netdev - Re: [PATCH] net: asix: fix uninit value in asix_mdio

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]

Date:   Sat, 14 Aug 2021 01:29:05 +0300
From:   Pavel Skripkin <paskripkin@...il.com>
To:     Andrew Lunn <andrew@...n.ch>
Cc:     davem@...emloft.net, kuba@...nel.org, linux@...pel-privat.de,
        himadrispandya@...il.com, linux-usb@...r.kernel.org,
        netdev@...r.kernel.org, linux-kernel@...r.kernel.org,
        syzbot+a631ec9e717fb0423053@...kaller.appspotmail.com
Subject: Re: [PATCH] net: asix: fix uninit value in asix_mdio_read

On 8/14/21 1:23 AM, Andrew Lunn wrote:
> On Fri, Aug 13, 2021 at 07:01:08PM +0300, Pavel Skripkin wrote:
>> Syzbot reported uninit-value in asix_mdio_read(). The problem was in
>> missing error handling. asix_read_cmd() should initialize passed stack
>> variable smsr, but it can fail in some cases. Then while condidition
>> checks possibly uninit smsr variable.
>> 
>> Since smsr is uninitialized stack variable, driver can misbehave,
>> because smsr will be random in case of asix_read_cmd() failure.
>> Fix it by adding error cheking and just continue the loop instead of
>> checking uninit value.
>> 
>> Fixes: 8a46f665833a ("net: asix: Avoid looping when the device is disconnected")
>> Reported-and-tested-by: syzbot+a631ec9e717fb0423053@...kaller.appspotmail.com
>> Signed-off-by: Pavel Skripkin <paskripkin@...il.com>
>> ---
>>  drivers/net/usb/asix_common.c | 8 +++++++-
>>  1 file changed, 7 insertions(+), 1 deletion(-)
>> 
>> diff --git a/drivers/net/usb/asix_common.c b/drivers/net/usb/asix_common.c
>> index ac92bc52a85e..572ca3077f8f 100644
>> --- a/drivers/net/usb/asix_common.c
>> +++ b/drivers/net/usb/asix_common.c
>> @@ -479,7 +479,13 @@ int asix_mdio_read(struct net_device *netdev, int phy_id, int loc)
>>  		usleep_range(1000, 1100);
>>  		ret = asix_read_cmd(dev, AX_CMD_STATMNGSTS_REG,
>>  				    0, 0, 1, &smsr, 0);
>> -	} while (!(smsr & AX_HOST_EN) && (i++ < 30) && (ret != -ENODEV));
>> +		if (ret == -ENODEV) {
>> +			break;
>> +		} else if (ret < 0) {
>> +			++i;
>> +			continue;
>> +		}
>> +	} while (!(smsr & AX_HOST_EN) && (i++ < 30));
> 
> No ret < 0, don't you end up with a double increment of i? So it will
> only retry 15 times, not 30?
> 
> Humm.
> 
> If ret < 0 is true, smsr is uninitialized? The continue statement
> causes a jump into the condition expression, where we evaluate smsr &
> AX_HOST_EN. Isn't this just as broken as the original version?
> 
>        Andrew
> 

Yes, you are right, I missed that, sorry. I will rewrote this loop into 
for loop in v2.

Im wondering why this wrong patch passed KMSAN testing...



With regards,
Pavel Skripkin