From mboxrd@z Thu Jan 1 00:00:00 1970 From: Peter Subject: Re: Java Problems Date: Sat, 13 Aug 2005 09:29:39 +1200 Message-ID: <42FD14C3.9000203@rimuhosting.com> References: <42FD03F7.4080901@rimuhosting.com> Mime-Version: 1.0 Content-Type: text/plain; charset=ISO-8859-1; format=flowed Content-Transfer-Encoding: 7bit Return-path: In-Reply-To: <42FD03F7.4080901@rimuhosting.com> List-Unsubscribe: , List-Post: List-Help: List-Subscribe: , Sender: xen-devel-bounces@lists.xensource.com Errors-To: xen-devel-bounces@lists.xensource.com Cc: xen-devel@lists.xensource.com, Reuben Kabel List-Id: xen-devel@lists.xenproject.org Oh. And to just simplify this test case a little more I bypass Java altogether and use this little C gem. It also triggers the problem. #include float s_i = 0.75; void test(float i) { if(i!=s_i) { printf("error\n"); } } int main(char **args) { printf("Starting\n"); long i=0; for (i = 0;i<100000000; i++) { test(s_i); } return 0; } Starting errorStarting Starting Starting Starting errorerrorerrorerrorStarting errorStarting errorStarting Starting Starting Starting errorerrorerrorStarting Peter wrote: > Hi. > > I don't have a 2.0.7 to test on at the moment. I will try again when I do. > > I got rid of the object allocation inside the loop. And I retested. I > don't hit the problem with int's on the call stack. I do hit the > problem with floats and doubles. > > The problem is quite easily reproducible (for me). Is someone there > with knowledge of register/stack handle able to take a look with gdb or > which ever tool you use? > > Regards, Peter > http://rimuhosting.com - Xen VPS Hosting >