On Sun, Jan 31, 2021, 7:34 AM Matias N. <mat...@imap.cc> wrote: > I was refering to the wait for the signal on the worker thread, which > would simply be waiting on a semaphore/condition variable which indicates > work (pending notification) is available. > > See here for how I changed how nimBLE uses SIGEV_THREAD due to the > aforementioned issue: > > > https://github.com/apache/mynewt-nimble/blob/241911bd02b9bd3445f0194dab4539281fa3f849/porting/npl/nuttx/src/os_callout.c > > In this case I'm using a condition variable. > > Any signals which are not desired from the thread can be masked, so no > reason for it to be interrupted. >
Masking the signals seems to be basically reimplementing what sigwaitinfo does by allowing you to filter what signals are used. What about sigwaitinfo do you think is adding overhead, maybe that will help me understand better. --Brennan >
