On 02/17/2015 05:49 PM, Maxim Uvarov wrote:
On 02/17/2015 06:35 PM, Taras Kondratiuk wrote:
On 02/17/2015 05:26 PM, Maxim Uvarov wrote:
On 02/17/2015 06:09 PM, Taras Kondratiuk wrote:
On 02/17/2015 05:04 PM, Maxim Uvarov wrote:
On 02/17/2015 05:50 PM, Taras Kondratiuk wrote:
On 02/17/2015 04:42 PM, Maxim Uvarov wrote:
d)
loop of term and init functions as test for memory / resource leaks.
Current agreement is to have only one global init call per
process, so
we can't run them in cycle.
So we can not call global_term / global_init in the loop?
We can't.
If we can not call that in the loop then having global_term is really
strange. How can we know that it works or not?
As I thought before global_term is needed to shutdown everything
correctly so it should be possible to do another init.
Right. To do another init in a *new* process.
So you need it to be able to restart application several times.
If it's new process why do we then free all pools and buffers for
linux-generic? We can just terminate application and everything will be
cleaned.
Linux-generic doesn't free pools in global_term. It just checks whether
all pools are closed to make sure that application did clean-up
correctly. Buffers are flushed from local cache, but that is in
local_term.
_______________________________________________
lng-odp mailing list
[email protected]
http://lists.linaro.org/mailman/listinfo/lng-odp
