linux-2.6.16 emulation: linux-sun-jdk1.6.0 program problem

Peter Kostouros kpeter at melbpc.org.au
Fri Mar 9 06:25:01 UTC 2007


Jung-uk Kim wrote:
> On Tuesday 06 March 2007 11:37 am, Jung-uk Kim wrote:
>   
>> On Tuesday 06 March 2007 08:06 am, Alexander Leidinger wrote:
>>     
>>> Quoting Peter Kostouros <kpeter at melbpc.org.au> (from Sat, 03 Feb
>>> 2007
>>>
>>> 21:29:42 +1100):
>>>       
>>>> Hi
>>>>
>>>> On my system (CURRENT as of 03FEB2007) linux-sun-jdk1.5.0 and
>>>> linux-sun-jdk1.6.0 applications do not run under linux 2.6.16
>>>> emulation whereas they do under linux 2.4.2 emulation.
>>>>         
>>> Could you please update to a recent current and check if this
>>> problem is still there or not?
>>>       
>> As far as I know, they don't work, i.e., they crash with internal
>> error.  Both require sched_getaffinity (and maybe more) and that is
>> not implemented.  I believe JDK breakage is serious blocker for the
>> 2.6.16.  Since we cannot support the syscall directly (i.e., we
>> don't have user process affinity), we may have to fake it.
>>     
>
> FYI, it's fixed in perforce.
>
> Jung-uk Kim
>
>   
Hi

I applied linuxolator-p4.diff to CURRENT (as of about 24 hours ago). 
Unfortunately the patch did not apply cleanly, so the rejected files 
were replaced by appropriate p4 files. Unfortunately I still had the 
same problem that I originally reported.

Over this code base, I copied what I thought were the relevant files 
from p4 that provided support for (linux_)sched_getaffinity as mentined 
by jkim. I had positive results with this setup! However, I also have 
some odd observations:

1. Compiling using javac ends in a signal 9 even though it looks like 
the program successfully completed compilation.
2. Running sample applications, e.g.,  demo/jfc/SwingSet2/SwingSet2.jar, 
results in an empty pane, where it was expected to be populated with 
controls (not sure if this is a toolkit problem); applications also core 
dump.
3. Running ftpserver (java application) displays a GUI window but 
terminates abruptly when the server is started (syscall epoll_create not 
implemented).

Everything seems to run OK with compat.linux.osrelease to 2.4.2.

Note that due to the nature of the build, it could be that some files 
are not correctly synchronised, so I will wait for the next patch set.

Let me know if you would like further information or want me to try 
anything else.

-- 

Regards

Peter

As always the organisation disavows knowledge of this email



More information about the freebsd-emulation mailing list