summaryrefslogtreecommitdiffstats
path: root/fs/stack.c
diff options
context:
space:
mode:
authorErez Zadok <ezk@cs.sunysb.edu>2009-12-03 21:56:09 -0500
committerAl Viro <viro@zeniv.linux.org.uk>2009-12-17 10:58:17 -0500
commit1b8ab8159ef8f818f870a1d2e3b6953d80eefd3f (patch)
tree97fdfc87493d5077c4c510e4defedccc483477af /fs/stack.c
parent9afa2fb6c13501e5b3536d15344fce4e5442c469 (diff)
downloadlinux-1b8ab8159ef8f818f870a1d2e3b6953d80eefd3f.tar.bz2
VFS/fsstack: handle 32-bit smp + preempt + large files in fsstack_copy_inode_size
Copy the inode size and blocks from one inode to another correctly on 32-bit systems with CONFIG_SMP, CONFIG_PREEMPT, or CONFIG_LBDAF. Use proper inode spinlocks only when i_size/i_blocks cannot fit in one 32-bit word. Signed-off-by: Hugh Dickins <hugh.dickins@tiscali.co.uk> Signed-off-by: Erez Zadok <ezk@cs.sunysb.edu> Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
Diffstat (limited to 'fs/stack.c')
-rw-r--r--fs/stack.c54
1 files changed, 51 insertions, 3 deletions
diff --git a/fs/stack.c b/fs/stack.c
index 0e20e43ad740..4a6f7f440658 100644
--- a/fs/stack.c
+++ b/fs/stack.c
@@ -7,10 +7,58 @@
* This function cannot be inlined since i_size_{read,write} is rather
* heavy-weight on 32-bit systems
*/
-void fsstack_copy_inode_size(struct inode *dst, const struct inode *src)
+void fsstack_copy_inode_size(struct inode *dst, struct inode *src)
{
- i_size_write(dst, i_size_read((struct inode *)src));
- dst->i_blocks = src->i_blocks;
+ loff_t i_size;
+ blkcnt_t i_blocks;
+
+ /*
+ * i_size_read() includes its own seqlocking and protection from
+ * preemption (see include/linux/fs.h): we need nothing extra for
+ * that here, and prefer to avoid nesting locks than attempt to keep
+ * i_size and i_blocks in sync together.
+ */
+ i_size = i_size_read(src);
+
+ /*
+ * But if CONFIG_LBDAF (on 32-bit), we ought to make an effort to
+ * keep the two halves of i_blocks in sync despite SMP or PREEMPT -
+ * though stat's generic_fillattr() doesn't bother, and we won't be
+ * applying quotas (where i_blocks does become important) at the
+ * upper level.
+ *
+ * We don't actually know what locking is used at the lower level;
+ * but if it's a filesystem that supports quotas, it will be using
+ * i_lock as in inode_add_bytes(). tmpfs uses other locking, and
+ * its 32-bit is (just) able to exceed 2TB i_size with the aid of
+ * holes; but its i_blocks cannot carry into the upper long without
+ * almost 2TB swap - let's ignore that case.
+ */
+ if (sizeof(i_blocks) > sizeof(long))
+ spin_lock(&src->i_lock);
+ i_blocks = src->i_blocks;
+ if (sizeof(i_blocks) > sizeof(long))
+ spin_unlock(&src->i_lock);
+
+ /*
+ * If CONFIG_SMP or CONFIG_PREEMPT on 32-bit, it's vital for
+ * fsstack_copy_inode_size() to hold some lock around
+ * i_size_write(), otherwise i_size_read() may spin forever (see
+ * include/linux/fs.h). We don't necessarily hold i_mutex when this
+ * is called, so take i_lock for that case.
+ *
+ * And if CONFIG_LBADF (on 32-bit), continue our effort to keep the
+ * two halves of i_blocks in sync despite SMP or PREEMPT: use i_lock
+ * for that case too, and do both at once by combining the tests.
+ *
+ * There is none of this locking overhead in the 64-bit case.
+ */
+ if (sizeof(i_size) > sizeof(long) || sizeof(i_blocks) > sizeof(long))
+ spin_lock(&dst->i_lock);
+ i_size_write(dst, i_size);
+ dst->i_blocks = i_blocks;
+ if (sizeof(i_size) > sizeof(long) || sizeof(i_blocks) > sizeof(long))
+ spin_unlock(&dst->i_lock);
}
EXPORT_SYMBOL_GPL(fsstack_copy_inode_size);