linux-kernel - Re: [PATCH] wireless: ath6kl: fix out of bound from length.

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: <20220721070707.GF2316@kadam>
Date:   Thu, 21 Jul 2022 10:07:07 +0300
From:   Dan Carpenter <dan.carpenter@...cle.com>
To:     xiaolinkui <xiaolinkui@...il.com>
Cc:     kvalo@...nel.org, davem@...emloft.net, edumazet@...gle.com,
        kuba@...nel.org, pabeni@...hat.com, gustavoars@...nel.org,
        quic_jjohnson@...cinc.com, keescook@...omium.org, johan@...nel.org,
        linux-wireless@...r.kernel.org, netdev@...r.kernel.org,
        linux-kernel@...r.kernel.org, Linkui Xiao <xiaolinkui@...inos.cn>
Subject: Re: [PATCH] wireless: ath6kl: fix out of bound from length.

On Thu, Jul 21, 2022 at 11:21:58AM +0800, xiaolinkui wrote:
> From: Linkui Xiao <xiaolinkui@...inos.cn>
> 
> If length from debug_buf.length is 4294967293 (0xfffffffd), the result of
> ALIGN(size, 4) will be 0.
> 
> 	length = ALIGN(length, 4);
> 
> In case of length == 4294967293 after four-byte aligned access, length will
> become 0.
> 
> 	ret = ath6kl_diag_read(ar, address, buf, length);
> 
> will fail to read.

It looks like "length" is untrustworthy.  Generally, I kind of distrust
all endian data by default, but I dug a bit deeper and I don't trust it.

Unfortunately, if "length" is larger than ATH6KL_FWLOG_PAYLOAD_SIZE
(1500) then we are screwed.  Can you add a check for that instead?
Please check my work on this because I didn't look *super* carefully.

No need to make any changes to the types, just add the upper bounds
check on ATH6KL_FWLOG_PAYLOAD_SIZE.  The type changes didn't fix the bug
on 32 bit systems anyway...

regards,
dan carpenter