On Fri, Aug 23, 2019 at 04:55:19PM +0200, Henry Jensen wrote:
> Not sure if ports@ is correct for this. I am trying to build e2guardian[0]
> 5.3.3 on OpenBSD in order to replace dansguardian[1], which does exists
> in ports but is derelict.
>
>
> There is one obstacle. In FatController.cpp the program uses
> sigtimedwait(), which doesn't exist on OpenBSD. The code snippet reads:
>
> timeout.tv_sec = 5;
> timeout.tv_nsec = (long) 0;
> rc = sigtimedwait(&signal_set, NULL, &timeout);
> if (rc < 0) {
> if (errno != EAGAIN) {
> syslog(LOG_INFO, "%sUnexpected error from sigtimedwait() %d
> %s", thread_id.c_str(), errno, strerror(errno));
> }
> } else {
> if (rc == SIGUSR1)
> gentlereload = true;
> if (rc == SIGTERM)
> ttg = true;
> if (rc == SIGHUP)
> gentlereload = true;
> #ifdef DGDEBUG
> std::cerr << "signal:" << rc << std::endl;
> #endif
> if (o.logconerror) {
> syslog(LOG_INFO, "%ssigtimedwait() signal %d recd:",
> thread_id.c_str(), rc);
> }
> }
>
>
> I am able to read code, but I am not a programmer. What would be the
> appropriate way to replace sigtimedwait() here?
>
>
> Henry
>
> [0] http://e2guardian.org/cms/index.php
> [1] https://openports.pl/path/www/dansguardian
Hi,
We have sigtimedwait disabled in the kernel. What is left is to gather
what info userland expects to be given back from the kernel. I started
on this quest a while ago, but for now I suspended it.
99% of the use-cases I found had NULL for info (just like you do
bellow) and just like I did in mine. So I enabled that in a custom
kernel as a temporary fix for my personal use. The point of this being
that what we have works just fine for NULL info.
Paul
