diff options
| author | Egor Tensin <Egor.Tensin@gmail.com> | 2023-06-12 01:42:08 +0200 |
|---|---|---|
| committer | Egor Tensin <Egor.Tensin@gmail.com> | 2023-06-13 01:37:08 +0200 |
| commit | 48ce9170b057ddd2165b0239a92aede15849f7a3 (patch) | |
| tree | c9928af6202081d9521107f1dc0ae362f54a6adc /src/signal.h | |
| parent | log: refactoring (diff) | |
| download | cimple-48ce9170b057ddd2165b0239a92aede15849f7a3.tar.gz cimple-48ce9170b057ddd2165b0239a92aede15849f7a3.zip | |
use signalfd to stop on SIGTERM
Is this an overkill? I don't know.
The thing is, correctly intercepting SIGTERM (also SIGINT, etc.) is
incredibly tricky. For example, before this commit, my I/O loops in
server.c and worker.c were inherently racy.
This was immediately obvious if you tried to run the tests. The tests
(especially the Valgrind flavour) would run a worker, wait until it
prints a "Waiting for a new command" line, and try to kill it using
SIGTERM. The problem is, the global_stop_flag check could have already
been executed by the worker, and it would hang forever in recv().
The solution seems to be to use signalfd and select()/poll(). I've never
used either before, but it seems to work well enough - at least the very
same tests pass and don't hang now.
Diffstat (limited to 'src/signal.h')
| -rw-r--r-- | src/signal.h | 9 |
1 files changed, 9 insertions, 0 deletions
diff --git a/src/signal.h b/src/signal.h index 4f1c280..e3f5897 100644 --- a/src/signal.h +++ b/src/signal.h @@ -8,6 +8,8 @@ #ifndef __SIGNAL_H__ #define __SIGNAL_H__ +#include "event_loop.h" + #include <signal.h> extern volatile sig_atomic_t global_stop_flag; @@ -18,4 +20,11 @@ int signal_block_stops(void); int signal_restore(const sigset_t *new); +int signalfd_create(const sigset_t *); +void signalfd_destroy(int fd); + +int signalfd_add_to_event_loop(int fd, struct event_loop *, event_loop_handler handler, void *arg); + +int signalfd_listen_for_stops(void); + #endif |