Bug#206238: (no subject)

["Kevin B. Hendricks" <kevin.hendricks@sympatico.ca>]

Hi,

And google was right!  Good catch!

This is the source tree to the Blackdown PPC Linux jdk 1.3.1 build that Jan 
uses:

[kbhend@base1 wcvs]$ cd jdk1.3.x
[kbhend@base1 jdk1.3.x]$ find ./ -name "*" -exec grep "nested memory" {} \;
        fprintf(stderr, "OUCH: nested memory code, to %d 
levels.\n",NestLevel);
Binary file ./build/linux-ppc/tmp/java/hpi/green_threads/obj/malloc.o matches
Binary file ./build/linux-ppc/tmp/java/hpi/green_threads/obj_g/malloc.o 
matches
Binary file ./build/linux-ppc/lib/ppc/green_threads/libhpi.so matches
Binary file ./build/linux-ppc/lib/ppc/green_threads/libhpi_g.so matches


The funny thing is green_threads has not bee used for a long time on ppc 
Linux.  So I think somehow the green_threads libhpi.so (which has to wrapper 
malloc for user land threads like green threads to work) is being found and 
linked into the build of the C++ code for idlc resulting in this funny error 
message.

I would suggest that Jan search his machine for libhpi.so and remove them from 
the standard library paths so that this library is no longer found.

The funny thing is I never would have guessed to search the Java source tree 
since idlc is not built with java source code and does not use java at all!!!

Nice catch!

Thanks,

Kevin
  
On Friday 29 August 2003 18:31, Matthew Garrett wrote:
> Google suggests that the error message is being generated by Java.
>
> --
> Matthew Garrett | mjg59@srcf.ucam.org