| Commit message (Collapse) | Author | Age | Files | Lines |
|
|
|
|
|
| |
orderly released, avoid use after free.
Detected via valgrind.
|
|
|
|
|
|
|
|
|
|
| |
absolute timeout_time, avoiding deadlocks
Classes affected incl. API change using fraction for relative time
- latch
- ringbuffer
- service_runner
- simple_timer
|
|
|
|
| |
false if 'running'; Add and expose mutex and condition_variable for 'shall_stop' change, allowing clients to wait or interrupt by such event (e.g. timer)
|
| |
|
|
|
|
| |
CaptureRefInvocationFunc; Add function call operator.
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
fix using service_shutdown_timeout_ms(), fix API & doc
Remove service_end_post_notify (data race leak)
- a post lock service_runner callback at EOL may introduce a critical data race memory leak condition,
i.e. use after free off-thread.
Fix using service_shutdown_timeout_ms()
- Have stop() and join() return boolean true only if successful.
- A timeout results in return false
Add join()
- Allow user to wait until service ends,
- which helps avoiding pulling resources from running service
Fix API & doc
- Doc: Clarify relationship of service_shutdown_timeout_ms, stop() and join().
- Simplify getter method-names, remove 'get_'.
|
|
|
|
| |
shall only proceed if !running
|
|
user services
|