Re: [dpdk-dev] [RFC v2] eal: simplify the implementation of rte_ctrl_thread_create
From: Stephen Hemminger <stephen@networkplumber.org>
Date: 2021-08-25 21:58:00
On Wed, 25 Aug 2021 15:23:08 +0000 Honnappa Nagarahalli [off-list ref] wrote:
<snip>quoted
On 2021-08-02 07:16, Honnappa Nagarahalli wrote:quoted
The current described behaviour of rte_ctrl_thread_create is rigid which makes the implementation of the function complex. The behavior is abstracted to allow for simplified implementation.Have you considered using a POSIX condition variable instead of atomics for synchronization?No, I have not considered. The current implementation is complex because of the error handling of pthread_barrier_xxx APIs. I am thinking similar complexity will come in with the condition variable functions. I am using atomic variable in this patch to synchronize which does not need much error handling which simplifies the implementation. <snip>
The problem with condition variables is that it then adds a mutex as well.