On Sun, Mar 09, 2025 at 10:10:29AM +0100, git_AT_suckless.org wrote:
> commit a70d5d2429abf8dcb70a8817990975dc9a621d27
> Author: NRK <nrk_AT_disroot.org>
> AuthorDate: Fri Sep 9 14:50:12 2022 +0600
> Commit: Hiltjo Posthuma <hiltjo_AT_codemadness.org>
> CommitDate: Sun Mar 9 10:02:34 2025 +0100
>
> simplify post-lock cmd and die if it fails
>
> this patch does two things:
>
> 0. simplify the code by using posix_spawn()
> 1. unify the behavior of what happens if the post-lock cmd fails.
>
> currently, if `fork()` fails, slock will die without locking the screen.
> HOWEVER if `execvp()` fails it prints a message to stderr (which the
> user cannot see since the screen has been locked already) and only exits
> the child while the parent locks the screen.
>
> to reproduce:
>
> # slock some_bin_that_doesnt_exist
>
> this behavior is inconsistent, if the idea is that post-lock cmd is
> _not_ important then we shouldn't `die()` on `fork()` failure either.
> and if we assume that the post-lock cmd _is_ important, then we should
> die on exec failure as well.
>
> this patch assumes the latter and calls `die()` if `posix_spawn()`
> fails.
>
> diff --git a/slock.c b/slock.c
> index b2f14e3..b5a9b04 100644
> --- a/slock.c
> +++ b/slock.c
> _AT_@ -13,6 +13,7 @@
> #include <stdio.h>
> #include <string.h>
> #include <unistd.h>
> +#include <spawn.h>
> #include <sys/types.h>
> #include <X11/extensions/Xrandr.h>
> #include <X11/keysym.h>
> _AT_@ -376,15 +377,12 @@ main(int argc, char **argv) {
>
> /* run post-lock command */
> if (argc > 0) {
> - switch (fork()) {
> - case -1:
> - die("slock: fork failed: %s\n", strerror(errno));
> - case 0:
> - if (close(ConnectionNumber(dpy)) < 0)
> - die("slock: close: %s\n", strerror(errno));
> - execvp(argv[0], argv);
> - fprintf(stderr, "slock: execvp %s: %s\n", argv[0], strerror(errno));
> - _exit(1);
> + pid_t pid;
> + extern char **environ;
> + int err = posix_spawnp(&pid, argv[0], NULL, NULL, argv, environ);
> + if (err) {
> + die("slock: failed to execute post-lock command: %s: %s\n",
> + argv[0], strerror(err));
> }
> }
>
>
Hi NRK,
Thanks for the patch! It looks good and I've pushed it to master.
--
Kind regards,
Hiltjo
Received on Sun Mar 09 2025 - 10:16:13 CET