From 3a326a2ce88e71d00ac0d133e314a3342a7709f8 Mon Sep 17 00:00:00 2001 From: Ingo Molnar Date: Mon, 10 Apr 2006 15:18:35 +0200 Subject: [PATCH] introduce a "kernel-internal pipe object" abstraction separate out the 'internal pipe object' abstraction, and make it usable to splice. This cleans up and fixes several aspects of the internal splice APIs and the pipe code: - pipes: the allocation and freeing of pipe_inode_info is now more symmetric and more streamlined with existing kernel practices. - splice: small micro-optimization: less pointer dereferencing in splice methods Signed-off-by: Ingo Molnar Update XFS for the ->splice_read/->splice_write changes. Signed-off-by: Jens Axboe --- fs/xfs/linux-2.6/xfs_file.c | 8 ++++---- fs/xfs/linux-2.6/xfs_lrw.c | 4 ++-- fs/xfs/linux-2.6/xfs_lrw.h | 4 ++-- fs/xfs/linux-2.6/xfs_vnode.h | 4 ++-- 4 files changed, 10 insertions(+), 10 deletions(-) (limited to 'fs/xfs') diff --git a/fs/xfs/linux-2.6/xfs_file.c b/fs/xfs/linux-2.6/xfs_file.c index ae4c4754ed31..269721af02f3 100644 --- a/fs/xfs/linux-2.6/xfs_file.c +++ b/fs/xfs/linux-2.6/xfs_file.c @@ -252,7 +252,7 @@ xfs_file_sendfile_invis( STATIC ssize_t xfs_file_splice_read( struct file *infilp, - struct inode *pipe, + struct pipe_inode_info *pipe, size_t len, unsigned int flags) { @@ -266,7 +266,7 @@ xfs_file_splice_read( STATIC ssize_t xfs_file_splice_read_invis( struct file *infilp, - struct inode *pipe, + struct pipe_inode_info *pipe, size_t len, unsigned int flags) { @@ -279,7 +279,7 @@ xfs_file_splice_read_invis( STATIC ssize_t xfs_file_splice_write( - struct inode *pipe, + struct pipe_inode_info *pipe, struct file *outfilp, size_t len, unsigned int flags) @@ -293,7 +293,7 @@ xfs_file_splice_write( STATIC ssize_t xfs_file_splice_write_invis( - struct inode *pipe, + struct pipe_inode_info *pipe, struct file *outfilp, size_t len, unsigned int flags) diff --git a/fs/xfs/linux-2.6/xfs_lrw.c b/fs/xfs/linux-2.6/xfs_lrw.c index 90cd314acbaa..74a52937f208 100644 --- a/fs/xfs/linux-2.6/xfs_lrw.c +++ b/fs/xfs/linux-2.6/xfs_lrw.c @@ -338,7 +338,7 @@ ssize_t xfs_splice_read( bhv_desc_t *bdp, struct file *infilp, - struct inode *pipe, + struct pipe_inode_info *pipe, size_t count, int flags, int ioflags, @@ -380,7 +380,7 @@ xfs_splice_read( ssize_t xfs_splice_write( bhv_desc_t *bdp, - struct inode *pipe, + struct pipe_inode_info *pipe, struct file *outfilp, size_t count, int flags, diff --git a/fs/xfs/linux-2.6/xfs_lrw.h b/fs/xfs/linux-2.6/xfs_lrw.h index eaa5659713fb..55c689a86ad2 100644 --- a/fs/xfs/linux-2.6/xfs_lrw.h +++ b/fs/xfs/linux-2.6/xfs_lrw.h @@ -94,9 +94,9 @@ extern ssize_t xfs_sendfile(struct bhv_desc *, struct file *, loff_t *, int, size_t, read_actor_t, void *, struct cred *); extern ssize_t xfs_splice_read(struct bhv_desc *, struct file *, - struct inode *, size_t, int, int, + struct pipe_inode_info *, size_t, int, int, struct cred *); -extern ssize_t xfs_splice_write(struct bhv_desc *, struct inode *, +extern ssize_t xfs_splice_write(struct bhv_desc *, struct pipe_inode_info *, struct file *, size_t, int, int, struct cred *); diff --git a/fs/xfs/linux-2.6/xfs_vnode.h b/fs/xfs/linux-2.6/xfs_vnode.h index 6f1c79a28f8b..88b09f186289 100644 --- a/fs/xfs/linux-2.6/xfs_vnode.h +++ b/fs/xfs/linux-2.6/xfs_vnode.h @@ -174,9 +174,9 @@ typedef ssize_t (*vop_sendfile_t)(bhv_desc_t *, struct file *, loff_t *, int, size_t, read_actor_t, void *, struct cred *); typedef ssize_t (*vop_splice_read_t)(bhv_desc_t *, struct file *, - struct inode *, size_t, int, int, + struct pipe_inode_info *, size_t, int, int, struct cred *); -typedef ssize_t (*vop_splice_write_t)(bhv_desc_t *, struct inode *, +typedef ssize_t (*vop_splice_write_t)(bhv_desc_t *, struct pipe_inode_info *, struct file *, size_t, int, int, struct cred *); typedef int (*vop_ioctl_t)(bhv_desc_t *, struct inode *, struct file *, -- cgit v1.2.3 From e50bd16fe49689bc5fb54fca5ed8b568dfba65c6 Mon Sep 17 00:00:00 2001 From: Nathan Scott Date: Tue, 11 Apr 2006 15:10:45 +1000 Subject: [XFS] Fix superblock validation regression for the zero imaxpct case. Thanks to kjamieson for noticing. SGI-PV: 951661 SGI-Modid: xfs-linux-melb:xfs-kern:25675a Signed-off-by: Nathan Scott --- fs/xfs/xfs_mount.