No, all signal trigger currently contain two path: 1. append the signal to the pending list, and call the signal action through up_signal_xxx 2. call up_switch_context to wake up the target thread directly
like https://github.com/apache/nuttx/pull/17357/files#diff-a444aee5a4c951cdbad0c465c6afdc301f9fdb23a30200ee0e6a9191c031a4cfR554-R599 The partial disable just need skip the first path and keep the second one. The full disable need skip both path, On Fri, Nov 21, 2025 at 11:25 AM Gregory Nutt <[email protected]> wrote: > For the case of waking up a thread, any current use of signals would have > > to be replaced with a different IPC, as sem_post() that you suggest. > > > > There a also lot of hidden uses of signals in the OS for notifications > and > > things like device driver timeouts. Those uses would not be simple to > > replace. > > > > Would this mean that we would have special versions of notifications and > device drivers for systems that do not support signals. Can we manage > that? >
