(xulrunner)

===> Leaving ``all'' after barrier for xulrunner-21.0
9035.74s 1372.18s 2:53:38.83 99% 0+0k
80+31214io 60pf+0w 0.04s 1.10s 2:53:38.83 0% 0+0k 2+1883io 0pf+0w
m08@makoto 13:59:18/130516(..wip/xulrunner)% 

seems to forget the MAKE_JOBS=4 line in /etc/mk.conf
(firefox)

===> Leaving ``all'' after barrier for firefox-21.0
175.46s 135.22s 4:59.60 103% 0+0k 1693+10232io 1396pf+0w
0.01s 0.23s 4:59.60 0% 0+0k 1+42io 0pf+0w
m08@makoto 14:27:42/130516(..wip/firefox)% 

Strange enough, the very first run after make package-install starts firefox,

m08@makoto 14:28:47/130516(..wip/firefox)% firefox
int XREMain::XRE_main(int, char**, const nsXREAppData*): 
   /export/CHROOT/usr/pkgsrc/wip/xulrunner/work/mozilla-release/toolkit/xre/nsAppRunner.cpp:3911

(process:28092): GLib-CRITICAL **: g_slice_set_config: assertion `sys_page_size == 0' failed
int XREMain::XRE_main(int, char**, const nsXREAppData*): 
   /export/CHROOT/usr/pkgsrc/wip/xulrunner/work/mozilla-release/toolkit/xre/nsAppRunner.cpp:3916
m08@makoto 14:28:52/130516(..wip/firefox)% 

but help says it is 20.0.1. The second invocation does not start.

m08@makoto 14:32:18/130516(..wip/firefox)% firefox                               
int XREMain::XRE_main(int, char**, const nsXREAppData*): 
  /export/CHROOT/usr/pkgsrc/wip/xulrunner/work/mozilla-release/toolkit/xre/nsAppRunner.cpp:3911

(process:19187): GLib-CRITICAL **: g_slice_set_config: assertion `sys_page_size == 0' failed
int XREMain::XRE_main(int, char**, const nsXREAppData*): 
  /export/CHROOT/usr/pkgsrc/wip/xulrunner/work/mozilla-release/toolkit/xre/nsAppRunner.cpp:3916
int XREMain::XRE_main(int, char**, const nsXREAppData*): 
  /export/CHROOT/usr/pkgsrc/wip/xulrunner/work/mozilla-release/toolkit/xre/nsAppRunner.cpp:4013

The first one did not get the line 4013.

--- toolkit/xre/nsAppRunner.cpp.orig	2013-05-12 04:21:47.000000000 +0900
+++ toolkit/xre/nsAppRunner.cpp	2013-05-16 10:03:05.000000000 +0900
@@ -3908,11 +3908,12 @@ XREMain::XRE_main(int argc, char* argv[]
   // Disable the slice allocator, since jemalloc already uses similar layout
   // algorithms, and using a sub-allocator tends to increase fragmentation.
   // This must be done before g_thread_init() is called.
+fprintf(stderr, "%s: %s:%d\n",__func__, __FILE__,__LINE__);
   g_slice_set_config(G_SLICE_CONFIG_ALWAYS_MALLOC, 1);
 #endif
   g_thread_init(NULL);
 #endif
-
+fprintf(stderr, "%s: %s:%d\n",__func__, __FILE__,__LINE__);
   // init
   bool exit = false;
   int result = XRE_mainInit(aAppData, &exit);
@@ -4009,7 +4010,7 @@ XREMain::XRE_main(int argc, char* argv[]
   XRE_DeinitCommandLine();
 
   SAMPLER_SHUTDOWN();
-
+fprintf(stderr, "%s: %s:%d\n",__func__, __FILE__,__LINE__);
   return NS_FAILED(rv) ? 1 : 0;
 }