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: <554b5c77-12ba-7b85-a65a-f1fefa5b6a51@gmail.com>
Date:   Sat, 11 Jun 2022 11:15:26 +0530
From:   Shreenidhi Shedi <yesshedi@...il.com>
To:     Greg KH <gregkh@...uxfoundation.org>
Cc:     arnd@...db.de, linux-kernel@...r.kernel.org,
        Shreenidhi Shedi <sshedi@...are.com>
Subject: Re: [PATCH 1/2] char: lp: ensure that index has not exceeded LP_NO

On 10/06/22 8:00 pm, Greg KH wrote:
> On Fri, Jun 10, 2022 at 07:12:02PM +0530, Shreenidhi Shedi wrote:
>> On 10/06/22 6:58 pm, Greg KH wrote:
>>> On Fri, Jun 03, 2022 at 06:30:39PM +0530, Shreenidhi Shedi wrote:
>>>> From: Shreenidhi Shedi <sshedi@...are.com>
>>>>
>>>> After finishing the loop, index value can be equal to LP_NO and lp_table
>>>> array is of size LP_NO, so this can end up in accessing an out of bound
>>>> address in lp_register function.
>>>>
>>>> Signed-off-by: Shreenidhi Shedi <sshedi@...are.com>
>>>> ---
>>>>  drivers/char/lp.c | 2 +-
>>>>  1 file changed, 1 insertion(+), 1 deletion(-)
>>>>
>>>> diff --git a/drivers/char/lp.c b/drivers/char/lp.c
>>>> index 0e22e3b0a..d474d02b6 100644
>>>> --- a/drivers/char/lp.c
>>>> +++ b/drivers/char/lp.c
>>>> @@ -972,7 +972,7 @@ static void lp_attach(struct parport *port)
>>>>  			if (port_num[i] == -1)
>>>>  				break;
>>>>
>>>> -		if (!lp_register(i, port))
>>>> +		if (i < LP_NO && !lp_register(i, port))
>>>>  			lp_count++;
>>>
>>> How can this ever be needed?  Look at the check further up for the check
>>> of lp_count which prevents this from every going too large.
>>>
>>> So how can an address be accessed out of bound here?
>>>
>>> thanks,
>>>
>>> greg k-h
>>
>> Thanks for the review. Assume lp_count is less than LP_NO now and we enter the for loop
>> and for some reason for loop exits after i reaching the value LP_NO
> 
> Wait, how can that happen?  That's what I am saying, the loop will never
> reach that value from what I can tell.
> 
> Yes, this whole thing should be moved to something more sane like an
> idr structure, but as-is, it seems correct to me.
> 
> Have you tested the code with that many devices to see if it really can
> overflow?
> 
> thanks,
> 
> greg k-h

No, I did not actually test it with real hardware but I ran a 
static analyzer check on this file and it also thinks the same.

```
Checking drivers/char/lp.c ...
drivers/char/lp.c:926:10: error: inconclusive: Array 'lp_table[8]' accessed at index 8, which is out of bounds. [arrayIndexOutOfBounds]
 lp_table[nr].dev = parport_register_dev_model(port, "lp",
         ^
drivers/char/lp.c:971:17: note: Assuming that condition 'i<8' is not redundant
  for (i = 0; i < LP_NO; i++)
                ^
drivers/char/lp.c:975:20: note: Calling function 'lp_register', 1st argument 'i' value is 8
  if (!lp_register(i, port))
                   ^
drivers/char/lp.c:926:10: note: Array index out of bounds
 lp_table[nr].dev = parport_register_dev_model(port, "lp",
         ^
drivers/char/lp.c:928:14: error: inconclusive: Array 'lp_table[8]' accessed at index 8, which is out of bounds. [arrayIndexOutOfBounds]
 if (lp_table[nr].dev == NULL)
             ^
drivers/char/lp.c:971:17: note: Assuming that condition 'i<8' is not redundant
  for (i = 0; i < LP_NO; i++)
                ^
drivers/char/lp.c:975:20: note: Calling function 'lp_register', 1st argument 'i' value is 8
  if (!lp_register(i, port))
                   ^
drivers/char/lp.c:928:14: note: Array index out of bounds
 if (lp_table[nr].dev == NULL)
```

-- 
Shedi

Powered by blists - more mailing lists

Powered by Openwall GNU/*/Linux Powered by OpenVZ