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'fs/xfs') diff --git a/fs/xfs/xfs_mount.c b/fs/xfs/xfs_mount.c index 049fabb7f7e0..c0b1c2906880 100644 --- a/fs/xfs/xfs_mount.c +++ b/fs/xfs/xfs_mount.c @@ -270,7 +270,7 @@ xfs_mount_validate_sb( (sbp->sb_blocklog - sbp->sb_inodelog != sbp->sb_inopblog) || (sbp->sb_rextsize * sbp->sb_blocksize > XFS_MAX_RTEXTSIZE) || (sbp->sb_rextsize * sbp->sb_blocksize < XFS_MIN_RTEXTSIZE) || - (sbp->sb_imax_pct > 100 || sbp->sb_imax_pct < 1))) { + (sbp->sb_imax_pct > 100 /* zero sb_imax_pct is valid */))) { xfs_fs_mount_cmn_err(flags, "SB sanity check 1 failed"); return XFS_ERROR(EFSCORRUPTED); } -- cgit v1.2.3 From 8272145c05c6d01a34f5114357c5e8093fb66472 Mon Sep 17 00:00:00 2001 From: Nathan Scott Date: Tue, 11 Apr 2006 15:10:55 +1000 Subject: [XFS] Fix a writepage regression where we accidentally stopped honouring nonblock mode with the new IO path code (since 2.6.16). SGI-PV: 951662 SGI-Modid: xfs-linux-melb:xfs-kern:25676a Signed-off-by: Nathan Scott --- fs/xfs/linux-2.6/xfs_aops.c | 18 +++++++++--------- 1 file changed, 9 insertions(+), 9 deletions(-) (limited to 'fs/xfs') diff --git a/fs/xfs/linux-2.6/xfs_aops.c b/fs/xfs/linux-2.6/xfs_aops.c index 6cbbd165c60d..4d191ef39b67 100644 --- a/fs/xfs/linux-2.6/xfs_aops.c +++ b/fs/xfs/linux-2.6/xfs_aops.c @@ -870,12 +870,14 @@ xfs_page_state_convert( pgoff_t end_index, last_index, tlast; ssize_t size, len; int flags, err, iomap_valid = 0, uptodate = 1; - int page_dirty, count = 0, trylock_flag = 0; + int page_dirty, count = 0; + int trylock = 0; int all_bh = unmapped; - /* wait for other IO threads? */ - if (startio && (wbc->sync_mode == WB_SYNC_NONE && wbc->nonblocking)) - trylock_flag |= BMAPI_TRYLOCK; + if (startio) { + if (wbc->sync_mode == WB_SYNC_NONE && wbc->nonblocking) + trylock |= BMAPI_TRYLOCK; + } /* Is this page beyond the end of the file? */ offset = i_size_read(inode); @@ -956,15 +958,13 @@ xfs_page_state_convert( if (buffer_unwritten(bh)) { type = IOMAP_UNWRITTEN; - flags = BMAPI_WRITE|BMAPI_IGNSTATE; + flags = BMAPI_WRITE | BMAPI_IGNSTATE; } else if (buffer_delay(bh)) { type = IOMAP_DELAY; - flags = BMAPI_ALLOCATE; - if (!startio) - flags |= trylock_flag; + flags = BMAPI_ALLOCATE | trylock; } else { type = IOMAP_NEW; - flags = BMAPI_WRITE|BMAPI_MMAP; + flags = BMAPI_WRITE | BMAPI_MMAP; } if (!iomap_valid) { -- cgit v1.2.3 From 1fc5d959d88a5f77aa7e4435f6c9d0e2d2236704 Mon Sep 17 00:00:00 2001 From: David Chinner Date: Tue, 11 Apr 2006 15:11:12 +1000 Subject: [XFS] Fix inode reclaim scalability regression. When a filesystem has millions of inodes cached and has sparse cluster population, removing inodes from the cluster hash consumes excessive amounts of CPU time. Reduce the CPU cost by making removal O(1) via use of a double linked list for the hash chains. SGI-PV: 951551 SGI-Modid: xfs-linux-melb:xfs-kern:25683a Signed-off-by: David Chinner Signed-off-by: Nathan Scott --- fs/xfs/xfs_iget.c | 29 ++++++++++++----------------- fs/xfs/xfs_inode.h | 1 + 2 files changed, 13 insertions(+), 17 deletions(-) (limited to 'fs/xfs') diff --git a/fs/xfs/xfs_iget.c b/fs/xfs/xfs_iget.c index bb33113eef9f..b53854325266 100644 --- a/fs/xfs/xfs_iget.c +++ b/fs/xfs/xfs_iget.c @@ -421,7 +421,10 @@ finish_inode: ip->i_chash = chlnew; chlnew->chl_ip = ip; chlnew->chl_blkno = ip->i_blkno; + if (ch->ch_list) + ch->ch_list->chl_prev = chlnew; chlnew->chl_next = ch->ch_list; + chlnew->chl_prev = NULL; ch->ch_list = chlnew; chlnew = NULL; } @@ -723,23 +726,15 @@ xfs_iextract( ASSERT(ip->i_cnext == ip && ip->i_cprev == ip); ASSERT(ip->i_chash != NULL); chm=NULL; - for (chl = ch->ch_list; chl != NULL; chl = chl->chl_next) { - if (chl->chl_blkno == ip->i_blkno) { - if (chm == NULL) { - /* first item on the list */ - ch->ch_list = chl->chl_next; - } else { - chm->chl_next = chl->chl_next; - } - kmem_zone_free(xfs_chashlist_zone, chl); - break; - } else { - ASSERT(chl->chl_ip != ip); - chm = chl; - } - } - ASSERT_ALWAYS(chl != NULL); - } else { + chl = ip->i_chash; + if (chl->chl_prev) + chl->chl_prev->chl_next = chl->chl_next; + else + ch->ch_list = chl->chl_next; + if (chl->chl_next) + chl->chl_next->chl_prev = chl->chl_prev; + kmem_zone_free(xfs_chashlist_zone, chl); + } else { /* delete one inode from a non-empty list */ iq = ip->i_cnext; iq->i_cprev = ip->i_cprev; diff --git a/fs/xfs/xfs_inode.h b/fs/xfs/xfs_inode.h index 39ef9c36ea55..3b544db1790b 100644 --- a/fs/xfs/xfs_inode.h +++ b/fs/xfs/xfs_inode.h @@ -189,6 +189,7 @@ typedef struct xfs_ihash { */ typedef struct xfs_chashlist { struct xfs_chashlist *chl_next; + struct xfs_chashlist *chl_prev; struct xfs_inode *chl_ip; xfs_daddr_t chl_blkno; /* starting block number of * the cluster */ -- cgit v1.2.3 From 58829e490ee805f1c8b3009abc90e2a1a7a0d278 Mon Sep 17 00:00:00 2001 From: David Chinner Date: Tue, 11 Apr 2006 15:11:20 +1000 Subject: [XFS] Fix an inode use-after-free durin an unpin. When reclaiming inodes that have been unlinked, we may need to execute transactions during reclaim. By the time the transaction has hit the disk, the linux inode and xfs vnode may already have been freed so we can't reference them safely. Use the known xfs inode state to determine if it is safe to reference the vnode and linux inode during the unpin operation. SGI-PV: 946321 SGI-Modid: xfs-linux-melb:xfs-kern:25687a Signed-off-by: David Chinner Signed-off-by: Nathan Scott --- fs/xfs/xfs_inode.c | 27 ++++++++++++++++++++------- 1 file changed, 20 insertions(+), 7 deletions(-) (limited to 'fs/xfs') diff --git a/fs/xfs/xfs_inode.c b/fs/xfs/xfs_inode.c index 48146bdc6bdd..94b60dd03801 100644 --- a/fs/xfs/xfs_inode.c +++ b/fs/xfs/xfs_inode.c @@ -2732,16 +2732,29 @@ xfs_iunpin( ASSERT(atomic_read(&ip->i_pincount) > 0); if (atomic_dec_and_test(&ip->i_pincount)) { - vnode_t *vp = XFS_ITOV_NULL(ip); + /* + * If the inode is currently being reclaimed, the + * linux inode _and_ the xfs vnode may have been + * freed so we cannot reference either of them safely. + * Hence we should not try to do anything to them + * if the xfs inode is currently in the reclaim + * path. + * + * However, we still need to issue the unpin wakeup + * call as the inode reclaim may be blocked waiting for + * the inode to become unpinned. + */ + if (!(ip->i_flags & (XFS_IRECLAIM|XFS_IRECLAIMABLE))) { + vnode_t *vp = XFS_ITOV_NULL(ip); - /* make sync come back and flush this inode */ - if (vp) { - struct inode *inode = vn_to_inode(vp); + /* make sync come back and flush this inode */ + if (vp) { + struct inode *inode = vn_to_inode(vp); - if (!(inode->i_state & I_NEW)) - mark_inode_dirty_sync(inode); + if (!(inode->i_state & I_NEW)) + mark_inode_dirty_sync(inode); + } } - wake_up(&ip->i_ipin_wait); } } -- cgit v1.2.3 From 8c0b5113a55c698f3190ec85925815640f1c2049 Mon Sep 17 00:00:00 2001 From: Nathan Scott Date: Tue, 11 Apr 2006 15:12:45 +1000 Subject: [XFS] Fix utime(2) in the case that no times parameter was passed in. SGI-PV: 949858 SGI-Modid: xfs-linux-melb:xfs-kern:25717a Signed-off-by: Jes Sorensen Signed-off-by: Nathan Scott --- fs/xfs/linux-2.6/xfs_iops.c | 3 +-- 1 file changed, 1 insertion(+), 2 deletions(-) (limited to 'fs/xfs') diff --git a/fs/xfs/linux-2.6/xfs_iops.c b/fs/xfs/linux-2.6/xfs_iops.c index 149237304fb6..2e2e275c786f 100644 --- a/fs/xfs/linux-2.6/xfs_iops.c +++ b/fs/xfs/linux-2.6/xfs_iops.c @@ -673,8 +673,7 @@ xfs_vn_setattr( if (ia_valid & ATTR_ATIME) { vattr.va_mask |= XFS_AT_ATIME; vattr.va_atime = attr->ia_atime; - if (ia_valid & ATTR_ATIME_SET) - inode->i_atime = attr->ia_atime; + inode->i_atime = attr->ia_atime; } if (ia_valid & ATTR_MTIME) { vattr.va_mask |= XFS_AT_MTIME; -- cgit v1.2.3 From 019ff2d57b0bbe77d1eca19f5b634e5e7ff2a0b8 Mon Sep 17 00:00:00 2001 From: Nathan Scott Date: Tue, 11 Apr 2006 15:45:05 +1000 Subject: [XFS] Fix a problem in aligning inode allocations to stripe unit boundaries. SGI-PV: 951862 SGI-Modid: xfs-linux-melb:xfs-kern:25726a Signed-off-by: Nathan Scott --- fs/xfs/xfs_ialloc.c | 15 ++++++++------- 1 file changed, 8 insertions(+), 7 deletions(-) (limited to 'fs/xfs') diff --git a/fs/xfs/xfs_ialloc.c b/fs/xfs/xfs_ialloc.c index 4eeb856183b1..deddbd03c166 100644 --- a/fs/xfs/xfs_ialloc.c +++ b/fs/xfs/xfs_ialloc.c @@ -158,9 +158,10 @@ xfs_ialloc_ag_alloc( */ agi = XFS_BUF_TO_AGI(agbp); newino = be32_to_cpu(agi->agi_newino); - if(likely(newino != NULLAGINO)) { - args.agbno = XFS_AGINO_TO_AGBNO(args.mp, newino) + - XFS_IALLOC_BLOCKS(args.mp); + args.agbno = XFS_AGINO_TO_AGBNO(args.mp, newino) + + XFS_IALLOC_BLOCKS(args.mp); + if (likely(newino != NULLAGINO && + (args.agbno < be32_to_cpu(agi->agi_length)))) { args.fsbno = XFS_AGB_TO_FSB(args.mp, be32_to_cpu(agi->agi_seqno), args.agbno); args.type = XFS_ALLOCTYPE_THIS_BNO; @@ -182,8 +183,8 @@ xfs_ialloc_ag_alloc( * Set the alignment for the allocation. * If stripe alignment is turned on then align at stripe unit * boundary. - * If the cluster size is smaller than a filesystem block - * then we're doing I/O for inodes in filesystem block size + * If the cluster size is smaller than a filesystem block + * then we're doing I/O for inodes in filesystem block size * pieces, so don't need alignment anyway. */ isaligned = 0; @@ -192,7 +193,7 @@ xfs_ialloc_ag_alloc( args.alignment = args.mp->m_dalign; isaligned = 1; } else if (XFS_SB_VERSION_HASALIGN(&args.mp->m_sb) && - args.mp->m_sb.sb_inoalignmt >= + args.mp->m_sb.sb_inoalignmt >= XFS_B_TO_FSBT(args.mp, XFS_INODE_CLUSTER_SIZE(args.mp))) args.alignment = args.mp->m_sb.sb_inoalignmt; @@ -220,7 +221,7 @@ xfs_ialloc_ag_alloc( if ((error = xfs_alloc_vextent(&args))) return error; } - + /* * If stripe alignment is turned on, then try again with cluster * alignment. -- cgit v1.2.3 From 7b04d7170e9af805cac19f97b28fff10db897893 Mon Sep 17 00:00:00 2001 From: Jeff Dike Date: Mon, 10 Apr 2006 22:53:27 -0700 Subject: [PATCH] Add GFP_NOWAIT Introduce GFP_NOWAIT, as an alias for GFP_ATOMIC & ~__GFP_HIGH. This also changes XFS, which is the only in-tree user of this idiom that I could find. The XFS piece is compile-tested only. Signed-off-by: Jeff Dike Acked-by: Nathan Scott Signed-off-by: Andrew Morton Signed-off-by: Linus Torvalds --- fs/xfs/linux-2.6/xfs_buf.c | 2 +- include/linux/gfp.h | 2 ++ 2 files changed, 3 insertions(+), 1 deletion(-) (limited to 'fs/xfs') diff --git a/fs/xfs/linux-2.6/xfs_buf.c b/fs/xfs/linux-2.6/xfs_buf.c index 9fb0312665ca..26fed0756f01 100644 --- a/fs/xfs/linux-2.6/xfs_buf.c +++ b/fs/xfs/linux-2.6/xfs_buf.c @@ -182,7 +182,7 @@ free_address( { a_list_t *aentry; - aentry = kmalloc(sizeof(a_list_t), GFP_ATOMIC & ~__GFP_HIGH); + aentry = kmalloc(sizeof(a_list_t), GFP_NOWAIT); if (likely(aentry)) { spin_lock(&as_lock); aentry->next = as_free_head; diff --git a/include/linux/gfp.h b/include/linux/gfp.h index 7851e6b520cf..3ac452945a7d 100644 --- a/include/linux/gfp.h +++ b/include/linux/gfp.h @@ -57,6 +57,8 @@ struct vm_area_struct; __GFP_NOFAIL|__GFP_NORETRY|__GFP_NO_GROW|__GFP_COMP| \ __GFP_NOMEMALLOC|__GFP_HARDWALL) +/* This equals 0, but use constants in case they ever change */ +#define GFP_NOWAIT (GFP_ATOMIC & ~__GFP_HIGH) /* GFP_ATOMIC means both !wait (__GFP_WAIT not set) and use emergency pool */ #define GFP_ATOMIC (__GFP_HIGH) #define GFP_NOIO (__GFP_WAIT) -- cgit v1.2.3 From cbb7e577e732f576b9f399bc2600bdc0626c68dc Mon Sep 17 00:00:00 2001 From: Jens Axboe Date: Tue, 11 Apr 2006 14:57:50 +0200 Subject: [PATCH] splice: pass offset around for ->splice_read() and ->splice_write() We need not use ->f_pos as the offset for the file input/output. If the user passed an offset pointer in through sys_splice(), just use that and leave ->f_pos alone. Signed-off-by: Jens Axboe --- fs/splice.c | 86 ++++++++++++++++++++++---------------------- fs/xfs/linux-2.6/xfs_file.c | 12 ++++--- fs/xfs/linux-2.6/xfs_lrw.c | 14 ++++---- fs/xfs/linux-2.6/xfs_lrw.h | 4 +-- fs/xfs/linux-2.6/xfs_vnode.h | 12 +++---- include/linux/fs.h | 12 +++---- 6 files changed, 74 insertions(+), 66 deletions(-) (limited to 'fs/xfs') diff --git a/fs/splice.c b/fs/splice.c index e50a460239dd..5d3eda64703b 100644 --- a/fs/splice.c +++ b/fs/splice.c @@ -231,8 +231,9 @@ static ssize_t move_to_pipe(struct pipe_inode_info *pipe, struct page **pages, } static int -__generic_file_splice_read(struct file *in, struct pipe_inode_info *pipe, - size_t len, unsigned int flags) +__generic_file_splice_read(struct file *in, loff_t *ppos, + struct pipe_inode_info *pipe, size_t len, + unsigned int flags) { struct address_space *mapping = in->f_mapping; unsigned int offset, nr_pages; @@ -241,8 +242,8 @@ __generic_file_splice_read(struct file *in, struct pipe_inode_info *pipe, pgoff_t index; int i, error; - index = in->f_pos >> PAGE_CACHE_SHIFT; - offset = in->f_pos & ~PAGE_CACHE_MASK; + index = *ppos >> PAGE_CACHE_SHIFT; + offset = *ppos & ~PAGE_CACHE_MASK; nr_pages = (len + offset + PAGE_CACHE_SIZE - 1) >> PAGE_CACHE_SHIFT; if (nr_pages > PIPE_BUFFERS) @@ -348,8 +349,9 @@ fill_it: * * Will read pages from given file and fill them into a pipe. */ -ssize_t generic_file_splice_read(struct file *in, struct pipe_inode_info *pipe, - size_t len, unsigned int flags) +ssize_t generic_file_splice_read(struct file *in, loff_t *ppos, + struct pipe_inode_info *pipe, size_t len, + unsigned int flags) { ssize_t spliced; int ret; @@ -358,12 +360,12 @@ ssize_t generic_file_splice_read(struct file *in, struct pipe_inode_info *pipe, spliced = 0; while (len) { - ret = __generic_file_splice_read(in, pipe, len, flags); + ret = __generic_file_splice_read(in, ppos, pipe, len, flags); if (ret <= 0) break; - in->f_pos += ret; + *ppos += ret; len -= ret; spliced += ret; @@ -561,7 +563,7 @@ typedef int (splice_actor)(struct pipe_inode_info *, struct pipe_buffer *, * to the wanted destination. See pipe_to_file/pipe_to_sendpage above. */ static ssize_t move_from_pipe(struct pipe_inode_info *pipe, struct file *out, - size_t len, unsigned int flags, + loff_t *ppos, size_t len, unsigned int flags, splice_actor *actor) { int ret, do_wakeup, err; @@ -573,7 +575,7 @@ static ssize_t move_from_pipe(struct pipe_inode_info *pipe, struct file *out, sd.total_len = len; sd.flags = flags; sd.file = out; - sd.pos = out->f_pos; + sd.pos = *ppos; if (pipe->inode) mutex_lock(&pipe->inode->i_mutex); @@ -656,9 +658,7 @@ static ssize_t move_from_pipe(struct pipe_inode_info *pipe, struct file *out, kill_fasync(&pipe->fasync_writers, SIGIO, POLL_OUT); } - out->f_pos = sd.pos; return ret; - } /** @@ -674,12 +674,12 @@ static ssize_t move_from_pipe(struct pipe_inode_info *pipe, struct file *out, */ ssize_t generic_file_splice_write(struct pipe_inode_info *pipe, struct file *out, - size_t len, unsigned int flags) + loff_t *ppos, size_t len, unsigned int flags) { struct address_space *mapping = out->f_mapping; ssize_t ret; - ret = move_from_pipe(pipe, out, len, flags, pipe_to_file); + ret = move_from_pipe(pipe, out, ppos, len, flags, pipe_to_file); /* * If file or inode is SYNC and we actually wrote some data, sync it. @@ -715,9 +715,9 @@ EXPORT_SYMBOL(generic_file_splice_write); * */ ssize_t generic_splice_sendpage(struct pipe_inode_info *pipe, struct file *out, - size_t len, unsigned int flags) + loff_t *ppos, size_t len, unsigned int flags) { - return move_from_pipe(pipe, out, len, flags, pipe_to_sendpage); + return move_from_pipe(pipe, out, ppos, len, flags, pipe_to_sendpage); } EXPORT_SYMBOL(generic_splice_sendpage); @@ -726,9 +726,8 @@ EXPORT_SYMBOL(generic_splice_sendpage); * Attempt to initiate a splice from pipe to file. */ static long do_splice_from(struct pipe_inode_info *pipe, struct file *out, - size_t len, unsigned int flags) + loff_t *ppos, size_t len, unsigned int flags) { - loff_t pos; int ret; if (unlikely(!out->f_op || !out->f_op->splice_write)) @@ -737,22 +736,21 @@ static long do_splice_from(struct pipe_inode_info *pipe, struct file *out, if (unlikely(!