fat: fix fake_offset handling on error path

For the root directory, .  and ..  are faked (using dir_emit_dots()) and
ctx->pos is reset from 2 to 0.

A corrupted root directory could cause fat_get_entry() to fail, but
->iterate() (fat_readdir()) reports progress to the VFS (with ctx->pos
rewound to 0), so any following calls to ->iterate() continue to return
the same entries again and again.

The result is that userspace will never see the end of the directory,
causing e.g.  'ls' to hang in a getdents() loop.

[hirofumi@mail.parknet.co.jp: cleanup and make sure to correct fake_offset]
Reported-by: Vegard Nossum <vegard.nossum@oracle.com>
Tested-by: Vegard Nossum <vegard.nossum@oracle.com>
Signed-off-by: Richard Weinberger <richard.weinberger@gmail.com>
Signed-off-by: OGAWA Hirofumi <hirofumi@mail.parknet.co.jp>
Cc: <stable@vger.kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

fs/fat/dir.c

@@ -610,9 +610,9 @@ parse_record:
 		int status = fat_parse_long(inode, &cpos, &bh, &de,
 					    &unicode, &nr_slots);
 		if (status < 0) {
-			ctx->pos = cpos;
+			bh = NULL;
 			ret = status;
-			goto out;
+			goto end_of_dir;
 		} else if (status == PARSE_INVALID)
 			goto record_end;
 		else if (status == PARSE_NOT_LONGNAME)
@@ -654,8 +654,9 @@ parse_record:
 	fill_len = short_len;
 
 start_filldir:
-	if (!fake_offset)
-		ctx->pos = cpos - (nr_slots + 1) * sizeof(struct msdos_dir_entry);
+	ctx->pos = cpos - (nr_slots + 1) * sizeof(struct msdos_dir_entry);
+	if (fake_offset && ctx->pos < 2)
+		ctx->pos = 2;
 
 	if (!memcmp(de->name, MSDOS_DOT, MSDOS_NAME)) {
 		if (!dir_emit_dot(file, ctx))
@@ -681,14 +682,19 @@ record_end:
 	fake_offset = 0;
 	ctx->pos = cpos;
 	goto get_new;
+
 end_of_dir:
-	ctx->pos = cpos;
+	if (fake_offset && cpos < 2)
+		ctx->pos = 2;
+	else
+		ctx->pos = cpos;
 fill_failed:
 	brelse(bh);
 	if (unicode)
 		__putname(unicode);
 out:
 	mutex_unlock(&sbi->s_lock);
+
 	return ret;
 }