On 12/8/05, Richard Fish <[EMAIL PROTECTED]> wrote: > On 12/8/05, Mark Knecht <[EMAIL PROTECTED]> wrote: > > On 12/8/05, Ernie Schroder <[EMAIL PROTECTED]> wrote: > > > On Thursday 08 December 2005 10:05, a tiny voice compelled Mark Knecht to > > > write: > > > > The problem I see periodically is that at time the java_vm will go to > > > > 100% CPU and I have to kill it by hand. > > > > > > > > > But you only show one instance of java_vm, right? > > > > As far as I know, yes, I only have one copy. I do not see the multiple > > copies that you are seeing. > > This could be a nptl vs posixthreads issue. nptl threads show up as > only a single process. For those having this problem, do you use > "nptlonly" for glibc? If not, anyone willing to try remerging glibc > with that flag? > > -Richard
I think you hit it Richard. On my AMD64 machine the main build, for 64-bit, was using nptl. I was not seeing multiple copies of java_vm. After seeing you email I did a bit of work in a new chroot environment and noticed I was getting multiple copies of java_vm there. I looked and found that when I built this yesterday I didn't set up any specific USE flags and hence didn't have nptl/nptlonly. I added those this evening and so far I am not getting multiple copies of java_vm. Good call! Note that I have never seen the specific problem Ernie discussed with mplayer, but possibly this will help him. cheers, Mark -- [email protected] mailing list
