Add interface for file hole punching.

This adds an interface to "punch holes" (deallocate space) in VFS
files. The interface is identical to the Solaris VOP_SPACE interface.
This interface is necessary for TRIM support on file vdevs.

This is implemented using Linux fallocate(FALLOC_FL_PUNCH_HOLE), which
was introduced in 2.6.38. For a brief time before 2.6.38 this was done
using the truncate_range inode operation, which was quickly deprecated.
This patch only supports FALLOC_FL_PUNCH_HOLE.

This adds support for the truncate_range() inode operation to
VOP_SPACE() for file hole punching. This API is deprecated and removed
in 3.5, so it's only useful for old kernels.

On tmpfs, the truncate_range() inode operation translates to
shmem_truncate_range(). Unfortunately, this function expects the end
offset to be inclusive and aligned to the end of a page. If it is not,
the kernel will stop with a BUG_ON().

This patch fixes the issue by adapting to the constraints set forth by
shmem_truncate_range().

Signed-off-by: Brian Behlendorf <[email protected]>
Closes #168

3 files changed, 81 insertions, 0 deletions

diff --git a/config/spl-build.m4 b/config/spl-build.m4
index b7ab54d32..0c7a03cf1 100644
--- a/config/spl-build.m4
+++ b/config/spl-build.m4
@@ -72,6 +72,7 @@ AC_DEFUN([SPL_AC_CONFIG_KERNEL], [
 	SPL_AC_4ARGS_VFS_RENAME
 	SPL_AC_VFS_FSYNC
 	SPL_AC_2ARGS_VFS_FSYNC
+	SPL_AC_INODE_TRUNCATE_RANGE
 	SPL_AC_FS_STRUCT_SPINLOCK
 	SPL_AC_CRED_STRUCT
 	SPL_AC_GROUPS_SEARCH
@@ -1993,6 +1994,27 @@ AC_DEFUN([SPL_AC_2ARGS_VFS_FSYNC], [
 ])
 
 dnl #
+dnl # 3.5 API change,
+dnl # inode_operations.truncate_range removed
+dnl # (deprecated in favor of FALLOC_FL_PUNCH_HOLE)
+dnl #
+AC_DEFUN([SPL_AC_INODE_TRUNCATE_RANGE], [
+	AC_MSG_CHECKING([whether truncate_range() inode operation is available])
+	SPL_LINUX_TRY_COMPILE([
+		#include <linux/fs.h>
+	],[
+		struct inode_operations ops;
+		ops.truncate_range = NULL;
+	],[
+		AC_MSG_RESULT(yes)
+		AC_DEFINE(HAVE_INODE_TRUNCATE_RANGE, 1,
+			[truncate_range() inode operation is available])
+	],[
+		AC_MSG_RESULT(no)
+	])
+]))
+
+dnl #
 dnl # 2.6.33 API change. Also backported in RHEL5 as of 2.6.18-190.el5.
 dnl # Earlier versions of rwsem_is_locked() were inline and had a race
 dnl # condition.  The fixed version is exported as a symbol.  The race
diff --git a/include/sys/vnode.h b/include/sys/vnode.h
index 03462bd70..99614d55b 100644
--- a/include/sys/vnode.h
+++ b/include/sys/vnode.h
@@ -66,6 +66,9 @@
 #define FNODSYNC	0x10000 /* fsync pseudo flag */
 #define FNOFOLLOW	0x20000 /* don't follow symlinks */
 
+#define F_FREESP	11 	/* Free file space */
+
+
 /*
  * The vnode AT_ flags are mapped to the Linux ATTR_* flags.
  * This allows them to be used safely with an iattr structure.
@@ -185,6 +188,8 @@ extern int vn_remove(const char *path, uio_seg_t seg, int flags);
 extern int vn_rename(const char *path1, const char *path2, int x1);
 extern int vn_getattr(vnode_t *vp, vattr_t *vap, int flags, void *x3, void *x4);
 extern int vn_fsync(vnode_t *vp, int flags, void *x3, void *x4);
+extern int vn_space(vnode_t *vp, int cmd, struct flock *bfp, int flag,
+    offset_t offset, void *x6, void *x7);
 extern file_t *vn_getf(int fd);
 extern void vn_releasef(int fd);
 extern int vn_set_pwd(const char *filename);
@@ -197,6 +202,7 @@ void spl_vn_fini(void);
 #define VOP_SEEK				vn_seek
 #define VOP_GETATTR				vn_getattr
 #define VOP_FSYNC				vn_fsync
+#define VOP_SPACE				vn_space
 #define VOP_PUTPAGE(vp, o, s, f, x1, x2)	((void)0)
 #define vn_is_readonly(vp)			0
 #define getf					vn_getf
diff --git a/module/spl/spl-vnode.c b/module/spl/spl-vnode.c
index 2e55b007b..f5fc65d26 100644
--- a/module/spl/spl-vnode.c
+++ b/module/spl/spl-vnode.c
@@ -25,6 +25,7 @@
 \*****************************************************************************/
 
 #include <sys/vnode.h>
+#include <linux/falloc.h>
 #include <spl-debug.h>
 
 #ifdef SS_DEBUG_SUBSYS
@@ -510,6 +511,58 @@ int vn_fsync(vnode_t *vp, int flags, void *x3, void *x4)
 } /* vn_fsync() */
 EXPORT_SYMBOL(vn_fsync);
 
+int vn_space(vnode_t *vp, int cmd, struct flock *bfp, int flag,
+    offset_t offset, void *x6, void *x7)
+{
+	int error = EOPNOTSUPP;
+	SENTRY;
+
+	if (cmd != F_FREESP || bfp->l_whence != 0)
+		SRETURN(EOPNOTSUPP);
+
+	ASSERT(vp);
+	ASSERT(vp->v_file);
+	ASSERT(bfp->l_start >= 0 && bfp->l_len > 0);
+
+#ifdef FALLOC_FL_PUNCH_HOLE
+	if (vp->v_file->f_op->fallocate) {
+		error = -vp->v_file->f_op->fallocate(vp->v_file,
+		    FALLOC_FL_KEEP_SIZE | FALLOC_FL_PUNCH_HOLE,
+		    bfp->l_start, bfp->l_len);
+		if (!error)
+			SRETURN(0);
+	}
+#endif
+
+#ifdef HAVE_INODE_TRUNCATE_RANGE
+	if (vp->v_file->f_dentry && vp->v_file->f_dentry->d_inode &&
+	    vp->v_file->f_dentry->d_inode->i_op &&
+	    vp->v_file->f_dentry->d_inode->i_op->truncate_range) {
+		off_t end = bfp->l_start + bfp->l_len;
+		/*
+		 * Judging from the code in shmem_truncate_range(),
+		 * it seems the kernel expects the end offset to be
+		 * inclusive and aligned to the end of a page.
+		 */
+		if (end % PAGE_SIZE != 0) {
+			end &= ~(off_t)(PAGE_SIZE - 1);
+			if (end <= bfp->l_start)
+				SRETURN(0);
+		}
+		--end;
+
+		vp->v_file->f_dentry->d_inode->i_op->truncate_range(
+			vp->v_file->f_dentry->d_inode,
+			bfp->l_start, end
+		);
+		SRETURN(0);
+	}
+#endif
+
+	SRETURN(error);
+}
+EXPORT_SYMBOL(vn_space);
+
 /* Function must be called while holding the vn_file_lock */
 static file_t *
 file_find(int fd)