(out->f_mode & FMODE_WRITE))) return -EBADF; - pos = out->f_pos; - - ret = rw_verify_area(WRITE, out, &pos, len); + ret = rw_verify_area(WRITE, out, ppos, len); if (unlikely(ret < 0)) return ret; - return out->f_op->splice_write(pipe, out, len, flags); + return out->f_op->splice_write(pipe, out, ppos, len, flags); } /* * Attempt to initiate a splice from a file to a pipe. */ -static long do_splice_to(struct file *in, struct pipe_inode_info *pipe, - size_t len, unsigned int flags) +static long do_splice_to(struct file *in, loff_t *ppos, + struct pipe_inode_info *pipe, size_t len, + unsigned int flags) { - loff_t pos, isize, left; + loff_t isize, left; int ret; if (unlikely(!in->f_op || !in->f_op->splice_read)) @@ -761,28 +759,27 @@ static long do_splice_to(struct file *in, struct pipe_inode_info *pipe, if (unlikely(!(in->f_mode & FMODE_READ))) return -EBADF; - pos = in->f_pos; - - ret = rw_verify_area(READ, in, &pos, len); + ret = rw_verify_area(READ, in, ppos, len); if (unlikely(ret < 0)) return ret; isize = i_size_read(in->f_mapping->host); - if (unlikely(in->f_pos >= isize)) + if (unlikely(*ppos >= isize)) return 0; - left = isize - in->f_pos; + left = isize - *ppos; if (unlikely(left < len)) len = left; - return in->f_op->splice_read(in, pipe, len, flags); + return in->f_op->splice_read(in, ppos, pipe, len, flags); } -long do_splice_direct(struct file *in, struct file *out, size_t len, - unsigned int flags) +long do_splice_direct(struct file *in, loff_t *ppos, struct file *out, + size_t len, unsigned int flags) { struct pipe_inode_info *pipe; long ret, bytes; + loff_t out_off; umode_t i_mode; int i; @@ -820,6 +817,7 @@ long do_splice_direct(struct file *in, struct file *out, size_t len, */ ret = 0; bytes = 0; + out_off = 0; while (len) { size_t read_len, max_read_len; @@ -829,7 +827,7 @@ long do_splice_direct(struct file *in, struct file *out, size_t len, */ max_read_len = min(len, (size_t)(PIPE_BUFFERS*PAGE_SIZE)); - ret = do_splice_to(in, pipe, max_read_len, flags); + ret = do_splice_to(in, ppos, pipe, max_read_len, flags); if (unlikely(ret < 0)) goto out_release; @@ -840,7 +838,7 @@ long do_splice_direct(struct file *in, struct file *out, size_t len, * must not do the output in nonblocking mode as then we * could get stuck data in the internal pipe: */ - ret = do_splice_from(pipe, out, read_len, + ret = do_splice_from(pipe, out, &out_off, read_len, flags & ~SPLICE_F_NONBLOCK); if (unlikely(ret < 0)) goto out_release; @@ -898,6 +896,7 @@ static long do_splice(struct file *in, loff_t __user *off_in, size_t len, unsigned int flags) { struct pipe_inode_info *pipe; + loff_t offset, *off; pipe = in->f_dentry->d_inode->i_pipe; if (pipe) { @@ -906,12 +905,13 @@ static long do_splice(struct file *in, loff_t __user *off_in, if (off_out) { if (out->f_op->llseek == no_llseek) return -EINVAL; - if (copy_from_user(&out->f_pos, off_out, - sizeof(loff_t))) + if (copy_from_user(&offset, off_out, sizeof(loff_t))) return -EFAULT; - } + off = &offset; + } else + off = &out->f_pos; - return do_splice_from(pipe, out, len, flags); + return do_splice_from(pipe, out, off, len, flags); } pipe = out->f_dentry->d_inode->i_pipe; @@ -921,11 +921,13 @@ static long do_splice(struct file *in, loff_t __user *off_in, if (off_in) { if (in->f_op->llseek == no_llseek) return -EINVAL; - if (copy_from_user(&in->f_pos, off_in, sizeof(loff_t))) + if (copy_from_user(&offset, off_in, sizeof(loff_t))) return -EFAULT; - } + off = &offset; + } else + off = &in->f_pos; - return do_splice_to(in, pipe, len, flags); + return do_splice_to(in, off, pipe, len, flags); } return -EINVAL; diff --git a/fs/xfs/linux-2.6/xfs_file.c b/fs/xfs/linux-2.6/xfs_file.c index 269721af02f3..c847416f6d10 100644 --- a/fs/xfs/linux-2.6/xfs_file.c +++ b/fs/xfs/linux-2.6/xfs_file.c @@ -252,6 +252,7 @@ xfs_file_sendfile_invis( STATIC ssize_t xfs_file_splice_read( struct file *infilp, + loff_t *ppos, struct pipe_inode_info *pipe, size_t len, unsigned int flags) @@ -259,13 +260,14 @@ xfs_file_splice_read( vnode_t *vp = vn_from_inode(infilp->f_dentry->d_inode); ssize_t rval; - VOP_SPLICE_READ(vp, infilp, pipe, len, flags, 0, NULL, rval); + VOP_SPLICE_READ(vp, infilp, ppos, pipe, len, flags, 0, NULL, rval); return rval; } STATIC ssize_t xfs_file_splice_read_invis( struct file *infilp, + loff_t *ppos, struct pipe_inode_info *pipe, size_t len, unsigned int flags) @@ -273,7 +275,7 @@ xfs_file_splice_read_invis( vnode_t *vp = vn_from_inode(infilp->f_dentry->d_inode); ssize_t rval; - VOP_SPLICE_READ(vp, infilp, pipe, len, flags, IO_INVIS, NULL, rval); + VOP_SPLICE_READ(vp, infilp, ppos, pipe, len, flags, IO_INVIS, NULL, rval); return rval; } @@ -281,13 +283,14 @@ STATIC ssize_t xfs_file_splice_write( struct pipe_inode_info *pipe, struct file *outfilp, + loff_t *ppos, size_t len, unsigned int flags) { vnode_t *vp = vn_from_inode(outfilp->f_dentry->d_inode); ssize_t rval; - VOP_SPLICE_WRITE(vp, pipe, outfilp, len, flags, 0, NULL, rval); + VOP_SPLICE_WRITE(vp, pipe, outfilp, ppos, len, flags, 0, NULL, rval); return rval; } @@ -295,13 +298,14 @@ STATIC ssize_t xfs_file_splice_write_invis( struct pipe_inode_info *pipe, struct file *outfilp, + loff_t *ppos, size_t len, unsigned int flags) { vnode_t *vp = vn_from_inode(outfilp->f_dentry->d_inode); ssize_t rval; - VOP_SPLICE_WRITE(vp, pipe, outfilp, len, flags, IO_INVIS, NULL, rval); + VOP_SPLICE_WRITE(vp, pipe, outfilp, ppos, len, flags, IO_INVIS, NULL, rval); return rval; } diff --git a/fs/xfs/linux-2.6/xfs_lrw.c b/fs/xfs/linux-2.6/xfs_lrw.c index 74a52937f208..67efe3308980 100644 --- a/fs/xfs/linux-2.6/xfs_lrw.c +++ b/fs/xfs/linux-2.6/xfs_lrw.c @@ -338,6 +338,7 @@ 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, @@ -360,7 +361,7 @@ xfs_splice_read( int error; error = XFS_SEND_DATA(mp, DM_EVENT_READ, BHV_TO_VNODE(bdp), - infilp->f_pos, count, + *ppos, count, FILP_DELAY_FLAG(infilp), &locktype); if (error) { xfs_iunlock(ip, XFS_IOLOCK_SHARED); @@ -368,8 +369,8 @@ xfs_splice_read( } } xfs_rw_enter_trace(XFS_SPLICE_READ_ENTER, &ip->i_iocore, - pipe, count, infilp->f_pos, ioflags); - ret = generic_file_splice_read(infilp, pipe, count, flags); + pipe, count, *ppos, ioflags); + ret = generic_file_splice_read(infilp, ppos, pipe, count, flags); if (ret > 0) XFS_STATS_ADD(xs_read_bytes, ret); @@ -382,6 +383,7 @@ 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, @@ -403,7 +405,7 @@ xfs_splice_write( int error; error = XFS_SEND_DATA(mp, DM_EVENT_WRITE, BHV_TO_VNODE(bdp), - outfilp->f_pos, count, + *ppos, count, FILP_DELAY_FLAG(outfilp), &locktype); if (error) { xfs_iunlock(ip, XFS_IOLOCK_EXCL); @@ -411,8 +413,8 @@ xfs_splice_write( } } xfs_rw_enter_trace(XFS_SPLICE_WRITE_ENTER, &ip->i_iocore, - pipe, count, outfilp->f_pos, ioflags); - ret = generic_file_splice_write(pipe, outfilp, count, flags); + pipe, count, *ppos, ioflags); + ret = generic_file_splice_write(pipe, outfilp, ppos, count, flags); if (ret > 0) XFS_STATS_ADD(xs_write_bytes, ret); diff --git a/fs/xfs/linux-2.6/xfs_lrw.h b/fs/xfs/linux-2.6/xfs_lrw.h index 55c689a86ad2..8f4539952350 100644 --- a/fs/xfs/linux-2.6/xfs_lrw.h +++ b/fs/xfs/linux-2.6/xfs_lrw.h @@ -93,11 +93,11 @@ extern ssize_t xfs_write(struct bhv_desc *, struct kiocb *, extern