linux-kernel - Re: [PATCH] gfs2 FS: Fix UBSAN array-index-out-of-bounds in __gfs2_iomap

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Message-ID: <97abb511-9918-4df5-e421-835b4ecd324c@gmail.com>
Date:   Wed, 15 Mar 2023 13:12:47 +0400
From:   Ivan Orlov <ivan.orlov0322@...il.com>
To:     rpeterso@...hat.com, agruenba@...hat.com
Cc:     cluster-devel@...hat.com, linux-kernel@...r.kernel.org,
        skhan@...uxfoundation.org, himadrispandya@...il.com,
        linux-kernel-mentees@...ts.linuxfoundation.org,
        syzbot+45d4691b1ed3c48eba05@...kaller.appspotmail.com
Subject: Re: [PATCH] gfs2 FS: Fix UBSAN array-index-out-of-bounds in
 __gfs2_iomap_get

On 3/15/23 13:06, Ivan Orlov wrote:
> Syzkaller reported the following issue:
> 
> UBSAN: array-index-out-of-bounds in fs/gfs2/bmap.c:901:46
> index 11 is out of range for type 'u64 [11]'
> CPU: 0 PID: 5067 Comm: syz-executor164 Not tainted 6.1.0-syzkaller-13031-g77856d911a8c #0
> Hardware name: Google Google Compute Engine/Google Compute Engine, BIOS Google 10/26/2022
> Call Trace:
>   <TASK>
>   __dump_stack lib/dump_stack.c:88 [inline]
>   dump_stack_lvl+0x1b1/0x290 lib/dump_stack.c:106
>   ubsan_epilogue lib/ubsan.c:151 [inline]
>   __ubsan_handle_out_of_bounds+0xe0/0x110 lib/ubsan.c:282
>   __gfs2_iomap_get+0x4a4/0x16e0 fs/gfs2/bmap.c:901
>   gfs2_iomap_get fs/gfs2/bmap.c:1399 [inline]
>   gfs2_block_map+0x28f/0x7f0 fs/gfs2/bmap.c:1214
>   gfs2_write_alloc_required+0x441/0x6e0 fs/gfs2/bmap.c:2322
>   gfs2_jdesc_check+0x1b9/0x290 fs/gfs2/super.c:114
>   init_journal+0x5a4/0x22c0 fs/gfs2/ops_fstype.c:804
>   init_inodes+0xdc/0x340 fs/gfs2/ops_fstype.c:889
>   gfs2_fill_super+0x1bb2/0x2700 fs/gfs2/ops_fstype.c:1247
>   get_tree_bdev+0x400/0x620 fs/super.c:1282
>   gfs2_get_tree+0x50/0x210 fs/gfs2/ops_fstype.c:1330
>   vfs_get_tree+0x88/0x270 fs/super.c:1489
>   do_new_mount+0x289/0xad0 fs/namespace.c:3145
>   do_mount fs/namespace.c:3488 [inline]
>   __do_sys_mount fs/namespace.c:3697 [inline]
>   __se_sys_mount+0x2d3/0x3c0 fs/namespace.c:3674
>   do_syscall_x64 arch/x86/entry/common.c:50 [inline]
>   do_syscall_64+0x3d/0xb0 arch/x86/entry/common.c:80
>   entry_SYSCALL_64_after_hwframe+0x63/0xcd
> RIP: 0033:0x7f2c63567aca
> Code: 83 c4 08 5b 5d c3 66 2e 0f 1f 84 00 00 00 00 00 c3 66 2e 0f 1f 84 00 00 00 00 00 0f 1f 44 00 00 49 89 ca b8 a5 00 00 00 0f 05 <48> 3d 01 f0 ff ff 73 01 c3 48 c7 c1 c0 ff ff ff f7 d8 64 89 01 48
> RSP: 002b:00007ffd0e3a28d8 EFLAGS: 00000282 ORIG_RAX: 00000000000000a5
> RAX: ffffffffffffffda RBX: 0000000000000003 RCX: 00007f2c63567aca
> RDX: 0000000020037f40 RSI: 0000000020037f80 RDI: 00007ffd0e3a28e0
> RBP: 00007ffd0e3a28e0 R08: 00007ffd0e3a2920 R09: 0000000000043350
> R10: 0000000002000011 R11: 0000000000000282 R12: 0000000000000004
> R13: 00005555567192c0 R14: 00007ffd0e3a2920 R15: 0000000000000000
>   </TASK>
> 
> This issue is caused by the 'while' loop in the '__gfs2_iomap_get' function,
> which increments 'height' var until it matches the condition. If height is
> larger or equal to 'sdp->sd_heightsize' array size (which is GFS2_MAX_META_HEIGHT
> + 1), the array-index-out-of-bounds issue occurs. Therefore we need to add extra
> condition to the while loop, which will prevent this issue.
> 
> Additionally, if 'height' var after the while ending is equal to GFS2_MAX_META_HEIGHT,
> the 'find_metapath' call right after the loop will break (because it assumes that
> 'height' parameter will not be larger than the size of metapath's mp_list array length,
> which is GFS2_MAX_META_HEIGHT). So, we need to check the 'height' after the loop ending,
> and if its value is too big we need to break the execution of the function, and return
> a proper error if it is too big.
> 
> Tested via syzbot.
> 
> Reported-by: syzbot+45d4691b1ed3c48eba05@...kaller.appspotmail.com
> Link: https://syzkaller.appspot.com/bug?id=42296ea544c870f4dde3b25efa4cc1b89515d38e
> Signed-off-by: Ivan Orlov <ivan.orlov0322@...il.com>
> ---
>   fs/gfs2/bmap.c | 10 +++++++++-
>   1 file changed, 9 insertions(+), 1 deletion(-)
> 
> diff --git a/fs/gfs2/bmap.c b/fs/gfs2/bmap.c
> index eedf6926c652..9b7358165f96 100644
> --- a/fs/gfs2/bmap.c
> +++ b/fs/gfs2/bmap.c
> @@ -895,8 +895,16 @@ static int __gfs2_iomap_get(struct inode *inode, loff_t pos, loff_t length,
>   	iomap->length = len << inode->i_blkbits;
>   
>   	height = ip->i_height;
> -	while ((lblock + 1) * sdp->sd_sb.sb_bsize > sdp->sd_heightsize[height])
> +
> +	while (height <= GFS2_MAX_META_HEIGHT
> +		&& (lblock + 1) * sdp->sd_sb.sb_bsize > sdp->sd_heightsize[height])
>   		height++;
> +
> +	if (height > GFS2_MAX_META_HEIGHT) {
> +		ret = -ERANGE;
> +		goto unlock;
> +	}
> +
>   	find_metapath(sdp, lblock, mp, height);
>   	if (height > ip->i_height || gfs2_is_stuffed(ip))
>   		goto do_alloc;

Upd. I made a typo in the second paragraph of patch description: the 
'find_metapath' call will break if height is larger than 
GFS2_MAX_META_HEIGHT, not equal to. So this I check in corresponding 
condition right after the while loop.