linux-kernel - Re: [PATCH 4.19 33/75] f2fs: fix to avoid accessing xattr across the boundary

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] [day] [month] [year] [list]

Message-ID: <d77e58a9-561d-7fbd-bee8-592c7d21df64@huawei.com>
Date:   Thu, 20 Jun 2019 09:46:35 +0800
From:   Chao Yu <yuchao0@...wei.com>
To:     Pavel Machek <pavel@....cz>
CC:     <linux-kernel@...r.kernel.org>,
        Randall Huang <huangrandall@...gle.com>,
        Jaegeuk Kim <jaegeuk@...nel.org>,
        Sasha Levin <sashal@...nel.org>
Subject: Re: [PATCH 4.19 33/75] f2fs: fix to avoid accessing xattr across the
 boundary

Hi Pavel,

On 2019/6/19 20:32, Pavel Machek wrote:
> Hi!
> 
>> When we traverse xattr entries via __find_xattr(),
>> if the raw filesystem content is faked or any hardware failure occurs,
>> out-of-bound error can be detected by KASAN.
>> Fix the issue by introducing boundary check.
> 
> Ok, so this prevents fs corruption from causing problems,
> 
>> @@ -340,7 +347,11 @@ static int lookup_all_xattrs(struct inode *inode, struct page *ipage,
>>  	else
>>  		cur_addr = txattr_addr;
>>  
>> -	*xe = __find_xattr(cur_addr, index, len, name);
>> +	*xe = __find_xattr(cur_addr, last_txattr_addr, index, len, name);
>> +	if (!*xe) {
>> +		err = -EFAULT;
>> +		goto out;
>> +	}
> 
> Is -EFAULT suitable here? We do not have userspace passing pointers to us, we
> have fs corruption. -EUNCLEAN?

Oh, right, f2fs uses -EFAULT as error number to indicate filesystem is corrupted
all the time, we need to fix it to follow other generic fs.

> 
> Should it do some kind of printk to let the user know fs is corrupted, and mark
> it as needing fsck?

Agreed, let me add it. :)

Thanks,

> 
> Thanks,
> 									Pavel
> .
>