[kaffe] Bug Report: BigInteger.toString() fails for large numbers

Alan Eliasen eliasen at mindspring.com
Wed Apr 21 16:15:03 PDT 2004


> In message "Re: [kaffe] Bug Report: BigInteger.toString() fails for large
numbers"
>     on 04/04/20, Guilhem Lavaux <guilhem at kaffe.org> writes:
>
>>Thanks for the excellent bug report. I fear this is the return of the
>>GCTest regression failure. After having passed the code through gdb it
>>seems kaffe hangs in getHeavyLock while trying to do some garbage
>>collection. The most interresting is that we have a pointer for
>>lk->holder which is beyond stack limits. So there is definitely
>>something wrong here.

Ito Kazumitsu wrote:
> I am afraid the change of kaffe/kaffevm/locks.c to Revision 1.48
> did something wrong.

   Thank you--this does affect the behavior.  I reverted the change made in
locks.c revision 1.48 (only one line was changed in that revision) and the
first regression test that I posted here (the class called BigIntTest) now works.

   Unfortunately, it doesn't fix the second testcase that I posted here (the
one called BigIntTest2).  That still fails at the same place (when numbers get
above 2^16,000,000; I tried increasing heap and stack space with no effect)
with a segmentation fault.

   Reverting this change may be an improvement, but I don't know of any
side-effects of making this change.  Is there anything else I can try to get
the second testcase to work?  (I'm aching to get this working; Kaffe with GMP
is so much faster than Sun's VM for this that I really want to announce it as
a good alternative.)

-- 
  Alan Eliasen                 | "You cannot reason a person out of a
  eliasen at mindspring.com       |  position he did not reason himself
  http://futureboy.homeip.net/ |  into in the first place."
                               |     --Jonathan Swift




More information about the kaffe mailing list