Commit f3914758 authored by Evgeniy Dushistov's avatar Evgeniy Dushistov Committed by Linus Torvalds
Browse files

[PATCH] ufs: missed brelse and wrong baseblk

This patch fixes two bugs, which introduced by previous patches:

1) Missed "brelse"

2) Sometimes "baseblk" may be wrongly calculated, if i_size is equal to
   zero, which lead infinite cycle in "mpage_writepages".
Signed-off-by: default avatarEvgeniy Dushistov <>
Signed-off-by: default avatarAndrew Morton <>
Signed-off-by: default avatarLinus Torvalds <>
parent 96710b29
......@@ -269,20 +269,17 @@ out:
* We can come here from ufs_writepage or ufs_prepare_write,
* locked_page is argument of these functions, so we already lock it.
static void ufs_change_blocknr(struct inode *inode, unsigned int count,
unsigned int oldb, unsigned int newb,
struct page *locked_page)
static void ufs_change_blocknr(struct inode *inode, unsigned int baseblk,
unsigned int count, unsigned int oldb,
unsigned int newb, struct page *locked_page)
unsigned int blk_per_page = 1 << (PAGE_CACHE_SHIFT - inode->i_blkbits);
sector_t baseblk;
struct address_space *mapping = inode->i_mapping;
pgoff_t index, cur_index = locked_page->index;
unsigned int i, j;
struct page *page;
struct buffer_head *head, *bh;
baseblk = ((i_size_read(inode) - 1) >> inode->i_blkbits) + 1 - count;
UFSD("ENTER, ino %lu, count %u, oldb %u, newb %u\n",
inode->i_ino, count, oldb, newb);
......@@ -439,7 +436,8 @@ unsigned ufs_new_fragments(struct inode * inode, __fs32 * p, unsigned fragment,
result = ufs_alloc_fragments (inode, cgno, goal, request, err);
if (result) {
ufs_change_blocknr(inode, oldcount, tmp, result, locked_page);
ufs_change_blocknr(inode, fragment - oldcount, oldcount, tmp,
result, locked_page);
*p = cpu_to_fs32(sb, result);
*err = 0;
......@@ -175,6 +175,7 @@ ufs_clear_frags(struct inode *inode, sector_t beg,
for (++beg; beg < end; ++beg) {
bh = sb_getblk(inode->i_sb, beg);
ufs_clear_frag(inode, bh);
return res;
Supports Markdown
0% or .
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment