[Swan-dev] Pluto memory consumption

[Andrew Cagney]

On 22 March 2017 at 11:27, D. Hugh Redelmeier <hugh at mimosa.com> wrote:
> Off the top of my head (I'm really short of time; sorry):
>
> 1) leaks of O(1) are of no consequence.  There is only a small fixed
>    number of these events (I think) and so they are not a leak problem.

True, it makes book keeping harder.

> 2) I think that there was a prefix in the name of an allocation that
>    meant that the leak detective should ignore it.

I'd not do that here.  When these events fire they seem to allocate a
new pointer.  For instance:

$ grep EVENT_SHUNT_SCAN west.pluto.log
| event_schedule_tv: new EVENT_SHUNT_SCAN-pe at 0x7f5fda66c388
| inserting event EVENT_SHUNT_SCAN, timeout in 20.000000 seconds
| handling event EVENT_SHUNT_SCAN
| event_schedule_tv: new EVENT_SHUNT_SCAN-pe at 0x7f5fda6750f8
| inserting event EVENT_SHUNT_SCAN, timeout in 20.000000 seconds
| delete_pluto_event: release EVENT_SHUNT_SCAN-pe at 0x7f5fda66c388
leak: EVENT_SHUNT_SCAN, item size: 32