4.4-stable review patch. If anyone has any objections, please let me know.
------------------
From: Wei Fang fangwei1@huawei.com
commit b86e33075ed1909d8002745b56ecf73b833db143 upstream.
A dead loop can be triggered in f2fs_fiemap() using the test case as below:
... fd = open(); fallocate(fd, 0, 0, 4294967296); ioctl(fd, FS_IOC_FIEMAP, fiemap_buf); ...
It's caused by an overflow in __get_data_block(): ... bh->b_size = map.m_len << inode->i_blkbits; ... map.m_len is an unsigned int, and bh->b_size is a size_t which is 64 bits on 64 bits archtecture, type conversion from an unsigned int to a size_t will result in an overflow.
In the above-mentioned case, bh->b_size will be zero, and f2fs_fiemap() will call get_data_block() at block 0 again an again.
Fix this by adding a force conversion before left shift.
Signed-off-by: Wei Fang fangwei1@huawei.com Acked-by: Chao Yu yuchao0@huawei.com Signed-off-by: Jaegeuk Kim jaegeuk@kernel.org Cc: Guenter Roeck linux@roeck-us.net Signed-off-by: Greg Kroah-Hartman gregkh@linuxfoundation.org
--- fs/f2fs/data.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-)
--- a/fs/f2fs/data.c +++ b/fs/f2fs/data.c @@ -721,7 +721,7 @@ static int __get_data_block(struct inode if (!ret) { map_bh(bh, inode->i_sb, map.m_pblk); bh->b_state = (bh->b_state & ~F2FS_MAP_FLAGS) | map.m_flags; - bh->b_size = map.m_len << inode->i_blkbits; + bh->b_size = (u64)map.m_len << inode->i_blkbits; } return ret; }