[ovs-dev] [PATCH] odp-util: Fix abort while formatting nsh actions.

[Ilya Maximets]

On 1/15/21 9:40 PM, Mark Gray wrote:
> On 12/01/2021 23:55, Ilya Maximets wrote:
>> On 1/11/21 7:22 PM, Mark Gray wrote:
>>> On 18/12/2020 13:34, Ilya Maximets wrote:
>>>> OVS should not exit if it cannot format NSH actions for the user.
>>>> It should just report the error like the other formatting functions do.
>>>>
>>>> Credit to OSS-Fuzz.
>>>>
>>>> Reported-at: https://bugs.chromium.org/p/oss-fuzz/issues/detail?id=21509
>>>> Fixes: 1fc11c5948cf ("Generic encap and decap support for NSH")
>>>> Signed-off-by: Ilya Maximets <i.maximets at ovn.org>
>>>> ---
>>>>  lib/odp-util.c | 3 ++-
>>>>  1 file changed, 2 insertions(+), 1 deletion(-)
>>>>
>>>> diff --git a/lib/odp-util.c b/lib/odp-util.c
>>>> index 879dea97e..bf19fa647 100644
>>>> --- a/lib/odp-util.c
>>>> +++ b/lib/odp-util.c
>>>> @@ -391,7 +391,8 @@ format_odp_push_nsh_action(struct ds *ds,
>>>>          break;
>>>>      }
>>>>      default:
>>>> -        OVS_NOT_REACHED();
>>>
>>> As we are not abort()ing, I wonder should we additionally log that this
>>> is unexpected. I am not sure that the problem would be would be entirely
>>> obvious from an output like the following. Also, it should probably be
>>> logged explicitly as an error.
>>
>> These functions (format_odp_*_action) creates the string for some other
>> parts of OVS.  Most likely this is already formatting for a log message
>> and in most cases this is an error message due to some problems with
>> this particular set of actions.  So, having one more extra log message
>> doesn't make much sense.  Anyway, none of the other similar functions
>> does that.
> 
> Ok, if you are happy it follows this same pattern, then leave as is.
> 
> Acked-by: Mark Gray <mark.d.gray at redhat.com>

Thanks!
Applied to master and backported down to 2.8.

Best regards, Ilya Maximets.