[omniORB] Assertion failed

[Sai-Lai Lo]

Helmut,

Your trace is completely normal. Before we look further, I want you to
disable the scavengers. If your process does not crash, then there may
still be problems with the scavengers to look into.

On the other hand, it could also be that in your application, you kept object 
references that have been released or other pointers to heap allocated
memory that have already been freed.

Also, it could well be a problem with the compiler's exception handling.
To get to the bottom of this, please use egcs-1.1.2 (compiled with
--enable-shared --enable-threads options). This would help us to recreate
your problem.


Sai-Lai


>>>>> Helmut Swaczinna writes:

> Sai-Lai,
> thanks for the patch, but I'm sorry, it didn't solve my problem :-( 
> Maybe I've got another problem. 

> These are the last trace messages from one of my server-processes, which 
> crashed today after applying the patch:

> outScavenger: shutting down idle connection
> outScavenger: shutting down idle connection
> outScavenger: shutting down idle connection
> inScavenger: scanning for idle incoming connections
> inScavenger: shutting down idle connection
> inScavenger: shutting down idle connection
> inScavenger: shutting down idle connection
> #### Communication failure. Connection closed.
> tcpSocketMT Worker thread: exits.
> tcpSocketStrand::~Strand() close socket no. 13

> It seems to me, that the processes crash after returning from a method
> invocation. This possibly points to a problem with the scavengers. But I'm
> not an expert.

> Helmut


> At 11:08 27.07.99 +0100, you wrote:
>> Helmut,
>> 
>> Bruce Visscher discovered another race condition in the scavenger a while
>> ago.  Together we've worked out a solution and hopefully is the last one to
>> fix (finger crossed).
>> 
>> The change is in omniORB-2.8.0pre1. I've attached below a patch that you
>> can apply to 2.7.1 Please test this and report if your problem is fixed.
>> 
>> As the fix involves changes to the public interfaces, you have to
>> recompile the whole runtime <top>/lib/* and all your stubs and
>> applications, just to be sure.
>> 
>> Sai-Lai
>> 





-- 
Sai-Lai Lo                                   S.Lo@uk.research.att.com
AT&T Laboratories Cambridge           WWW:   http://www.uk.research.att.com 
24a Trumpington Street                Tel:   +44 1223 343000
Cambridge CB2 1QA                     Fax:   +44 1223 313542
ENGLAND