; + need_i_mutex = 0; + } + + trace_xfs_file_direct_write(xip, count, *offset, ioflags); + ret = generic_file_direct_write(iocb, iovp, + &segs, pos, offset, count, ocount); + + /* + * direct-io write to a hole: fall through to buffered I/O + * for completing the rest of the request. + */ + if (ret >= 0 && ret != count) { + XFS_STATS_ADD(xs_write_bytes, ret); + + pos += ret; + count -= ret; + + ioflags &= ~IO_ISDIRECT; + xfs_iunlock(xip, iolock); + goto relock; + } + } else { + int enospc = 0; + ssize_t ret2 = 0; + +write_retry: + trace_xfs_file_buffered_write(xip, count, *offset, ioflags); + ret2 = generic_file_buffered_write(iocb, iovp, segs, + pos, offset, count, ret); + /* + * if we just got an ENOSPC, flush the inode now we + * aren't holding any page locks and retry *once* + */ + if (ret2 == -ENOSPC && !enospc) { + error = xfs_flush_pages(xip, 0, -1, 0, FI_NONE); + if (error) + goto out_unlock_internal; + enospc = 1; + goto write_retry; + } + ret = ret2; + } + + current->backing_dev_info = NULL; + + isize = i_size_read(inode); + if (unlikely(ret < 0 && ret != -EFAULT && *offset > isize)) + *offset = isize; + + if (*offset > xip->i_size) { + xfs_ilock(xip, XFS_ILOCK_EXCL); + if (*offset > xip->i_size) + xip->i_size = *offset; + xfs_iunlock(xip, XFS_ILOCK_EXCL); + } + + if (ret == -ENOSPC && + DM_EVENT_ENABLED(xip, DM_EVENT_NOSPACE) && !(ioflags & IO_INVIS)) { + xfs_iunlock(xip, iolock); + if (need_i_mutex) + mutex_unlock(&inode->i_mutex); + error = XFS_SEND_NAMESP(xip->i_mount, DM_EVENT_NOSPACE, xip, + DM_RIGHT_NULL, xip, DM_RIGHT_NULL, NULL, NULL, + 0, 0, 0); /* Delay flag intentionally unused */ + if (need_i_mutex) + mutex_lock(&inode->i_mutex); + xfs_ilock(xip, iolock); + if (error) + goto out_unlock_internal; + goto start; + } + + error = -ret; + if (ret <= 0) + goto out_unlock_internal; + + XFS_STATS_ADD(xs_write_bytes, ret); + + /* Handle various SYNC-type writes */ + if ((file->f_flags & O_DSYNC) || IS_SYNC(inode)) { + loff_t end = pos + ret - 1; + int error2; + + xfs_iunlock(xip, iolock); + if (need_i_mutex) + mutex_unlock(&inode->i_mutex); + + error2 = filemap_write_and_wait_range(mapping, pos, end); + if (!error) + error = error2; + if (need_i_mutex) + mutex_lock(&inode->i_mutex); + xfs_ilock(xip, iolock); + + error2 = xfs_fsync(xip); + if (!error) + error = error2; + } + + out_unlock_internal: + if (xip->i_new_size) { + xfs_ilock(xip, XFS_ILOCK_EXCL); + xip->i_new_size = 0; + /* + * If this was a direct or synchronous I/O that failed (such + * as ENOSPC) then part of the I/O may have been written to + * disk before the error occured. In this case the on-disk + * file size may have been adjusted beyond the in-memory file + * size and now needs to be truncated back. + */ + if (xip->i_d.di_size > xip->i_size) + xip->i_d.di_size = xip->i_size; + xfs_iunlock(xip, XFS_ILOCK_EXCL); + } + xfs_iunlock(xip, iolock); + out_unlock_mutex: + if (need_i_mutex) + mutex_unlock(&inode->i_mutex); + return -error; +} + STATIC ssize_t xfs_file_aio_read( struct kiocb *iocb, -- cgit v1.2.3 From 00258e36b2d33b1b5cef7b489e06c5e0a9df58b5 Mon Sep 17 00:00:00 2001 From: Christoph Hellwig Date: Mon, 15 Feb 2010 09:44:47 +0000 Subject: xfs: remove wrappers for read/write file operations Currently the aio_read, aio_write, splice_read and splice_write file operations are divided into a low-level routine doing all the