Ack.
Thanks,
Ramesh.
On 1/25/2017 5:19 PM, Anders Widell wrote:
> The *_hdl() functions are also good candidates for removal. :-)
>
> The timer destruction utilizes multiple mechanisms for
> synchronization: a flag called "tmr_destroying", and a selection
> object called tmr_destroy_syn_obj.
The *_hdl() functions are also good candidates for removal. :-)
The timer destruction utilizes multiple mechanisms for synchronization:
a flag called "tmr_destroying", and a selection object called
tmr_destroy_syn_obj. Presumably, these mechanisms are enough for
synchronizing the destruction
Hi Anders,
Across OpenSAF code, ncshm_<...>_hdl() functions are serving the similar
purpose.
The reason for not noticing inconsistency in handling ncslpg_<>
functions in sysf_tmr can be due to
1. Threads that are using timer calls are gracefully exited before the
main thread exits (or)
Let't not become sentimental about old broke code. As noted, this
mechanism is currently broken and doesn't do anything useful in its
current form. I doubt that it has ever worked during the history of the
OpenSAF project. And if it was really needed, it is doubtful why it
would only be needed
Hi Anders,
Quick comments.
The purpose of having ncslpg_<...>() functions in sysf_tmr is to
make sure to go through gracefull shutdown process.
sysfTmrDestroy() is suppose to wait till all threads using sysfTmr
to complete their tasks. Observed that ncslpg_destroy() should have
src/base/hj_hdl.c | 96 +-
src/base/ncs_hdl_pub.h | 22 +--
src/base/sysf_tmr.c| 34 ++---
3 files changed, 6 insertions(+), 146 deletions(-)
Remove the "local persistence guard" API since it is only referenced by