From 62752ee198dca9209b7dee504763e51b11e9e0ca Mon Sep 17 00:00:00 2001 From: Mark Fasheh Date: Tue, 17 Oct 2006 10:31:38 +0200 Subject: [PATCH] Take i_mutex in splice_from_pipe() The splice_actor may be calling ->prepare_write() and ->commit_write(). We want i_mutex on the inode being written to before calling those so that we don't race i_size changes. The double locking behavior is done elsewhere in splice.c, and if we eventually want _nolock variants of generic_file_splice_write(), fs modules might have to replicate the nasty locking code. We introduce inode_double_lock() and inode_double_unlock() to consolidate the locking rules into one set of functions. Signed-off-by: Mark Fasheh Signed-off-by: Jens Axboe --- fs/inode.c | 36 ++++++++++++++++++++++++++++++++++++ fs/splice.c | 24 +++++++++++------------- 2 files changed, 47 insertions(+), 13 deletions(-) (limited to 'fs') diff --git a/fs/inode.c b/fs/inode.c index d9a21d12292..26cdb115ce6 100644 --- a/fs/inode.c +++ b/fs/inode.c @@ -1306,6 +1306,42 @@ void wake_up_inode(struct inode *inode) wake_up_bit(&inode->i_state, __I_LOCK); } +/* + * We rarely want to lock two inodes that do not have a parent/child + * relationship (such as directory, child inode) simultaneously. The + * vast majority of file systems should be able to get along fine + * without this. Do not use these functions except as a last resort. + */ +void inode_double_lock(struct inode *inode1, struct inode *inode2) +{ + if (inode1 == NULL || inode2 == NULL || inode1 == inode2) { + if (inode1) + mutex_lock(&inode1->i_mutex); + else if (inode2) + mutex_lock(&inode2->i_mutex); + return; + } + + if (inode1 < inode2) { + mutex_lock_nested(&inode1->i_mutex, I_MUTEX_PARENT); + mutex_lock_nested(&inode2->i_mutex, I_MUTEX_CHILD); + } else { + mutex_lock_nested(&inode2->i_mutex, I_MUTEX_PARENT); + mutex_lock_nested(&inode1->i_mutex, I_MUTEX_CHILD); + } +} +EXPORT_SYMBOL(inode_double_lock); + +void inode_double_unlock(struct inode *inode1, struct inode *inode2) +{ + if (inode1) + mutex_unlock(&inode1->i_mutex); + + if (inode2 && inode2 != inode1) + mutex_unlock(&inode2->i_mutex); +} +EXPORT_SYMBOL(inode_double_unlock); + static __initdata unsigned long ihash_entries; static int __init set_ihash_entries(char *str) { diff --git a/fs/splice.c b/fs/splice.c index a567010b62a..c1072b6940c 100644 --- a/fs/splice.c +++ b/fs/splice.c @@ -713,6 +713,7 @@ ssize_t splice_from_pipe(struct pipe_inode_info *pipe, struct file *out, { int ret, do_wakeup, err; struct splice_desc sd; + struct inode *inode = out->f_mapping->host; ret = 0; do_wakeup = 0; @@ -722,8 +723,13 @@ ssize_t splice_from_pipe(struct pipe_inode_info *pipe, struct file *out, sd.file = out; sd.pos = *ppos; - if (pipe->inode) - mutex_lock(&pipe->inode->i_mutex); + /* + * The actor worker might be calling ->prepare_write and + * ->commit_write. Most of the time, these expect i_mutex to + * be held. Since this may result in an ABBA deadlock with + * pipe->inode, we have to order lock acquiry here. + */ + inode_double_lock(inode, pipe->inode); for (;;) { if (pipe->nrbufs) { @@ -797,8 +803,7 @@ ssize_t splice_from_pipe(struct pipe_inode_info *pipe, struct file *out, pipe_wait(pipe); } - if (pipe->inode) - mutex_unlock(&pipe->inode->i_mutex); + inode_double_unlock(inode, pipe->inode); if (do_wakeup) { smp_mb(); @@ -1400,13 +1405,7 @@ static int link_pipe(struct pipe_inode_info *ipipe, * grabbing by inode address. Otherwise two different processes * could deadlock (one doing tee from A -> B, the other from B -> A). */ - if (ipipe->inode < opipe->inode) { - mutex_lock_nested(&ipipe->inode->i_mutex, I_MUTEX_PARENT); - mutex_lock_nested(&opipe->inode->i_mutex, I_MUTEX_CHILD); - } else { - mutex_lock_nested(&opipe->inode->i_mutex, I_MUTEX_PARENT); - mutex_lock_nested(&ipipe->inode->i_mutex, I_MUTEX_CHILD); - } + inode_double_lock(ipipe->inode, opipe->inode); do { if (!opipe->readers) { @@ -1450,8 +1449,7 @@ static int link_pipe(struct pipe_inode_info *ipipe, i++; } while (len); - mutex_unlock(&ipipe->inode->i_mutex); - mutex_unlock(&opipe->inode->i_mutex); + inode_double_unlock(ipipe->inode, opipe->inode); /* * If we put data in the output pipe, wakeup any potential readers. -- cgit v1.2.3-18-g5258 From 6da61809822c22634a3de2dcb3c60283b836a88a Mon Sep 17 00:00:00 2001 From: Mark Fasheh Date: Tue, 17 Oct 2006 18:43:07 +0200 Subject: [PATCH] Introduce generic_file_splice_write_nolock() This allows file systems to manage their own i_mutex locking while still re-using the generic_file_splice_write() logic. OCFS2 in particular wants this so that it can order cluster locks within i_mutex. Signed-off-by: Mark Fasheh Signed-off-by: Jens Axboe --- fs/splice.c | 80 ++++++++++++++++++++++++++++++++++++++++++++++++++----------- 1 file changed, 66 insertions(+), 14 deletions(-) (limited to 'fs') diff --git a/fs/splice.c b/fs/splice.c index c1072b6940c..68e20e65c6e 100644 --- a/fs/splice.c +++ b/fs/splice.c @@ -707,13 +707,12 @@ out_ret: * key here is the 'actor' worker passed in that actually moves the data * to the wanted destination. See pipe_to_file/pipe_to_sendpage above. */ -ssize_t splice_from_pipe(struct pipe_inode_info *pipe, struct file *out, - loff_t *ppos, size_t len, unsigned int flags, - splice_actor *actor) +static ssize_t __splice_from_pipe(struct pipe_inode_info *pipe, + struct file *out, loff_t *ppos, size_t len, + unsigned int flags, splice_actor *actor) { int ret, do_wakeup, err; struct splice_desc sd; - struct inode *inode = out->f_mapping->host; ret = 0; do_wakeup = 0; @@ -723,14 +722,6 @@ ssize_t splice_from_pipe(struct pipe_inode_info *pipe, struct file *out, sd.file = out; sd.pos = *ppos; - /* - * The actor worker might be calling ->prepare_write and - * ->commit_write. Most of the time, these expect i_mutex to - * be held. Since this may result in an ABBA deadlock with - * pipe->inode, we have to order lock acquiry here. - */ - inode_double_lock(inode, pipe->inode); - for (;;) { if (pipe->nrbufs) { struct pipe_buffer *buf = pipe->bufs + pipe->curbuf; @@ -803,8 +794,6 @@ ssize_t splice_from_pipe(struct pipe_inode_info *pipe, struct file *out, pipe_wait(pipe); } - inode_double_unlock(inode, pipe->inode); - if (do_wakeup) { smp_mb(); if (waitqueue_active(&pipe->wait)) @@ -815,6 +804,69 @@ ssize_t splice_from_pipe(struct pipe_inode_info *pipe, struct file *out, return ret; } +ssize_t splice_from_pipe(struct pipe_inode_info *pipe, struct file *out, + loff_t *ppos, size_t len, unsigned int flags, + splice_actor *actor) +{ + ssize_t ret; + struct inode *inode = out->f_mapping->host; + + /* + * The actor worker might be calling ->prepare_write and + * ->commit_write. Most of the