work and one that implements the Linux file operations and does minimal argument wrapping. This is a leftover from the days of the vnode operations layer and can be removed to simplify the code a lot. Signed-off-by: Christoph Hellwig Reviewed-by: Dave Chinner Signed-off-by: Alex Elder --- fs/xfs/linux-2.6/xfs_file.c | 285 ++++++++++++++++++-------------------------- 1 file changed, 114 insertions(+), 171 deletions(-) (limited to 'fs/xfs/linux-2.6/xfs_file.c') diff --git a/fs/xfs/linux-2.6/xfs_file.c b/fs/xfs/linux-2.6/xfs_file.c index 51fc510828a4..1eb561a10e26 100644 --- a/fs/xfs/linux-2.6/xfs_file.c +++ b/fs/xfs/linux-2.6/xfs_file.c @@ -96,28 +96,34 @@ xfs_iozero( return (-status); } -ssize_t /* bytes read, or (-) error */ -xfs_read( - xfs_inode_t *ip, +STATIC ssize_t +xfs_file_aio_read( struct kiocb *iocb, const struct iovec *iovp, - unsigned int segs, - loff_t *offset, - int ioflags) + unsigned long nr_segs, + loff_t pos) { struct file *file = iocb->ki_filp; struct inode *inode = file->f_mapping->host; - xfs_mount_t *mp = ip->i_mount; + struct xfs_inode *ip = XFS_I(inode); + struct xfs_mount *mp = ip->i_mount; size_t size = 0; ssize_t ret = 0; + int ioflags = 0; xfs_fsize_t n; unsigned long seg; - XFS_STATS_INC(xs_read_calls); + BUG_ON(iocb->ki_pos != pos); + + if (unlikely(file->f_flags & O_DIRECT)) + ioflags |= IO_ISDIRECT; + if (file->f_mode & FMODE_NOCMTIME) + ioflags |= IO_INVIS; + /* START copy & waste from filemap.c */ - for (seg = 0; seg < segs; seg++) { + for (seg = 0; seg < nr_segs; seg++) { const struct iovec *iv = &iovp[seg]; /* @@ -134,17 +140,16 @@ xfs_read( xfs_buftarg_t *target = XFS_IS_REALTIME_INODE(ip) ? mp->m_rtdev_targp : mp->m_ddev_targp; - if ((*offset & target->bt_smask) || + if ((iocb->ki_pos & target->bt_smask) || (size & target->bt_smask)) { - if (*offset == ip->i_size) { - return (0); - } + if (iocb->ki_pos == ip->i_size) + return 0; return -XFS_ERROR(EINVAL); } } - n = XFS_MAXIOFFSET(mp) - *offset; - if ((n <= 0) || (size == 0)) + n = XFS_MAXIOFFSET(mp) - iocb->ki_pos; + if (n <= 0 || size == 0) return 0; if (n < size) @@ -161,7 +166,7 @@ xfs_read( int dmflags = FILP_DELAY_FLAG(file) | DM_SEM_FLAG_RD(ioflags); int iolock = XFS_IOLOCK_SHARED; - ret = -XFS_SEND_DATA(mp, DM_EVENT_READ, ip, *offset, size, + ret = -XFS_SEND_DATA(mp, DM_EVENT_READ, ip, iocb->ki_pos, size, dmflags, &iolock); if (ret) { xfs_iunlock(ip, XFS_IOLOCK_SHARED); @@ -172,9 +177,11 @@ xfs_read( } if (unlikely(ioflags & IO_ISDIRECT)) { - if (inode->i_mapping->nrpages) - ret = -xfs_flushinval_pages(ip, (*offset & PAGE_CACHE_MASK), - -1, FI_REMAPF_LOCKED); + if (inode->i_mapping->nrpages) { + ret = -xfs_flushinval_pages(ip, + (iocb->ki_pos & PAGE_CACHE_MASK), + -1, FI_REMAPF_LOCKED); + } mutex_unlock(&inode->i_mutex); if (ret) { xfs_iunlock(ip, XFS_IOLOCK_SHARED); @@ -182,10 +189,9 @@ xfs_read( } } - trace_xfs_file_read(ip, size, *offset, ioflags); + trace_xfs_file_read(ip, size, iocb->ki_pos, ioflags); - iocb->ki_pos = *offset; - ret = generic_file_aio_read(iocb, iovp, segs, *offset); + ret = generic_file_aio_read(iocb, iovp, nr_segs, iocb->ki_pos); if (ret > 0) XFS_STATS_ADD(xs_read_bytes, ret); @@ -193,20 +199,24 @@ xfs_read( return ret; } -ssize_t -xfs_splice_read( - xfs_inode_t *ip, +STATIC ssize_t +xfs_file_splice_read( struct file *infilp, loff_t *ppos, struct pipe_inode_info *pipe, size_t count, - int