On 18 April 2016 at 15:40, Martin Pieuchot <[email protected]> wrote:
> On 18/04/16(Mon) 12:30, Mike Belopuhov wrote:
>> On Mon, Apr 11, 2016 at 15:29 +0200, Martin Pieuchot wrote:
>> > Instead of rtfree(9)ing the cached route after using it, if it is a
>> > multipath one, free it before.
>> >
>> > Ok?
>> >
>>
>> This changes the case where ipforward_rt.ro_rt is NULL and
>> the route that we allocate is RTF_MPATH (is that possible?).
>> Before we would free it in the freert branch, but now we
>> keep it around.  Is that an oversight or an intended change?
>
> It's the intended change.  The end result is the same: the kernel always
> do a lookup if the previous result was an RTF_MPATH entry.
>

OK, AFAICT it's fine.  OK mikeb

Reply via email to