[dpdk-dev] [dpdk-stable] [PATCH] test/eventdev: fix sprintf with snprintf

Aaron Conole aconole at redhat.com
Wed Mar 13 15:35:54 CET 2019


Ferruh Yigit <ferruh.yigit at intel.com> writes:

> On 3/13/2019 1:43 PM, Aaron Conole wrote:
>> Ferruh Yigit <ferruh.yigit at intel.com> writes:
>> 
>>> On 3/12/2019 2:44 PM, Aaron Conole wrote:
>>>> "Parthasarathy, JananeeX M" <jananeex.m.parthasarathy at intel.com> writes:
>>>>
>>>>> Hi
>>>>>
>>>>>> -----Original Message-----
>>>>>> From: Parthasarathy, JananeeX M
>>>>>> Sent: Tuesday, February 19, 2019 6:33 PM
>>>>>> To: Aaron Conole <aconole at redhat.com>; Poornima, PallantlaX
>>>>>> <pallantlax.poornima at intel.com>
>>>>>> Cc: dev at dpdk.org; Pattan, Reshma <reshma.pattan at intel.com>; Rao, Nikhil
>>>>>> <nikhil.rao at intel.com>; stable at dpdk.org
>>>>>> Subject: RE: [dpdk-dev] [PATCH] test/eventdev: fix sprintf with snprintf
>>>>>>
>>>>>>
>>>>>>
>>>>>>> -----Original Message-----
>>>>>>> From: dev [mailto:dev-bounces at dpdk.org] On Behalf Of Aaron Conole
>>>>>>> Sent: Saturday, February 09, 2019 2:50 AM
>>>>>>> To: Poornima, PallantlaX <pallantlax.poornima at intel.com>
>>>>>>> Cc: dev at dpdk.org; Pattan, Reshma <reshma.pattan at intel.com>; Rao, Nikhil
>>>>>>> <nikhil.rao at intel.com>; stable at dpdk.org
>>>>>>> Subject: Re: [dpdk-dev] [PATCH] test/eventdev: fix sprintf with
>>>>>>> snprintf
>>>>>>>
>>>>>>> Pallantla Poornima <pallantlax.poornima at intel.com> writes:
>>>>>>>
>>>>>>>> sprintf function is not secure as it doesn't check the length of string.
>>>>>>>> More secure function snprintf is used.
>>>>>>>>
>>>>>>>> Fixes: 2a9c83ae3b ("test/eventdev: add multi-ports test")
>>>>>>>> Cc: stable at dpdk.org
>>>>>>>>
>>>>>>>> Signed-off-by: Pallantla Poornima <pallantlax.poornima at intel.com>
>>>>>>>> ---
>>>>>>>>  test/test/test_event_eth_rx_adapter.c | 3 ++-
>>>>>>>>  1 file changed, 2 insertions(+), 1 deletion(-)
>>>>>>>>
>>>>>>>> diff --git a/test/test/test_event_eth_rx_adapter.c
>>>>>>>> b/test/test/test_event_eth_rx_adapter.c
>>>>>>>> index 1d3be82b5..38f5c039f 100644
>>>>>>>> --- a/test/test/test_event_eth_rx_adapter.c
>>>>>>>> +++ b/test/test/test_event_eth_rx_adapter.c
>>>>>>>> @@ -479,7 +479,8 @@ adapter_multi_eth_add_del(void)
>>>>>>>>  	/* add the max port for rx_adapter */
>>>>>>>>  	port_index = rte_eth_dev_count_total();
>>>>>>>>  	for (; port_index < RTE_MAX_ETHPORTS; port_index += 1) {
>>>>>>>> -		sprintf(driver_name, "%s%u", "net_null", drv_id);
>>>>>>>> +		snprintf(driver_name, sizeof(driver_name), "%s%u", "net_null",
>>>>>>>> +				drv_id);
>>>>>>>>  		err = rte_vdev_init(driver_name, NULL);
>>>>>>>>  		TEST_ASSERT(err == 0, "Failed driver %s got %d",
>>>>>>>>  		driver_name, err);
>>>>>>>
>>>>>>> You call this a fix, but it's not possible for the value of drv_id to
>>>>>>> exceed '32' and the buffer size is plenty accommodating for that.  Did
>>>>>>> I miss something?  What is this fixing?
>>>>>>
>>>>>> It is better practice to use snprintf although in this case buffer will not overflow
>>>>>> as size is big enough to accommodate. The changes were done mainly to
>>>>>> replace sprintf to snprintf. Probably we can remove "fix" line as it is not issue in
>>>>>> this scenario.
>>>>>>
>>>>>> Thanks
>>>>>> M.P.Jananee
>>>>>
>>>>> Please suggest if we can remove "fix" line.
>>>>
>>>> This is a stylistic change, I don't think it's appropriate to call it a
>>>> fix, so I think you can remove the "Fixes" line.
>>>>
>>>> On further reflection, I actually think it will still be wrong.  If the
>>>> size buffer is ever changed, what will happen on truncation?  We don't
>>>> get an overflow any longer, but we still pass an invalid argument, so I
>>>> don't think this 'fix' is really even a fix.  It still has a bug -
>>>> albeit not one that immediately triggers SSP exception or stack
>>>> overflow.
>>>>
>>>> Makes sense?
>>>
>>> Hi Aaron,
>>>
>>> I see your point and I agree that existing code is not broken, it is functioning
>>> well as it is.
>>>
>>> But we are fixing a possible issue, or lets say fixing using less secure API
>>> although it doesn't cause any problem right now. Perhaps we can update the patch
>>> title slightly [1] but I am for keeping the fix and I think it makes sense to
>>> keep "Fixes" tag so that this update can be backported to stable trees.
>> 
>> I can get behind changing the sprintf to snprintf, since it is a better
>> API - but it needs to handle the return value properly (otherwise, in
>> this case we will specify an incorrect device).  I can even
>> understanding calling it a fix, it's metadata and is probably needed
>> from some kind of compliance anyway.
>> 
>> I also understand that this is in test suite, but people usually copy
>> code from test suites and that means the flaw at some point will be
>> propagated.  So I still think it should be a version which checks the
>> return code.  Otherwise in production if this is copied, and if I can
>> figure out how to overflow the counter knowing the buffer boundaries,
>> then there is a fixed device that will always be chosen.
>> 
>> I think it goes for all the other 's/sprintf\(/snprintf\)' replacements,
>> too.  Maybe I misunderstand something?
>
> These patches focus on preventing possible buffer overflow, the impact of
> possible truncation changes case by case I think, like for this case I don't see
> much benefit of adding return value check.
>
> For all cases I expect truncation trigger a functional error which should be
> already handled properly, like in this case 'rte_vdev_init()' will fail in
> second call if buffer is small.

And give the user a bad error ("I said net_null1038123825, not net_null10
- bug in dpdk!").

> There may be cases to check the return value, but that should be the case with
> 'sprintf' as well, changing API to 'snprintf' shouldn't require additional check
> by default.

I agree, that's true.  I think it's the right thing to do here, though.


More information about the dev mailing list