flags, - int ioflags) + unsigned int flags) { - xfs_mount_t *mp = ip->i_mount; + struct xfs_inode *ip = XFS_I(infilp->f_mapping->host); + struct xfs_mount *mp = ip->i_mount; + int ioflags = 0; ssize_t ret; XFS_STATS_INC(xs_read_calls); + + if (infilp->f_mode & FMODE_NOCMTIME) + ioflags |= IO_INVIS; + if (XFS_FORCED_SHUTDOWN(ip->i_mount)) return -EIO; @@ -234,22 +244,26 @@ xfs_splice_read( return ret; } -ssize_t -xfs_splice_write( - xfs_inode_t *ip, +STATIC ssize_t +xfs_file_splice_write( struct pipe_inode_info *pipe, struct file *outfilp, loff_t *ppos, size_t count, - int flags, - int ioflags) + unsigned int flags) { - xfs_mount_t *mp = ip->i_mount; - ssize_t ret; struct inode *inode = outfilp->f_mapping->host; + struct xfs_inode *ip = XFS_I(inode); + struct xfs_mount *mp = ip->i_mount; xfs_fsize_t isize, new_size; + int ioflags = 0; + ssize_t ret; XFS_STATS_INC(xs_write_calls); + + if (outfilp->f_mode & FMODE_NOCMTIME) + ioflags |= IO_INVIS; + if (XFS_FORCED_SHUTDOWN(ip->i_mount)) return -EIO; @@ -484,42 +498,43 @@ out_lock: return error; } -ssize_t /* bytes written, or (-) error */ -xfs_write( - struct xfs_inode *xip, +STATIC ssize_t +xfs_file_aio_write( struct kiocb *iocb, const struct iovec *iovp, - unsigned int nsegs, - loff_t *offset, - int ioflags) + unsigned long nr_segs, + loff_t pos) { struct file *file = iocb->ki_filp; struct address_space *mapping = file->f_mapping; struct inode *inode = mapping->host; - unsigned long segs = nsegs; - xfs_mount_t *mp; + struct xfs_inode *ip = XFS_I(inode); + struct xfs_mount *mp = ip->i_mount; ssize_t ret = 0, error = 0; + int ioflags = 0; xfs_fsize_t isize, new_size; int iolock; int eventsent = 0; size_t ocount = 0, count; - loff_t pos; int need_i_mutex; XFS_STATS_INC(xs_write_calls); - error = generic_segment_checks(iovp, &segs, &ocount, VERIFY_READ); + BUG_ON(iocb->ki_pos != pos); + + if (unlikely(file->f_flags & O_DIRECT)) + ioflags |= IO_ISDIRECT; + if (file->f_mode & FMODE_NOCMTIME) + ioflags |= IO_INVIS; + + error = generic_segment_checks(iovp, &nr_segs, &ocount, VERIFY_READ); if (error) return error; count = ocount; - pos = *offset; - if (count == 0) return 0; - mp = xip->i_mount; - xfs_wait_for_freeze(mp, SB_FREEZE_WRITE); if (XFS_FORCED_SHUTDOWN(mp)) @@ -535,30 +550,30 @@ relock: mutex_lock(&inode->i_mutex); } - xfs_ilock(xip, XFS_ILOCK_EXCL|iolock); + xfs_ilock(ip, XFS_ILOCK_EXCL|iolock); start: error = -generic_write_checks(file, &pos, &count, S_ISBLK(inode->i_mode)); if (error) { - xfs_iunlock(xip, XFS_ILOCK_EXCL|iolock); + xfs_iunlock(ip, XFS_ILOCK_EXCL|iolock); goto out_unlock_mutex; } - if ((DM_EVENT_ENABLED(xip, DM_EVENT_WRITE) && + if ((DM_EVENT_ENABLED(ip, DM_EVENT_WRITE) && !(ioflags & IO_INVIS) && !eventsent)) { int dmflags = FILP_DELAY_FLAG(file); if (need_i_mutex) dmflags |= DM_FLAGS_IMUX; - xfs_iunlock(xip, XFS_ILOCK_EXCL); - error = XFS_SEND_DATA(xip->i_mount, DM_EVENT_WRITE, xip, + xfs_iunlock(ip, XFS_ILOCK_EXCL); + error = XFS_SEND_DATA(ip->i_mount, DM_EVENT_WRITE, ip, pos, count, dmflags, &iolock); if (error) { goto out_unlock_internal; } - xfs_ilock(xip, XFS_ILOCK_EXCL); + xfs_ilock(ip, XFS_ILOCK_EXCL); eventsent = 1; /* @@ -568,33 +583,33 @@ start: * event prevents another call to XFS_SEND_DATA, which is * what allows the size to change in the first place. */ - if ((file->f_flags & O_APPEND) && pos != xip->i_size) + if ((file->f_flags & O_APPEND) && pos != ip->i_size) goto start; } if (ioflags & IO_ISDIRECT) { xfs_buftarg_t *target = - XFS_IS_REALTIME_INODE(xip) ? + XFS_IS_REALTIME_INODE(ip) ? mp->m_rtdev_targp : mp->m_ddev_targp; if ((pos & target->bt_smask) || (count & target->bt_smask)) { - xfs_iunlock(xip, XFS_ILOCK_EXCL|iolock); + xfs_iunlock(ip, XFS_ILOCK_EXCL|iolock); return XFS_ERROR(-EINVAL); } - if (!need_i_mutex && (mapping->nrpages || pos > xip->i_size)) { - xfs_iunlock(xip, XFS_ILOCK_EXCL|iolock); + if (!need_i_mutex && (mapping->nrpages || pos > ip->i_size)) { + xfs_iunlock(ip, XFS_ILOCK_EXCL|iolock); iolock = XFS_IOLOCK_EXCL; need_i_mutex = 1; mutex_lock(&inode->i_mutex); - xfs_ilock(xip, XFS_ILOCK_EXCL|iolock); + xfs_ilock(ip, XFS_ILOCK_EXCL|iolock); goto start; } } new_size = pos + count; - if (new_size > xip->i_size) - xip->i_new_size = new_size; + if (new_size > ip->i_size) + ip->i_new_size = new_size; if (likely(!(ioflags & IO_INVIS))) file_update_time(file); @@ -608,14 +623,14 @@ start: * to zero it out up to the new size. */ - if (pos > xip->i_size) { - error = xfs_zero_eof(xip, pos, xip->i_size); + if (pos > ip->i_size) { + error = xfs_zero_eof(ip, pos, ip->i_size); if (error) { - xfs_iunlock(xip, XFS_ILOCK_EXCL); + xfs_iunlock(ip, XFS_ILOCK_EXCL); goto out_unlock_internal; } } - xfs_iunlock(xip, XFS_ILOCK_EXCL); + xfs_iunlock(ip, XFS_ILOCK_EXCL); /* * If we're writing the file then make sure to clear the @@ -633,7 +648,7 @@ start: if ((ioflags & IO_ISDIRECT)) { if (mapping->nrpages) { WARN_ON(need_i_mutex == 0); - error = xfs_flushinval_pages(xip, + error = xfs_flushinval_pages(ip, (pos & PAGE_CACHE_MASK), -1, FI_REMAPF_LOCKED); if (error) @@ -642,16 +657,16 @@ start: if (need_i_mutex) { /* demote the lock now the cached pages are gone */ - xfs_ilock_demote(xip, XFS_IOLOCK_EXCL); + xfs_ilock_demote(ip, XFS_IOLOCK_EXCL); mutex_unlock(&inode->i_mutex); iolock = XFS_IOLOCK_SHARED; need_i_mutex = 0; } - trace_xfs_file_direct_write(xip, count, *offset, ioflags); + trace_xfs_file_direct_write(ip, count, iocb->ki_pos, ioflags); ret = generic_file_direct_write(iocb, iovp, - &segs, pos, offset, count, ocount); + &nr_segs, pos, &iocb->ki_pos, count, ocount); /* * direct-io write to a hole: fall through to buffered I/O @@ -664,7 +679,7 @@ start: count -= ret; ioflags &= ~IO_ISDIRECT; - xfs_iunlock(xip, iolock); + xfs_iunlock(ip, iolock); goto relock; } } else { @@ -672,15 +687,15 @@ start: ssize_t ret2 = 0; write_retry: - trace_xfs_file_buffered_write(xip, count, *offset, ioflags); - ret2 = generic_file_buffered_write(iocb, iovp, segs, - pos, offset, count, ret); + trace_xfs_file_buffered_write(ip, count, iocb->ki_pos, ioflags); + ret2 = generic_file_buffered_write(iocb, iovp, nr_segs, + pos, &iocb->ki_pos, count, ret); /* * if we just got an ENOSPC, flush the inode now we * aren't holding any page locks and retry *once* */ if (ret2 == -ENOSPC && !enospc) { - error = xfs_flush_pages(xip, 0, -1, 0, FI_NONE); + error = xfs_flush_pages(ip, 0, -1, 0, FI_NONE); if (error) goto out_unlock_internal; enospc = 1; @@ -692,27 +707,27 @@ write_retry: current->backing_dev_info = NULL; isize = i_size_read(inode); - if (unlikely(ret < 0 && ret != -EFAULT && *offset > isize)) - *offset = isize; - - if (*offset > xip->i_size) { - xfs_ilock(xip, XFS_ILOCK_EXCL); - if (*offset > xip->i_size) - xip->i_size = *offset; - xfs_iunlock(xip, XFS_ILOCK_EXCL); + if (unlikely(ret < 0 && ret != -EFAULT && iocb->ki_pos > isize)) + iocb->ki_pos = isize; + + if (iocb->ki_pos > ip->i_size) { + xfs_ilock(ip, XFS_ILOCK_EXCL); + if (iocb->ki_pos > ip->i_size) + ip->i_size = iocb->ki_pos; + xfs_iunlock(ip, XFS_ILOCK_EXCL); } if (ret == -ENOSPC && - DM_EVENT_ENABLED(xip, DM_EVENT_NOSPACE) && !