ssize_t xfs_sendfile(struct bhv_desc *, struct file *, loff_t *, int, size_t, read_actor_t, void *, struct cred *); -extern ssize_t xfs_splice_read(struct bhv_desc *, struct file *, +extern ssize_t xfs_splice_read(struct bhv_desc *, struct file *, loff_t *, struct pipe_inode_info *, size_t, int, int, struct cred *); extern ssize_t xfs_splice_write(struct bhv_desc *, struct pipe_inode_info *, - struct file *, size_t, int, int, + struct file *, loff_t *, size_t, int, int, struct cred *); #endif /* __XFS_LRW_H__ */ diff --git a/fs/xfs/linux-2.6/xfs_vnode.h b/fs/xfs/linux-2.6/xfs_vnode.h index 88b09f186289..2a8e16c22353 100644 --- a/fs/xfs/linux-2.6/xfs_vnode.h +++ b/fs/xfs/linux-2.6/xfs_vnode.h @@ -173,11 +173,11 @@ typedef ssize_t (*vop_write_t)(bhv_desc_t *, struct kiocb *, typedef ssize_t (*vop_sendfile_t)(bhv_desc_t *, struct file *, loff_t *, int, size_t, read_actor_t, void *, struct cred *); -typedef ssize_t (*vop_splice_read_t)(bhv_desc_t *, struct file *, +typedef ssize_t (*vop_splice_read_t)(bhv_desc_t *, struct file *, loff_t *, struct pipe_inode_info *, size_t, int, int, struct cred *); typedef ssize_t (*vop_splice_write_t)(bhv_desc_t *, struct pipe_inode_info *, - struct file *, size_t, int, int, + struct file *, loff_t *, size_t, int, int, struct cred *); typedef int (*vop_ioctl_t)(bhv_desc_t *, struct inode *, struct file *, int, unsigned int, void __user *); @@ -284,10 +284,10 @@ typedef struct vnodeops { rv = _VOP_(vop_write, vp)((vp)->v_fbhv,file,iov,segs,offset,ioflags,cr) #define VOP_SENDFILE(vp,f,off,ioflags,cnt,act,targ,cr,rv) \ rv = _VOP_(vop_sendfile, vp)((vp)->v_fbhv,f,off,ioflags,cnt,act,targ,cr) -#define VOP_SPLICE_READ(vp,f,pipe,cnt,fl,iofl,cr,rv) \ - rv = _VOP_(vop_splice_read, vp)((vp)->v_fbhv,f,pipe,cnt,fl,iofl,cr) -#define VOP_SPLICE_WRITE(vp,f,pipe,cnt,fl,iofl,cr,rv) \ - rv = _VOP_(vop_splice_write, vp)((vp)->v_fbhv,f,pipe,cnt,fl,iofl,cr) +#define VOP_SPLICE_READ(vp,f,o,pipe,cnt,fl,iofl,cr,rv) \ + rv = _VOP_(vop_splice_read, vp)((vp)->v_fbhv,f,o,pipe,cnt,fl,iofl,cr) +#define VOP_SPLICE_WRITE(vp,f,o,pipe,cnt,fl,iofl,cr,rv) \ + rv = _VOP_(vop_splice_write, vp)((vp)->v_fbhv,f,o,pipe,cnt,fl,iofl,cr) #define VOP_BMAP(vp,of,sz,rw,b,n,rv) \ rv = _VOP_(vop_bmap, vp)((vp)->v_fbhv,of,sz,rw,b,n) #define VOP_OPEN(vp, cr, rv) \ diff --git a/include/linux/fs.h b/include/linux/fs.h index 162c6e57307a..3de2bfb2410f 100644 --- a/include/linux/fs.h +++ b/include/linux/fs.h @@ -1039,8 +1039,8 @@ struct file_operations { int (*check_flags)(int); int (*dir_notify)(struct file *filp, unsigned long arg); int (*flock) (struct file *, int, struct file_lock *); - ssize_t (*splice_write)(struct pipe_inode_info *, struct file *, size_t, unsigned int); - ssize_t (*splice_read)(struct file *, struct pipe_inode_info *, size_t, unsigned int); + ssize_t (*splice_write)(struct pipe_inode_info *, struct file *, loff_t *, size_t, unsigned int); + ssize_t (*splice_read)(struct file *, loff_t *, struct pipe_inode_info *, size_t, unsigned int); }; struct inode_operations { @@ -1613,13 +1613,13 @@ extern void do_generic_mapping_read(struct address_space *mapping, loff_t *, read_descriptor_t *, read_actor_t); /* fs/splice.c */ -extern ssize_t generic_file_splice_read(struct file *, +extern ssize_t generic_file_splice_read(struct file *, loff_t *, struct pipe_inode_info *, size_t, unsigned int); extern ssize_t generic_file_splice_write(struct pipe_inode_info *, - struct file *, size_t, unsigned int); + struct file *, loff_t *, size_t, unsigned int); extern ssize_t generic_splice_sendpage(struct pipe_inode_info *pipe, - struct file *out, size_t len, unsigned int flags); -extern long do_splice_direct(struct file *in, struct file *out, + struct file *out, loff_t *, size_t len, unsigned int flags); +extern long do_splice_direct(struct file *in, loff_t *ppos, struct file *out, size_t len, unsigned int flags); extern void -- cgit v1.2.3