[ovs-dev] [PATCH v3 ovn 4/4] controller: binding: add memory accounting for local_lports and related_lports

[Lorenzo Bianconi]

> On 10/14/21 12:49, Dumitru Ceara wrote:
> > On 10/14/21 12:45 PM, Lorenzo Bianconi wrote:
> >>> On 9/25/21 12:19 AM, Lorenzo Bianconi wrote:
> >>>> Introduce memory accounting for:
> >>>> - binding local_lports map
> >>>> - binding related_lports map
> >>>>
> >>>> Signed-off-by: Lorenzo Bianconi <lorenzo.bianconi at redhat.com>
> >>>> ---
> >>>>  controller/binding.c        | 59 +++++++++++++++++++++++++++++++++++++
> >>>>  controller/binding.h        |  5 ++++
> >>>>  controller/ovn-controller.c |  5 ++--
> >>>>  3 files changed, 67 insertions(+), 2 deletions(-)
> >>>>
> >>>> diff --git a/controller/binding.c b/controller/binding.c
> >>>> index 661b4bb24..ec1e8bec7 100644
> >>>> --- a/controller/binding.c
> >>>> +++ b/controller/binding.c
> >>>> @@ -39,6 +39,31 @@
> >>>>  
> >>>>  VLOG_DEFINE_THIS_MODULE(binding);
> >>>>  
> >>>> +static uint64_t local_lports_usage;
> >>>> +static uint64_t related_lports_usage;
> >>>> +
> >>>> +static void
> >>>> +sset_mem_update(uint64_t *usage, struct sset *set,
> >>>> +                const char *name, bool erase)
> >>>> +{
> >>>> +    struct sset_node *node = sset_find(set, name);
> >>>> +
> >>>> +    if (!node && !erase) { /* add new element */
> >>>> +        *usage += (sizeof *node + strlen(name));
> >>>> +    } else if (node && erase) { /* remove an element */
> >>>> +        *usage -= (sizeof *node + strlen(name));
> >>>> +    }
> >>>> +}
> >>>> +
> >>>> +static void
> >>>> +sset_mem_clear(struct sset *set, uint64_t *usage)
> >>>> +{
> >>>> +    const char *name;
> >>>> +    SSET_FOR_EACH (name, set) {
> >>>> +        sset_mem_update(usage, set, name, true);
> >>>> +    }
> >>>> +}
> >>>> +
> >>>
> >>> I'm not too sure about this.  These functions are very generic.  I think
> >>> this should be part of the sset code itself.  Maybe a sset should just
> >>> automatically (or if configured) track all the memory it owns?  What do
> >>> you think?
> >>
> >> +Ilya
> >>
> >> I guess it would be a nice to have. sset are widely used but I do not think
> >> this patch will introduce a huge overhead.
> >>
> >> Regarding the specific patch we can remove it from the series in order to
> >> unlock it or we can have a local routines for the moment and then remove them
> >> when we have sset support. What do you think?
> > 
> > I don't really expect the local_lports and related_lports ssets to grow
> > too big so I think we can probably wait with this patch until we have a
> > more generic solution, maybe directly in sset.
> 
> I'm not sure about this.  Calling strlen for every add/remove
> sounds costly for a generic sset.

ok, so do you prefer to have some just local or add a switch to enable/disable it?

> 
> > 
> > Regards,
> > Dumitru
> > 
>