(ioflags & IO_INVIS)) { - xfs_iunlock(xip, iolock); + DM_EVENT_ENABLED(ip, DM_EVENT_NOSPACE) && !(ioflags & IO_INVIS)) { + xfs_iunlock(ip, iolock); if (need_i_mutex) mutex_unlock(&inode->i_mutex); - error = XFS_SEND_NAMESP(xip->i_mount, DM_EVENT_NOSPACE, xip, - DM_RIGHT_NULL, xip, DM_RIGHT_NULL, NULL, NULL, + error = XFS_SEND_NAMESP(ip->i_mount, DM_EVENT_NOSPACE, ip, + DM_RIGHT_NULL, ip, DM_RIGHT_NULL, NULL, NULL, 0, 0, 0); /* Delay flag intentionally unused */ if (need_i_mutex) mutex_lock(&inode->i_mutex); - xfs_ilock(xip, iolock); + xfs_ilock(ip, iolock); if (error) goto out_unlock_internal; goto start; @@ -729,7 +744,7 @@ write_retry: loff_t end = pos + ret - 1; int error2; - xfs_iunlock(xip, iolock); + xfs_iunlock(ip, iolock); if (need_i_mutex) mutex_unlock(&inode->i_mutex); @@ -738,17 +753,17 @@ write_retry: error = error2; if (need_i_mutex) mutex_lock(&inode->i_mutex); - xfs_ilock(xip, iolock); + xfs_ilock(ip, iolock); - error2 = xfs_fsync(xip); + error2 = xfs_fsync(ip); if (!error) error = error2; } out_unlock_internal: - if (xip->i_new_size) { - xfs_ilock(xip, XFS_ILOCK_EXCL); - xip->i_new_size = 0; + if (ip->i_new_size) { + xfs_ilock(ip, XFS_ILOCK_EXCL); + ip->i_new_size = 0; /* * If this was a direct or synchronous I/O that failed (such * as ENOSPC) then part of the I/O may have been written to @@ -756,89 +771,17 @@ write_retry: * file size may have been adjusted beyond the in-memory file * size and now needs to be truncated back. */ - if (xip->i_d.di_size > xip->i_size) - xip->i_d.di_size = xip->i_size; - xfs_iunlock(xip, XFS_ILOCK_EXCL); + if (ip->i_d.di_size > ip->i_size) + ip->i_d.di_size = ip->i_size; + xfs_iunlock(ip, XFS_ILOCK_EXCL); } - xfs_iunlock(xip, iolock); + xfs_iunlock(ip, iolock); out_unlock_mutex: if (need_i_mutex) mutex_unlock(&inode->i_mutex); return -error; } -STATIC ssize_t -xfs_file_aio_read( - struct kiocb *iocb, - const struct iovec *iov, - unsigned long nr_segs, - loff_t pos) -{ - struct file *file = iocb->ki_filp; - int ioflags = 0; - - BUG_ON(iocb->ki_pos != pos); - if (unlikely(file->f_flags & O_DIRECT)) - ioflags |= IO_ISDIRECT; - if (file->f_mode & FMODE_NOCMTIME) - ioflags |= IO_INVIS; - return xfs_read(XFS_I(file->f_path.dentry->d_inode), iocb, iov, - nr_segs, &iocb->ki_pos, ioflags); -} - -STATIC ssize_t -xfs_file_aio_write( - struct kiocb *iocb, - const struct iovec *iov, - unsigned long nr_segs, - loff_t pos) -{ - struct file *file = iocb->ki_filp; - int ioflags = 0; - - BUG_ON(iocb->ki_pos != pos); - if (unlikely(file->f_flags & O_DIRECT)) - ioflags |= IO_ISDIRECT; - if (file->f_mode & FMODE_NOCMTIME) - ioflags |= IO_INVIS; - return xfs_write(XFS_I(file->f_mapping->host), iocb, iov, nr_segs, - &iocb->ki_pos, ioflags); -} - -STATIC ssize_t -xfs_file_splice_read( - struct file *infilp, - loff_t *ppos, - struct pipe_inode_info *pipe, - size_t len, - unsigned int flags) -{ - int ioflags = 0; - - if (infilp->f_mode & FMODE_NOCMTIME) - ioflags |= IO_INVIS; - - return xfs_splice_read(XFS_I(infilp->f_path.dentry->d_inode), - infilp, ppos, pipe, len, flags, ioflags); -} - -STATIC