Date: Thu, 6 Oct 2005 14:36:50 -0700 (PDT) From: Nick Johnson <freebsd@spatula.net> To: Tomaz Borstnar <tomaz.borstnar@over.net> Cc: freebsd-java@freebsd.org Subject: Re: jdk15 and SMP: are libmap.conf hacks needed on FreeBSD 5.4 and higher (both i386 and amd64)? Message-ID: <20051006142719.M44305@turing.morons.org> In-Reply-To: <6.2.3.4.0.20051006225641.0abea1d0@193.189.169.9>
index | next in thread | previous in thread | raw e-mail
On Thu, 6 Oct 2005, Tomaz Borstnar wrote: > Hello! > > Did not find anything recent enough in archives so I decided to ask > here. So here it goes... I want to utilize SMP on FreeBSD 5.4 (both i386 and > amd64) as much as possible with jdk1.5. So I was wondering whether I still > need to mess with libmap.conf to map threading libs to kse or not? It appears > that I do not need to do that since my limited testing showed that machine > was easily overloaded by severely hammering JBoss/jdk15 combination. Also top > -H shows many java threads in ksesig and similar states. So does that mean I > should keep the defaults and be happy with them? In my experience, the Java port simply does not work reliably with libkse at all. I've opened a number of bug reports on libekse for things which are trivial to reproduce and really quite obvious problems, but nobody's touched them that I know of. It appears that at least some of these problems are caused by bugs in signal handling and forking. See for example: http://www.freebsd.org/cgi/query-pr.cgi?pr=threads/84778 http://www.freebsd.org/cgi/query-pr.cgi?pr=threads/85112 The only way I've been able to get stable behaviour from the FreeBSD JVM is to map to libc_r, which of course isn't going to take advantage of SMP at all. This is what I have in libmap.conf: [/usr/local/jdk1.4.2/] libpthread.so.1 libc_r.so.5 libpthread.so libc_r.so I've gone so far as to offer to give money to the FreeBSD Foundtain to pay for someone to look at these problems, but got no response.home | help
Want to link to this message? Use this
URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?20051006142719.M44305>
