On Feb 8, 2005, at 5:32 PM, Jens Vagelpohl wrote:
I honestly don't remember where that whole "LD_ASSUME_KERNEL" stuff originated, but it seems to be one of those myths that somehow never die.
There was, at one point, a problem. That point in time has long since passed. A small python program that exhibited the problem is here. <https://bugzilla.redhat.com/bugzilla/show_bug.cgi?id=97637 Red Hat 9 originally shipped with a version of glibc that they called "glibc-2.3.2-11.9.i386.rpm" which exhibited the problem. Later updates to RH9 supplied a fix. The current update to RH9's glibs is available at <ftp://ftp.redhat.com/pub/redhat/linux/updates/9/en/os/i386/glibc -2.3.2-27.9.7.i386.rpm> This, of course, only affects the bugs underlying pthreads implementation. The Python level threading bugs in Python >= 2.2b1 and Python < 2.4 that can cause Zope to hang are not effected by LD_ASSUME_KERNEL.