linux-kernel - Re: [PATCH] USB: realtek_cr: fix return check for dma functions

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:   Tue, 11 Aug 2020 11:54:28 -0700
From:   Tom Rix <trix@...hat.com>
To:     Alan Stern <stern@...land.harvard.edu>
Cc:     gregkh@...uxfoundation.org, acozzette@...hmc.edu,
        linux-usb@...r.kernel.org, usb-storage@...ts.one-eyed-alien.net,
        linux-kernel@...r.kernel.org
Subject: Re: [PATCH] USB: realtek_cr: fix return check for dma functions


On 8/11/20 10:53 AM, Alan Stern wrote:
> On Tue, Aug 11, 2020 at 10:29:29AM -0700, Tom Rix wrote:
>> On 8/11/20 9:03 AM, Alan Stern wrote:
>>> On Tue, Aug 11, 2020 at 08:15:05AM -0700, trix@...hat.com wrote:
>>>> From: Tom Rix <trix@...hat.com>
>>>>
>>>> clang static analysis reports this representative problem
>>>>
>>>> realtek_cr.c:639:3: warning: The left expression of the compound
>>>>   assignment is an uninitialized value. The computed value will
>>>>   also be garbage
>>>>     SET_BIT(value, 2);
>>>>     ^~~~~~~~~~~~~~~~~
>>>>
>>>> value is set by a successful call to rts51x_read_mem()
>>>>
>>>> 	retval = rts51x_read_mem(us, 0xFE77, &value, 1);
>>>> 	if (retval < 0)
>>>> 		return -EIO;
>>>>
>>>> A successful call to rts51x_read_mem returns 0, failure can
>>>> return positive and negative values.  This check is wrong
>>>> for a number of functions.  Fix the retval check.
>>>>
>>>> Fixes: 065e60964e29 ("ums_realtek: do not use stack memory for DMA")
>>>> Signed-off-by: Tom Rix <trix@...hat.com>
>>>> ---
>>>>  drivers/usb/storage/realtek_cr.c | 36 ++++++++++++++++----------------
>>>>  1 file changed, 18 insertions(+), 18 deletions(-)
>>>>
>>>> diff --git a/drivers/usb/storage/realtek_cr.c b/drivers/usb/storage/realtek_cr.c
>>>> index 3789698d9d3c..b983753e2368 100644
>>>> --- a/drivers/usb/storage/realtek_cr.c
>>>> +++ b/drivers/usb/storage/realtek_cr.c
>>>> @@ -481,16 +481,16 @@ static int enable_oscillator(struct us_data *us)
>>>>  	u8 value;
>>>>  
>>>>  	retval = rts51x_read_mem(us, 0xFE77, &value, 1);
>>>> -	if (retval < 0)
>>>> +	if (retval != STATUS_SUCCESS)
>>>>  		return -EIO;
>>> Instead of changing all these call sites, wouldn't it be a lot easier 
>>> just to change rts51x_read_mem() to make it always return a negative 
>>> value (such as -EIO) when there's an error?
>>>
>>> Alan Stern
>> I thought about that but there was already existing (retval != 
>> STATUS_SUCCESS) checks for these calls.
> The only values that routine currently returns are 
> USB_STOR_TRANSPORT_ERROR, -EIO, and 0.  None of the callers distinguish 
> between the first two values, so you can just change the first to the 
> second.
>
> Note that STATUS_SUCCESS is simply 0.

Yes, i noted all of these already. My change is consistent with the existing correct checks.  consistency is important.  returning a neg value to reuse the exiting check should mean the STATUS_SUCCESS != 0 checks are changed to neg check.  i can do this larger change if required.

Tom

>
> Alan Stern
>