xfs: byte range granularity for XFS_IOC_ZERO_RANGE
authorDave Chinner <dchinner@redhat.com>
Thu, 29 Nov 2012 04:26:33 +0000 (15:26 +1100)
committerBen Myers <bpm@sgi.com>
Thu, 29 Nov 2012 20:21:46 +0000 (14:21 -0600)
XFS_IOC_ZERO_RANGE simply does not work properly for non page cache
aligned ranges. Neither test 242 or 290 exercise this correctly, so
the behaviour is completely busted even though the tests pass.

Fix it to support full byte range granularity as was originally
intended for this ioctl.

Signed-off-by: Dave Chinner <dchinner@redhat.com>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Ben Myers <bpm@sgi.com>
fs/xfs/xfs_file.c
fs/xfs/xfs_vnodeops.c
fs/xfs/xfs_vnodeops.h

index 400b187595bb5454383362d9e60af40a5ab2aa35..67284edb84d74d2156283c3c34e1b28d179b4861 100644 (file)
@@ -86,7 +86,7 @@ xfs_rw_ilock_demote(
  *     valid before the operation, it will be read from disk before
  *     being partially zeroed.
  */
-STATIC int
+int
 xfs_iozero(
        struct xfs_inode        *ip,    /* inode                        */
        loff_t                  pos,    /* offset in file               */
index 26880793feca9a5e50ff94f1d68f075701e900fc..d95f565a390e8bc006b708b734b6107620a8d0aa 100644 (file)
@@ -2095,6 +2095,73 @@ xfs_free_file_space(
        return error;
 }
 
+
+STATIC int
+xfs_zero_file_space(
+       struct xfs_inode        *ip,
+       xfs_off_t               offset,
+       xfs_off_t               len,
+       int                     attr_flags)
+{
+       struct xfs_mount        *mp = ip->i_mount;
+       uint                    granularity;
+       xfs_off_t               start_boundary;
+       xfs_off_t               end_boundary;
+       int                     error;
+
+       granularity = max_t(uint, 1 << mp->m_sb.sb_blocklog, PAGE_CACHE_SIZE);
+
+       /*
+        * Round the range of extents we are going to convert inwards.  If the
+        * offset is aligned, then it doesn't get changed so we zero from the
+        * start of the block offset points to.
+        */
+       start_boundary = round_up(offset, granularity);
+       end_boundary = round_down(offset + len, granularity);
+
+       ASSERT(start_boundary >= offset);
+       ASSERT(end_boundary <= offset + len);
+
+       if (!(attr_flags & XFS_ATTR_NOLOCK))
+               xfs_ilock(ip, XFS_IOLOCK_EXCL);
+
+       if (start_boundary < end_boundary - 1) {
+               /* punch out the page cache over the conversion range */
+               truncate_pagecache_range(VFS_I(ip), start_boundary,
+                                        end_boundary - 1);
+               /* convert the blocks */
+               error = xfs_alloc_file_space(ip, start_boundary,
+                                       end_boundary - start_boundary - 1,
+                                       XFS_BMAPI_PREALLOC | XFS_BMAPI_CONVERT,
+                                       attr_flags);
+               if (error)
+                       goto out_unlock;
+
+               /* We've handled the interior of the range, now for the edges */
+               if (start_boundary != offset)
+                       error = xfs_iozero(ip, offset, start_boundary - offset);
+               if (error)
+                       goto out_unlock;
+
+               if (end_boundary != offset + len)
+                       error = xfs_iozero(ip, end_boundary,
+                                          offset + len - end_boundary);
+
+       } else {
+               /*
+                * It's either a sub-granularity range or the range spanned lies
+                * partially across two adjacent blocks.
+                */
+               error = xfs_iozero(ip, offset, len);
+       }
+
+out_unlock:
+       if (!(attr_flags & XFS_ATTR_NOLOCK))
+               xfs_iunlock(ip, XFS_IOLOCK_EXCL);
+       return error;
+
+}
+
 /*
  * xfs_change_file_space()
  *      This routine allocates or frees disk space for the given file.
@@ -2120,10 +2187,8 @@ xfs_change_file_space(
        xfs_fsize_t     fsize;
        int             setprealloc;
        xfs_off_t       startoffset;
-       xfs_off_t       end;
        xfs_trans_t     *tp;
        struct iattr    iattr;
-       int             prealloc_type;
 
        if (!S_ISREG(ip->i_d.di_mode))
                return XFS_ERROR(EINVAL);
@@ -2172,31 +2237,20 @@ xfs_change_file_space(
        startoffset = bf->l_start;
        fsize = XFS_ISIZE(ip);
 
-       /*
-        * XFS_IOC_RESVSP and XFS_IOC_UNRESVSP will reserve or unreserve
-        * file space.
-        * These calls do NOT zero the data space allocated to the file,
-        * nor do they change the file size.
-        *
-        * XFS_IOC_ALLOCSP and XFS_IOC_FREESP will allocate and free file
-        * space.
-        * These calls cause the new file data to be zeroed and the file
-        * size to be changed.
-        */
        setprealloc = clrprealloc = 0;
-       prealloc_type = XFS_BMAPI_PREALLOC;
-
        switch (cmd) {
        case XFS_IOC_ZERO_RANGE:
-               prealloc_type |= XFS_BMAPI_CONVERT;
-               end = round_down(startoffset + bf->l_len, PAGE_SIZE) - 1;
-               if (startoffset <= end)
-                       truncate_pagecache_range(VFS_I(ip), startoffset, end);
-               /* FALLTHRU */
+               error = xfs_zero_file_space(ip, startoffset, bf->l_len,
+                                               attr_flags);
+               if (error)
+                       return error;
+               setprealloc = 1;
+               break;
+
        case XFS_IOC_RESVSP:
        case XFS_IOC_RESVSP64:
                error = xfs_alloc_file_space(ip, startoffset, bf->l_len,
-                                               prealloc_type, attr_flags);
+                                               XFS_BMAPI_PREALLOC, attr_flags);
                if (error)
                        return error;
                setprealloc = 1;
index 91a03fa3814f4b295bd614f9e91aa92d77e3b874..5163022d98089b9b6eb2fb4bdbd9df8f0861b92b 100644 (file)
@@ -49,6 +49,7 @@ int xfs_attr_remove(struct xfs_inode *dp, const unsigned char *name, int flags);
 int xfs_attr_list(struct xfs_inode *dp, char *buffer, int bufsize,
                int flags, struct attrlist_cursor_kern *cursor);
 
+int xfs_iozero(struct xfs_inode *, loff_t, size_t);
 int xfs_zero_eof(struct xfs_inode *, xfs_off_t, xfs_fsize_t);
 int xfs_free_eofblocks(struct xfs_mount *, struct xfs_inode *, bool);