ssize_t -xfs_file_splice_write( - struct pipe_inode_info *pipe, - struct file *outfilp, - loff_t *ppos, - size_t len, - unsigned int flags) -{ - int ioflags = 0; - - if (outfilp->f_mode & FMODE_NOCMTIME) - ioflags |= IO_INVIS; - - return xfs_splice_write(XFS_I(outfilp->f_path.dentry->d_inode), - pipe, outfilp, ppos, len, flags, ioflags); -} - STATIC int xfs_file_open( struct inode *inode, -- cgit v1.2.3 From fd3200bef7d66ed3924f72c79a465fb7ff85478a Mon Sep 17 00:00:00 2001 From: Christoph Hellwig Date: Mon, 15 Feb 2010 09:44:48 +0000 Subject: xfs: remove wrapper for the fsync file operation Currently the fsync file operation is divided into a low-level routine doing all the work and one that implements the Linux file operation and does minimal argument wrapping. This is a leftover from the days of the vnode operations layer and can be removed to simplify the code a bit, as well as preparing for the implementation of an optimized fdatasync which needs to look at the Linux inode state. Signed-off-by: Christoph Hellwig Signed-off-by: Alex Elder --- fs/xfs/linux-2.6/xfs_file.c | 140 ++++++++++++++++++++++++++++++++++++-------- 1 file changed, 117 insertions(+), 23 deletions(-) (limited to 'fs/xfs/linux-2.6/xfs_file.c') diff --git a/fs/xfs/linux-2.6/xfs_file.c b/fs/xfs/linux-2.6/xfs_file.c index 1eb561a10e26..6c283b7be8ab 100644 --- a/fs/xfs/linux-2.6/xfs_file.c +++ b/fs/xfs/linux-2.6/xfs_file.c @@ -35,6 +35,7 @@ #include "xfs_dir2_sf.h" #include "xfs_dinode.h" #include "xfs_inode.h" +#include "xfs_inode_item.h" #include "xfs_bmap.h" #include "xfs_error.h" #include "xfs_rw.h" @@ -96,6 +97,120 @@ xfs_iozero( return (-status); } +/* + * We ignore the datasync flag here because a datasync is effectively + * identical to an fsync. That is, datasync implies that we need to write + * only the metadata needed to be able to access the data that is written + * if we crash after the call completes. Hence if we are writing beyond + * EOF we have to log the inode size change as well, which makes it a + * full fsync. If we don't write beyond EOF, the inode core will be + * clean in memory and so we don't need to log the inode, just like + * fsync. + */ +STATIC int +xfs_file_fsync( + struct file *file, + struct dentry *dentry, + int datasync) +{ + struct xfs_inode *ip = XFS_I(dentry->d_inode); + struct xfs_trans *tp; + int error = 0; + int log_flushed = 0; + + xfs_itrace_entry(ip); + + if (XFS_FORCED_SHUTDOWN(ip->i_mount)) + return -XFS_ERROR(EIO); + + xfs_iflags_clear(ip, XFS_ITRUNCATED); + + /* + * We always need to make sure that the required inode state is safe on + * disk. The inode might be clean but we still might need to force the + * log because of committed transactions that haven't hit the disk yet. + * Likewise, there could be unflushed non-transactional changes to the + * inode core that have to go to disk and this requires us to issue + * a synchronous transaction to capture these changes correctly. + * + * This code relies on the assumption that if the i_update_core field + * of the inode is clear and the inode is unpinned then it is clean + * and no action is required. + */ + xfs_ilock(ip, XFS_ILOCK_SHARED); + + if (ip->i_update_core) { + /* + * Kick off a transaction to log the inode core to get the + * updates. The sync transaction will also force