Differential D5890
net/ocserv: update to 0.11.1 Authored by cpm on Apr 8 2016, 11:10 PM. Tags None Referenced Files
Subscribers None
Details
Changelog: http://lists.infradead.org/pipermail/openconnect-devel/2016-March/003551.html ocserv port build/install fine but core dump if it fork to background. http://lists.infradead.org/pipermail/openconnect-devel/2016-March/003552.html
Diff Detail
Event TimelineComment Actions Regarding this, http://lists.infradead.org/pipermail/openconnect-devel/2016-March/003552.html, what's the next step if there is a runtime error? Are we going to wait for an upstream release or patch to fix it?
Comment Actions I'm working in a patch to submit upstream which fix the invalid fd. Also, I'm in contact with the ocserv developer. Comment Actions The port fails to build at this point Making all in doc
gmake[4]: Entering directory '/usr/ports/net/ocserv/work/ocserv-0.11.1/doc'
gsed 's/@subheading \(.*\)/@*\n@var{\1}\n@*/' \
../src/ocserv-args.def > ../src/ocserv-args.def.tmp && \
autogen -L../src -DMAN_SECTION=8 -Tagman-cmd.tpl \
../src/ocserv-args.def.tmp && \
rm -f ../src/ocserv-args.def.tmp
gsed 's/@subheading \(.*\)/@*\n@var{\1}\n@*/' \
../src/ocpasswd/args.def > ../src/ocpasswd/args.def.tmp && \
autogen -L../src -DMAN_SECTION=8 -Tagman-cmd.tpl \
../src/ocpasswd/args.def.tmp && \
rm -f ../src/ocpasswd/args.def.tmp
gsed 's/@subheading \(.*\)/@*\n@var{\1}\n@*/' \
../src/occtl/args.def > ../src/occtl/args.def.tmp && \
autogen -L../src -DMAN_SECTION=8 -Tagman-cmd.tpl \
../src/occtl/args.def.tmp && \
rm -f ../src/occtl/args.def.tmp
sed -I -e 's/^\.NOP //' occtl.8
sed -I -e 's/^\.NOP //' ocpasswd.8
sed: ocpasswd.8: No such file or directory
Makefile:1145: recipe for target 'ocpasswd.8' failed
gmake[4]: *** [ocpasswd.8] Error 1
gmake[4]: *** Waiting for unfinished jobs....
sed -I -e 's/^\.NOP //' ocserv.8
gmake[4]: Leaving directory '/usr/ports/net/ocserv/work/ocserv-0.11.1/doc'
Makefile:1011: recipe for target 'all-recursive' failed
gmake[3]: *** [all-recursive] Error 1
gmake[3]: Leaving directory '/usr/ports/net/ocserv/work/ocserv-0.11.1'
Makefile:952: recipe for target 'all' failed
gmake[2]: *** [all] Error 2
gmake[2]: Leaving directory '/usr/ports/net/ocserv/work/ocserv-0.11.1'
===> Compilation failed unexpectedly.
Try to set MAKE_JOBS_UNSAFE=yes and rebuild before reporting the failure to
the maintainer.
*** Error code 1'MAKE_JOBS_UNSAFE=yes' has effect Comment Actions Ok, then it would be appropriate to mention it in the commit message like this (you can use triple backticks for code quotes here): - Update to 0.11.1 release - Convert LICENSE= "GPLxx # or later" to "GPLxx+" - Add dependency on libev - Regenerate patches - Add MAKE_JOBS_UNSAFE; port attempts to modify man page before generation sed: ocpasswd.8: No such file or directory Changelog: http://lists.infradead.org/pipermail/openconnect-devel/2016-March/003551.html Approved by: junovitch (mentor) Differential Revision: D5890 Perfect. Once we factor in a plan to address the runtime issue I'll mark this approved. Keep up the good work. Comment Actions Understood!
The following patch fixes the bad file descriptor that prevents ocserv running in the background. --- src/main.c.orig 2016-03-06 11:14:35 UTC +++ src/main.c @@ -77,7 +77,7 @@ static void listen_watcher_cb (EV_P_ ev_ int syslog_open = 0; sigset_t sig_default_set; -struct ev_loop *loop; +struct ev_loop *loop = NULL; /* EV watchers */ ev_io ctl_watcher; @@ -147,8 +147,9 @@ int y; #elif defined(IP_RECVDSTADDR) /* *BSD */ if (family == AF_INET) { y = 1; - if (setsockopt(fd, IPPROTO_IP, IP_RECVDSTADDR, - (const void *)&y, sizeof(y)) < 0) + if (family == AF_INET && + setsockopt(fd, IPPROTO_IP, IP_RECVDSTADDR, + (const void *)&y, sizeof(y)) < 0) perror("setsockopt(IP_RECVDSTADDR) failed"); } #endif @@ -661,13 +662,15 @@ void clear_lists(main_server_st *s) ctl_handler_deinit(s); main_ban_db_deinit(s); - /* clear libev state */ - ev_io_stop (loop, &ctl_watcher); - ev_io_stop (loop, &sec_mod_watcher); - ev_child_stop (loop, &child_watcher); - ev_timer_stop(loop, &maintainance_watcher); - /* free memory by the event loop */ - ev_loop_destroy (loop); + if (loop) { + /* clear libev state */ + ev_io_stop (loop, &ctl_watcher); + ev_io_stop (loop, &sec_mod_watcher); + ev_child_stop (loop, &child_watcher); + ev_timer_stop(loop, &maintainance_watcher); + /* free memory by the event loop */ + ev_loop_destroy (loop); + } } /* A UDP fd will not be forwarded to worker process before this number of @@ -1165,12 +1168,6 @@ int main(int argc, char** argv) memset(&creds, 0, sizeof(creds)); - loop = EV_DEFAULT; - if (loop == NULL) { - fprintf(stderr, "could not initialise libev\n"); - exit(1); - } - /* main pool */ main_pool = talloc_init("main"); if (main_pool == NULL) { @@ -1253,6 +1250,12 @@ int main(int argc, char** argv) exit(1); } + loop = EV_DEFAULT; + if (loop == NULL) { + fprintf(stderr, "could not ititialise libev\n"); + exit(1); + } + mslog(s, NULL, LOG_INFO, "initialized %s", PACKAGE_STRING); /* chdir to our chroot directory, to allow opening the sec-mod Comment Actions cpm send me a shorter patch which would fix the run-time problem as well. I tested it, it works. From my point of view it can be updated. Comment Actions Is there something wrong with the upstream on this? At least right now we can fetch from distcache but we will immediately break the port if we update right now. => ocserv-0.11.1.tar.xz doesn't seem to exist in /usr/ports/distfiles/. => Attempting to fetch ftp://ftp.infradead.org/pub/ocserv/ocserv-0.11.1.tar.xz fetch: ftp://ftp.infradead.org/pub/ocserv/ocserv-0.11.1.tar.xz: Protocol error => Attempting to fetch http://distcache.FreeBSD.org/ports-distfiles/ocserv-0.11.1.tar.xz fetch: http://distcache.FreeBSD.org/ports-distfiles/ocserv-0.11.1.tar.xz: Not Found => Couldn't fetch it - please try to retrieve this => port manually into /usr/ports/distfiles/ and try again. Comment Actions Once we can actually fetch it, the update, patch to handle the runtime regression in the new version and commit message all look fine. | ||||||||||||||||||||||||||||||||||||||