[Mono-osx] anybody noticed a mach port leak?

[Allan Hsu]

On Jun 8, 2006, at 7:56 PM, Miguel de Icaza wrote:

>> I'm having a hard time tracking down a mach port leak in imeem. Some
>> roundabout debugging leads me to believe it's somewhere in the
>> networking-related portions of either our managed code or the Mono
>> runtime. Has anybody else noticed their mach port counts rising over
>> time? If so, what does your application do?
>
> A test case narrowing the problem would be the first step.

On a whim, I tried watching the port count with the racy.c code from  
bug 77470 and it seems that it causes a similar (possibly the same)  
port leak. I've filed a bug here:

http://bugs.ximian.com/show_bug.cgi?id=78615

I've also identified a few spots in the darwin_stop_world.c where it  
looks like mach_port_deallocate() isn't being called where it should  
be. I've written a patch, but I had to leave the office before Mono  
finished building. I have a scenario in my mind where the bug I filed  
and the mach_port_deallocate() thing are both related. I'll know more  
tomorrow.

	-Allan

--
Allan Hsu <allan at counterpop dot net>
1E64 E20F 34D9 CBA7 1300  1457 AC37 CBBB 0E92 C779