the log. + */ + xfs_iunlock(ip, XFS_ILOCK_SHARED); + tp = xfs_trans_alloc(ip->i_mount, XFS_TRANS_FSYNC_TS); + error = xfs_trans_reserve(tp, 0, + XFS_FSYNC_TS_LOG_RES(ip->i_mount), 0, 0, 0); + if (error) { + xfs_trans_cancel(tp, 0); + return -error; + } + xfs_ilock(ip, XFS_ILOCK_EXCL); + + /* + * Note - it's possible that we might have pushed ourselves out + * of the way during trans_reserve which would flush the inode. + * But there's no guarantee that the inode buffer has actually + * gone out yet (it's delwri). Plus the buffer could be pinned + * anyway if it's part of an inode in another recent + * transaction. So we play it safe and fire off the + * transaction anyway. + */ + xfs_trans_ijoin(tp, ip, XFS_ILOCK_EXCL); + xfs_trans_ihold(tp, ip); + xfs_trans_log_inode(tp, ip, XFS_ILOG_CORE); + xfs_trans_set_sync(tp); + error = _xfs_trans_commit(tp, 0, &log_flushed); + + xfs_iunlock(ip, XFS_ILOCK_EXCL); + } else { + /* + * Timestamps/size haven't changed since last inode flush or + * inode transaction commit. That means either nothing got + * written or a transaction committed which caught the updates. + * If the latter happened and the transaction hasn't hit the + * disk yet, the inode will be still be pinned. If it is, + * force the log. + */ + xfs_iunlock(ip, XFS_ILOCK_SHARED); + if (xfs_ipincount(ip)) { + if (ip->i_itemp->ili_last_lsn) { + error = _xfs_log_force_lsn(ip->i_mount, + ip->i_itemp->ili_last_lsn, + XFS_LOG_SYNC, &log_flushed); + } else { + error = _xfs_log_force(ip->i_mount, + XFS_LOG_SYNC, &log_flushed); + } + } + } + + if (ip->i_mount->m_flags & XFS_MOUNT_BARRIER) { + /* + * If the log write didn't issue an ordered tag we need + * to flush the disk cache for the data device now. + */ + if (!log_flushed) + xfs_blkdev_issue_flush(ip->i_mount->m_ddev_targp); + + /* + * If this inode is on the RT dev we need to flush that + * cache as well. + */ + if (XFS_IS_REALTIME_INODE(ip)) + xfs_blkdev_issue_flush(ip->i_mount->m_rtdev_targp); + } + + return -error; +} + STATIC ssize_t xfs_file_aio_read( struct kiocb *iocb, @@ -755,7 +870,8 @@ write_retry: mutex_lock(&inode->i_mutex); xfs_ilock(ip, iolock); - error2 = xfs_fsync(ip); + error2 = -xfs_file_fsync(file, file->f_path.dentry, + (file->f_flags & __O_SYNC) ? 0 : 1); if (!error) error = error2; } @@ -826,28 +942,6 @@ xfs_file_release( return -xfs_release(XFS_I(inode)); } -/* - * We ignore the datasync flag here because a datasync is effectively - * identical to an fsync. That is, datasync implies that we need to write - * only the metadata needed to be able to access the data that is written - * if we crash after the call completes. Hence if we are writing beyond - * EOF we have to log the inode size change as well, which makes it a - * full fsync. If we don't write beyond EOF, the inode core will be - * clean in memory and so we don't need to log the inode, just like - * fsync. - */ -STATIC int -xfs_file_fsync( - struct file *file, - struct dentry *dentry, - int datasync) -{ - struct xfs_inode *ip = XFS_I(dentry->d_inode); - - xfs_iflags_clear(ip, XFS_ITRUNCATED); - return -xfs_fsync(ip); -} - STATIC int xfs_file_readdir( struct file *filp, -- cgit v1.2.3 From 66d834ea603d61bd90fedad90300ca91c5bba0a3 Mon Sep 17 00:00:00 2001 From: Christoph Hellwig Date: Mon, 15 Feb 2010 09:44:49 +0000 Subject: xfs: implement optimized fdatasync Allow us to track