fsck.f2fs: fix potential stack overflow issue

In fsck_chk_inode_blk, we will allocate 256 bytes memory in stack before
traversing sub-directory recursively, it's not safe, in order to avoid
potential stack overflow, use malloc instead.

Signed-off-by: Chao Yu <yuchao0@huawei.com>
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>

diff --git a/fsck/fsck.c b/fsck/fsck.c
index ec88712..11b8b0b 100644 (file)
--- a/fsck/fsck.c
+++ b/fsck/fsck.c
@@ -633,7 +633,7 @@ void fsck_chk_inode_blk(struct f2fs_sb_info *sbi, u32 nid,
        u64 i_size = le64_to_cpu(node_blk->i.i_size);
        u64 i_blocks = le64_to_cpu(node_blk->i.i_blocks);
        int ofs = get_extra_isize(node_blk);
-       unsigned char en[F2FS_NAME_LEN + 1];
+       unsigned char *en;
        int namelen;
        unsigned int idx = 0;
        int need_fix = 0;
@@ -838,6 +838,9 @@ check:
                }
        }
 skip_blkcnt_fix:
+       en = malloc(F2FS_NAME_LEN + 1);
+       ASSERT(en);
+
        namelen = convert_encrypted_name(node_blk->i.i_name,
                                        le32_to_cpu(node_blk->i.i_namelen),
                                        en, file_enc_name(&node_blk->i));
@@ -879,6 +882,9 @@ skip_blkcnt_fix:
                        }
                }
        }
+
+       free(en);
+
        if (ftype == F2FS_FT_SYMLINK && i_blocks && i_size == 0) {
                DBG(1, "ino: 0x%x i_blocks: %lu with zero i_size",
                                                nid, (unsigned long)i_blocks);