+1 by me.

> On Jan 8, 2019, at 7:09 AM, Stefan Sperling <s...@apache.org> wrote:
> 
> See https://svn.apache.org/r1663375 for a related fix by covener.
> Which, by the way, should probably be backported to 2.4; I see a failure
> on a buildbot which deliberately builds with non-threaded APR to ensure
> that this configuration remains in SVN's test matrix:
> https://ci.apache.org/builders/svn-bb-openbsd/builds/272/steps/Build/logs/stdio
> 
> This patch hasn't even been compile tested but looks obvious enough.
> Fine to commit?
> 
> Index: modules/test/mod_dialup.c
> ===================================================================
> --- modules/test/mod_dialup.c (revision 1850735)
> +++ modules/test/mod_dialup.c (working copy)
> @@ -107,7 +107,9 @@ dialup_callback(void *baton)
>     dialup_baton_t *db = (dialup_baton_t *)baton;
>     conn_rec *c = db->r->connection;
> 
> +#if APR_HAS_THREADS
>     apr_thread_mutex_lock(db->r->invoke_mtx);
> +#endif
> 
>     status = dialup_send_pulse(db);
> 
> @@ -115,7 +117,9 @@ dialup_callback(void *baton)
>         ap_mpm_register_timed_callback(apr_time_from_sec(1), dialup_callback, 
> baton);
>     }
>     else if (status == DONE) {
> +#if APR_HAS_THREADS
>         apr_thread_mutex_unlock(db->r->invoke_mtx);
> +#endif
>         ap_finalize_request_protocol(db->r);
>         ap_process_request_after_handler(db->r);
>         return;
> @@ -127,7 +131,9 @@ dialup_callback(void *baton)
>         ap_die(status, db->r);
>     }
> 
> +#if APR_HAS_THREADS
>     apr_thread_mutex_unlock(db->r->invoke_mtx);
> +#endif
> 
>     ap_mpm_resume_suspended(c);
> }

Reply via email to