the difference between timestamp and size updates by using mark_inode_dirty from the I/O completion code, and checking the VFS inode flags in xfs_file_fsync. Signed-off-by: Christoph Hellwig Reviewed-by: Dave Chinner Signed-off-by: Alex Elder --- fs/xfs/linux-2.6/xfs_file.c | 23 ++++++++++++----------- 1 file changed, 12 insertions(+), 11 deletions(-) (limited to 'fs/xfs/linux-2.6/xfs_file.c') diff --git a/fs/xfs/linux-2.6/xfs_file.c b/fs/xfs/linux-2.6/xfs_file.c index 6c283b7be8ab..43f9554adaac 100644 --- a/fs/xfs/linux-2.6/xfs_file.c +++ b/fs/xfs/linux-2.6/xfs_file.c @@ -97,16 +97,6 @@ xfs_iozero( return (-status); } -/* - * We ignore the datasync flag here because a datasync is effectively - * identical to an fsync. That is, datasync implies that we need to write - * only the metadata needed to be able to access the data that is written - * if we crash after the call completes. Hence if we are writing beyond - * EOF we have to log the inode size change as well, which makes it a - * full fsync. If we don't write beyond EOF, the inode core will be - * clean in memory and so we don't need to log the inode, just like - * fsync. - */ STATIC int xfs_file_fsync( struct file *file, @@ -139,7 +129,18 @@ xfs_file_fsync( */ xfs_ilock(ip, XFS_ILOCK_SHARED); - if (ip->i_update_core) { + /* + * First check if the VFS inode is marked dirty. All the dirtying + * of non-transactional updates no goes through mark_inode_dirty*, + * which allows us to distinguish beteeen pure timestamp updates + * and i_size updates which need to be caught for fdatasync. + * After that also theck for the dirty state in the XFS inode, which + * might gets cleared when the inode gets written out via the AIL + * or xfs_iflush_cluster. + */ + if (((dentry->d_inode->i_state & I_DIRTY_DATASYNC) || + ((dentry->d_inode->i_state & I_DIRTY_SYNC) && !datasync)) && + ip->i_update_core) { /* * Kick off a transaction to log the inode core to get the * updates. The sync transaction will also force the log. -- cgit v1.2.3 From 024910cbac323ab2e5ad6d7fa7958799b04b9728 Mon Sep 17 00:00:00 2001 From: Christoph Hellwig Date: Wed, 17 Feb 2010 19:34:57 +0000 Subject: xfs: fix inode pincount check in fsync We need to hold the ilock to check the inode pincount safely. While we're at it also remove the check for ip->i_itemp->ili_last_lsn, a pinned inode always has it set. Signed-off-by: Christoph Hellwig Reviewed-by: Dave Chinner Signed-off-by: Alex Elder --- fs/xfs/linux-2.6/xfs_file.c | 13 ++++--------- 1 file changed, 4 insertions(+), 9 deletions(-) (limited to 'fs/xfs/linux-2.6/xfs_file.c') diff --git a/fs/xfs/linux-2.6/xfs_file.c b/fs/xfs/linux-2.6/xfs_file.c index 43f9554adaac..42dd3bcfba6b 100644 --- a/fs/xfs/linux-2.6/xfs_file.c +++ b/fs/xfs/linux-2.6/xfs_file.c @@ -180,17 +180,12 @@ xfs_file_fsync( * disk yet, the inode will be still be pinned. If it is, * force the log. */ - xfs_iunlock(ip, XFS_ILOCK_SHARED); if (xfs_ipincount(ip)) { - if (ip->i_itemp->ili_last_lsn) { - error = _xfs_log_force_lsn(ip->i_mount, - ip->i_itemp->ili_last_lsn, - XFS_LOG_SYNC, &log_flushed); - } else { - error = _xfs_log_force(ip->i_mount, - XFS_LOG_SYNC, &log_flushed); - } + error = _xfs_log_force_lsn(ip->i_mount, + ip->i_itemp->ili_last_lsn, + XFS_LOG_SYNC, &log_flushed); } + xfs_iunlock(ip, XFS_ILOCK_SHARED); } if (ip->i_mount->m_flags & XFS_MOUNT_BARRIER) { -- cgit v1.2.3