[Mono-bugs] [Bug 75424][Maj] Changed - wapi caused build hangs

bugzilla-daemon at bugzilla.ximian.com bugzilla-daemon at bugzilla.ximian.com
Fri Jul 1 06:30:44 EDT 2005


Please do not reply to this email- if you want to comment on the bug, go to the
URL shown below and enter your comments there.

Changed by dick at ximian.com.

http://bugzilla.ximian.com/show_bug.cgi?id=75424

--- shadow/75424	2005-07-01 03:39:46.000000000 -0400
+++ shadow/75424.tmp.13874	2005-07-01 06:30:44.000000000 -0400
@@ -78,6 +78,28 @@
 I've committed a patch to run semdel on 'make clean'.  Aborted builds
 aren't a problem, I think, since the .wapi directory will still be
 around.  The problem is deleting the .wapi directory without releasing
 the resources in it.  I think that nuking the build tree is a rare
 enough occurance (apart from the buildbot, that is :-) that we can let
 a few resources leak.
+
+------- Additional Comments From dick at ximian.com  2005-07-01 06:30 -------
+Except that buildbot will still fall over every now and then because
+of this leak.
+
+Running semdel inside mono-wrapper will work, so long as you can
+guarantee that only one mono process is accessing that set of shared
+memory.  It doesn't matter that sequential processes have a different
+semaphore, because there still will be no cross-process contention. 
+Alternatively, running semdel after the mcs build in all-local might
+be a better place if you can preserve the mcs failure status for make.
+
+Currently, semdel will exit(-1) if it can't attach the shared memory.
+ That is easily changed to fit better in the Makefiles.
+
+Regarding debugging, it won't make any difference to me - I always
+debug with the installed runtime that I just built, not from the
+source tree.
+
+Buildbot question: when it runs the nunit tests, does it use the build
+tree shared memory or the file in the user's home dir?
+


More information about the mono-bugs mailing list