time, these expect i_mutex to + * be held. Since this may result in an ABBA deadlock with + * pipe->inode, we have to order lock acquiry here. + */ + inode_double_lock(inode, pipe->inode); + ret = __splice_from_pipe(pipe, out, ppos, len, flags, actor); + inode_double_unlock(inode, pipe->inode); + + return ret; +} + +/** + * generic_file_splice_write_nolock - generic_file_splice_write without mutexes + * @pipe: pipe info + * @out: file to write to + * @len: number of bytes to splice + * @flags: splice modifier flags + * + * Will either move or copy pages (determined by @flags options) from + * the given pipe inode to the given file. The caller is responsible + * for acquiring i_mutex on both inodes. + * + */ +ssize_t +generic_file_splice_write_nolock(struct pipe_inode_info *pipe, struct file *out, + loff_t *ppos, size_t len, unsigned int flags) +{ + struct address_space *mapping = out->f_mapping; + struct inode *inode = mapping->host; + ssize_t ret; + int err; + + ret = __splice_from_pipe(pipe, out, ppos, len, flags, pipe_to_file); + if (ret > 0) { + *ppos += ret; + + /* + * If file or inode is SYNC and we actually wrote some data, + * sync it. + */ + if (unlikely((out->f_flags & O_SYNC) || IS_SYNC(inode))) { + err = generic_osync_inode(inode, mapping, + OSYNC_METADATA|OSYNC_DATA); + + if (err) + ret = err; + } + } + + return ret; +} + +EXPORT_SYMBOL(generic_file_splice_write_nolock); + /** * generic_file_splice_write - splice data from a pipe to a file * @pipe: pipe info -- cgit v1.2.3-18-g5258 From 8c34e2d63231d4bf4852bac8521883944d770fe3 Mon Sep 17 00:00:00 2001 From: Jens Axboe Date: Tue, 17 Oct 2006 19:43:22 +0200 Subject: [PATCH] Remove SUID when splicing into an inode Originally from Mark Fasheh generic_file_splice_write() does not remove S_ISUID or S_ISGID. This is inconsistent with the way we generally write to files. Signed-off-by: Mark Fasheh Signed-off-by: Jens Axboe --- fs/splice.c | 19 +++++++++++++++---- 1 file changed, 15 insertions(+), 4 deletions(-) (limited to 'fs') diff --git a/fs/splice.c b/fs/splice.c index 68e20e65c6e..49fb9f12993 100644 --- a/fs/splice.c +++ b/fs/splice.c @@ -845,6 +845,10 @@ generic_file_splice_write_nolock(struct pipe_inode_info *pipe, struct file *out, ssize_t ret; int err; + err = remove_suid(out->f_dentry); + if (unlikely(err)) + return err; + ret = __splice_from_pipe(pipe, out, ppos, len, flags, pipe_to_file); if (ret > 0) { *ppos += ret; @@ -883,12 +887,21 @@ generic_file_splice_write(struct pipe_inode_info *pipe, struct file *out, loff_t *ppos, size_t len, unsigned int flags) { struct address_space *mapping = out->f_mapping; + struct inode *inode = mapping->host; ssize_t ret; + int err; + + err = should_remove_suid(out->f_dentry); + if (unlikely(err)) { + mutex_lock(&inode->i_mutex); + err = __remove_suid(out->f_dentry, err); + mutex_unlock(&inode->i_mutex); + if (err) + return err; + } ret = splice_from_pipe(pipe, out, ppos, len, flags, pipe_to_file); if (ret > 0) { - struct inode *inode = mapping->host; - *ppos += ret; /* @@ -896,8 +909,6 @@ generic_file_splice_write(struct pipe_inode_info *pipe, struct file *out, * sync it. */ if (unlikely((out->f_flags & O_SYNC) || IS_SYNC(inode))) { - int err; - mutex_lock(&inode->i_mutex); err = generic_osync_inode(inode, mapping, OSYNC_METADATA|OSYNC_DATA); -- cgit v1.2.3-18-g5258