Re: runtime error with own *.so

[Jochen Rassler <jochen at cornmania dot de>]

Hi.

Am Mittwoch, den 02.02.2005, 10:39 +0530 schrieb Ranjit Mathew:
> Bryce McKinlay wrote:
> > This error is often caused by an inconsistent build and/or runtime 
> > environment: mixing .so/.o files compiled with different versions of 
> > GCJ, for example. When the BC-ABI is enabled by default, this issue 
> > should go away.

This may be the problem. There is little native c++ code mixed with java
code for the *.so.
And, as you may guess, the c++ code is compiled with gcc 3.3 while the
java code is with gcj 4.0.

Never thought about this mixing. Thanks for the hint ;-)
> 
> FWIW, on some rare occasions, I see the same
> error with the Thread_Wait_2 testcase during
> my regular bootstrap+testsuite cycles.
> 
> Unfortunately, when I try to run it manually
> everything succeeds. The machine has 1GB of
> physical RAM and 512MB of swap. It could therefore
> be a "bogus" size argument to the GC as Hans
> says, but I can't be completely sure. :-/
> 
> (The RAM on the box has been tested with
> various memory testers including the vendor's
> diagnostics utility as well as Memtest86 and
> there are no problems.)

Do you mean there is a problem with too much RAM? My machine also has
1GB of physical RAM and 1GB of swap. I thought the thing mentioned by
Hans was meant to cause problems, if too few memory is there, but you
sound as if too much memory causes this problem?

Thanks
	Jochen