linux-kernel - Re: [PATCH v2] f2fs: avoid out-of-range memory access

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: <65e4ad7b-ffbc-d5c9-9a0f-0532f4c4f5a9@huawei.com>
Date:   Thu, 4 Jul 2019 15:11:27 +0800
From:   Chao Yu <yuchao0@...wei.com>
To:     Ocean Chen <oceanchen@...gle.com>, <jaegeuk@...nel.org>,
        <linux-f2fs-devel@...ts.sourceforge.net>,
        <linux-kernel@...r.kernel.org>
Subject: Re: [PATCH v2] f2fs: avoid out-of-range memory access

Hi Ocean,

On 2019/7/3 23:03, Ocean Chen wrote:
> Hi Yu Chao,
> 
> The cur_data_segno only was checked in mount process. In terms of
> security concern, it's better to check value before using it. I know the

Could you explain more about security concern.. Do you get any report from user
or tools that complaining f2fs issue/codes?

I'm not against sanity check for basic core data of filesystem in run-time, but,
in order to troubleshoot root cause of this issue we can trigger panic directly
to dump more info under F2FS_CHECK_FS macro.

So, maybe we can change as below?

blk_off = le16_to_cpu(ckpt->cur_data_blkoff[i]);
+if (blk_off > ENTRIES_IN_SUM) {
+	f2fs_bug_on(1);
+	f2fs_put_page(page, 1);
+	return -EFAULT;
+}

Thanks,

> risk is low. IMHO, it can be safer.
> BTW, I found we can only check blk_off before for loop instead of
> checking 'j' in each iteratoin.
> 
> On Wed, Jul 03, 2019 at 10:07:11AM +0800, Chao Yu wrote:
>> Hi Ocean,
>>
>> If filesystem is corrupted, it should fail mount due to below check in
>> f2fs_sanity_check_ckpt(), so we are safe in read_compacted_summaries() to access
>> entries[0,blk_off], right?
>>
>> 	for (i = 0; i < NR_CURSEG_DATA_TYPE; i++) {
>> 		if (le32_to_cpu(ckpt->cur_data_segno[i]) >= main_segs ||
>> 			le16_to_cpu(ckpt->cur_data_blkoff[i]) >= blocks_per_seg)
>> 			return 1;
>>
>> Thanks,
>>
>> On 2019/7/2 16:05, Ocean Chen wrote:
>>> blk_off might over 512 due to fs corrupt.
>>> Use ENTRIES_IN_SUM to protect invalid memory access.
>>>
>>> v2:
>>> - fix typo
>>> Signed-off-by: Ocean Chen <oceanchen@...gle.com>
>>> ---
>>>  fs/f2fs/segment.c | 2 ++
>>>  1 file changed, 2 insertions(+)
>>>
>>> diff --git a/fs/f2fs/segment.c b/fs/f2fs/segment.c
>>> index 8dee063c833f..a5e8af0bd62e 100644
>>> --- a/fs/f2fs/segment.c
>>> +++ b/fs/f2fs/segment.c
>>> @@ -3403,6 +3403,8 @@ static int read_compacted_summaries(struct f2fs_sb_info *sbi)
>>>  
>>>  		for (j = 0; j < blk_off; j++) {
>>>  			struct f2fs_summary *s;
>>> +			if (j >= ENTRIES_IN_SUM)
>>> +				return -EFAULT;
>>>  			s = (struct f2fs_summary *)(kaddr + offset);
>>>  			seg_i->sum_blk->entries[j] = *s;
>>>  			offset += SUMMARY_SIZE;
>>>
> .
>